Anda di halaman 1dari 465

History of Linguistics 2005


General Editor
Zentrum für Allgemeine Sprachwissenschaft,
Typologie und Universalienforschung
Schützenstrasse 18, D-10117 Berlin


Advisory Editorial Board

Cristina Altman (São Paulo); Lia Formigari (Rome)

Gerda Haßler (Potsdam); John E. Joseph (Edinburgh)
Barbara Kaltz (Aix-en-Provence); Douglas A. Kibbee (Urbana, Ill.)
Hans-Josef Niederehe (Trier); Emilio Ridruejo (Valladolid)
Otto Zwartjes (Amsterdam)

Volume 112

Douglas A. Kibbee (ed.)

History of Linguistics 2005

Selected papers from the Tenth International Conference on the History of
the Language Sciences (ICHoLS X), 1–5 September 2005, Urbana-Champaign, Illinois
History of Linguistics 2005
Selected papers from the tenth International
Conference on the History of the Language Sciences
(ICHOLS X), 1–5 September 2005, Urbana-Champaign, Illinois

edited by
Douglas A. Kibbee
University of Illinois at Urbana-Champaign


4- The paper used in this publication meets the minimum requirements of American Na-
tional Standard for Information Sciences — Permanence of Paper for Printed Library
Materials, ANSI Z39.48–1984.

Library of Congress Cataloging-in-Publication Data

History of Linguistics 2005 : selected papers from the Tenth International Conference on the History of the
Language Sciences (ICHoLS X), 1–5 September 2005, Urbana-Champaign, Illinois / edited by Douglas A.
   p.   cm. -- (Amsterdam studies in the theory and history of linguistic science. Series III, Studies in
the history of the language sciences, ISSN 0304-0720; v. 112)
 Includes bibliographical references and index.
  1. Linguistics--History--Congresses.
P61 .I57   2007
410.9-dc22 2007037110
ISBN 978 90 272 4603 5 (alk. paper)

© 2007 – John Benjamins B.V.

No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means,
without written permission from the publisher.
John Benjamins Publishing Company • P.O. Box 36224 • 1020 me Amsterdam • The Netherlands
John Benjamins North America • P.O. Box 27519 • Philadelphia, PA 19118-0519 • USA

Acknowledgements ix
Douglas Kibbee
The Natural: Its Meanings and Functions in the History of Linguistic
Thought 1
John E. Joseph
On Grammatical Gender as an Arbitrary and Redundant Category 24
Marcin Kilarski
Penser l’espace, penser l’espèce: Modélisation des affinités
linguistiques 37
Carita Klippi
On the Origins of the Participle as a Part of Speech 50
Pierre Swiggers & Alfons Wouters
Grammar as a Liberal Art in Antiquity 67
Anneli Luhtala
Priscian’s Pedagogy: A Critique of the Institutio de nomine et
pronomine et verbo 80
Daniel J. Taylor
L’horizon de retrospection du Mithridate de Conrad Gessner (1555) 89
Bernard Colombat
Montaigne’s View of Skepticism and Language in the Essais 103
Danilo Marcondes
Competing Models for a 17th Century Universal Language: A Study of
the Dispute Between George Dalgarno and John Wilkins 112
Joseph L. Subbiondo
La notion d’unité sonore dans les grammaires françaises des 17ème et
18ème siècles 120
Jean-Marie Fournier
Une “Grammaire générale et raisonnée” en 1651 (1635?): Description
et intérpretation d’une découverte empirique 131
Sylvain Auroux & Francine Mazière

‘Analogy’: The History of a Concept and a Term from the 17th to the
19th Century 156
Gerda Hassler
Une écriture de l’histoire: La Lettre à M. Pinglin sur l’histoire de la
science grammaticale 169
Valérie Raby
Quels facteurs (linguistiques ou historiques) considérer dans l’accord
en français? Étude de certains cas dans le Journal de la langue
française (1784) d’Urbain Domergue 183
E-Jung Choi
Nicolas Beauzée: La clé inexploitée de la phonétique française 197
Christophe Rey
Colonialism, Scientific Expeditions and Linguistics in 19th Century
Brazil 212
Cristina Altman
The Concept of Civilization in Historic Brazilian Linguistics 228
Eduardo Guimarães
The European Linguistic Tradition and Early Missionary Grammars in
Central and South America 236
Manuel Breva-Claramonte
Steinthal and the Limits of Etymology: The Special Case of Chinese 252
T. Craig Christy
An Epistemological Assessment of the Neogrammarian Movement 262
Jean Leroux
Privileged Languages and Others in the History of Historical-
Comparative Linguistics 274
Hans Henrich Hock
The Nationalist Turn: Dutch Linguistics and German Philosophy in the
Eighteenth and Early Nineteenth Centuries 288
Gijsbert Rutten
Représentations de l’autre: L’italien et les Italiens dans quelques
dictionnaires bilingues des XVIIe-XVIIIe siècles 308
Nadia Minerva
L’utile et l’agréable dans les méthodes familières et autres ouvrages
utilisés pour l’apprentissage du français aux Pays-Bas (XVIIIe-XIXe
siècles) 321
Marie-Christine Kok Escalle

La reformulation dans la lexicographie des XVIe-XIXe siècles:

L’emergence de la syntaxe française 333
Rachele Raus
Words and Concepts for Child Language Learning in Late Nineteenth
versus Late Twentieth Century America 344
Margaret Thomas
La lexicologie, un savoir scolarisable? 356
Sonia Branca-Rosoff & Dan Savatovsky
Aspects de la linguistique prescriptive: Les perceptions des
vocabulaires de specialité à travers des dictionnaires français (XIXe
et XXe s.) 372
Danielle Candel
Semantique et analogie dans la tradition grammaticale arabe: La valeur
des formes verbales 386
Georgine Ayoub
Meaning by Collocation: The Firthian Filiation of Corpus Linguistics 404
Jacqueline Léon
Kristeva on the Encyclopedists: Linguistics, Semanalysis, and the
Epistemology of Enlightenment Science 416
Katherine Arens
La preuve de Gaifman: Réflexions sur la méthode de construction des
grammaires catégorielles 432
Beatrice Godart-Wendling
Name Index 441
Subject Index 447

September 1-5, 2005, linguists from twenty-five countries gathered on the

campus of the University of Illinois at Urbana-Champaign to share their
passion for the history of their discipline. The study of the history of the
language sciences does not promise to follow the old chestnut “those who
ignore history are bound to repeat it”. Rather, as each period in the history of
the discipline has chosen to focus on different key questions, the study of that
history promises to open our eyes to the variety of interesting questions that
can be asked, and answered – taking off the blinders of contemporary
Over the course of the conference a hundred presentations attempted to do
just that. This volume is a distillation of many fine contributions. The first
thanks, therefore, go to the scholars who shed light on the many different
approaches to the study of language. From the initial welcome by Chancellor
Richard Herman to the concluding banquet in the Levis Center, we enjoyed
five days of good weather, good company and good cheer.
The ability to bring these scholars to this oasis of learning in the midst of
the corn and soybean fields of Illinois is due to the generous contributions of
many units, including the College of Liberal Arts and Sciences, the Foreign
Language Building Fund (now the School of Literatures, Cultures and
Linguistics), the Department of French, the Department of Germanic
Languages and Literatures, the Department of Linguistics, the Department of
Spanish, Italian and Portuguese, the Department of Slavic Languages and
Literatures, and the Program for the Study of Religion.
These financial contributions were matched by personal contributions on
the part of many faculty, staff, and graduate students at the University of
Illinois. Graduate students from a half-dozen campus departments volunteered
to serve at the reception desk for the conference, helping our guests in many
different languages: Antje Muntenda, Aimee Alnet, Laura Fisher, Jessica
Miller, Daria Kabanova, Anita Saalfeld, Awa Sarr, Chris Hodge, Errol O’Neill,
Nicola Dach and Marco Shappeck gave unstintingly of their time and energy.
The organization of the student volunteers and many other aspects of the
planning and execution of the conference fell into the extremely competent
hands of E-Jung Choi, then completing her doctorate in French Linguistics and
now an assistant professor at Western Washington University. Publicity for
the conference was organized by Rick Partin, and the conference poster was
designed by Jason Stewart. As the conference dates coincided with my
assumption of new duties, much of the last minute panic management fell to

the incomparable Marita Romine, in her first days as my administrative

assistant. After the conference, Elizabeth Blount has done an outstanding job
in the preparation of the conference volume, identifying errors, tracking down
life dates, and managing the tedious process of creating camera-ready copy.
We are also grateful to the staff at John Benjamins Publishing Company, and
notably to Anke de Looper and Paul Peranteau. The support of the Benjamins
family for the history of the language sciences is a great gift. Konrad Koerner,
although he could not be present at the conference, is always a presence among
those of us working in this field. My thanks to all of the above are deep and
The contributions of one person to my career and to my life are beyond
measure, and for this I dedicate this book to my wife, Jo Kibbee.

Douglas Kibbee
University of Illinois at Urbana-Champaign
August 2007

University of Edinburgh


Attempts at distinguishing one part of language from another on the basis

of naturalness, or certain other criteria that recapitulate the dichotomy of
something that is grounded versus something that is not, can be regarded as
normative, and ultimately tautological. A historical continuity exists linking
such attempts across the centuries, despite changes in terminology and shifts in
the balance of the sometimes contradictory conceptions of what in language is
natural. This paper examines the history of eleven conceptual features
traditionally associated with linguistic naturalness: mimesis, orality,
physicality, rationality, simplicity, unplannedness, rusticity, musicality, purity,
systematicity and efficiency. The aim is to uncover some of the underlying
methodological and ideological assumptions of modern linguistics by
examining how these concepts have been deployed. This offers a way forward
beyond the limits which the implicit tautology has imposed upon our
explanatory and analytic imaginations.

Everything is natural.
Mary Catherine Bateson (1995:10)

1. Initiation into the problem

Throughout history, the study of language has relied on various conceptual
devices to reduce the apparently limitless diversity of actual linguistic
production, so as to narrow the phenomena to be accounted for into something
tractable, about which simple rules could be convincingly stated. Many of
these devices have been grounded in a belief that language can be divided into
what is basic or elemental on the one hand, and what is derived or compounded
from the basic elements on the other; and in some cases this has led to the view
that the basic elements are more ‘natural’ in one sense or another, and that the
origins of language must lie with them. Since the 19th century this conceptual
process has been driven still further by a need for perceived scientific validity,
which has appeared to demand even greater narrowing of the phenomena, in

order for generalisations to be made that are watertight enough for any
exceptions to be cogently disposed of as flukes.
Although a miniscule proportion of the population ever studies linguistics
and encounters these beliefs in their scientific form, everyone deals with the
basic concepts by virtue of how, in every culture, linguistic production is
hierarchised into good and bad, right and wrong, standard and non-standard.
Following Taylor’s (2000) model of reflexivity, one can see how the basic idea
of interpreting forms of language in this way becomes part of a speaker’s
linguistic habitus (not Taylor’s term, but a medieval one revived by Bourdieu)
through everyday metacomments like “Don’t say brung, say brought”.
Children who are native speakers of English spend twelve or more years in
English classes, where much of the work consists of teaching them the
standard language, particularly for purposes of writing, where using their
‘natural’ dialect (insofar as the standard deviates from it) would be interpreted
as a mark of insufficient education.
Linguistics is supposed to stand above this standard–dialect division that is
fundamental to the general cultural discourse on language. First-year students
are trained to pooh-pooh the prescriptivism of naïve language mavens, as
opposed to the linguist’s purely scientific descriptivism. Yet in their analytical
work, linguists rarely follow through the logic of this doctrinal point by sorting
out those forms that belong to the standard language, which have a history of
deliberate production and have to be learned in place of the forms acquired
‘naturally’, which native speakers are naturally inclined to use. The two types
of forms ought, in theory, to bear very different relationships to any Universal
Grammar (UG) that may underlie an individual’s language faculty. But the
descriptivism of modern linguistics consists of treating standard and non-
standard forms indifferently.
I was struck by the seeming inconsistency when studying Romance
linguistics in the mid-1970s. Our work was clearly laid out for us, just as it had
been for our predecessors for well over a century: to analyse and trace the
history of the Romance dialects. We were by no means uninterested in the
synchronic state of the dialects, but knew that surviving ‘uncontaminated’
speakers were few and disappearing by the day (especially those with enough
teeth left for phonetic transcription to be reliable). The ‘contamination’ we
were trained to avoid meant principally exposure to the standard language,
though other dialects could also be a problem, e.g. through speakers marrying
out of their own village.
We were taught that passive exposure — listening to the radio, watching
movies or TV — was not likely to induce contamination, since active language
use alone had an impact on the individual’s langue. But men who had served in
national military divisions, or anyone who had lived a considerable time
outside their native village, were highly suspect. As for cities, we were referred

to studies of urban vernaculars but never encouraged to take them seriously, at

least not in our ultimate quest of reconstructing the ‘diasystem’ and the ‘proto-
dialect’ which had given rise to it. Cities were just too messy. Urban ways of
speaking, like urban ways of living, were not natural like out on the farm. City
folk had been educated (hence contaminated), and represented a variety of
dialect origins, so their speech was not a homogeneous and stable system
amenable to study in our historico-scientific vein. We also learned structuralist
and generativist types of analysis in which this distinction was simply ignored,
and variationist work aimed at relocating the old city–country dyad within our
increasingly urbanised context, as lying between classes and styles. 1

2. The normative nature of naturalness

As Bateson (1995) has rightly claimed, no family of words has given rise
to more intellectual problems than nature and its derivatives. Where language
is concerned, using the term ‘natural’ implies a dichotomy between forms of
speech that precede rational thought and wilful choice (and are perhaps what
gave rise to thought, reason and will), and forms of speech (and writing) that
are the product of such thought and choice. This dichotomy is inherently
paradoxical, since language is itself the means by which we recognise the
presence or absence of reason and wilful (as opposed to instinctive) agency,
and the medium in which reason and agency have their existence. So, as
Epicurus already asked in the 3rd century BCE, how could language be the
product of reason and wilful agency, when it is required if they are to manifest
Of the various solutions to the paradox that have been offered through the
ages, the most enduring has been Epicurus’s own, laid out in a letter addressed
to Herodotus. He posited a two-stage emergence of language, starting with a
natural stage in which proto-languages arose directly from the body, differing
according to ethnicity, thus also explaining how multiple languages could exist
and still be grounded in nature. Although spoken not by reason but by natural
constraint, these rough languages provided a basis upon which rational thought
and social agreement could be constructed. People then became aware of the

There were always voices of protest. My late teacher Ernst Pulgram (1915–2005), despite
having been brought up in this tradition, did not have the typical Romantic commitment to the
unique nobility of peasant dialects. He found language mixture as interesting and as natural as
dialect purity, and, like the Prague School, found literary language indistinguishable in linguis-
tic interest from ‘natural’ language. His intellectual contribution to Romance historical studies
included what we might call his ‘social uniformitarianism’ — a belief that, throughout human
history, our common bodily needs and functions have not so much given universal structure to
our language as presented us with universal opportunities for creating and reproducing the
social stratifications that seem to be the most universal need of all. But he was marginalised for
such beliefs, which did not accord with the received ideas of mid-20th-century linguistics.

flaws in their language, such as ambiguity or lack of conciseness, and

introduced improvements by common consent within the ethnos.
The 17th-century neo-Epicureanism of Gassendi (1592-1655), Hobbes
(1588-1679) and Locke (1632-1704) was passed on, mainly via Locke, to
Condillac (1714-1780) and others in the 18th century, and thence into linguistic
science as it developed in the 19th and 20th, with a shift from ontogenetic to
phylogenetic formulation of linguistic naturalness. Now the natural would be
equated with an ‘unconscious’ mind, and all the ‘real’ existence of language
located there, with any products of conscious thought relegated to the waste
bin of phenomena that, being wilful in nature, do not show the sort of
regularity that makes them amenable to scientific investigation. As for the
paradox of the origin of language, it was likewise ruled axiomatically out of
scientific consideration; but at the same time it seemed to disappear as a
problem with the general acceptance of the theory of evolution. The
development of language would be tied to the evolution of the human brain,
with the details to be worked out in due course. As for how language might
emerge directly from the structure of the brain, which does not differ across
ethnicities or cultures — and yet thousands of different languages exist — a
solution would eventually be offered by Chomsky, who posited that there is in
fact only one human language, with what we normally refer to as ‘languages’
being merely dialects of it, and with the ‘dialectal’ differences insignificant
compared with the fundamental unity that is not apparent on the surface but
can be worked out with enough theoretical ingenuity.
Linguistics has become comfortable with this view of things to the extent
that some linguists reject out of hand any suggestion that our understanding of
language might be improved by refining our dichotomised view of the ‘natural’
toward something more subtle, a continuum perhaps. Nonetheless, the
questions are worth raising, not least because of how fully adjacent sciences
have problematised the concept of consciousness, and to a lesser extent
naturalness. Let us consider some easy and hard cases of natural causes, as
defined in the traditional way. If a hurricane felled a tree, or I died of a heart
attack, those would be easy cases of natural causation. If the same tree were
cut down, or if I died from being shot through the heart, those would be
equally easy non-natural cases. But what if I were an obese chain-smoker who
had ignored years of doctors’ warnings to change my behaviour? This is
harder, demanding a careful assessment of the nature of my addictions. Even
the hurricane may be less obvious than at first appears. ICHoLS X took place
just after the American Gulf Coast was struck by Hurricane Katrina, the
unusual force of which was blamed on the warming of the seas by carbon
dioxide emissions that humans are responsible for (along with flatulent cows).
In both cases, the line between natural causation and its opposites is fuzzy.
We seem loath to accept that nature can be so cruel without a little help from

her human friends, while, in the other direction, addiction is an explanatory

means of ‘naturalising’ the choices people make that run contrary to their own
interests and often to their own express will, whether it is addiction to
cigarettes or to automobiles. The dichotomy of the natural and its opposites
masks the fact that, for example, when lightning strikes a lightning rod, an
implement deliberately erected in order to attract the strike away from a
building, the human intervention has exploited natural properties of electricity
and metal in such a way that it makes little sense to try classifying the strike
one way or the other. On a continuum, it would be just about in the middle.
It is useful to be able to distinguish degrees of naturalness in this way —
though one must bear in mind that, on the one hand, ultimately “everything is
natural”, and on the other, that nature is a cultural construct. There may be a
nature in itself and for itself, but we experience it through a culturally-
mediated lens, and since cultures differ so greatly in how they conceive of
nature, the choice of any one concept is an implicitly normative one.
Moreover, to characterise any human action as other than natural, even by
degrees, is necessarily a normative statement. If it were a descriptive
statement, then, taken literally, the action in question could not occur at all, or
else would have to be taken as supernatural. For instance, without mechanical
assistance, it is impossible for me to fly, because of the natural configuration of
the human body. If you were to see me flying through the air, you would
assume that it is through some conjuror’s mechanical contrivance, the
alternative being that it is supernatural. But when actions are described as
‘unnatural’, the meaning is usually not that they are impossible, but that they
are contrary to the proper function of whatever is acting or being acted upon.
‘Proper function’, in evolutionary discourse, means the specific purpose for
which something has evolved, as opposed any other side purposes to which it
may be put.
So long as the proper function of any thing or action is open to debate,
ascriptions of unnaturalness are grounded in value judgements. Indeed, they
may carry a strong connotation of immorality. When the Bible applies the
word ‘unnatural’ to sexual acts (whether hetero-, homo- or solitary) performed
with no intent to procreate, it means that the proper function of sex is to
reproduce. If so used the ‘reproductive’ organs produce sexual pleasure, well
and good, but to use them as though pleasure were their proper function is
perverse and immoral. Millions of Christians, Jews, Muslims and followers of
other faiths hold this belief firmly; millions more do not. I use it as an example
because I expect that most of my readers are dubious enough about the Biblical
view to see this as a clear case in which ascribing a ‘proper function’ to a
human action makes the resultant determination of what is unnatural into a
normative statement, masquerading as a descriptive one. As such I believe it is
not exceptional, but typical of statements about the naturalness of any human

action, including language, or its component parts. To characterise a human

action as either natural or unnatural is an inherently normative act, based on
some prior conception of what constitutes the norm, usually grounded in some
further prior conception of proper function.
With regard to language, we face the particular problem at the heart of
Plato’s Cratylus: that ultimately there is no sure way of distinguishing what is
linguistically natural from what is the result of historical accident. Socrates is
committed to the view that words have originated through the art of mimesis,
their proper function being to capture and reproduce some part of the essence
of what they designate. Yet he is forced to admit that, once constructed, the
language passes into the care of hoi polloi, the masses, who care only about the
‘vulgar’ function of communicating with their fellows, for which conventional
words suffice just as well as naturally mimetic ones. Even words that were
originally mimetic have undergone change from their original form in the
mouths of the masses, who care only about communicating and sounding nice.
Hence we cannot take the study of language as a guide to the true nature of the
universe. This is clearly disappointing to Socrates, who believes words should
be naturally and deterministically bound to their meanings, but who finally
concedes that, as actually used, they are not so bound.

3. Opposites of the natural through history

The word physis, usually translated as ‘nature’, is at the centre of the
debate in the Cratylus, its opposite being nomos, the word for a conventional
law. In later Greek thought, thesis would be the term generally counterpoised
to physis. Joseph (1990) considers what the shift may have signified, but the
basic dichotomy of what is natural — grounded in something beyond wilful
human choice — versus what is not so grounded, remained intact. It has since
endured across long centuries in which the term ‘nature’ itself has sometimes
receded to a less prominent place in the discourse, in favour of other terms or
concepts that nevertheless continue to occupy the ‘grounded’ pole in the
grounded versus non-grounded dichotomy. Such, at least, is my hypothesis —
for the challenge of a historical project like this one is that it must continually
question whether the ‘same concept’ is maintained across stretches of time in
which practical and intellectual contexts are shifting. This is so regardless of
whether the same terms are maintained or new terms introduced; indeed, the
danger of falling victim to a credulous intellectual uniformitarianism is greater
when the terms have not shifted than when they have. A shift requires a prima
facie demonstration that a fundamental unity has been maintained, which is
liable to be taken for granted when the terminology has stayed constant. Nor is
this true only diachronically. In any given period, including the present, people
may be using the same term to denote quite different concepts, as the
discussion of ‘natural’ in the preceding section has already indicated.

The dichotomies of groundedness for language have differed principally

according to what the ‘proper function’ of language has been conceived as
being. The Socratic–Platonic view, shared by the great religious traditions, as
well as by secular philosophy through most of its history, is that the proper
function of language is to embody truth, and so to constitute knowledge in the
true sense. The idea that this, rather than, say, communication, is the proper
function of language is so deeply embedded in thinking about language, across
cultures and across history, that relatively few people have ever been aware of
it, let alone questioned it. Concern with linguistic correctness is a function of
this concept — and remember that ‘the correctness of words’ is the traditional
thematic designation of the Cratylus. If embodying truth is the proper function
of language, then, properly, one would expect one form for every meaning,
making any variation in form ‘unnatural’ in the normal, functional sense of the
word. This belief motivates the concern with enforcing standard usage
discussed in §1. Even when, in the modern period, concern with communica-
tion comes to the fore, it cannot be neatly separated from the truth function
(normally referred to as ‘representation’, though this term itself embeds a
particular concept of the relationship between language and reality); measures
to improve communication tend overwhelmingly to impose the ‘correct’ form,
rather than the one in most general use.
Another key aspect of how nature has been conceived in the context of
language is shown by the semantic narrowing of physis as it was borrowed into
Western European languages, where ‘physical’ and its congeners indicate what
is material rather than ideal or spiritual or conceptual. This development has
run directly counter to the position taken by Socrates and Plato, for whom the
realm of nature was precisely that of the ideal, existing in the sky, and at best
merely reflected in the palpable ‘reality’ (so-called) of ever-changing things
that we can sense. The illusions which our senses produce start with the
impression that the things around us are solid and permanent, when in reality
they are conglomerations of atoms and are always deteriorating, though more
slowly than it is possible for us to see. Although this link between the bodily
senses and error was inherited by Christianity along with the overall Platonic
vision of the ideal, it had to contend with the basic belief that Jesus had taken
on human bodily nature — hence the traditional Christian ambivalence toward
the body as, on the one hand, the source of carnal sinfulness, and on the other,
God’s greatest earthly creation (see Joseph 2005a; 2006, §2.5). The idea that
language might itself be grounded in the body is not Platonic, but Aristotle
provides a model of how mental sensations give rise to utterances, and he
considers what such sensations have in common with those of animals, as well
as why our language is different from the sounds they make. The difference is
articulation and reason. But he does not address the question of why, if
language is indeed grounded in mental sensations that are the same for all

humans and continuous with those of other sentient beings, different languages
should exist. This is the question Epicurus answered by positing that languages
originate directly from the body, as differentiated by the nature of the ethnos,
itself partly shaped by environment.
From Galen through to the 19th century, the theory of the bodily humours
was the foundation of understanding the operations of the body, and indeed the
mind, as in Descartes’ late work Les passions de l’âme (1649). Unfortunately
the historical centrality of this theory has been marginalised in the two
centuries since it was abandoned, so that still today historians of ideas are wont
to ignore it, distorting intellectual history as a result. Joseph (2004a) examines
how, in Shakespeare’s The Merchant of Venice, the language used by various
characters can be seen to correspond with what they themselves or other
characters have to say about the disposition of their humours. Exclamations in
particular were thought to be generated by the excess of a particular humour,
and the act of exclaiming mitigated the excess. Even the use of Germanic
words and simple sentences can be seen to correlate with certain humours,
Latin–Romance words and complex syntax with others.
Such beliefs would become more explicit in the 18th century. Condillac’s
Essai sur l’origine des connoissances humaines (1746) locates the origins of
human language and reason in the transition from natural signs (as when
smoke signifies fire, or screaming signifies pain) to the artificial signs of
language. The crucial change occurred when the ‘language of action’,
atemporal and synthetic, evolved into conventional language, which is
successive, signifying in bits spread across time. The sounds make up the
word, the words make up the sentence, the sentences make up the discourse.
Both in the species and in the individual, it is the acquisition of conventional
language that makes thought itself temporal and forces us to analyse our
experience rather than simply taking it in as a synthetic whole. Rousseau’s
(1712-1778) Essai sur l’origine des langues (1761) insists, contrary to
Condillac, that although “needs dictated the first gestures, [...] the passions
drew out the first words”, which in fact must have been sung, since accents,
tones and rhythm are the features of language that issue naturally from the
throats of infants. These features continue to predominate in the ‘southern’
language type represented by Chinese and Arabic, which are also more natural
in the prominence they give to vowels. Consonants, an intrusion of the
analytical, have spread with phonetic writing, first in the northern languages
and only latterly to the southern. Moreover, the original language consisted
entirely of tropes, images, and other forms of figurative speech, with the force
of onomatopoeia felt continually. Again these continue to be more
characteristic of the southern type than the northern. Herder’s Abhandlung
über den Ursprung der Sprache (1770) opposes both Condillac and Rousseau,
asserting that “the former turned animals into men and the latter men into

animals”. He rejects any attempt to draw an analogy from the ‘naturalness’ of

animal language to human language. Human language was invented, once man
was “placed in the state of reflection which is peculiar to him”, and when the
mind in this state of reflection was “for the first time given full freedom of
The history of how such views of the natural and its opposites became part
of modern linguistics in the 19th and 20th centuries is too long and complex to
detail here; the main lines at least are traced in Joseph (2000, Chaps. 4–6). I
would however like to mention one particular concept, which has figured more
prominently than any other in debates over the naturalness of language in the
20th century: the notion of the ‘mark’. It was formulated by Jakobson (1896-
1982) and Trubetzkoy starting in 1930, in the wake of their attempts to analyse
language structure along the lines of Saussure’s (1857-1913) model of a
system whose elements are purely relational in character (see Joseph et al.
2001, Chap. 2). Despite trying very hard indeed to devise analytical systems
that followed this ideal, they concluded that language was not actually that
way. Rather, they saw its fundamental principle as being that elements are
ranked relative to one another according to functionality, where the ‘proper
function’ of language is taken to be communication, with other interpersonal
effects occupying an important secondary role. Another version of markedness
came to prominence with Chomsky & Halle (1968); built upon the foundation
of the Jakobsonian one, it differed from it in that the internal mechanics of the
system — specifically its systemic economy — came to matter at least as much
as any functionality. In the 1970s this approach gave rise to various
phonological schools, some of which used ‘natural’ to designate themselves.
But for some of these, ‘natural’ meant a dependence on the more bodily
considerations of phonetics, for others Jakobsonian functionality, and for
others still Chomskyan–Hallean economy. From the late 1970s on, the same
systemic economy would be put by Chomsky at the centre of his syntactic
analysis, and a dichotomy of ‘core’ and ‘periphery’ would be put in place to do
the traditional work of any dichotomy of something natural and something not.

4. Conceptual features of ‘natural’ language

This section will look back across the centuries at eleven features
traditionally associated with ‘natural’ language. Although views have not been
consistent, the resulting picture may help shed light on the underlying concept
of language that informs ideas about naturalness. The selection is not complete,
but a first attempt at teasing out distinctive themes from within discourses that
have always interwoven them. As my comments will show, some pairs within
the eleven seem perhaps to be collapsible into one, and it may ultimately be
desirable to collapse them (or indeed to unfold a single item into two or more).
But in each case there is some evidence that they are conceived of separately.

The fact that some of the features directly contrast with others may help us
understand why discourses involving the naturalness of language, covertly or
overtly, have never followed a continuous, smooth path of historical
development, but one strewn with ambivalence and contradiction.

4.1 Mimetic
The idea that language is naturally mimetic or iconic means that what we
say is structured at least in part by what we mean. It assumes that the proper
function of language is the representation of extra-linguistic reality, and that
this reality leaves its traces in the language that represents it. Insofar as
language is not mimetic, it is not natural. Classically, this is the view
associated with Plato’s teacher Cratylus in the dialogue named for him, where
he makes the even stronger argument that if language is not mimetic it fails to
qualify as language at all, but is merely noise. Mimesis would have a long life
in discourses on the origin of language, though it would eventually be
incorporated into the broader theory of ‘natural signs’ (see §3). In 20th-century
linguistics, mimesis, now called ‘iconicity’ or ‘isomorphism’ would enjoy a
notable re-emergence beginning with the writings of Jakobson, who believed
that the structure of languages reproduces that of the universe of phenomena
which language represents.

4.2 Oral
How language is conceived as naturally oral, with writing, sign language
and any other medium constituting a mere representation of the ‘real’ language
that can exist only in the presence of the speaker, was brought to light most
fully by Derrida (1930-2004) (see Joseph et al. 2001, Chap. 13). The matter is
actually more complex than at first appears, because the purely oral already
begins to be mitigated with the introduction of consonants — articulated
sounds being, according to Aristotle, the vehicle of reason, as opposed to the
minimally articulated vowels that we share with animals. The idea that vowels
are the most basic and natural elements of the sound system of any language
would persist into modern phonology, where one even encounters the idea that
the maximally open vowel, /a/, is the most natural linguistic sound of all. A
further complexity results from cultural ambivalence toward writing. Although
treated as derivative of oral language and even parasitic upon it, and
representing the absence of the speaker, writing is simultaneously regarded as
a technological advance of unparalleled power in the development of cultures
— away from nature. Hence the ambivalence: it is in the nature of cultures to
be moving constantly away from the ‘nature’ they have constructed for
themselves, while simultaneously regretting the move and re-constructing the

The oral stands partially in contrast to the mimetic, in the sense that those
meanings which can be reproduced mimetically in oral speech are limited.
Visual media like writing and sign language allow greater scope for direct
mimesis through the picturing of things and concepts. Francis Bacon was
among the first modern Europeans to take seriously the notion that Chinese
character writing represents natural signs and physical objects rather than
conventionalised, language-specific signs and concepts, or could be adapted to
represent them and thereby form a truly universal language. In the 17th century
various schemes for a ‘real character’ were launched, notably those by
Comenius (1592-1670), Wilkins (1614-1672) and Dalgarno (c.1619-1687).
Behind all these schemes was a desire to return to the natural and original
language of man, which conventional languages had obfuscated.

4.3 Physical, bodily, emotional

Those schools of phonology which locate the naturalness of a sound
system, or of its analysis, in facts of articulatory or acoustic phonetics are
grounding it in the physical structure of the body. Similarly with Chomsky’s
conception of UG as a ‘language organ’, a physical part of the human brain: its
physicality is what makes it natural, innate and real. The venerable idea that
the structure of language originates from the configuration of the body and is
grounded in it has been discussed in §2 and §3 with reference to Aristotle,
Epicurus, Galen, Shakespeare, Condillac, Rousseau and Herder. It gains much
of its power from the firm cultural belief in racial and ethnic differences, which
appear at first glance to correspond closely to linguistic boundaries. To a
linguist it is obvious that the link is indirect at best, so we can easily forget
how powerfully it functions in cultures the world over, including our own.
Ethnic belonging forms a compelling part of people’s identity, and language a
central arena for the performance of that identity (see Joseph 2004b).
At various times from the 16th through the 20th centuries the idea has
emerged that manual signs represent the most original and natural form of
language (Rée 1999 gives an excellent overview). This conflicted with the
oral-as-natural concept, but was strengthened by a belief in the greater mimesis
of sign language (as discussed in §4.2), and a sense that, being more fully
bodily than oral language is, it is more universal. The concept also manifests
itself in a firm cultural belief that ‘body language’ is always truthful even when
a speaker is lying (see §4.6). The idea that the most natural language derives
directly from bodily passions has a modern reflex in the view, expressed by
Jespersen (1860-1943), that educated standard-language speakers will, in
moments of emotional turmoil or physical distress, revert from the standard
language to the local dialect of their childhood, which is thus shown to be their
natural idiom. “It is also no rare thing that a man who day by day in peaceful
circumstances speaks the Standard language irreproachably, may in moments

of passion slip back into the dialect of his childhood […]” (Jespersen 1925:75–
Also connected to this concept is a widespread belief that the original and
natural state of language was to be concrete, and that abstractions, elements of
language lacking a directly sensible referent, are less natural than those that
have one. This is implicit in the treatment of language by Locke and the other
neo-Epicurean empiricists of his time. It would find its way too into 19th-
century historical linguistics. Bréal (1832-1915) locates linguistic ‘superiority’
in the movement from the concrete to the abstract:

If one had to say wherein lies the superiority of the IE languages, I would not look for
it in the grammatical mechanism, or the compounds, or even the syntax: I think it is
elsewhere. It is in the ease these languages have, and have had since the most ancient
recorded period, of creating abstract nouns. (Bréal 1897:273; my translation, as are all
that follow) 2

Meillet (1866-1936) will make a parallel assertion about the effect of

‘civilisation’ on language structure, in a paper on ‘The concrete character of
the word’, when he credits the French with a capacity for abstraction that was
denied to their Roman forebears by the morphological structure of Latin:

French has an invariable word loup ‘wolf’, the form of which is always the same,
whatever sentence the word appears in, however the animal is envisaged […]. In
Latin, on the contrary, there is really no word that means ‘wolf’; if you want to say
that ‘the wolf has come’, the form will be lupus; if you see some wolves, lupos […],
etc. No one of these forms can be considered as being the name of the ‘wolf’ any
more than the others. […A] Roman was not capable of naming ‘the wolf in itself’
[…]. The universal tendency of language, over the course of civilisation, has been to
give the noun a character progressively more independent of all its particular uses.
(Meillet 1936[1923]:2.11-13]) 3

The similarity between Bréal and Meillet is that the former links abstraction
with superiority and the latter with civilisation, which for him implies

“S’il fallait dire où réside la supériorité des langues indo-européennes, je ne la chercherais
pas dans le mécanisme grammatical, ni dans les composés, ni même dans la syntaxe: je crois
qu’elle est ailleurs. Elle est dans la facilité qu’ont ces langues, et depuis les temps les plus
anciens que nous connaissons, à créer des noms abstraits.”
“Le français a un mot “loup” invariable, dont la forme est toujours la même, quelle que soit la
phrase où ce mot figure, quelle que soit la façon dont on envisage l’animal […]. En latin au
contraire, il n’y a à vrai dire aucun mot qui signifie “loup”; si l’on veut dire que “le loup est
venu”, on aura la forme: lupus; si l’on voit des loups: lupos […], etc. On ne peut pas
considérer l’une quelconque de ces formes comme étant le nom du “loup” plutôt que les autres.
[…U]n Romain n’était pas capable de nommer “le loup en soi” […]. La tendance universelle
du langage, au cours de la civilisation, a été de donner au nom un caractère de plus en plus
indépendant de tous ses emplois particuliers.”

superiority. Yet Meillet is contradicting Bréal’s assertion that the IE languages

have always been characterised by ease of abstraction.

4.4 Rational
This is the feature most directly in conflict with certain others, notably
‘physical, bodily, emotional’ (§4.2) and ‘unplanned’ (§4.6). The conflict helps
explain cultural ambivalence and disagreement about what is ‘natural’ in
language. From either a Platonic or a Christian point of view, the true nature of
language is so wholly natural as to be fundamentally inseparable from reason
itself. Joseph (2005a) looks at Christian views from Augustine (354-430 CE)
to Gregory the Great (540-604 CE) to Thomas Aquinas (1225-1274) on what
the true nature of language is, given that it belongs not only to humans but to
God and the angels, purely spiritual beings without bodies. Whatever is bodily
in language is clearly secondary, either an obstacle to perfect communication
(as for Gregory) or merely a vehicle for expressing our own ultimately trivial
wills, rather than the will of God, which is one with truth and knowledge.
Modern analytic philosophy, however secularised it may be, has not strayed far
from this basic dichotomy between logic and ‘ordinary language’; philosophers
would however be likely to characterise the latter as ‘natural’ (but erroneous)
and the mathematical formulation of the former as ‘true’, which is to say
‘natural’ in the Platonic and traditional Christian sense. On the other hand, as
noted in §4.2, Aristotle sought a bodily account of rationality, including in
language, and the Western medical and psychological tradition, with its theory
of the humours, only divorced the two in modern times (see Wollock 1997).
A hallmark of modern thought has been its ambivalence toward the
rational, even among those intellectuals who are its high priests. Naturalness is
generally attributed to those who lack education — savages, children and
women being the classic triad, with the peasantry or working class of one’s
own country counting among the savages, and with all treated as the equivalent
of children. As noted above, there is a lingering regret for the lost paradise that
those naturals are constructed as inhabiting. But a ‘civilised’ person who
shows signs of ‘natural’ behaviour, including linguistically, will be treated as
deviant, for example in the psychiatric discourse of the 1920s–40s examined in
Joseph (2005b). A widely used test for schizophrenia (which covered a far
wider range of mental disorders than the term denotes in common parlance
today) was based on the theory that the language of schizophrenics revealed an
excess of ‘concrete’ thinking. For the psychiatrist Goldstein (1944:22–27),

[T]here is no question that a very great concreteness is characteristic for the behavior
of schizophrenics. […] This concreteness also finds its expression in the language of
schizophrenics. Analysis reveals that many of the very strange words which the
patients use become understandable when considered in relation to the concrete
situation which the patient experiences at the moment […]. In their language there is

an absence of generic words which signify categories or classes. […I]n the color-
sorting test one of our patients picked out various shades of green, but in doing so he
named them as peacock green, emerald green, taupe green, bright green, equet green,
bell green, baby green. He refused to say that all might be called green. […] This
individual character of the word is characteristic of the language of schizophrenics, in
general. According to the specific way in which the patient experiences a certain
object or situation, a definite property or aspect of the object or situation becomes the
basis for the choice of words.

This patient’s condition is reminiscent of what Meillet said about the ancient
Romans in §4.3, except that the concreteness which represents the natural state
in Meillet’s historical analysis here occupies the unnatural pole. Nor was this
diagnosis purely theoretical. It would have immediate and profound conse-
quences for Goldstein’s patient, including being locked away and subjected to
radical drug and electric shock therapy, and maybe lobotomy, which in the
worst case could result in near-total loss of cerebral function.

4.5 Simple
The greater naturalness of what is simple in language over what is complex
or compounded manifests itself in many forms. In rhetoric, a style like the
Senecan, featuring short sentences with uncomplicated syntax, is reckoned to
be more natural than the Ciceronian. Words are analysed into stem and ending,
verbs are described as nouns with the addition of time — all leading to a
conception of the original and natural state of language as consisting of
uninflected nouns, or rather words indistinguishable by class as nouns, verbs,
adjectives etc. With John Webb’s The Antiquity of China (1678) this leads to
Chinese being identified as the most original and natural of extant languages.
In many cases, an objective criterion can be cited for determining what is
simple and what is not, e.g. a smaller phonological repertoire, or a smaller
number of morphemes to express a parallel notion or perform a parallel
function. Voltaire’s Dictionnaire philosophique (1764) defines the ‘genius’ of
a language as “its aptitude for saying in the shortest and most harmonious way
possible what other languages express less happily”. Jakobson (1963)
extrapolated from universals formulated by Greenberg the idea that categories
which are ‘unmarked’ semantically are never expressed by a morphologically
longer form than their ‘marked’ counterparts. For example, plurals are either
longer than singulars (English ear, ears) or the same length (man, men), but
the plural is never shorter than the singular. For Jakobson, this is also a case of
isomorphism, since plurality is conceptually ‘larger’ than singularity. Of
course, counterexamples are not hard to find: in the Old French masculine
nominative, for instance, the singular took a final –s while the plural was
unmarked. Some ad hoc device or other (like ‘markedness reversal’) would
have to be conjured up to explain away — or naturalise — the exceptions.

In other cases there is no objective, countable criterion, and here

judgements of simplicity and naturalness are more obviously normative and
ideological. Phonologists treat back unrounded vowels as less natural than
back rounded or front unrounded vowels on the basis not of feature counts, but
of the distribution of sounds among the world’s languages. They may also
construct explanations of why the features [back] and [rounded] work against
each other acoustically, and the explanations may be valid, but they are still ex
post facto to a normative view that the sounds lack naturalness. Basic word
order is another example. The idea that SUBJECT–VERB–OBJECT is the simplest
and most natural word order because it accords mimetically with the temporal
order of actions in the physical world had its classical formulation in Rivarol’s
De l’universalité de la langue française (1784), within a discourse very much
focussed on the ‘natural’ in language (and its doublet ‘naïf’, on which see
Kibbee 1995):

What distinguishes our language from the ancient and modern languages is the order
and construction of sentences. This order must always be direct and necessarily clear.
French first names the subject of the discourse, then the verb which is the action, and
finally the object of this action; this is the logic natural to all men; this is what
constitutes common sense. 4

One need only be a weak Whorfian to wonder whether SVO is so self-evidently

a matter of natural logic and common sense for speakers of non-SVO
languages, and if it is, why the majority of the world’s languages do not have
this order. Yet as late as 1947 Dauzat calls SVO “l’ordre logique”, and the basic
view is still alive and well, not only among those who still consider French to
be the natural language of man, but among generative linguists, who since
Kayne (1994) have formed a consensus that SVO is the setting of our innate
UG, from which any other order has to be derived.

4.6 Unplanned
The belief that the natural state of language is found in its least deliberate,
least ‘conscious’ production is an ancient one, but it would not come to the
fore until the Romantic period, after which it would attain steadily greater
prominence into the 20th century. It remains a point of dogma for many
linguists (apart from Chomskyans) that valid data must come, wherever
possible, from ‘naturally occurring conversation’. Historically, this concept
overlaps so much with the equation of the natural with the bodily (§4.3) that
logic might militate for folding them together; but for clarity of presentation
“Ce qui distingue notre langue des langues anciennes et modernes, c’est l’ordre et la con-
struction de la phrase. Cet ordre doit toujours être direct et nécessairement clair. Le français
nomme d’abord le sujet du discours, ensuite le verbe qui est l’action, et enfin l’objet de cette
action; voilà la logique naturelle à tous les hommes; voilà ce qui constitue le sens commun.”

and discussion I am treating them separately. Anything that is planned is

complex and false in the sense that, although it appears to be proceeding
spontaneously, it has actually been worked out in advance on a different time-
scale, in order to achieve some end. Romanticism celebrates the spontaneity of
the childlike and primitive as being in direct communion with the true, and it
redefines the beautiful accordingly. Language in its natural state will be
directly expressive of the soul. Within the dyad of mind and body, planning
belongs to the mind, that theatre in which the future is staged and written for.
I noted in §4.3 that ‘body language’ is taken to be more natural and truthful
than oral language. In part this is because most of us do not plan our body
language the way we do our verbal language; it happens less consciously. The
link between unplanned language and truth has legal recognition to the extent
that courts admit lie detector evidence, where the ‘body language’ takes the
form of physical movements that are detectable only to a machine.
The notion of ‘unconscious’ language production is, of course, a modern
one. Joseph (2006, §5.1–2) looks at how swearing, traditionally regarded as
contrary to reason, shifted from being seen as unnatural when it was assumed
always to be wilful, to natural when even clerical writers gradually came to
admit that at least some occurrences were ‘automatic’. §6.5 of the same book
examines how Chomsky’s theory treats the interpretation of utterances as
natural just so long as they take place with no conscious effort, otherwise
marginalising them as ‘imposed’.
The conception of the natural as unplanned has too many other reflexes in
modern linguistic thought to enumerate here, but they include all that was said
in §1 above about the marginalisation of standard languages, themselves the
product of planning. Linguistics treats ‘correct’ or ‘proper’ language as a
deliberately created social artefact, and any deviation from it as having a
natural explanation — unless the deviation occurs in ‘literary’ writing. Many
linguists feel an instinctive repulsion at the introduction of data taken from a
literary source, as if the ‘literary’ formed some clearly defined category of
texts so deliberate in their planning as not really to be language at all.

4.7 Rural, connected to the land

Williams (1973) demonstrated how the rural and the urban have been
constructed in modern thought quite along the lines of the classic nature–other
dichotomy. So it is not surprising that, insofar as forms of language can be
linked to ‘the soil’, they will be regarded as more natural than those having to
do with the relatively abstract life of the city. This ideology is continuous with
those of the physical, the simple and the unplanned, and also with the unmixed
and pure (§4.10 below). My comments in §1 about my training in Romance
linguistics reflect how much this ideology of the soil endured in that subfield at
least into the 1980s. It also had reflexes in the 18th-century ideas of Rousseau

and Montesquieu (1689-1755) on how climate and environment shape

language, in the 19th-century discourse of language and nationalism, for
instance with Fichte (on whom see Joseph 2004b:109–115), and more recently
in the discourse on the spread of ‘world’ languages at the expense of smaller
community ones. Here anxiety is expressed that speakers of the latter are being
robbed of their natural cultural heritage, the language they should properly
speak and insist on having their children educated in; and when the speakers
themselves make different choices, it is attributed to a false consciousness
spread by hegemonic forces driven by interests that run counter to the natural
way of things.

4.8 Musical, poetic

The idea of language being naturally musical, as well as poetic (in terms
both of sound structure and of meaning being metaphorical rather than direct),
has been discussed in §3 with reference to Rousseau. It would abide in modern
linguistics at least through Jespersen (1922:483–484), where it gets
modernised by the addition of a Darwinian element:

[T]he genesis of language is not to be sought in the prosaic, but in the poetic side of
life; the source of speech is not gloomy seriousness, but merry play and youthful
hilarity. And among the emotions which were most powerful in eliciting outbursts of
music and of song, love must be placed in the front rank. To the feeling of love,
which has left traces of its vast influence on countless points in the evolution of
organic nature, are due not only, as Darwin has shown, the magnificent colours of
birds and flowers […]; it inspired many of the first songs, and through them was
instrumental in bringing about human language.

The poetic nature of language was however in some conflict with the powerful
concept of it as unplanned, which explains the marginalisation of the poetic
concept in later linguistic thought.

4.9 Unmixed, pure

While admitting that all languages have undergone some degree of outside
influence, linguists and non-linguists alike have tended to imagine languages
as naturally inward-looking, complete in themselves (this relates to systema-
ticity, §4.10), and taking on elements from other languages only in case of
need (a lexical gap) or under duress after conquest. For Humboldt (1767-
1835), a language is superior insofar as it is pure within its particular type — a
sign that its original genius and nature have been relatively undiluted. As
Jespersen (1922:191) points out, “some earlier scholars denied categorically
the existence of mixed languages”, and he rightly singles out Hugo Schuchardt
(1842-1927) for having challenged such views. Creolists like Schuchardt have
been inclined to see mixture as natural to language, but they are in the

minority; and pidgins, which as we saw in §4.5 are treated as natural on

account of their simplicity, will instead be described in very different terms
when their mixed nature is the issue. The ties between language and ethnic
identity are bound up with this concept, and how people think about the
naturalness of language mixture often reflects what they think of racial
Bi- and multilingualism too tend to be treated as less natural than
monolingualism, even though, globally speaking, they are the norm. To quote
the great Dane one last time:

It is, of course, an advantage for a child to be familiar with two languages: but
without doubt the advantage may be, and generally is, purchased too dear. […T]he
brain effort required to master two languages instead of one certainly diminishes the
child’s power of learning other things which might and ought to be learnt. (Jespersen

Factors of ethnic identity again enter. In October 2004 I took part in a call-in
programme on BBC Asian Radio with two young British men of South Asian
origin, one of whom was there to express his ‘cultural cringe’ at hearing other
South Asians speak English in an ethnically marked way. The second man was
arguing that, on the contrary, your English should express “what you really
are”. As caller after caller kept condemning the first man — one even asking
him why he didn’t go have his skin bleached white (see also Armour 2001) —
I found myself genuinely surprised at the passions aroused in so many by this
issue of ethnic performance in a language that outsiders might reckon not to be
‘theirs’, but the relic of a colonial imposition. Ideas about where the true nature
of both language and people lie are by no means confined to linguists.

4.10 Systematic
How systematic it is in the nature of language to be was at the centre of the
ancient debate between analogists and anomalists. The analogists of
Alexandria left us the legacy of grammar, that form of systematic analysis of
language that has done more than anything else to perpetuate and spread the
belief that, at its most natural, language is a system — maybe even a living
organism — that holds together perfectly. One of the principal results has been
a tendency from ancient times to the present to treat whatever is classified as
lexical as being less natural than what is grammatical or functional. Hence
Humboldt’s conclusion that these structural levels are where the true nature of
a language lies, whereas the word stock — much less amenable to systematic
analysis — is relatively marginal in this regard. This also connects to the
concept of the natural as unmixed and pure, since language mixture occurs
much more readily and frequently in lexicon than in phonology, morphology
or syntax. Joseph (1995; 2000, Chap. 4) traces this history, showing too how

verbs and nouns have been treated as differing from one another in naturalness
whenever one was seen as more bound up with grammar than the other.
Still today, Chomskyan linguistics treats the lexicon as a repository for
whatever is intractable for a systematic analysis. Within syntax, Chomsky &
Lasnik (1977) introduced the distinction between ‘core grammar’ and ‘peri-
phery’, with the former made up of all those elements that proceed directly
from innate UG and therefore are perfectly systematic, while the latter are
deviant in some way. Joseph (2000, Chap. 6) looks at Chomsky’s explanatory
devices, alongside those of Jakobson mentioned above in §4.5, to see how they
rely on an ideology of the natural as systematic, while rather facilely tossing
anything that doesn’t fit into an ad hoc waste bin (see also Thomas 2003).

4.11 Efficient
This category overlaps significantly with ‘simple’ (§4.5), but they are
sometimes in conflict, since the most efficient way of doing something is not
always the simplest. As Jakobson noted, what is simpler for a speaker to
produce may be more difficult for a hearer to process. The naturalness of
efficiency proceeds from the idea that the role which language has played in
human evolution must have been so central that the evolution of language,
indeed its very nature, has to be driven by forces connected to survival —
efficiency of communication on the one hand, economy of mental structure on
the other. The more effectively (quickly and accurately) communication can
take place in a language in moments of conflict with other groups, the better
the chance of survival. The Darwinian perspective would give a new lease of
life to the mimesis concept, since it could be argued that language is more
efficient if mimetic than purely arbitrary; and we have seen it applied to the
concept of language as naturally musical and poetic (§4.8).
If two languages differed in that one demanded more intense mental
processing than the other, speakers of the less demanding one would poten-
tially have an evolutionary advantage in that more of their mental processing
would be available for solving problems. Linguists tend not to believe that
existing languages differ from one another in this way, but within any
language, some structures are taken to be simpler and more basic, others
complex and derived, and the former are treated as more ‘natural’ in the
various senses discussed heretofore. Also, economy of operation and economy
of explanation are taken to be isomorphic — this is the principle known as
Occam’s razor, which says that, given a choice of alternative accounts of the
same phenomenon, the most economical one is the naturally correct one.

5. Conclusions
I have tried to show how attempts to distinguish one part of language from
another on the basis of naturalness, or certain other criteria that recapitulate the

dichotomy of something that is grounded versus something that is not, can be

regarded as normative, and ultimately tautological; and that a historical
continuity exists linking such attempts across the centuries, despite changes in
terminology and shifts in the balance of the sometimes contradictory
conceptions of what in language is natural. My purpose has not been to suggest
that these attempts have been misguided. Rather I believe that we can learn
something important and useful about the underlying methodological and
ideological assumptions of modern linguistics by examining how these notions
were deployed, and that this offers a way forward beyond the limits which the
implicit tautology has imposed upon our explanatory and analytic imagina-
tions. In terms of my own work, it has led me to open up greater space for the
study and understanding of what is individual, constructed, artificial,
seemingly unpatterned and resistant to explanation.
I close with a quote that brings home how much continuity exists in
linguistic thought from the Cratylus to the present day. It is from an interview
in which Chomsky was asked about the idealisation of language that is at the
heart of the generative project, which, for many non-generativists, removes it
too far from the domain of real language as it is actually produced. This was
his reply:

[I]dealization is a very misleading term, because it really means a move towards

reality. When you talk about idealization or abstraction, it is an effort to find the
reality. When you roll a ball down a frictionless plane, that is called an idealization,
but what you are really doing is finding the real principle by which things attract one
another. It is the phenomena that are a nuisance: they are unreal in a way, because
they are too complicated. Reality hides behind the phenomena, as it were, so you have
to get rid of a lot of the phenomena to find it. (Chomsky, in Dillinger & Palácio

It could be Socrates speaking. In saying this I do not mean to suggest that,

insofar as Chomsky is a Platonist, he is wrong. On the contrary, I agree with
what he says here, in fact I consider it a perfect summary of the paradox faced
by every attempt to analyse language. The challenge is to make the phenomena
tractable — this must be done if anything is to be said about them at all. Yet
how we limit them is precisely where all the tekhne, the art, of linguistics lies.
Language is different from a ball rolling down a plane because the boundary
between the ‘real principle’ and the annoying ‘phenomena’ is blurred by the
speaker’s will agency. The speaker corresponds to the plane in Chomsky’s
analogy, and language to the ball. The speaker’s wilful agency is the friction.
Abstracting it away, are we left with what is really real? Whatever the answer,
a more classic example of naturalising the problem would, I think, be
impossible to find.


Armour, William S. 2001. “‘This Guy Is Japanese Stuck in a White Man’s

Body’: A discussion of meaning making, identity slippage, and cross-cul-
tural adaptation”. Journal of Multilingual and Multicultural Development
Bateson, Mary Catherine. 1995. “On the Naturalness of Things”. How Things
Are: A science tool-kit for the mind ed. by John Brockman & Katinka
Matson, 9-16. New York: William Morrow & Co.; London: Weidenfeld &
Bréal, Michel. 1897. Essai de sémantique: Science des significations. Paris:
Chomsky, Noam & Morris Halle. 1968. The Sound Pattern of English. New
York: Harper & Row.
Chomsky, Noam & Howard Lasnik. 1977. “Filters and Control”. Linguistic
Inquiry 8.425–504.
Condillac, Étienne Bonnot, abbé de. 1746. Essai sur l’origine des
connoissances humaines: ouvrage où l’on réduit à un seul principe tout ce
qui concerne l’entendement humain. Amsterdam: P. Mortier. (English
transl., Essay on the Origin of Human Knowledge, by Hans Aarsleff,
Cambridge: Cambridge University Press, 2001.)
Dauzat, Albert. 1947. Le génie de la langue française. Rev. ed. Paris: Payot.
Dillinger, Mike & Adair Palácio. 1997. “Generative Linguistics: Development
and perspectives. An interview with Noam Chomsky”. Chomsky no Brasil /
Chomsky in Brazil, Revista de Documentação de Estudos em Lingüística
Teórica e Aplicada (D.E.L.T.A.), vol. 13, no. especial, 159–193.
Goldstein, Kurt. 1944. “Methodological Approach to the Study of Schizo-
phrenic Thought Disorder”. Language and Thought in Schizophrenia ed.
by Jacob S. Kasanin, 14–40. Berkeley & Los Angeles: University of
California Press.
Herder, Johann Gottfried von. 1772 [1770]. Abhandlung über den Ursprung
der Sprache, welche den von der Königl. Academie der Wissenschaften auf
das Jahr 1770 gesetzten Preis erhalten hat. Berlin: Chr. Friedrich Voss.
(English transl. of first two parts by Alexander Gode, in On the Origin of
Language: Two essays by Jean-Jacques Rousseau and Johann Gottfried
Herder, 85–166, New York: Frederick Ungar, 1966.)
Jakobson, Roman. 1963. “Implications of Language Universals for
Linguistics”. Universals of Language ed. by Joseph H. Greenberg, 263–
278. Cambridge, Mass.: MIT Press.
Jespersen, Otto. 1922. Language: Its nature, development and origin. London:
George Allen & Unwin; New York: Henry Holt.
Jespersen, Otto. 1925. Mankind, Nation, and Individual from a Linguistic
Point of View. Cambridge, Mass.: Harvard University Press.

Joseph, John E. 1990. “The Abandonment of nómos in Greek Linguistic

Thought”. Historiographia Linguistica 17.1–13.
Joseph, John E. 1995. “Natural Grammar, Arbitrary Lexicon: An enduring
parallel in the history of linguistic thought”. Language and Communication
Joseph, John E. 2000. Limiting the Arbitrary: Linguistic naturalism and its
opposites in Plato’s Cratylus and modern theories of language. Amsterdam
& Philadelphia: John Benjamins.
Joseph, John E. 2004a. “Body, Passions and Race in Classical Theories of
Language and Emotion”. Emotion in Dialogic Interaction: Advances in the
complex ed. by Edda Weigand, 73–96. Amsterdam & Philadelphia: John
Joseph, John E. 2004b. Language and Identity: National, Ethnic, Religious.
Basingstoke & New York: Palgrave Macmillan.
Joseph, John E. 2005a. “The Tongues of Men and of Angels: Knowledge,
inner speech and diglossia in medieval linguistic thought”. Flores
Grammaticae: Essays in memory of Vivien Law ed. by Nicola McLelland
& Andrew R. Linn, 119–139. Münster: Nodus.
Joseph, John E. 2005b. “Concrete Language and Madness in Psychiatry and
Linguistic Thought, 1938–46”. Paper read to the annual meeting of the
North American Association for the History of the Language Sciences,
Oakland, Cal., 6–9 January 2005.
Joseph, John E. 2006. Language and Politics. Edinburgh: Edinburgh Univer-
sity Press.
Joseph, John E., Nigel Love & Talbot J. Taylor. 2001. Landmarks in Linguistic
Thought II: The Western Tradition in the Twentieth Century. London &
New York: Routledge.
Kayne, Richard S. 1994. The Antisymmetry of Syntax. Cambridge, Mass.: MIT
Kibbee, Douglas A. 1995. “Dictionaries and Usage in 17th-Century France: Le
naïf françois”. History of Linguistics 1993: Papers from ICHoLS VI ed. by
Kurt R. Jankowsky et al., 167–177. Amsterdam & Philadelphia: John
Meillet, Antoine. 1923. “Le caractère concret du mot”. Journal de
Psychologie, 246ff. (Repr. in Linguistique historique et linguistique
générale, vol. II, 9–13. Paris: H. Champion, 1936.)
Rée, Jonathan. 1999. I See a Voice: A philosophical history of language,
deafness and the senses. London: HarperCollins.
Rivarol, Antoine de. 1784. De l’Universalité de la langue française; discours
qui a remporté le prix à l’Académie de Berlin. Berlin: [s. n.], et se trouve à
Paris chez Bailly et chez Dessenne.
Rousseau, Jean-Jacques. 1781 [1761]. “Essai sur l’origine des langues, où il est
parlé de la mélodie et de l’imitation musicale”. Traités sur la musique,

210–225. Genève: [s. n.]. (English transl. by John H. Moran, in On the

Origin of Language: Two essays by Jean-Jacques Rousseau and Johann
Gottfried Herder, 1–74, New York: Frederick Ungar, 1966.)
Taylor, Talbot J. 2000. “Language Constructing Language: The implications of
reflexivity for linguistic theory”. Language Sciences 22.483–499.
Thomas, Margaret. 2003. “Plato’s Problems and Plato’s Problem”. Language
and Communication 23.81–91.
Voltaire [François-Marie Arouet le jeune]. 1764. Dictionnaire philosophique,
portatif. London: [s. n.]. (Engl. transl., Philosophical Dictionary: A
compendium, by Wade Baskin, London: P. Owen, 1961.)
Williams, Raymond. 1973. The Country and the City. London: Chatto &
Wollock, Jeffrey. 1997. The Noblest Animate Motion: Speech, physiology, and
medicine in pre-Cartesian linguistic thought. Amsterdam & Philadelphia:
John Benjamins.

Adam Mickiewicz University, Poznań


In this paper I give an overview of tendencies in the research on

grammatical gender within the Western linguistic tradition. More specifically, I
focus on the recurring claims concerning the supposed semantic arbitrariness,
and formal and non-functional character of this category. Representative
examples are given from every period of linguistic activity, from the ancient
Greek scholars up to contemporary descriptive and typological studies.
Particular attention is given to the most influential works, e.g. those of the
Neogrammarians and European as well as American structuralists within 19th
and 20th century scholarship. While examples have been drawn mainly from
the research on Indo-European, the tendencies described are also indicative of
the research on other families and on systems traditionally referred to as “noun
classes”. Finally, I consider these claims in the light of the evidence that is now
available of the semantic regularity of gender, its discourse functions and
cognitive correlates.

The mere mention of linguistic gender has always stimulated heated

discussions, regardless of whether semantic, formal, functional, synchronic or
diachronic aspects were concerned. 1 In this paper I would like to consider
tendencies in the history of research on gender whose common ground consists
in the insistence on its arbitrariness.
We’ll begin with an overview of the tendencies, which will then be
illustrated with examples from the Western linguistic tradition, together with
counter-arguments offered by contemporary scholars. Generally, these
conflicting approaches can be analysed in terms of such distinctions as
meaning vs. form, regular vs. arbitrary and functional vs. redundant. Due to the
limitations of the present paper and the amount of relevant literature, I would

I would like to thank Peter Trudgill and Yishai Tobin for reading an earlier version of this
paper, providing encouragement and sharing ideas with me.

like to consider only the most representative treatments, as examples of the

attitudes which were characteristic of a given period. Finally, I will briefly
review the evidence that is now available of the semantic regularity of gender,
its functions and cognitive correlates.
In the most characteristic claim that has been made, grammatical gender
has been described as arbitrary, typically in terms of its semantic content –
either referential (with respect to entities in the real world) or lexical (with
respect to the lexical meaning of the noun). Less frequently, a lack of
regularity has been attributed to formal (morphological or phonological)
assignment principles. As regards semantic arbitrariness, examples cited most
frequently involve nouns from the lower parts of the animacy hierarchy, i.e.
inanimates. Other common examples include “problematic” animates, such as
epicenes, which denote both males and females but take only one agreement
form, e.g., G. maus ‘mouse’ f., and hybrids, such as G. Mädchen ‘girl’ n.,
which take more than one agreement form (here n. or f.), and finally sex-
differentiable nouns, by virtue of their relatively small share in the lexicon. It is
a notable feature of many approaches that they accentuate the lack of complete
regularity in assignment, also in the face of the outcomes of morphological or
phonological change, as in the contrast between m. and f. genders in Romance
and c. and n. in mainland Scandinavian, or cross-linguistic differences in the
expression of the sex differentiation.
The insistence on semantic arbitrariness of gender may be reflected in the
treatment of formal assignment criteria and its function. Gender has thus been
regarded as a system of formal classification, motivated only, or primarily, on
morphological or phonological grounds. In terms of its function, it has been
described as a “luxury”, a category which is largely redundant or non-
functional with respect to its role in grammar and human communication,
especially in comparison with other grammatical categories such as number or
tense, and in view of its supposed non-universal character.
The early Greek scholars addressed the issues that were to form the core of
the discussion of gender until modern times. These generally concerned the
inadequacy of the correspondence between natural and grammatical gender, as
exemplified in the status of n. gender, the assignment of inanimates and
individual irregularities in the expression of the category.
In view of the lack of correspondence with the sex distinction, n. gender
was defined by Aristotle (384-322 BCE) as τὸ μεταξύ “that which is between
m. and f.”, and οὐδέτερον “neuter”, i.e. “neither of the two”, by the Stoics and
Dionysius Thrax (c. 170-c. 90 BCE), in contrast to the earlier notional
definition as σκε͂υος “thing” by Protagoras (c. 490-c. 420 BCE). Cases of
irregularities in the expression of gender were provided by epicenes as well as
assignments which were thought of as incongruous, based on proposed
principles. Aristotle tells us that Protagoras suggested that m. gender was more

appropriate for the f. nouns μῆνις ‘anger’ and πήληξ ‘helmet’ (De sophisticis
elenchis 173b 20). In turn, Socrates (469-399 BCE), as described by
Aristophanes, is said to have proposed the f. form ἀλεκτρύαινα ‘hen’ instead
of ἀλεκτρυών ‘cock, hen’ (Nubes 660-80). Aristotle himself discussed
solecisms in the use of epicenes and n. nouns due to the supposed
“intermediate” qualities of n. gender (De sophisticis elenchis 174a 1). In his
later attack on analogists, Sextus Empiricus (fl. 200 CE) considered the
assignment of m. or f. gender to inanimates and the presence of epicenes, as
examples of the anomalies in the relation between natural and grammatical
gender (Adversus grammaticos 142-153).
These accounts can be viewed in terms of the dichotomies of nature vs.
convention and analogy vs. anomaly. Among naturalists, for whom categories
were a regular reflection of the object’s qualities, gender was understood as a
classification that reflects natural gender, while conventionalists, in the face of
irregularities in language, regarded the correlation as conventional. We can
here draw an analogy between the arguments of the two sides and the
arguments given by modern linguists: thus a naturalist position can generally
be seen in the insistence on semantic or extralinguistic motivation of gender as
a type of nominal classification, while a conventionalist position would
correspond to the insistence on its semantic arbitrariness. The attempts at
bringing in line the correspondence between natural and grammatical gender
can be seen as a reflection of the parallel dichotomy of analogy vs. anomaly,
with an often prescriptive emphasis on regularity between meaning and form,
or an emphasis on the presence of irregularities in inflectional morphology.
Fully notional criteria were applied by Priscian (d. ca. 530) in the form of
the distinction active vs. passive, based on the properties of the sexes and their
roles in procreation (Keil 1855-59). Owing to the influence of his Institutiones
grammaticae, this interpretation of gender reappeared throughout the Middle
Ages. Thus in the grammars of the Modistae, e.g., De Modis significandi by
Martin of Dacia (d. 1304) and Grammatica speculativa by Thomas of Erfurt
(fl. 1300), the three genders are distinguished through the opposition between
active and passive properties of males and females, with n. gender signifying
an indeterminate property. At the same time, gender was relegated by Ockham
(c. 1285-c. 1349) to a meaningless and unnecessary category, as opposed to
those, such as number, which reflect distinctions in the universal mental
language (Summa I.3); we will later encounter comparable statements in, e.g.,
20th century structuralist studies.
If we now look at the grammars of the 17th century, Arnauld (1612-1694)
and Lancelot (1615-1695) identified the role of gender in the expression of the

relation between the noun and the adjective.2 While the assignment of animates
is attributed to sex and typical occupations, that of inanimate nouns is said to
proceed on an arbitrary basis, “by pure caprice and a habit without reason”
(Arnauld & Lancelot 1975[1660]:77), an observation based on different
assignments in related languages and gender change within the same language.
In addition, the Port Royal scholars criticized Greek and Roman grammarians
for not treating n. gender in rational terms, i.e. as the absence of the two
properties, and for the addition of common and epicene genders. This approach
can be contrasted with the notional accounts in the 17th and 18th centuries by,
e.g., Antoine Court de Gébelin (1774), and James Harris (1751). 3
The period between the late 18th and late 19th century was dominated by the
controversy over the origin of Indo-European gender. We will first consider
the dominant notional view and then look at the counter-arguments. According
to a succession of scholars (Herder 1772, Adelung 1783, Humboldt (e.g.,
Humboldt 1827), and Grimm 1890), the origin of the category was sought in
imagination and personification, in an extension of natural gender to
inanimates, based on semantic oppositions such as active vs. passive. 4 It is
important to emphasize here the role which was attributed especially by
Humboldt (1767-1835) to perception and categorization, which shape
The insistence on the development of Indo-European grammatical gender
from natural gender and the often almost exclusive role assigned to semantic
criteria were eventually questioned by the Neogrammarians. The article by
Karl Brugmann (1889) explicitly stated their arguments. While Brugmann
(1849-1919) admitted the presence of a connection between grammatical and
natural gender, he rejected the possibility of the development of grammatical
gender from natural gender by way of personification, which he regarded as
historically secondary, instead attributing the development of f. gender to a
morphological reanalysis of certain suffixes. The resulting exchange with
Gustav Roethe (1859-1926) illustrates a range of motifs found elsewhere in the
accounts of the nature of gender, particularly in the denial of – or insistence
upon – its semantic regularity, function and universal character. 5
Gender featured prominently in the late 18th and throughout the 19th
century in discussions about the supposed value of grammatical categories and
their role in the development of a language or a nation. On the one hand,

Thus, “… the masculine or feminine gender of a word does not properly concern its
signification, but merely says that grammatically it is such that it should be joined to the
adjective in the masculine or feminine ending” (Arnauld & Lancelot 1975[1660]:78).
Arnauld and Lancelot would certainly have disapproved of the abundance of genders
introduced in English by 17th and 18th century grammarians (see Baron 1986:95-97).
Analogous views were voiced by, e.g., Pott (1836), Bindseil (1838), and Bopp (1857).
For an account of the debate see, e.g., Royen (1929:42-141) and Kilarski (2006).

grammatical gender was described as an unnecessary component of language

which it is best to dispose of (Adelung 1783) or a luxury (Gabelentz 1891).
The loss of grammatical gender, as in English, was therefore approved of
(Bleek 1872), 6 particularly in view of the belief that gender appears only in
primitive languages and societies, as “a useless encumbrance” which is
“perhaps inevitable at that early childish stage of the human intelligence”
(Farrar 1865:212). On the other hand, in the spirit of the age, the presence of
grammatical gender, as a typical feature of morphologically complex
languages, was regarded as an indication of linguistic development, as a result
of which isolating, agglutinating and inflectional languages which have lost it
were viewed as less developed (Pott 1836, Bopp 1857, Grimm 1890).
Likewise, the presence of gender was considered a marker of cultural
development. The ideal was commonly regarded to have been reached in Indo-
European, gender being a prerogative of its speakers, their aesthetic and
scientific creativity (Gabelentz 1891), and their whole civilization (Miklosich
1868-74, Lepsius 1880). 7
The explanations proposed in 19th century scholarship, particularly the
opposition between a semantic and formal motivation for gender, reappeared in
the 20th century, to be discussed within structural and anthropological
linguistics, and later within purely formal perspectives as opposed to cultural,
cognitive and typological ones. We’ll first consider examples of descriptive
studies in which gender was described in predominantly formal terms,
following the Saussurean precept of the arbitrariness of the linguistic sign.
Several scholars voiced their disillusionment with the possibility of stating
assignment rules, suggesting that gender plays a less central role among
grammatical categories, remaining “un luxe linguistique, sans relation avec la
logique” [a linguistic luxury, with no connection to logic] (Bally 1935:66).
According to, e.g., Jespersen (1924:228), “[i]t is certainly impossible to find
any single governing-principle in this chaos.” The loss of grammatical gender
contributes then towards a more analytic structure and communicative
efficiency, in the “perfect stage of complete genderlessness” (Jespersen
1922:347). 8 Likewise, in American structuralism, gender was defined by
Bloomfield (1887-1949) (1933) in terms of arbitrary classes, a claim which he

“The more logical arrangement which some Northern TEUTONIC nations (particularly the
English) have adopted, by removing all nouns in which sex cannot be distinguished from the
masculine and feminine classes (or genders), has its undoubted advantages” (Bleek 1872:79).
Nevertheless, Steinthal (1823-1899) regarded the presence of gender inflection in the verb in
Slavonic languages as evidence of a weaker Sprachgeist, adding that “besides, it’s the Slavs
who seem to me the most sluggish amongst the Indo-Germanic tribes” (Steinthal 1865:297).
Like Steinthal (1865) (cf. fn. 7), Meillet (1931) was critical of Slavonic, on account of the
introduction of distinctions based on animacy, as opposed to the loss or reduction of the
category in “advanced” languages such as English and Romance. In contrast, this development
was interpreted by Hjelmslev (1956) as contributing towards greater transparency.

illustrated unhesitatingly with Algonquian and Indo-European. 9 In his

distrustful stance towards regularity in gender assignment, as well as the
methodology of non-formal explanations generally, Bloomfield’s account can
be contrasted with those of Sapir (1884-1939) and Whorf (1897-1941), to
which we will return shortly.
More impressionistically, Baudouin de Courtenay (1845-1929) (1929)
regarded grammatical gender as a historical accident, comparable to a
deformity or a bad habit, and held it responsible for a range of human
afflictions, including nightmares, pathological behaviour, erotic and religious
delusions and sadism. Further, he associated the presence of the category in
Indo-European and Semitic with the cruelty and acts of savagery committed by
the speakers of these languages. Consequently, as it also contributes towards
literary and scientific creativity, gender emerges as friend and foe, blessing and
Semantic, social and psychological factors were explicitly rejected by
Fodor (1920- ) (1959) and Ibrahim (d. 1994) (1973). Fodor contrasted gender
with other grammatical categories, describing it as superfluous and having no
relation to conceptual categories. 10 If we follow these arguments and reject
social and psychological correlates, then grammatical gender may well appear
to be “an accident of linguistic history” (Ibrahim 1973:50). Following Fodor,
Ibrahim asserted a formal nature of gender, reduced to a “semantically...empty
category” (1973:53), and, in a like manner, contrasted earlier theories of
gender based on personification with modern empirical ones, where linguistic
structure is described as shaping – rather than being shaped by – thought
As regards the diachrony of gender, several studies suggested that a
semantic classification was only a secondary development in the origin of
Indo-European gender, to be triggered originally by a formal change. An initial
phonological change, as suggested by Lehmann (1916- ) (1958), was to result
in a set of consonantal endings and tripartite congruence, “though without any
reference to sex or natural gender” (Lehmann 1958:196). Similarly,
Kuryłowicz (1964) argued against non-linguistic factors, and related the two
successive splits in the Indo-European gender system to syntactic and
phonological changes. The explanation offered by Fodor (1959) was based on

“The gender-categories of most Indo-European languages ... do not agree with anything in the
practical world, and this is true of most such classes. … [t]here seems to be no practical
criterion by which the gender of a noun in German, French, or Latin could be determined”
(Bloomfield 1933:271, 280).
“Natural gender, value differentiation and the habits and forms of thinking of the community
speaking the language cannot bring about the category of gender, because the content of
thought cannot be transferred to its external form, the system of language” (Fodor 1959:213).

a combination of syntactic, morphological and phonological factors; his

insistence on the semantic arbitrariness of gender has already been mentioned.
These accounts of the nature and origin of gender can be contrasted with
structuralist studies influenced by late 19th and early 20th century sociological,
psychological and anthropological theory. The category of gender was here
treated as a primarily social and psychological phenomenon, its development
resulting from an interplay between structural and extralinguistic factors, as
emphasized by several French and Dutch scholars, e.g., de la Grasserie (1906),
Meillet (1921, 1931), and Royen (1929), and within American anthropology
by, e.g., Sapir (Sapir & Swadesh 1946) and Whorf (1945). Further analogies
were made with religious beliefs (Nieuwenhuis 1935, Wienold 1967), the
matriarchal and nomadic periods in the history of the Indo-Europeans (Schmidt
1926, Havers 1960), as well as climatic and geographical correlates
(Nieuwenhuis 1935). Supposed male and female features in the assignment of
inanimates, particularly the active vs. passive distinction, were considered by,
e.g., Martinet (1957) and Meillet (1921). An analogy between gender and the
creation of the world was made by, among others, Baudouin de Courtenay
(1929). Finally, the origin of individual genders was attributed to less tangible
factors such as “magical energy” (Wensinck 1927) and the male sexual drive
(Sperber 1912).
If we now turn to the research on gender over the last three decades or so,
we can see a substantial contribution based not only on previously undescribed
languages but also on languages which have long been at the centre of
attention. This has made it possible to state with more confidence the common
formal and semantic properties of gender systems.
The attitudes that we have traced since antiquity also recur in contemporary
studies, for example as regards the supposed arbitrariness of gender in German
(e.g., Maratsos 1979). 11 Examples used in this context are in line with those
used in earlier accounts, including Mädchen ‘girl’ n. Thus Allan (1977:290-
291) emphasizes the role of the derivational suffix -chen over the
characteristics of a referent, while Braunmüller (2000:44) uses the noun to
demonstrate that “only grammatical rules operate in languages with a three-
gender system [where the] use of gender is thus restricted to the morphological
and syntactic level”. 12

“The classification is arbitrary. No underlying rationale can be guessed at [...] The presence
of such systems in a human cognitive system constitutes by itself excellent testimony to the
occasional nonsensibleness of the species. Not only was this system devised by humans, but
generation after generation of children peaceably relearns it” (Maratsos 1979:232).
This is however a double-edged sword, since in such hybrid nouns the competition between
semantic and formal rules may be resolved in favour of semantics, depending among others on
the distance between the controller noun and target forms (see Corbett 1991: Chp. 8).

In view of the supposed obscure function of natural gender other than in 3rd
person sg., and grammatical gender in languages with a small number of
genders, Trudgill (1999) refers to these phenomena as afunctional historical
baggage, or “linguistic male nipples”. Considering the increase in language
contact in the modern world, Trudgill (1999:149) claims that it is “not unlikely
that languages with large numbers of afunctional grammatical devices will
become less numerous, and indeed it is not entirely impossible that linguistic
gender, except perhaps for natural gender in the third person, will one day
disappear from the languages of the world, never to return.”
As the most grammaticalized type of nominal classification, gender
inevitably has a weaker correlation with semantic properties than noun classes
and classifiers. In consequence, it has been described in this context as
“semantically empty” (Allan 1977:290-291) or arbitrary beyond the sex
distinction, e.g., by Grinevald (2004) 13 and Moravcsik (1978). 14 Comparable
claims with respect to semantic content have also been made about the other
types. 15
In contrast, the insistence on semantic or referential motivation underlies
the arguments for a common semantic basis in all gender languages, as
suggested by Greenberg (1966), Aksenov (1984) and Corbett (1991), as well as
for the presence of assignment rules in individual languages, e.g., by Köpcke
& Zubin (1984) on German, 16 and Tucker, Lambert & Rigault (1977) on
French. The common semantic principles and organization of nominal
classification types have been explored within the UNITYP project (e.g., Seiler
& Lehmann 1982), and in the more recent contributions in Aikhenvald (2000),
Senft (2000) and Unterbeck & Rissanen (2000).
These descriptive and typological studies suggest that formally based
approaches to gender are largely inadequate. In the first place, they ignore the
presence of a semantic basis, which is reflected in the primary properties of
animacy and sex, as well as such secondary criteria as physical properties and
countability. Further, the rejection of cognitive motivation and the insistence
on an incidental nature of gender disallows the possibility of a more universal

“Gender systems … have either two or three classes … to which most nouns are assigned
arbitrarily, beyond the recognition of sex differences for animate nouns” (Grinevald
“Gender thus includes distinctions related to animacy, humanness, sex, or any other
qualitative property of nominal referents, as well as distinctions that are not correlated with
any such semantic property – such as the masculine-feminine-neuter distinction in GERMAN
or other INDOEUROPEAN languages or the semantically equally non-interpretable
distinctions on which noun classification in BANTU languages is based” (Moravcsik
See, e.g., Richardson (1967:373) on noun classes, and Lehman (1979:165) on classifiers.
It is suggestive that Zubin & Köpcke (1986) base their analysis of taxonomies in the German
lexicon on Grimm’s (1890) account of m., f. and n. genders.

study of gender in relation to cognitive categories, and belies the possibility of

the development of gender essentially in any language. In addition, the
observed correlates with social parameters suggest that gender is “probably the
only grammatical category which directly phenomena and
shifts” (Aikhenvald 2000:347). Likewise, metaphorical extensions of animate
genders to inanimates point towards correlations with cultural parameters. 17
Finally, as regards the function of gender, evidence against its non-functional
character is provided by semantic roles, e.g., in the organization of lexicon, and
discourse functions, e.g., in reference tracking (for an overview see
Aikhenvald 2000: Chp. 12).
Despite the different cultural and philosophical background, the treatments
of gender in different periods of linguistic thought show remarkable
similarities. One feels tempted to think of how the motifs we have considered,
i.e. semantic arbitrariness, and formal and non-functional character of gender,
will resurface in future accounts of the category, with new data made available,
but also in the face of the impending loss of so many endangered languages.
It has been repeatedly stressed that it is important to study gender and the
other nominal classification systems, as they “offer a ‘unique window’ into
studying how humans construct representations of the world and encode them
into their languages” (Aikhenvald 2000:307). I hope to have shown that in turn
they offer a unique window into studying how humans construct
representations of linguistic structure and encode them into their grammars.


Adelung, Johann Christoph. 1783. “Von dem Geschlechte der Substantive”.

Magazin für die Deutsche Sprache 1:4.3-20.
Aikhenvald, Alexandra Y. 2000. Classifiers: A Typology of Noun
Categorization Devices. Oxford: Oxford University Press.
Aksenov, A.T. 1984. “K probleme èkstralingvističeskoj motivacii
grammatičeskoj kategorii roda”. Voprosy jazykoznanija 33:1.14-25.
Allan, Keith. 1977. “Classifiers”. Language 53.285-311.
Aristophanes. 1970. Clouds. Ed. by Kenneth James Dover. Oxford: Oxford
University Press.
Aristotle. 1984. “De sophisticis elenchis”. The Complete Works of Aristotle.
The Revised Oxford Translation. Ed. by Jonathan Barnes. Princeton:
Princeton University Press.

See, e.g., Pawley (2002) on English, and Dixon (1982) on Dyirbal. Admittedly, there are
problems with psychological and social correlates, particularly as regards the lack of evidence
and circularity in explanation, cf. Aikhenvald (2000:341, 345), Pawley (2002:135).

Arnauld, Antoine & Claude Lancelot. 1975[1660]. General and Rational

Grammar: The Port-Royal Grammar. 2nd ed. Ed. and transl. by Jacques
Rieux & Bernard E. Rollin. The Hague: Mouton.
Bally, Charles. 1935. Le langage et la vie. 2nd ed. Zurich: Max Niehans
Baron, Dennis E. 1986. Grammar and Gender. New Haven: Yale University
Baudouin de Courtenay, Jan. 1929. “Einfluß der sprache auf weltanschauung
und stimmung”. Prace Filologiczne 14.185-256.
Bindseil, Heinrich Ernst. 1838. Abhandlungen zur allgemeinen vergleichenden
Sprachlehre. I. Physiologie der Stimm- und Sprachlaute. II. Ueber die
verschiedenen Bezeichnungsweisen des Genus in den Sprachen. Hamburg:
Friedrich Perthes.
Blank, David L. 1998. Sextus Empiricus: Against the Grammarians (Adversus
Mathematicos I). Oxford: Clarendon.
Bleek, Wilhelm H.I. 1872. “The Concord, the Origin of Pronouns, and the
Formation of Classes or Genders of Nouns”. The Journal of the
Anthropological Institute of Great Britain and Ireland 1.lxiv-xc.
Bloomfield, Leonard. 1933. Language. New York: Holt, Rinehart & Winston.
Bopp, Franz. 1857. Vergleichende Grammatik des Sanskrit, Send,
Armenischen, Griechischen, Lateinischen, Litauischen, Altslavischen,
Gothischen und Deutschen. 2nd ed. Berlin: Dümmler.
Braunmüller, Kurt. 2000. “Gender in North Germanic: A diasystematic and
functional approach”. Unterbeck & Rissanen 2000.25-53.
Brugmann, Karl. 1889. “Das Nominalgeschlecht in den indogermanischen
Sprachen”. Internationale Zeitschrift für allgemeine Sprachwissenschaft
Corbett, Greville G. 1991. Gender. Cambridge: Cambridge University Press.
Court de Gébelin, Antoine. 1774. Monde primitif. Part 2: Grammaire
universelle. Paris: by the author, Boudet & Valleyre.
Dixon, R.M.W. 1982. “Noun Classes”. Where Have All the Adjectives Gone?
and other essays in semantics and syntax ed. by R.M.W. Dixon, 157-184.
Berlin: Mouton.
Farrar, Frederic W. 1865. Chapters on Language. London: Longmans &
Fodor, István. 1959. “The Origin of Grammatical Gender”. Lingua 8.1-41,
Gabelentz, Georg von der. 1891. Die Sprachwissenschaft. Ihre Aufgaben,
Methoden und bisherigen Ergebnisse. Leipzig: Tauchnitz.
Grasserie, Raoul de la. 1906. De la catégorie du genre. Paris: Ernest Leroux.
Greenberg, Joseph H. 1966. “Language Universals”. Current Trends in
Linguistics ed. by Thomas S. Sebeok, vol. 3: Theoretical foundations, 61-
112. The Hague: Mouton.

Grimm, Jacob. 1890. Deutsche Grammatik. 2nd ed.; 1st ed. 1831. Ed. by Gustav
Roethe & Edward Schröder. Vol. 3. Güthersloh: C. Bertelsmann.
Grinevald, Colette. 2004. “Classifiers”. Morphologie: Ein internationales
Handbuch zur Flexion und Wortbildung ed. by Geert E. Booij et al., vol. 2,
1016-1031. Berlin & New York: Walter de Gruyter.
Harris, James. 1751. Hermes: Or, a philosophical inquiry concerning
language and universal grammar. London: H. Woodfall.
Havers, Wilhelm. 1960. “Zur Lehre vom Genus”. Die Sprache 6.151-155.
Herder, Johann Gottfried. 1772. Abhandlung über den Ursprung der Sprache.
Berlin: Akademie der Wissenschaften.
Hjelmslev, Louis. 1956. “Animé et inanimé, personnel et nonpersonnel”.
Travaux de l’Institut de Linguistique 1.155-199.
Humboldt, Wilhelm von. 1827. Lettre à M. Abel-Rémusat sur la nature des
formes grammaticales en général et sur la génie de la langue chinoise en
particulier. Paris: Dondey-Dupré.
Ibrahim, Muhammad H. 1973. Grammatical Gender: Its origin and develop-
ment. The Hague: Mouton.
Jespersen, Otto. 1922. Language: Its nature, development, and origin. London:
Allen and Unwin.
Jespersen, Otto. 1924. The Philosophy of Grammar. London: Allen and
Keil, Heinrich. 1855-59. Grammatici latini. Prisciani grammatici Caesariensis
institutionum grammaticarum libri xviii. Vol. II-III. Leipzig: B.G. Teubner.
Kemp, Alan. 1986. “The Technē Grammatikē of Dionysius Thrax: English
translation with introduction and notes”. Historiographia Linguistica 13:2-
Kilarski, Marcin. 2006. “Die Materie gegen den Geist: On 19th century German
linguistic epistolography”. IFAtuation: A life in IFA. A Festschrift for
Professor Jacek Fisiak on the Occasion of His 70th birthday ed. by
Katarzyna Dziubalska-Kołaczyk, 385-400. Poznań: Wydawnictwo
Naukowe UAM.
Köpcke, Klaus-Michael & David A. Zubin. 1984. “Sechs Prinzipien für die
Genuszuweisung im Deutschen: Ein Beitrag zur natürlichen
Klassifikation”. Linguistische Berichte 93.26-50.
Kuryłowicz, Jerzy. 1964. The Inflectional Categories of Indo-European.
Heidelberg: Carl Winter.
Lehman, F.K. 1979. “Aspects of a Formal Theory of Noun Classifiers”.
Studies in Language 3.153-180.
Lehmann, Winfred P. 1958. “On Earlier Stages of the Indo-European Nominal
Inflection”. Language 34:2.179-202.
Lepsius, Karl Richard. 1880. Nubische Grammatik. Mit einer Einleitung über
die Völker und Sprachen Afrika’s. Berlin: Wilhelm Hertz.

Loux, Michael. 1974. Ockham’s Theory of Terms. A Translation of Part One

of Ockham’s Summa logicae. Notre Dame: University of Notre Dame
Maratsos, Michael P. 1979. “Learning How and When to use Pronouns and
Determiners”. Language Acquisition: Studies in first language develop-
ment ed. by Paul Fletcher & Michael Garman, 225-240. Cambridge:
Cambridge University Press.
Martin of Dacia. 1961[c. 1270]. “De Modis Significandi”. Martini de Dacia
Opera ed. by Heinrich Roos, 1-118. Copenhagen: G.E.C. Gad.
Martinet, André. 1957. “Le genre féminin en indo-européen: Examen
fonctionnel du problème”. Bulletin de la Société de Linguistique de Paris
Meillet, Antoine. 1921. “La catégorie du genre et les conceptions
indoeuropéennes”. Linguistique historique et linguistique générale ed. by
Antoine Meillet, 211-229. Paris: É. Champion.
Meillet, Antoine. 1931. “Essai de chronologie des langues indo-européennes”.
Bulletin de la Société de Linguistique de Paris 33:1.1-28.
Miklosich, Franz von. 1868-74. Vergleichende Grammatik der Slavischen
Sprachen. Vol. 4: Syntax. Wien: Wilhelm Braumüller.
Moravcsik, Edith A. 1978. “Agreement”. Universals of Human Language ed.
by Joseph H. Greenberg, Charles A. Ferguson & Edith A. Moravcsik, vol.
4: Syntax, 25-140. Stanford: Stanford University Press.
Nieuwenhuis, A.W. 1935. “Die Entstehung des Sexualwortgenus männlich-
weiblich als Kulturelement des Sexualtotemismus und seiner Formen in
den Ausstrahlungsgebieten in Amerika, Australien und Ozeanien”.
Internationales Archiv für Ethnographie 33:1.35-98.
Pawley, Andrew. 2002. “Using ‘He’ and ‘She’ for Inanimate Referents in
English: Questions of grammar and world view”. Ethnosyntax: Explora-
tions in Grammar and Culture ed. by Nick J. Enfield, 110-137. Oxford:
Oxford University Press.
Pott, August Friedrich. 1836. Etymologische Forschungen auf dem Gebiete der
Indo-Germanischen Sprachen. Zweiter Theil. Grammatischer Lautwechsel
und Wortbildung. Lemgo: im Verlage der Meyerschen Hof Buchhandlung.
Richardson, Irvine. 1967. “Linguistic Evolution and Bantu Noun Class
Systems”. La classification nominale dans les langues négro-africaines,
373-390. Paris: CNRS.
Royen, Gerlach. 1929. Die nominalen Klassifikationssysteme in den Sprachen
der Erde. Historisch-kritische Studie, mit besonderer Berücksichtigung des
Indogermanischen. Mödling bei Wien: Anthropos.
Sapir, Edward & Morris Swadesh. 1946. “American Indian Grammatical
Categories”. Word 2.103-111.
Schmidt, Wilhelm. 1926. Die Sprachfamilien und Sprachenkreise der Erde.
Heidelberg: Winter.

Seiler, Hansjakob & Christian Lehmann, eds. 1982. Apprehension: Das

sprachliche Erfassen von Gegenständen. Vol. 1: Bereich und Ordnung der
Phänomene. Tübingen: Gunter Narr.
Senft, Gunter, ed. 2000. Systems of Nominal Classification. Cambridge:
Cambridge University Press.
Sperber, Hans. 1912. “Über den Einfluß sexueller Momente auf Entstehung
und Entwicklung der Sprache”. Imago 1.405-453.
Steinthal, Heymann. 1865. “Die genera des nomen”. Beiträge zur
Vergleichenden Sprachforschung 1.292-307.
Thomas of Erfurt. 1972[c. 1300]. Grammatica speculativa. Ed. and transl. by
Geoffrey L. Bursill-Hall. London: Longman.
Trudgill, Peter. 1999. “Language Contact and the Function of Linguistic
Gender”. Poznań Studies in Contemporary Linguistics 35.133-152.
Tucker, G.R., W.E. Lambert & A.A. Rigault. 1977. The French Speaker’s Skill
with Grammatical Gender: An example of rule-governed behavior. The
Hague: Mouton.
Unterbeck, Barbara & Matti Rissanen, eds. 2000. Gender in Grammar and
Cognition. Part 1. Approaches to gender. Part 2. Manifestations of gender.
Berlin: Mouton de Gruyter.
Wensinck, Arent Jan. 1927. Some Aspects of Gender in the Semitic Languages.
Amsterdam: Koninklijke Akademie van Wetenschappen.
Whorf, Benjamin Lee. 1945. “Grammatical Categories”. Language 21:1.1-11.
Wienold, Götz. 1967. Genus und Semantik. Meisenheim am Glan: Anton Hain.
Zubin, David A. & Klaus-Michael Köpcke. 1986. “Gender and Folk
Taxonomy: The indexical relation between grammatical and lexical
categorization”. Noun Classes and Categorization ed. by Colette Craig,
139-180. Amsterdam: John Benjamins.

Université de Tampere


This article aims at giving an outline of the recurrent debates carried on not
only within Indo-European but also within Finno-Ugrian comparative
linguistics whose arguments are either in favor of the family tree model or the
wave theory model. From the very early doubts surrounding its foundations at
the end of the 19th century until the present day, comparative grammar has
been dominated by these two rival (but also complementary) ways of
patterning language affinities. However, linguistic modelization does not take
place in a void, but is influenced by a complex network of interdisciplinary
transactions which reflect the concepts of ‘space’ and ‘species’. While
establishing relationships between languages, one may observe that their
modelization tries to be a faithful translation of the often opaque historical
reality. Despite the alleged objective criteria, a linguist may turn out to be a
dupe of his own convictions and conceptions, in which case the best model
seems only a matter of choice. Even if one should admit that facts are always
theory-laden, models and theories sanctioned by the history of linguistics
might be considered to have been made according to natural measures.

1. Introduction
La France de la fin du 19e siècle connaît un moment épistémologique
intéressant – elle est témoin de la naissance institutionnelle et intellectuelle de
nombre de disciplines. La linguistique ne se trouve pas dans un paysage
désertique, mais plutôt à un carrefour d’idées où les disciplines se ravitaillent
mutuellement. Sortis de leurs domaines scientifiques primitifs, l’espèce
biologique et l’espace géographique constituent des objets d’imitation qui
articulent les controverses concernant la modélisation des concepts-clefs, qu’il
s’agisse de la politologie, de l’anthropologie, de la sociologie ou de la
linguistique (voir ex. Blanckaert 1995, Guillo 2000). La modélisation de la
langue en tant qu’espèce, centrale dans le cadre de la grammaire historico-
comparative, cède son monopole à une modélisation concurrente, pertinente
pour la dialectologie, qui prend l’espace pour repère. Plutôt que d’un

changement paradigmatique, il est question d’un changement de perspective

qui ouvre d’autres horizons sur la langue, sans jamais être une alternative
exclusive lorsqu’il s’agit d’établir des relations entre les langues. Tant le
modèle de l’arbre généalogique que le modèle de la vague, qui reprend
l’ancien modèle diffusionniste, cherchent à être, malgré une part de facticité, la
traduction plausible d’une réalité souvent opaque en raison de la pénurie de
données historiques. Aujourd’hui encore, les débats qui gravitent autour
d’enjeux implicites dans les deux modèles témoignent de l’absence d’une
modélisation par excellence, apte à exprimer la vie délicate de la langue une
fois pour toutes.

2. La crise de la linguistique finno-ougrienne

Soit un détour. La remise en cause de l’un des composants théoriques
fondamentaux de la grammaire comparée témoigne de la crise de la
linguistique finno-ougrienne moderne. Symptôme d’épuisement d’un
paradigme, un débat intéressant a surgi entre les spécialistes de la linguistique
finno-ougrienne au sujet du statut de l’arbre généalogique (voir Mikone 1996,
1997; Künnap 1997, 1998a, 1998b; Laakso 1998, 1999; Wiik 1996). Les uns
entendent couper l’arbre généalogique en soutenant qu’il s’agit d’un modèle
démodé, dont la structure ne reflète pas la différenciation historique réelle des
langues. Au lieu de décrire la réalité linguistique complexe moyennant une
image estropiée, on a proposé à la place du diagramme arborescent soit un
buisson (voir Häkkinen 1983, 1984; voir aussi Anttila 1989[1972]), soit une
vague, soit un arbre inversé (voir Wiik 1995). Les défenseurs du modèle
arborescent estiment à leur tour que le principe de descendance que l’arbre fait
valoir est non seulement nécessaire sur le plan méthodologique, mais
généralement vrai sur le plan ontologique (Itkonen 1999b). En outre, dans la
perspective historiographique, le modèle arborescent respecte le “principe de
conservatisme épistémologique”, selon lequel “la meilleure théorie
épistémologique [et mutatis mutandis le meilleur modèle] est celle qui permet
de conserver le maximum d’acquis cognitifs produits par le développement
historique” (Auroux 1998:224). En dépit de nombreuses tentatives pour le
couper, l’arbre a réussi à conserver une posture solide.
Les opposants modernes de l’arbre généalogique cherchent à balayer les
acquis incontournables de la grammaire historique et comparée en s’appuyant
sur une théorie des contacts (Wiik 1995, 1996, 1997, 2002). Les tenants de la
théorie des contacts estiment qu’il est plus plausible de classer le finnois plutôt
du côté du suédois que du côté des langues ouraliennes éloignées en raison des
affinités génétiques et culturelles dues à la contiguïté géographique (Östman et
Raukko 1995). La similarité entre le finnois et le suédois qui se dégage de la
structure typologique s’explique par un jeu de substrats, de superstrats et
d’adstrats. Grâce à des échanges interdisciplinaires avec l’archéologie, la
génétique, l’anthropologie, la géologie etc., on a pu reconstituer un nouveau

passé pour les langues finno-ougriennes qui, avant l’infiltration graduelle des
langues indo-européennes, auraient occupé un vaste espace géographique de
l’Europe du Sud-Ouest jusqu’en Sibérie occidentale, et qui auraient laissé une
trace de substrat dans les langues indo-européennes (Wiik 2002). La
grammaire comparée finno-ougrienne traditionnelle a également intégré la
théorie des contacts, mais les emprunts d’une langue à l’autre n’affectent pas
pour autant la place du finnois dans l’arbre généalogique finno-ougrien. 1
En réintroduisant l’espace dans la modélisation linguistique, les critiques
de l’arbre généalogique ont eu la bonne volonté de remplacer le rigorisme de la
grammaire comparée par un relativisme pluraliste qui prône la coexistence de
plusieurs points de vue sur la langue. Par leur critique, ils croient montrer que
c’est la théorie, en l’occurrence le modèle arborescent, qui surdétermine les
faits, que la réalité linguistique est plus complexe que ce que l’arbre sous-
entend. Ils estiment que les défenseurs de l’arbre ignorent la complexité de
l’évolution et de la nature des langues. La surdétermination des faits par la
théorie est effectivement flagrante si l’on fait une interprétation simpliste et
naïve de l’arbre. Les défenseurs de l’arbre ont eux-mêmes souligné le
raisonnement défectueux de cette lecture littérale (Itkonen 1999b:138):
1) La langue souche est homogène (parce qu’elle est représentée par un seul point).
2) Les langues filles ont perdu tout contact mutuel (parce que les bifurcations
laissent un espace vide sur le papier entre les branches).
3) L’arbre généalogique est le seul modèle qui représente l’évolution linguistique
(parce qu’il se tient debout solitaire sur le papier).

L’accentuation délibérée de la naïveté de ces arguments montre les opposants à

l’arbre généalogique sous un jour ridicule. Les critiques de l’arbre partent de
bases erronées, s’ils ne comprennent pas sa nature métaphorique reconnue dans
le cadre de la linguistique historico-comparative (Itkonen 1999a, 1999b). Par là
même, l’intitulé “Penser l’espace, penser l’espèce” s’ancre dans une
problématique épistémologique plus générale qui se plie autour de différentes
représentations de la réalité linguistique, lesquelles finissent par prendre le
dessus pour devenir elles-mêmes la réalité.
Il est intéressant de constater chez les opposants à l’arbre une absence de
références aux précédents historiques, soit en raison de l’indifférence à l’égard
de l’histoire de la linguistique, soit en raison d’une ignorance totale des
fondements historiques de ce débat, d’où une récurrence du questionnement
qui anima la grammaire comparée indo-européenne lors de sa première crise.
Le débat moderne au sein de la linguistique finno-ougrienne n’est que la

Selon la vue traditionnelle, l’emprunt entre les systèmes grammaticaux de deux différentes
langues n’affecte que les sous-systèmes particuliers, alors que l’emprunt entre les dialectes ou
les langues apparentées proches peut concerner tous les éléments en raison de l’identité
grammaticale des systèmes (Itkonen 1966:129).

variante de cette crise. Nous allons établir le bilan des enjeux extra- et
intralinguistiques de ce débat dans le contexte intellectuel français du 19e

3. L’espèce naturaliste
La rencontre de la linguistique avec l’histoire naturelle au 19e siècle
sensibilisa définitivement le linguiste non seulement à la continuité temporelle,
mais aussi à la représentation discontinue de la temporalité. L’expression de
l’espèce évolutive n’est clairement formulée que dans le transformisme de
Lamarck (1744-1829) au début du 19e siècle. L’adhésion initiale de Lamarck à
l’essentialisme suivant les canons scientifiques de l’époque évolue vers une
conception évolutive de l’espèce par le biais du nominalisme. Lamarck estime
que, d’une part, les distributions d’un taxinomiste relèvent de l’art de la pensée
et dérivent des facultés cognitives de l’homme, mais que, d’autre part, le
devoir d’un taxinomiste est de “rechercher dans nos distributions l’ordre même
qui est propre à la nature” (Lamarck 1809:81). La nomenclature de l’art
naturaliste se veut une image fidèle des rapports naturels, mais les
délimitations strictes entre les catégories avoisinantes sont exposées à
Les espèces […] rangées en séries et rapprochées d’après la considération de leurs
rapports naturels, présentent, avec celles qui les avoisinent, des différences si légères
qu’elles se nuancent, et que ces espèces se confondent, en quelque sorte, les unes avec les
autres, ne laissant presque aucun moyen de fixer par l’expression, les petites différences
qui les distinguent. (Lamarck 1809:103)

La difficulté d’appréhender l’espèce est palpable dans les énoncés de Lamarck,

car “il n’y a réellement dans la nature que des individus” (Lamarck 1802:97). Or,
étant donné que la science non seulement n’opère pas par l’intermédiaire des
individus, mais doit déterminer son entité principale, cette réduction nominaliste
s’avère insuffisante:
Cependant pour faciliter l’étude et la connoissance de ces corps, je donne le nom
d’espèce à toute collection d’individus qui, pendant une longue durée, se ressemblent
tellement par toutes leurs parties comparées entr’elles, que ces individus ne présentent
que de petites différences accidentelles, que, dans les végétaux, la reproduction par
graines fait disparaître. (Lamarck 1802:100-101)

La situation des espèces dans un arbre généalogique fait valoir leur lignée
évolutive à partir d’une souche commune ainsi que les rapports que les espèces
entretiennent les unes avec les autres. L’arbre traduit les discontinuités entre
les espèces qui seraient dues à une dynamique endogène et auto-organisatrice.
Le modèle passe sous silence deux facteurs: en premier lieu, il tait le
comportement des populations vivantes dans leur milieu naturel, en second
lieu, il omet que la réalité est sans doute beaucoup plus buissonnante. Lamarck

a toutefois une forte intuition de la complexité réelle de la généalogie que l’on

a du mal à valoriser dans nos modèles descriptifs:
les animaux […] forment une série rameuse, irrégulièrement graduée, et qui n’a point de
discontinuité dans ses parties, ou qui, du moins n’en a pas toujours eu, s’il est vrai que,
par suite de quelques espèces perdues, il s’en trouve quelque part. Il en résulte que les
espèces qui terminent chaque rameau de la série générale tiennent, au moins d’un côté, à
d’autres espèces voisines qui se nuancent avec elles. (Lamarck 1809:104)

Le foisonnement réel des branchements montre que plusieurs espèces

parentes “cohabitèrent souvent dans la même aire géographique et connurent
de multiples interactions” (Gould 1997[1996]:90). Ce questionnement
rapidement esquissé du côté de la biologie émergeante montre au linguiste
qu’il n’est pas seul lorsqu’il s’agit de faire un compromis entre les faits et la
théorie. Or, le linguiste se trouve dans une position peu avantageuse par
rapport à l’historien de la nature. Lorsque manquent les données linguistiques
historiques concernant les évolutions parallèles, tous les modèles s’avèrent
insuffisants. Par son principe de l’hérédité des caractères acquis, Lamarck tient
à souligner que pour échapper à l’extinction, une espèce doit s’adapter à son
environnement en se transformant elle-même suivant les circonstances. En ce
qui concerne la langue, les transformations dues à l’environnement, à savoir
celles qui sont issues des contacts de langues dans l’espace, se superposent aux
transformations endogènes, les seules nécessaires (voir Lass 1997:207-209).
En conséquence, l’espèce semble objectivement être en position de force par
rapport à l’espace.

4. De l’espace géographique à l’espèce géographique

D’emblée, il semble que l’espèce et l’espace soient incompatibles, mais
comment les géographes, les véritables spécialistes de l’espace, voient-ils le
rapport entre ces deux concepts? Selon Roger Brunet (1997:301), “on n’a pas à
couper l’espace géographique à sa guise, il se découpe tout seul”. Cette attitude
discontinuiste moderne n’est pas celle de la première école française de
géographie qui s’intéressa dès sa formation (1870-1914) au caractère continu
de l’espace géographique, mais aussi au rapport qu’entretient l’homme avec
l’espace. Si la surface terrestre est constituée d’infimes gradations, elle permet
une diffusion à passages progressifs des inventions, des traits culturels, des
habitudes etc. (voir Gay 1995; Berdoulay 1995). Or, outre les discontinuités
physiques objectives, l’homme est le générateur majeur des discontinuités
spatiales. Les barrières se constituent en barrières une fois seulement qu’elles
sont perçues comme telles – l’espace géographique devient un espace vécu (Di
Méo et Veyret 2002). Le positivisme, épistémologie dominante du 19e siècle,
cède progressivement pour faire place au conventionnalisme qui situe l’homme
au centre de la science. Le désir positiviste se confond toutefois avec les faits
construits. Lucien Gallois (1857-1941), l’une des figures majeures de la

première école française de géographie, estime que, malgré ses apparences

artificielles, la nomenclature des noms de pays possède un bien-fondé naturel
géologique qui fragmente objectivement la surface terrestre:
Les frontières sont des réalités, même quand elles ne répondent pas à des groupements de
sympathie et d’intérêts que la force ne saurait imposer à la conscience des peuples. Le
seul besoin de limiter son champ d’étude conduit d’ailleurs souvent le géographe à
conserver les cadres établis. […] il faut nécessairement adopter les divisions qui se
prêteront le mieux au classement et à l’interprétation des faits, c’est-à-dire les emprunter
à la nature elle-même. Ces divisions sont des régions naturelles. […] Des noms y sont
restés attachés depuis des siècles à des régions qui n’ont jamais correspondu à des
divisions politiques ou administratives. Ce sont les noms de pays. Il semble bien qu’ils
désignent de véritables régions naturelles et l’on a pu dire qu’il suffisait de les recueillir
avec soin pour retrouver du même coup les divisions rationnelles du sol que l’instinct
populaire, devançant la science, aurait depuis longtemps aperçues. (Gallois 1908:2)

Si le découpage de la nature dans ses articulations spatiales et spécifiques

n’a pas toujours été senti comme une évidence, qu’en est-il de l’espace et de
l’espèce linguistiques? La transposition d’une lecture littérale du modèle
arborescent contribue à engendrer un débat au sujet de la nature et de la réalité
des catégories linguistiques.
La discontinuité géographique participe à une spéciation. Le débat sur les
dialectes qui se déclenche entre Graziadio Isaia Ascoli (1829-1907) et Paul
Meyer (1840-1913) est au cœur de cette problématique (voir Ascoli 1876,
1878; Meyer 1875, 1876). La frange qui dérange le système que constitue
l’espace dialectal français se traduit par la découverte d’un dialecte de
transition entre la langue d’oc et la langue d’oïl permettant un rapprochement
avec une découverte d’une nouvelle espèce animale ou végétale. Ce supposé
observable qu’est le dialecte de transition en question, le franco-provençal, a
des traits communs avec des espèces préexistantes ce qui facilite son
identification dans le tableau du vivant. Mais au lieu d’en donner une
description anatomique et physiologique complète, on choisit de justifier son
statut d’espèce par les traits qui le distinguent de ses proches parents. A
l’intérieur de cette nouvelle espèce, il y a des sous-espèces que l’on explique
par certaines composantes particulières qui ont tendance à se reproduire.
Or, le linguiste crée et emploie des catégories, même si leur définition ne
relève pas du contenu de l’objet réel, mais du contenu que le linguiste a choisi
de leur donner. Pour Paul Meyer, le dialecte est une espèce artificielle, les
frontières dialectales étant absolument arbitraires. Il accorde une réalité aux
seuls traits dialectaux particuliers, et par ricochet nie l’existence de toute
Le dialecte est une espèce bien plutôt artificielle que naturelle […] toute définition de
dialecte est une definitio nominis et non une definitio rei. Or, si le dialecte est de sa nature
indéfini, on conçoit que les groupes qu’on en peut former ne sauraient être parfaitement
finis. De là vient qu’on pourra imaginer bien des manières de les grouper, chacune se

fondant sur un certain choix de faits linguistiques et aucune n’échappant à l’inconvénient

de tracer des circonscriptions là où la nature n’en fournit point. (Meyer 1875:295)

Dans son compte-rendu de ce débat, un Lucien Adam (1833-1918) pose que

l’idée de l’organisme linguistique, ou espèce linguistique, se greffe plutôt dans
l’esprit du linguiste que dans le monde du vivant, bien que par une nécessité
méthodologique on puisse faire un rapprochement avec les espèces naturelles:
[…] quand le linguiste groupe les langues par familles, il considère les langues en elles-
mêmes comme autant d’erga […]; il fait abstraction du temps, de l’espace, de l’histoire,
de la géographie, en un mot de tout ce qui n’est pas la langue elle-même. (Adam

L’entité aux contours bien dessinés située dans l’arbre généalogique est
mise au défi par un modèle alternatif, la théorie des ondes. La topographie du
continu spatial représentée par la vague est à même d’estomper la
configuration qui traduit le discontinu temporel. Les arbres généalogiques
représentant des lignées évolutives parallèles d’espèces disjointes ne sont pas
jugés aptes à rendre compte de l’évolution culturelle, où l’on ne saurait
détecter des frontières étanches. L’espace devient gagnant pour ce qui est de la
transmission des artefacts. Cependant, un repère temporel est présent dans les
deux modèles: l’origine à partir de laquelle la propagation s’enclenche. Dans
l’un des modèles, la descendance se hisse dans l’arbre à partir du point
d’origine, dans l’autre, elle se répand à la manière d’une tâche d’huile sur un
support cartographique. Le modèle évolutionniste accentue les hiatus dans le
continuum temporel, mais en raison de larges intervalles de la pénurie des
données, les points temporels ne sont que très relatifs. En revanche, dans le
modèle diffusionniste une histoire véritablement globale d’un phénomène se
fonde sur l’étendue terrestre qui est couverte d’un filet dont les maillons sont
en rapport continu les uns avec les autres. On est dans l’impossibilité d’établir
des ouvertures et des fermetures dans cet espace à transitions nuancées, si ce
n’est que par une chronologie vague dans laquelle les maillons entrent en
contact, d’où la conception que l’espace géographique est historisé. L’avantage
de la vague par rapport à l’arbre généalogique est de permettre de décrire les
emprunts entre les phénomènes contigus, et par conséquent, leur expansion
dans de vastes contrées, là où l’évolution du type biologique est localement
limitée et à la merci d’une écologie donnée.
La géographie linguistique est variable selon qu’elle s’inspire de
l’évolutionnisme ou du diffusionisme, selon qu’elle s’articule autour de
l’espèce, ou bien, de l’espace. La carte est une représentation iconique
approximative de la situation géographique réelle, mais les représentations de

l’espace contribuent à orienter son organisation. 2 La répartition de travaux

empiriques est fonction de la façon dont l’homme possède l’espace. La
corporalité de l’homme dans l’espace engendre du discontinu spécifique. Les
limites des langues, des dialectes et des parlers ne sont pas arbitraires, mais
vécues. La discontinuité est une manifestation d’une corporalité sédentaire,
d’un esprit de clocher, pour utiliser le terme saussurien, modifiable selon
l’échelle qui va d’un village jusqu’à l’Etat. En revanche, la représentation
topographique de la diffusion des phénomènes linguistiques à partir d’une aire
d’attraction est une traduction du mouvement continu de la corporalité de
l’homme à travers l’espace. L’espace organisateur des identités devient un
espace de contacts sous l’influence de la force d’intercourse, pour reprendre un
terme saussurien. Poussé par cette force, l’homme engendre à son passage des
discontinuités nuancées sans pour autant laisser un espace fini à l’égard de tous
les phénomènes linguistiques.
A la suite du linguiste allemand August Leskien (1840-1916), Antoine
Meillet (1866-1936) tente une synthèse du modèle évolutionniste et du modèle
diffusionniste en intégrant dans son tableau de l’histoire des langues indo-
européennes un dialogue positif entre l’espace et l’espèce. Les cartes
d’isoglosses des langues indo-européennes ne font pas référence à la diffusion
de traits à partir d’un point d’origine, mais sont iconiques avec la situation où
l’indo-européen est éclaté en dialectes. La disposition géographique de ces
dialectes les uns par rapport aux autres est non seulement conservée, mais aussi
renforcée si l’on en croit le témoignage des langues attestées. Les solidarités
systémiques ne sont pas vraies seulement à l’intérieur d’un système, mais aussi
entre les systèmes en contact. Dans de nombreux cas, les langues originaires de
la même souche représentent des évolutions de traits particuliers qui
convergent en raison de contiguïtés géographiques sans pour autant toucher à
l’écart qui maintient ces langues à part les unes des autres.
Se divisant en parlers répartis chacun sur un espace géographique bien
déterminé, l’indo-européen ne représentait qu’une unité relative, avant même
sa dispersion sur une étendue plus vaste, estime Meillet. L’indo-européen
constituait une langue commune, une langue dominante, résultant de
vicissitudes historiques extralinguistiques, mais cette langue a bien pu n’être

Selon A.-L. Terracher (1914:I), le terme de géographie linguistique désigne une telle
diversité de travaux “d’ordre et d’esprit” qu’il serait plutôt approprié de systématiser sous des
expressions nouvelles les études effectuées dans le cadre de la dialectologie gallo-romane. Il
propose de rediviser les travaux existants comme suit: 1) la cartographie linguistique qui
dresse des cartes, d’une part, des limites de dialectes (la méthode représentée par Ascoli), et
d’autre part, des caractères linguistiques (la méthode de Meyer); 2) la géologie linguistique qui
établit des cartes de mots en vue d’en examiner la stratigraphie historique (la méthode de J.
Gilliéron); 3) la géographie linguistique qui constitue une “explication de la répartition
topographique des faits du langage” au moyen des facteurs ethnographiques, géographiques et
historiques (la méthode de Terracher même) (ibid. I-XIV).

qu’un représentant conservé parmi de nombreuses langues communes dont

l’histoire a effacé les traces. Tant que cette langue commune n’aura pas
rencontré une langue commune rivale, la transition d’un dialecte ou d’un parler
à l’autre se fera d’une manière insensible. Il en est de même de nos langues
Aussi longtemps qu’il n’intervient pas d’extension d’une langue dominante, les dialectes
n’ont pas de limites définies, puisque chacune de leurs particularités a son aire propre; on
ne saurait où commencent les dialectes français du Nord et où finissent les dialectes
méridionaux […] mais il y a des zones intermédiaires. Seuls des accidents historiques
déterminent la création de frontières nettes: le langage de Paris tend à se répandre sur
toute la France; il va ainsi à la rencontre de la forme du toscan sur laquelle repose
l’italien littéraire et qui tend à se répandre sur toute l’Italie: il y a dès lors contact de deux
dialectes d’autrefois, et la limite peut être tracée avec précision, tandis que, entre le
parisien et le toscan, les parlers locaux présentent des transitions et se suivent les uns les
autres avec des dégradations insensibles. (Meillet 1964[1903]:53)

Dans le vaste espace qu’est celui des langues indo-européennes, les langues
ont entre elles un vague air de famille en raison d’une communauté d’origine.
Cet air de famille, non perçu par les locuteurs de ces langues, se rencontre sur
une échelle plus rétrécie, à savoir celle des parlers. Les différents parlers à
l’intérieur d’une certaine circonscription géographique dégagent un air de
ressemblance évident pour les locuteurs en raison de la contiguïté physique qui
pousse ces derniers à l’interaction.
La conception de la langue impliquée par le modèle arborescent est sous-
jacente au diffusionnisme. Au lieu de se restreindre aux traits dialectaux,
Meillet fait référence à la nomenclature établie par l’arbre généalogique de la
grammaire comparée, qui fait valoir les bifurcations au gré desquelles les
langues se distinguent les unes des autres chronologiquement. L’existence de
cette nomenclature sur le plan cartographique prouve que Meillet considère
que la grammaire comparée, dont les rapports entre les entités se modélisent
par l’intermédiaire de l’arbre, est un présupposé tacite du diffusionnisme dans
la mesure où, pour étudier deux traits dialectaux, le linguiste doit savoir que
ces traits dérivent de catégories différentes. Les fameuses isoglosses satem et
centum creusent un fossé entre deux dialectes naturels, qui à leur tour sont
divisés en parlers locaux dont la contiguïté est mise en évidence par le support
cartographique (Figure 1). A l’inverse des isoglosses qui “limitent un groupe
de localités pour un fait déterminé”, le dialecte naturel constitue un faisceau
d’isoglosses assemblés, à savoir une espèce géographique (Meillet 1908:4).

Figure 1. Les isoglosses.

Le modèle de la vague met en valeur les unifications, alors que celui de

l’arbre accentue les diversifications. L’union de ces deux modèles représente la
dynamique de la langue, d’où il s’ensuit que la conception d’une homogénéité
parfaite des catégories n’est que relative et idéale. Par sa notion de dialecte
naturel, Meillet insiste toutefois sur le fait que les traits dialectaux et les
dialectes ne relèvent pas de théories opposées, mais que les dialectes sont issus
de la coïncidence de plusieurs traits dialectaux qui se renforcent avec le temps
pour devenir des langues dans un espace déterminé. L’union de l’espèce
linguistique avec l’espace linguistique fait valoir que “les discontinuités ne
seraient pas uniquement des attributs internes des choses, ni de pures données
de la conscience humaine. Elles s’inscriraient dans les rapports qui unissent
l’homme au monde, à l’espace géographique” (Di Méo et Veyret 2002:15).

5. En guise de conclusion
Le recours à la métaphore sollicite une recherche de correspondances entre
la théorie et la réalité, mais plutôt qu’une transposition fidèle de la réalité, la
métaphore constitue “une représentation simplifiée de la réalité élaborée en
vue d’une démonstration ou d’une description” (Baudelle 2000). De ce fait, les
métaphores et les modèles sont plutôt un moyen constructiviste pour
conceptualiser l’essentiel d’une réalité à laquelle nous n’avons pas accès dans
tous ses détails. Ils permettent de donner une expression symbolique aux traits
de la réalité dont l’existence vaut d’être mise en avant. Henri Poincaré (1854-
1912) a exprimé au début du 20e siècle d’une manière éloquente que la science
n’est ni complètement naturelle, ni complètement factice:
[…] quelques personnes ont été frappées par ce caractère de libre convention qu’on
reconnaît dans certains principes fondamentaux des sciences. Elles ont voulu généraliser
outre mesure et en même temps elles ont oublié que la liberté n’est pas l’arbitraire. Elles
ont abouti ainsi à ce qu’on appelle nominalisme et elles se sont demandé si le savant

n’est pas dupe de ses définitions et si le monde qu’il croit découvrir n’est pas tout
simplement créé par son caprice. Dans ces conditions la science serait certaine, mais
dépourvue de portée. (Poincaré 1906:3)

[…] ce cadre où nous voulons tout faire rentrer, c’est donc nous qui l’avons fait; mais
nous ne l’avons pas fait au hasard, nous l’avons fait pour ainsi dire sur mesure et c’est
pour cela que nous pouvons y faire rentrer les faits sans dénaturer ce qu’ils ont
d’essentiel. (Poincaré 1906:5)

En sciences naturelles, les différents modèles s’effacent les uns les autres. En
linguistique, il n’en est rien. Les différentes formulations des affinités
linguistiques empruntées aux disciplines avoisinantes ne rendent pas caduques
les formulations alternatives, mais constituent plutôt un affinement de l’image
que nous faisons du lien entre les langues. La mise en parallèle de discours
espacés d’une centaine d’années – le débat moderne autour des fondements de
la linguistique finno-ougrienne et son antécédent dans la réflexion linguistique
de la fin du 19e siècle – montre que les représentations temporelle et spatiale
sont toutes aussi valables aujourd’hui qu’au début de la grammaire comparée.
Les références extralinguistiques, les échanges interdisciplinaires ne sont plus
toujours les mêmes, les contextes ne sont pas les mêmes, ni même les familles
linguistiques dont il est question. Cependant, les analyses qui expliquent le
mieux la réalité résistent au temps. C’est l’idéal de la véracité qui inspire les
protagonistes de ces débats.


Adam, Lucien. 1882. Les classifications, l’objet, la méthode, les conclusions

de la linguistique. Paris: Maisonneuve et Compagnie.
Anttila, Raimo. 1989[1972]. Historical and Comparative Linguistics. 2nd rev.
ed. Amsterdam & Philadelphia: John Benjamins.
Ascoli, Graziadio Isaia. 1876.“P. Meyer e il franco-provenzale”. Archivio
glottologico italiano 2.385-395.
Ascoli, Graziadio Isaia. 1878. “Schizzi franco-provenzali”. Archivio
glottologico italiano 3.61-120.
Auroux, Sylvain. 1998. La raison, le langage et les normes. Paris: Presses
Universitaires de France.
Baudelle, Guy. 2000. Géographie du peuplement. Paris: Armand Colin.
Berdoulay, Vincent. 1995. La formation de l’école française de géographie.
Paris: CTHS.
Blanckaert, Claude. 1995. “Fondements disciplinaires de l’anthropologie
française au XIXe siècle. Perspectives historiographiques”. Politix 29.31-
Brunet, Roger. 1997. “La discontinuité en géographie: origines et problèmes de
recherche”. L’espace géographique 4.297-308.

Di Méo, Guy & Yvette Veyret. 2002. “Problématiques, enjeux théoriques, et

épistémologiques pour la géographie”. Limites et discontinuités en
géographie ed. par L. Carroué et al., 5-22. Paris: Sedes.
Gallois, Lucien. 1908. Régions naturelles et noms de pays. Etude sur la région
parisienne. Paris: A. Colin.
Gay, Jean-Christophe. 1995. Les discontinuités spatiales. Paris: Economica.
Gould, Stephen Jay. 1997[1996]. L’éventail du vivant. Paris: Seuil.
Guillo, Dominique. 2000. “La sociologie d’inspiration biologique au XIXe siècle:
une science de l’ ‘organisation’ sociale”. Revue française de sociologie
Häkkinen, Kaisa. 1983. Suomen kielen vanhimmasta sanastosta ja sen
tutkimisesta. Turku: Turun yliopiston suomalaisen ja yleisen kielitieteen
laitoksen julkaisuja 17.
Häkkinen, Kaisa. 1984. “Wäre es schon an der Zeit den Stammbaum zu fällen”.
Ural-Altaische Jahrbücher Neue Folge 4.1-24.
Itkonen, Erkki. 1966. Kieli ja sen tutkimus. Helsinki: Werner Söderström.
Itkonen, Esa. 1999a. “There is Nothing Wrong with the Comparative Method”.
Kielitieteen kääntöpuoli. Kirjoituksia vuosilta 1963-1999 ed. par E.
Itkonen, 128-136. Turku: Publications in General Linguistics.
Itkonen, Esa. 1999b. “Sukupuu ja kontakti”. Kielitieteen kääntöpuoli.
Kirjoituksia vuosilta 1963-1999 ed. par E. Itkonen, 137-147. Turku:
Publications in General Linguistics.
Künnap, Ago. 1997. “Uralilaisten kielten jäljistä indoeurooppalaisissa
kielissä”. Virittäjä 1.79-83.
Künnap, Ago. 1998a. Breakthrough in Present-Day Uralistics. Tartu:
University of Tartu.
Künnap, Ago. 1998b. “Olemme väsyneet puihin”. Virittäjä 4.607-609.
Laakso, Johanna. 1998. “Tyvestä puuhun noustaan”. Virittäjä 4.610-612.
Laakso, Johanna. 1999. “Voiko väittää mitä vain?”. Virittäjä 4.637-639.
Lamarck, Jean-Baptiste. 1802. Recherches sur l’organisation des corps
vivants. Paris: Fayard.
Lamarck, Jean-Baptiste. 1809. Philosophie zoologique I-II. Paris: Dentu et
Lass, Roger. 1997. Historical Linguistics and Language Change. Cambridge:
Cambridge University Press.
Meillet, Antoine. 1964[1903]. Introduction à l’étude comparative des langues
indo-européennes. Alabama: University of Alabama Press.
Meillet, Antoine. 1908. Les dialectes indo-européens. Paris: Champion.
Meyer, Paul. 1875. “Compte-rendu de Archivio glottologico italiano”.
Romania IV.293-296.
Meyer, Paul. 1876. “Compte-rendu de Archivio glottologico italiano”.
Romania V.504-505.

Mikone, Eve. 1996. “Suomalais-ugrilaisia kieliä puhuvan väestön alkuperä”.

Virittäjä 3.401-419.
Mikone, Eve. 1997. “Tutkijan taustasta riippumatonta kritiikkiä”. Virittäjä
Östman, Jan-Ola et Raukko, Jarno. 1995. “The ‘Pragmareal’ Challenge to
Genetic Language Tree Models”. The Fenno-Baltic Cultural Area ed. par
S. Suhonen, 31-69. Helsinki: Castreianumin toimitteita.
Poincaré, Henri. 1906. La valeur de la science. Paris: Ernest Flammarion.
Terracher, Adolphe-Louis. 1914. Les aires morphologiques dans les parlers
populaires du Nord-Ouest de l’Angoumois (1800-1900). Paris: Honoré
Wiik, Kalevi. 1995. Itämerensuomalaisten kansojen ja kielten
syntykysymyksiä. Uusia kontaktiteoriaan perustuvia ratkaisuja. Non-
published study.
Wiik, Kalevi. 1996. “Alkuperäiset ajatukseni”. Virittäjä 4.590-598.
Wiik, Kalevi. 1997. Suomen väestön juuret: miksi lainasanatutkimus ei riitä?
Non-published study.
Wiik, Kalevi. 2002. Eurooppalaisten juuret. Jyväskylä: Atena.


Centre for the Historiography of Linguistics
KU Leuven (Belgium)


The participle is one of the eight parts of speech in ancient Greek

grammaticography. In spite of its constant inclusion in grammar manuals of
the ancient Greek tradition, its status is problematic given that the participle is
defined not in terms of proper and specific characteristics, but precisely in
terms of ‘participating’ in the nature of two other parts of speech. The
historical reality was that the status of the participle constituted a topic of
discussion from the philosophical-classificatory point of view (going back to
the logical and grammatical doctrines of the Stoics) and that it also formed the
object of grammatical debates (concerning the parts of speech of which a
number of combined characteristics are displayed by the participle). This
contribution offers a source-based study of the philosophically inspired and
grammatically oriented discussions concerning the status, the definition and
the classification of the participle in Greek antiquity.

1. Introduction
The topic of the present paper lies at the juncture of three long-term
research lines we have been pursuing over the past years:

(a) the investigation of the development of grammatical thought (and

grammatical praxis) in ancient Greece and Rome, and specifically in its
relationship with philosophical and rhetorical theories (cf. our publications
on “relational nouns” [Swiggers & Wouters 1995c; 1999] and on
Dionysius of Halicarnassus [1995b]);
(b) the study of the origins and subsequent evolution of grammatical
categorization, especially in the field of morphosyntax, with special focus
on the definition, classification and subdivision of the parts of speech (cf.
our publications on the noun [Lenoble, Swiggers & Wouters 2001]; the
pronoun [Lenoble, Swiggers & Wouters 2003]; the adverb [Swiggers &
Wouters 2002] and the conjunction [Swiggers & Wouters 1998]);

(c) the study of the didactic systematization of grammar teaching, in the

form of grammatical manuals (tekhnai grammatikai, artes grammaticae)
and other types of school texts (cf. Wouters 1979, 1999; Swiggers &
Wouters 1995a, 2000; Wouters 2005).

These lines of research have been pursued, while taking into account not
only the Greek and Roman world, but also later insights, and always with an
eye at a more general linguistic perspective, thus integrating philological
spade-work (e.g., on papyrological texts) with the methodological horizon of
the historian of linguistics (cf. Swiggers 1997).
Our topic here lends itself to a multiperspectival approach: philological,
grammatical, philosophical, all of these perspectives being subsumed under a
historiographical stand. In what follows we will try to shed some new light on
the complex issue of the origin and the early stages in the recognition of the
participle as a part of speech. Our approach will serve as an illustration of how
in recent years the historiography of ancient grammar has been revised and
refined; we hope that it can serve as an example of the pluridisciplinary
approach that is indispensable for the study of ancient grammatical and
linguistic thought.
Among these advances, two major insights must be singled out. On the
methodological-epistemological level there is the newly developed view of the
macro-evolution of ancient grammar, more particularly of the history of
grammatical categorization. The traditional view (exemplified by Steinthal
[1890-91], but also by Robins [1957, 1986]) consisted in claiming a linear
development from early philosophical reflections on language to the creation
of grammatical technography by the Alexandrian grammarians, and leading
towards a later synthesis, enriched by reflections on syntax (as in Apollonius
Dyscolus’s treatise Peri; suntavxew" of the second cent. CE). This view has
now been replaced by the reconstruction of a dynamic development of
grammar in constant interaction with rhetorical views as well as with didactic
concerns, whose practical orientation led to readaptations and refinements of
pieces of grammatical doctrine. Within this view, the so-called “unoriginality”
and “derived nature” of Roman grammar with respect to Greek grammar has
been reconsidered and justly qualified (see Taylor 1986).
Another important insight, deriving from spadework on the scholia on
Greek authors (principally Homer) by Alexandrian philologoi from the third
century BCE on, as well as from the edition and study of the papyrological
documentation (including “technical” treatises and school exercises) has been
the recognition of the necessity to rewrite the traditional account of the history
of the parts of speech system, i.e. the historiographical description of the
terminology, the classification and the analysis of the various parts and
subparts of speech. This has resulted not only in a number of newly formulated

views on the origin and development of certain parts of speech, 1 but also in the
abandonment, or at least playing down (cf. Ax 1996:117-118), of the
traditionally upheld controversy between analogists and anomalists about the
question to what degree the Greek language showed proportional regularity or
analogy, and to what extent it was characterised by irregularities, or anomalies.
The present case-study is the first attempt at proposing an historiographical
account of the early history of the participle. As noted by Viljamaa (1998:265),
in a study on the syntactic role of the participle as discussed by Apollonius
Dyscolus (fl. 2nd cent. CE) and Priscian (fl. c.500-530), much remains to be
investigated concerning the origins and development of this problematic part
of speech: “The history of the participle in the ars grammatica is obscure”. A
major point of interest in our study will be the fact that the autonomous status
accorded to this part of speech, which only exists by virtue of being a
combination of two other parts of speech, was the object of thorough, technical
discussion from the second century BCE on.

2. An historiographical crux: the status of the participle

At first sight, the history of the participle as a part of speech fits
unproblematically within the larger history of the parts of speech system,
which seems to be a natural process of ramification. This is the impression one
gets when taking a snapshot view at the history of ancient grammar and
philosophy of language, starting with Plato’s bipartition (noun/verb),
Aristotle’s tripartition (noun/verb/‘joining element’), the Stoic addition of the
‘article’ and the distinction they introduced between proper name and common
noun, and the subsequent establishment of the eight parts of speech by the
Alexandrian grammarians.
Philosophical reflection and grammatical praxis interacted from the third
century BCE on, when in the philological school of Alexandria the study of
literary texts (principally those of Homer) urged the creation of a
terminological apparatus and a classificatory scheme for the analysis of
(linguistic) phenomena dealt with in the edition and commentary of literary
works. As to the precise role and contribution of the Alexandrian philologists,
we are faced, when dealing with the history of the participle, with an
historiographical crux.
As a matter of fact, two conflicting retrospective accounts are presented in
a few (complete or partial) short stories of the parts of speech system
transmitted by Greek and Latin sources. 2 On the one hand, we have the

See, e.g., the recent studies on the o[noma kuvrion (Matthaios 1996), on the pronoun (Lallot
1995), on the adverb (Matthaios 2003) and on the conjunction (Swiggers & Wouters 1998).
In roughly chronological order: Dionysius of Halicarnassus (first cent. BCE), De
compositione verborum, 2.6 and De Demosthene 48; Quintilian (first cent. CE), Inst. Orat. I. 4.
17-21; Plutarch (first cent. CE), Plat. Quaest. X (cf. Wouters 1996); Schol. Dion. Thrac.,

summary by Quintilian, Inst. Orat. I. 4. 17-21 (first cent. CE), which attributes
to the Alexandrian grammarians the introduction of a system of eight parts of
Tum uidebit, ad quem hoc pertinet, quot et quae partes orationis, quamquam de
numero parum conuenit. Veteres enim, quorum fuerunt Aristoteles quoque atque
Theodectes, uerba modo et nomina et conuinctiones tradiderunt [...].
Paulatim a philosophis ac maxime Stoicis auctus est numerus, ac primum
conuinctionibus articuli adiecti, post praepositiones: nominibus appellatio, deinde
pronomen, deinde mixtum uerbo participium, ipsis uerbis aduerbia. Noster sermo
articulos non desiderat ideoque in alias partes orationis sparguntur, sed accedit
superioribus interiectio.
Alii tamen ex idoneis dumtaxat auctoribus octo partes secuti sunt, ut Aristarchus et
aetate nostra Palaemon, qui uocabulum siue appellationem nomini subiecerunt
tamquam speciem eius, at ii qui aliud nomen, aliud uocabulum faciunt, nouem [...].

“The teacher responsible will then need to consider how many parts of speech there
are, and what they are, although there is little agreement about the number. Earlier
writers, including also Aristotle and Theodectes, listed only verbs (verba), nouns
(nomina), and ‘convinctions’ (convinctiones) (= conjunctions) [...].
The philosophers, particularly the Stoics, gradually increased the number: to
‘convinctions’ were first added ‘articles’ (articuli), and then ‘prepositions’
(praepositiones); to ‘nouns’ was added the ‘appellation’ (appellatio), next the
‘pronoun’ (pronomen), and then the quasi-verbal ‘participle’ (participium); to ‘verbs’
were added ‘adverbs’ (adverbia). Our language does not need articles (articuli), and
these are therefore distributed among other parts of speech. In addition, however,
there is the ‘interjection’ (interiectio).
Some, belonging to the competent authorities, have gone as far as eight parts of
speech: so Aristarchus and, in our own day, Palaemon, who both put ‘vocable’ or
‘appellative’ under ‘noun’, as species of that genus. Those who distinguished
‘vocable’ from ‘noun’, make the total nine”. (translation of De Jonge 2005:13)

This corresponds with the 8 mevrh tou' lovgou which we find in the Tekhnê
grammatikê 3 ascribed to Dionysius Thrax (second – first cent. BCE.), viz. (a)

356.2-357.26 and 515.19–521.37 (Byzantine period?); Priscian (sixth cent. CE), G.L. II 54.5-
55.3 and 548.1-549.6, and the Ars anonyma Bernensis (cf. FDS no. 549).
We would like to point out that this specific sub-genre of grammatical texts is in urgent
need of thorough historiographical and linguistic-philological investigation. The texts of
Dionysius of Halicarnassus and Quintilian have recently been compared with each other by De
Jonge (2005) who considers the first text as “the prototype of the traditional western approach
to the history of linguistics” (2005:5). Some of these texts have already been exploited to some
extent by Hagius (1979:130-143), by Fehling (1958:53-54, in the discussion of Varro’s
idiosyncratic system of word classes), and by Matthaios (1999; 2002).
Tou' de; lovgou mevrh ejsti;n ojktwv: o[noma, rJh'ma, metochv, a[rqron, ajntwnumiva,
provqesi", ejpivrrhma, suvndesmo". hJ ga;r proshgoriva wJ" ei\do" tw'/ ojnovmati
uJpobevblhtai (G.G. I 1, 23. 1-3). “The parts of speech are eight [in number]: noun, verb,
participle, article, pronoun, preposition, adverb, conjunction. For the appellative is subsumed
under the noun, as [one of its] species”.

the noun, (b) the verb, (c) the participle, (d) the article, (e) the pronoun, (f) the
preposition, (g) the adverb, (h) the conjunction.
On the other hand, the sixth-century Latin grammarian Priscian offers a
somewhat different view. His text seems to suggest that it was Tryphon, a first-
century BCE Greek grammarian, who “set off” the participle as a part of
speech on its own.
(De participio). Quaesitum est tamen, an bene separaverint id ab aliis partibus
grammatici et primus Trypho, quem Apollonius quoque sequitur, maximus auctor
artis grammaticae. (G.L. II, 548.4-7)

“(On the participle). The question has been raised if the grammarians were correct in
separating it from the other parts (of speech) and the first (to treat the issue) was
Tryphon, who was followed also by Apollonius, the most competent author in
grammatical science”.

Among present-day historians of linguistics Pinborg (1975:116) and

Viljamaa (1998:266) have attributed to Priscian the statement that Tryphon
was the first to establish the participle as a separate part of speech.
It is here that recent progress in the study of ancient grammatical and
philological texts can offer a crucial contribution to historians of ancient
linguistics. Matthaios’s detailed study (1999) of Aristarch’s scholia on Homer
(and other Greek authors) has in fact shown, very convincingly, that

(a) the term metochv was used already by Aristarch in the third century
BCE, as a designation for the participle;
(b) Aristarch recognized the mixed nature of the participle as being a
verbo-nominal class;
(c) Aristarch took this very characteristic as the motivation for assigning a
separate place to the participle (cf. Matthaios 1999:420-430).

It is in the light of these findings that we should look again at the Tekhnê
grammatikê attributed to Aristarch’s disciple, Dionysius Thrax. Although the
authenticity problem 4 regarding this text is not at stake here, we want to point

(a) that the Tekhnê indeed has the term metochv as a designation for the
participle, which
(b) is considered to be a separate part of speech,
(c) the nature of which consists in being a combination of the properties of
two other parts of speech.

The short paragraph of the Tekhnê on the participle is thus fully in

For a status quaestionis, see Law & Sluiter (eds. 1995).

conformity with grammatical doctrine as elaborated in the third/second century

Metochv ejsti levxi" metevcousa th'" tw'n rJhmavtwn kai; th'" tw'n ojnomavtwn
ijdiovthto". Parevpetai de; aujth'/ taujta; a} kai; tw'/ ojnovmati kai; tw'/ rJhvmati
divca proswvpwn te kai; ejgklivsewn (G.G. I 1, 60.1-3).

“A participle is a word which participates in the proper nature of the verbs and that of
the nouns. It has the same accidents as the noun and the verb, except for persons and

The participle is defined here as a “word” (better: “word-type”), i.e. a levxi"

(this hyperonym is also used in the definition of the verb, of the pronoun, of
the preposition and of the conjunction, whereas in the definitions of the noun,
of the article and of the adverb, we find the syntagm mevro" lovgou); contrary to
all other word-types or word-classes, the participle has no proper definitory
features; other word-classes are defined in terms of

(a) their semantic-referential content; 5

(b) the content they express and specific formal characteristics; 6
(c) their place
– with respect to another word-class; 7
– with respect to other word-classes; 8
(d) their function
– with respect to another word-class; 9
Cf. the noun: [Onomav ejsti mevro" lovgou ptwtikovn, sw'ma h] pra'gma shmai'non, sw'ma
me;n oi|on livqo", pra'gma de; oi|on paideiva, koinw'" te kai; ijdivw" legovmenon, koinw'"
me;n oi|on a[nqrwpo" i{ppo", ijdivw" de; oi|on Swkravth" (§ 12).
“A noun is a part of speech, with case-inflection, signifying a (concrete) substance or a(n
abstract) thing — a (concrete) substance like ‘stone’, a(n abstract) thing like ‘education’ —,
taken in a common or particular sense — in a common sense, e.g., ‘man’, ‘horse’, in a
particular sense, e.g., ‘Socrates’”.
Cf. the verb: ÔRh'mav ejsti levxi" a[ptwto", ejpidektikh; crovnwn te kai; proswvpwn kai;
ajriqmw'n, ejnevrgeian h] pavqo" parista'sa (§ 13).
“A verb is a word without cases, accepting tenses, persons, and numbers, and signifying an
activity or an undergoing”.
Cf. the adverb: ?Epivrrhmav ejsti mevro" lovgou a[kliton, kata; rJhvmato" legovmenon h]
ejpilegovmenon rJhvmati. Tw'n de; ejpirrhmavtwn ta; mevn ejstin aJpla', ta; de; suvnqeta:
aJpla' me;n wJ" pavlai, suvnqeta de; wJ" provpalai (§ 19).
“An adverb is an uninflected part of speech, used with respect to a verb or added to a verb. Of
the adverbs some are simple, others are compound; simple, like pálai [‘long ago’], compound,
like própalai [‘very long ago’]”.
Cf. the preposition: Provqesiv" ejsti levxi" protiqemevnh pavntwn tw'n tou' lovgou merw'n
e[n te sunqevsei kai; suntavxei (§ 18).
“A preposition is a word which is placed before all the parts of speech, in compounding and in
Cf. the pronoun: ’Antwnumiva ejsti; levxi" ajnti; ojnovmato" paralambanomevnh, proswvpwn
wJrismevnwn dhlwtikhv (§ 17).

– with respect to larger semantic(-syntactic) sequences. 10

The participle has no specific (semantic or formal) characteristics: about its

semantics nothing is said, and as to its formal characteristics, the definition
states that it has no proper idiotês, but shares (most of) the characteristics of
the noun and the verb.
Dionysius’s treatment of the participle consists in stating its position as one
of the parts of speech, but at the same time it raises its non-specific
characteristics, and it leaves unmentioned a number of problems: (a) its
derived status (the fact that it is derived from a verbal lexeme); (b) its
grammatical functioning; (c) its complex semantics (involving, as a salient
fact, the logically converse relationship of active and passive).

3. Grammatical and philosophical views on the status of the participle

A closer examination of Priscian’s text is in order. Clearly, if one interprets
his words in the sense that he wants to attribute to Tryphon the establishing of
the participle as a separate part of speech, we must conclude that Priscian’s
statement is incorrect. The evidence from Aristarch’s scholia would easily
settle that question. But it may be that there is more at stake. As a matter of
fact, what did Priscian, who was rather well informed about the history of
Greek and Latin grammar, precisely mean by “an bene separaverint id ab aliis
partibus grammatici et primus Trypho”? Priscian speaks here of “bene
separare”, i.e. “to set off accurately”; this is not simply a matter of listing an
element in an inventory. Also, the very fact that he inserts an historiographical
account precisely at the point of dealing with the participle is an indication that
something more fundamental is at stake here. The “separation” to which
Priscian refers would then be, not the simple fact of recognizing the participle,
but of arguing for its separate status among the various word-classes.
If we adopt this line of interpretation, we have to reconsider the traditional
“linear development account” of ancient Greek grammaticography, a revision
which is also suggested by Matthaios’s work (1999, 2002). 11 Also, the
existence of various (traditions of) short retrospective stories of the parts of
speech system (cf. supra), both in Greek and Latin grammarians or rhetoricians
(such as Dionysius of Halicarnassus and Quintilian) points to a dynamic,
intertwined development of grammatical doctrines involving substantial
discussion about the system of parts of speech, with an interesting cross-
fertilization between grammatical and philosophical standpoints.

“A pronoun is a word used instead of a noun, and indicating definite persons”.

Cf. the conjunction: Suvndesmov" ejsti levxi" sundevousa diavnoian meta; tavxew" kai; to;
th'" eJrmhneiva" kechno;" dhlou'sa (§ 20).
“A conjunction is a word linking together the thought, with order, and showing the void of the
For the participle, see especially Matthaios (1999:425-431).

Let us take a closer look at these ‘doxographical’ accounts. One of the firm
statements we find in some of them is that the Stoics, while splitting up the
class of the “names” into the onomata (proper names) and prosegoriai
(common nouns), reunited under one heading: the verb and the participle. This
is what we read in Schol. Dion. Thrac. 356.7-16:
[...] OiJ ga;r S tw i k oi; filovsofoi lovgoi" tisi;n ejpereidovmenoi cwrivzousi tou'
ojnovmato" th;n proshgorivan, i[dion mevro" lovgou aujth;n levgonte" ei\nai: kai;
katalevgousin ou{tw ta; mevrh tou' lovgou: prw'ton o[noma, deuvteron
proshgoriva, trivton uJf’ e}n rJh'ma <kai;> metochv, to; me;n rJh'ma kathgovrhma
levgonte", th;n de; metoch;n e[gklima rJhvmato", o{ ejsti rJhvmato" paragwghv:
tevtarton uJf’ e}n a[rqron kai; ajntwnumiva, to; me;n favskonte" ajovriston a[rqron,
to; de; wJrismevnon a[rqron: kai; pevmpton uJf’ e}n provqesi" <kai;> suvndesmo",
th;n me;n proqetiko;n suvndesmon prosagoreuvonte", to;n de; uJpotaktiko;n
suvndesmon: ta; ga;r ejpirrhvmata ou[te lovgou ou[te ajriqmou' hjxivwsan,
parafuavdi kai; ejpifullivdi aujta; pareikavsante".

“[...] Relying upon certain arguments, the Stoic philosophers separate the appellative
from the noun, saying that it is a proper part of speech. And they recount the parts of
speech in this way: first there is the noun; second the appellative; third, under one
heading, the verb and the participle, calling the verb a categoreme [= predicate] and
the participle an inflected form of the verb, which is a derivation of the verb; fourth,
under one heading, the article and the pronoun, calling the first an indefinite article,
the second a definite article, and, fifth, under one heading, the preposition and the
conjunction, calling the first the prepositive conjunction, the second the postpositive
conjunction. But the adverbs they did not deem worthy to mention or to include in the
list, likening them to a side-growth or by-growth”.

and in Schol. Dion. Thrac. 518.17-22:

Tw'/ rJhvmati sunavptousi th;n metochvn, e[gklisin aujth;n rJhvmato" kalou'nte":
a[lloi de; meta; tw'n ojnomavtwn sunavptousi th;n metochvn: oujk ajxiou'si de; ijdiva/
paralambavnein th;n metochvn, levgonte" o{ti “pa'n mevro" lovgou qevlei e[cein
pavntw" prwtotuvpou" fwnav", hJ de; metoch; oujdevpote e[cei prwtovtupon
fwnhvn: ajpo; ga;r rJhvmato" paravgetai: oujkou'n ouj dunato;n ei\nai ijdiva/ mevro"

“They [= the Stoics] combine the participle with the verb, calling it an inflection of
the verb. But others combine the participle with the noun. They do not deem it worthy
to take the participle as a proper part of speech, saying that ‘every part of speech in all
ways ought to have prototype sounds, but that the participle never has a prototype
sound, for it is derived from a verb. Therefore it cannot be properly a part of speech’”.

According to these scholiasts’ account — note that Schol. Dion. Thrac.

356.7-16 describes the situation before Antipater’s introduction (2nd cent.
BCE) of the adverb into the (Stoic) list of parts of speech 12 — the Stoics split

Cf. Diogenes Laertius, VII 57: Tou' de; lovgou ejsti; mevrh lovgou, w{" fhsi Diogevnh" t’
ejn tw'/ Peri; fwnh'" kai; Cruvsippo", o[noma, proshgoriva, rJhm ' a, suvndesmo", a[rqron. JO
d’ jAntivpatro" kai; th;n mesovthta tivqhsin. “As stated by Diogenes in his treatise on

up the former onoma (name-noun) into onoma ([proper] name) and prosegoria
(noun), and they achieved a number of regroupings, as if they operated from an
extant larger list of parts of speech. Three regroupings are mentioned in the
first text: (1) the regrouping of the provqesi" (preposition) and suvndesmo"
(conjunction) under one heading syndesmos, with a subspecification into
proqetiko;" suvndesmo" (‘prepositive conjunction’), and uJpotaktiko;"
suvndesmo" (‘postpositive conjunction’); (2) the regrouping of the a[rqron
(article) and ajntwnumiva (pronoun) under one heading arthron, with a
subspecification into a[rqron ajovriston (‘indefinite article’) and a[rqron
wJrismevnon (‘definite article’); and, finally, (3) the regrouping, mentioned also
in the second text, of the rJh'ma (verb) and metochv (participle). In the latter case
there is not a common hyperonym, but a distinction is made between two
subclasses: the verb is called kathgovrhma (‘predicate’), the participle form is
called (in the first text) e[gklima rJhvmato" or (in the second text) e[gklisi"
rJhvmato" (an ‘inflected form’/‘inflection’ of the verb) in the two accounts, and
in one of them this is further specified as rJhvmato" paragwghv (‘derivation of
the verb’). The idea behind this seems to be that the participle, as a form
derived from the verbal lexeme, should not be given a separate status; the
formal relationship of derivation ranks higher than the semantic-syntactic
distinction between (full) predicative status (of the conjugated verb) and non(-
full)-predicative status of the participle.

4. “Participating in what?”: on the “proper” place of the participle

These are but two short retrospective accounts, which in summary seem to
imply that “the Stoic philosophers” reduced, on a number of semantico-
syntactic grounds or using morphological criteria, a previously existing longer
list (established by the Alexandrian philologists) of parts of speech, and that in
the case of the participle, they (re-)united the participle with the “verb”, of
which it is a derived form.
However, when taking into account the side-remarks in both accounts and
especially when looking at a larger set of doxographical accounts (see note 2),
a more complex, and an historiographically more interesting picture emerges.
We see that the participle was considered a candidate for being ranked under
three different (Stoic) parts of speech:

(I) the o[noma (i.e. the proper name in Stoic usage); this is what we learn from
a side-remark in Schol. Dion. Thrac. 518.18 (cf. supra: a[lloi de; meta; tw'n
ojnomavtwn sunavptousi th;n metochvn “But others combine the participle with
the noun”). It may also be that the first-century BCE grammarian Tyrannion

Language and by Chrysippus, there are five parts of speech: proper name, common noun, verb,
conjunction, article. To these Antipater adds another part: the ‘mean’ (= the adverb)”.

was using the term onoma in its meaning “proper name” when he labelled the
participles (onomata) metochika, but the fragmentary text could also be
interpreted as suggesting — in the Stoic tradition — a subspecification of the
“nominal” class into proper names (kuria onomata), common nouns (onomata
prosegorika) which allow thematic derivation (degrees of comparison for the
adjectives?), 13 and participial nouns (onomata metochika) which do not allow
such thematic derivation (cf. Wendel [1948:1817]).

Peri; tw'n merw'n tou' lovgou, ejn w|/ levgei (sc. oJ Turannivwn) a[toma me;n ei\nai
ta; kuvria ojnovmata, qematika; de; ta; proshgorikav, ajqevmata de; ta; metocikav
(Haas 1977, fr. 36).

“(A treatise) ‘On the parts of speech’, in which Tyrannion states that the proper
names are ‘individual’, 14 that the common nouns can be qematikav (‘primary
elements’) [i.e. they allow derivation], while the ‘participial’nouns (/participles)

(II) the proshgoriva (i.e. the common noun in Stoic usage); this is a kind of
regrouping that we find mentioned in Priscian, G.L. II, 548.7-17 (following the
passage quoted in § 2.):
S to ic i enim quomodo articulum et pronomen unam partem orationis accipiebant [...],
sic igitur [...] etiam participium aiebant appellationem esse reciprocam, id est
ajntanavklaston proshgorivan, hoc modo: ‘legens est lector’ et ‘lector legens’,
‘cursor est currens’ et ‘currens cursor’, ‘amator est amans’ et ‘amans amator’ [...].

“The Stoics, in the same way as they accepted the article and the pronoun to be one
part of speech [...], said that the participle is a reciprocal noun or an ajntanavklasto"
proshgoriva — in this way: ‘reading’ means the same as ‘reader’ and ‘reader’ the
same as ‘reading’; ‘runner’ means the same as ‘running’ and ‘running’ the same as
‘runner’, and ‘lover’ means the same as ‘loving’ and ‘loving’ the same as ‘lover’

and in Plutarch, Quaest. Plat. X 6:

JH de; kaloumevnh metochv, mi'gma rJhvmato" ou\sa kai; ojnovmato", kaq’ eJauth;n
me;n oujk e[stin, w{sper oujde; ta; koina; qhlukw'n kai; ajrrenikw'n ojnovmata,
suntavttetai d’ ejkeivnoi", ejfaptomevnh toi'" me;n crovnoi" tw'n rJhmavtwn tai'"
de; ptwvsesi tw'n ojnomavtwn. oiJ de; dialektikoi; ta; toiau'ta kalou'sin
ajntanaklavstou", oi|on oJ fronw'n ajnti; tou' fronivmou kai; oJ swfronw'n ajnti;
tou' swvfronov" ejstin, wJ" ojnomavtwn kai; proshgoriw'n duvnamin e[conta.

“And as for what is called the participle, since it is a mixture of the verb and the noun,
it does not exist of itself, as the nouns of common feminine and masculine gender do
not exist either; but it is grouped with those parts of speech, since through its tenses it

The adjective was not (yet) an autonomous part of speech, but a subclass of the noun (cf.
Dionysius Thrax, G.G. I 1, 34.3-35.2; and see Lallot 1992).
Cf. Matthaios (1996:72): “nur für Individuen, für nicht mehr teilbare Einheiten bestimmt”.

borders on the verbs and through its cases on the nouns. Terms of this kind, moreover,
are called reciprocals by the dialecticians on the ground that they have the force of
nouns, that is of appellatives, as for example ‘the reflecting’ instead of ‘the reflective’
and ‘the abstaining’ instead of ‘the abstinent’”.

In these two accounts the participles are reported to have been treated as a
specific subclass of the noun: the ajntanavklasto" proshgoriva or appellatio
reciproca. The relationship of semantic symmetry between participles and
common nouns is illustrated by Priscian with, a.o., the examples amator est
amans and amans est amator. Plutarch refers to the synonymy or
interchangeability between (present) participles and nouns (or adjectives) used
in a noun phrase with definite article: oJ fronw'n = oJ frovnimo" and oJ
swfronw'n = oJ swvfrwn.

(III) the rJh'ma (verb): this is the regrouping we found in the two scholiastic
texts quoted above (§ 3), viz. Schol. Dion. Thrac. 356.7-16 (esp. 10-13): uJf j
e}n rJh'ma <kai;> metochv, to; me;n rJh'ma kathgovrhma levgonte", th;n de;
metoch;n e[gklima rJhvmato", o{ ejsti rJhvmato" paragwghv (“under one
heading, the verb and the participle, calling the verb a categoreme [= predicate]
and the participle an inflected form of the verb, which is a derivation of the
verb”), and Schol. Dion. Thrac. 518.17-22 (esp. 17-18): tw'/ rJhvmati
sunavptousi th;n metochvn, e[gklisin aujth;n rJhvmato" kalou'nte" (“they
combine the participle with the verb, calling it an inflection of the verb”).
We also find this regrouping mentioned in two passages of Priscian’s
Institutiones, where he speaks of the participle as verbum participiale/casuale
and modus verbi casualis, viz. in G.L. II, 54.8-10:
Secundum Stoicos vero quinque sunt eius partes: nomen, appellatio, verbum,
pronomen sive articulus, coniunctio. nam participium connumerantes verbis
participiale verbum vocabant vel casuale [...].

“But according to the Stoics there are five (parts of speech): proper name, common
name, verb, pronoun or article, and conjunction. For they ranged the participle with
the verbs, calling it ‘participial verb’ or ‘verb with case inflection’ [...]”.

and in G.L. II, 549.1-6 (following the passage quoted under (II) above):
(participium aiebant) [...] vel nomen verbale vel modum verbi casualem. unde
videntur nostri ascivisse inter verba gerundia vel participialia, cum videantur ea
diversos assumere casus. Ideo autem participium separatim non tradebant partem
orationis, quod nulla alia pars orationis semper in derivatione est nullam propriam
positionem habens, nisi participium; ceterae enim partes primo in positione inventae
sunt, ad quam etiam derivativa aptantur.

“[...] or (it is) a verbal noun, or a casual mode of the verb. Therefore our Stoics seem
to have grouped the gerunds or participial forms with the verbs, although they appear
to take different cases. And so they did not transmit the participle as an independent

part of speech, because no other part of speech, except the participle, always appears
as a derived form, whereas the ‘primitive’ form is not attested. The other parts have in
first instance been created as original words, to which also the derived forms

5. Concluding remarks
As is clear from this short survey, the participle was an issue of
grammatical and linguistic-philosophical debate in Greek antiquity (cf. already
Robins [1986:27]). In fact, the topic of the participle was a crucial one in the
complex development of grammatical doctrine, which was not a matter of
linear ramification, nor of (systematic) reduction. Rather, as we learn also from
the papyrological tradition, 15 the gradual autonomization of
grammaticography, and the constitution of its core part, the system of merê tou
logou, involved a process of bricolage, of splitting and merging, and all this
occurred in a didactic context still largely influenced by philosophical views.
The Alexandrian eight parts of speech system, as propounded by Aristarch and
his disciples, did not immediately become “the canonical system”. This is
precisely what we can see in the case of the participle, taking full profit from
the extant body of doxographical texts: these retrospective accounts point to
the existence of a “fraction” of grammatikoi influenced by Stoic doctrine
(which, one should not forget, witnessed an evolution from the third century
BCE to the first centuries CE). The efforts of these grammarians concentrated
on using elements of Stoic philosophy in order to show that the more reduced
system of parts of speech (in fact, the historically prior one) of the Stoics was a
more perfect rearrangement (or systematization) of a (previous) more extensive
list of word classes (as established by grammarians with lesser philosophical
insight). Hence the insistence on aspects of reducibility.
It is interesting to note that the grammatical discussion concerning the
status of the participle went on for centuries during Greek antiquity. Adopting
an overall perspective, one notices that there were recurrent questionings of its
autonomous status among Latin grammarians, among Byzantine grammarians,
among grammarians in the Renaissance (cf. Swiggers 1998, 1999) and in
modern times, and it was only during the first half of the 20th century that the
participle was dropped as a separate part of speech in most of the Western
national traditions of grammaticography, although in some conservative school
grammar traditions the participle managed to survive.
What is interesting in the ancient debate concerning the status of the
participle is that the rejection of its autonomous status, while making
(apparently) reference to the old Stoic concept of “first imposition” (participles
as derived forms do not have a prwtovtupo" fwnhv), came to be based on

See, e.g., the papyrus P.Amh. 2.21 (fourth cent. CE) (= Wouters 1979, no. 14, ll. 13-15)
where the grammarian mentions two systems of 9 and 8 parts, respectively. He ultimately
prefers the latter — in his opinion more correct — system.

morphological considerations (an issue in which the early Stoics were not
interested — this fact might also point to an interesting evolution within Stoic
dialectics and philosophy of language!). As is clear from the scholiastic text of
Schol. Dion. Thrac. 356.7 - 357.26, later generations of grammarians
influenced by Stoic doctrine debated with Alexandrian grammarians, and
attacked them on their own grounds, using arguments based on analogia and
on inflection. To defend, for example, the (re-)separation of the o[noma and the
proshgoriva, the Stoics pointed to differences in the inflectional pattern (the
proper name Pavri", Pavrido" versus the noun mavnti", mavntio" and
mavntew"). Tryphon for his part adduced counterexamples (the noun mh'ni",
mhvnido" has the same inflection as the proper name Pavri", Pavrido") and
argued that if inflection was indeed the criterion for the merismov" (‘partition in
parts of speech’), “(participle) forms such as levgwn, levgonto" and (the noun)
levwn, levonto" would represent one part of speech, which is not the case”
(356.33-35). The latter argument obviously was perhaps not the strongest
possible one against the Stoics, who in fact proposed to combine the participle
and the noun …!
We can schematize the evolution as follows:

a. third cent. BCE: the (early) Stoic system of 5 parts of speech (o[noma, proshgoriva,
rJh'ma, suvndesmo", a[rqron) [later addition: mesovth" (adverb)]
b. second cent. BCE: the Alexandrian system of 8 parts (o[noma, rJh'ma, metochv, a[rqron,
ajntwnumiva, provqesi", ejpivrrhma, suvndesmo")
c. first cent. BCE: the ‘Stoics’ argue for reducing the Alexandrian system to theirs,
through the recombination of metochv + o[noma or + rJh'ma, of the a[rqron + ajntwnumiva,
of the provqesi" + suvndesmo", and, on the other hand, the separation (again) of the
o[noma and the proshgoriva
d. reaction of Tryphon (and other Alexandrian grammarians?) in defence of the system of
8 parts of speech.

We can now, in conclusion, take another look at Priscian’s statement

concerning Tryphon. Priscian’s text makes clear his important role in the
history of the participle. The philological question of how to “read” Priscian’s
elliptical text (viz. should we understand “et primus [quaesivit]” or “et primus
[separavit]”) 16 should be answered in the light of what we know about the
Alexandrian philologists and their grammatical distinctions and terminology.
Given our present state of knowledge concerning Aristarch’s metalanguage
and his grammatical system, we have to understand Priscian’s text as “Tryphon
was the first to inquire thoroughly concerning the issue of the [autonomy of
the] participle”. But how do we have to interpret the deeper sense of this
statement? Certainly, Tryphon did not cast doubts about the Alexandrian
system of parts of speech, since in the extant sources (cf. Schol. Dion. Thrac.

Cf. Matthaios (1999:420-421) for a survey.

356.21-23) he is presented as the defender of the system of eight parts. What

Tryphon apparently inquired about was the argumentation offered by those
grammarians who, under Stoic influence, ranked the participle under another
part of speech, an argumentation which he criticized (on morphological
grounds); as a defender of the eight parts of speech system, he also criticized
— in an opposite sense — the arguments of those who, again under Stoic
influence, held the view that the onoma kurion and the onoma prosegorikon
should be treated as two different parts of speech.
Tryphon thus did not establish the participle, but he defended its
autonomous status against those who wanted to merge it with another part of
speech. He probably was the first to develop a full-fledged argumentation on
the question. 17 Writing after him, Apollonius Dyscolus, the most perspicacious
of ancient Greek grammarians, did not reconsider the issue: for him, the
participle clearly was a separate part of speech, of which he meticulously
studied the syntactic functioning. But that is a matter to be dealt with on
another occasion.


Ax, W. 1996. “Pragmatic Arguments in Morphology: Varro’s Defence of

Analogy in Book 9 of his De Lingua Latina”. Ancient Grammar: Content
and Context ed. by P. Swiggers and A. Wouters, 105-119. Leuven & Paris:
De Jonge, C. 2005. “Dionysius of Halicarnassus as a Historian of Linguistics:
The History of the Theory of the ‘Parts of Speech’ in De compositione
verborum 2”. The Henry Sweet Society Bulletin 44.5-18.
De Velsen, A. 1853. Tryphonis grammatici Alexandrini fragmenta. Berlin:
Diogenes Laertius. Lives and Opinions of Eminent Philosophers. Greek text
with English translation by R.D. Hicks. London: Heinemann; Cambridge,
Mass.: Harvard University Press. 1925. (Repr. 1965, 2 vols.)
FDS = Hülser, K. 1987-8. Die Fragmente zur Dialektik der Stoiker. Stuttgart:
Frommann & Holzboog.
Fehling, D. 1958. “Varro und die grammatische Lehre von der Analogie und
der Flexion (Schluss)”. Glotta 36.52-100.
G.G. I, 1 = G. Uhlig. 1883. Dionysii Thracis Ars Grammatica (Grammatici
Graeci I, 1). Lipsiae: Teubner.
G.L. = Hertz, Martin, ed. 1855-1859. Institutiones grammaticae. Vols. 2 & 3 of
Grammatici latini, gen. ed. Heinrich Keil. Leipzig: Teubner. (Repr. 1961,
Hildesheim: Olms, 2 vols.)

Whether this took place in a special study on the participle, as de Velsen (1853:33) claims,
remains uncertain.

Haas, W. 1977. Die Fragmente der Grammatiker Tyrannion und Diokles

(=SGLG, 3, 79-184). Berlin & New York: de Gruyter.
Hagius, H. 1979. The Stoic Theory of the Parts of Speech. [Ph.D., Columbia
Lallot, J. 1992. “L’adjectif dans la tradition grammaticale grecque”. Histoire
Épistémologie Langage 14:1.25-35.
Lallot, J. 1995. “Qu’est-ce qu’un pronom? La réponse du grammairien grec”.
Mémoires de la Société de Linguistique de Paris, n.s. 3.11-26.
Law, V. & I. Sluiter, eds. 1995. Dionysius Thrax and the Techne Grammatike.
Münster: Nodus.
Lenoble, M., P. Swiggers & A. Wouters. 2001. “Étude comparative des
dénominations de catégories grammaticales dans les textes artigraphiques
latins de l’Antiquité”. Métalangage et terminologie linguistique. Actes du
colloque international de Grenoble (Université Stendhal - Grenoble III,
14-16 mai 1998) ed. by B. Colombat and M. Savelli, 275-291. Leuven,
Paris & Sterling: Peeters.
Lenoble, M., P. Swiggers & A. Wouters. 2003. “La structure des Artes
grammaticae latinae: l’exemple du pronom”. History of Linguistics 1999.
Selected papers from the Eighth International Conference on the History of
the Language Sciences (14-19 September 1999, Fontenay, St Cloud) ed. by
S. Auroux, 1-18. Amsterdam: Benjamins.
Matthaios, S. 1996. “Kuvrion o[noma. Zur Geschichte eines grammatischen
Terminus”. Ancient Grammar: Content and Context ed. by P. Swiggers
and A. Wouters, 55-77. Leuven & Paris: Peeters.
Matthaios, S. 1999. Untersuchungen zur Grammatik Aristarchs: Texte und
Interpretation zur Wortartenlehre. Göttingen: Vandenhoeck & Ruprecht.
Matthaios, S. 2002. “Neue Perspektiven für die Historiographie der antiken
Grammatik: Das Wortartensystem der Alexandriner”. Grammatical Theory
and Philosophy of Language in Antiquity ed. by P. Swiggers and A.
Wouters, 161-220. Leuven, Paris & Sterling: Peeters.
Matthaios, S. 2003. “Aristarch, die tevcnai grammatikaiv und Apollonios
Dyskolos zum Adverb. Zur Geschichte einer Wortkategorie in der antiken
griechischen Grammatik”. Göttinger Beiträge zur Sprachwissenschaft
Pinborg, J. 1975. “Classical Antiquity: Greece”. Current Trends in Linguistics
13.1: Historiography of Linguistics ed. by T.A. Sebeok, 69-126. The
Hague: Mouton.
Plutarch. Quaestiones Platonicae. Platônika Zêtêmata. Platonic Quaestiones.
Greek text with English translation by Harold Cherniss. (= Plutarch’s
Moralia, vol. XIII/1). London: Heinemann; Cambridge, Mass.: Harvard
University Press. 1976.
Robins, R.H. 1957. “Dionysius Thrax and the Western Grammatical
Tradition”. Transactions of the Philological Society 1957.67-106.

Robins, R.H. 1986. “The Techne Grammatike of Dionysius Thrax in its

Historical Perspective: The Evolution of the Traditional European Word
Class System”. Mots et parties du discours. Word and Word Classes. Wort
und Wortarten ed. by P. Swiggers and W. Van Hoecke, 9-37. Leuven &
Paris: Peeters.
Schol. Dion. Thrac. = A. Hilgard. 1901. Scholia in Dionysii Thracis artem
grammaticam (Grammatici Graeci I, 1). Lipsiae: Teubner.
Steinthal, H. 1890-91. Geschichte der Sprachwissenschaft bei den Griechen
und Römern. (2 vol.) Berlin: Dümmler.
Swiggers, P. 1997. Histoire de la pensée linguistique. Paris: P.U.F.
Swiggers, P. 1998. “Le statut du participe dans la grammaire française du
seizième siècle”. Et multum et multa. Festschrift für Peter Wunderli zum
60. Geburtstag ed. by E. Werner, R. Liver, Y. Stork and M. Nicklaus, 181-
195. Tübingen: Narr.
Swiggers, P. 1999. “Systématique et terminologie de la description du participe
dans la grammaire française préclassique”. Le français préclassique 6.13-
Swiggers, P. & A. Wouters. 1995a. “Technê et Empeiria: La dynamique de la
grammaire grecque dans l’Antiquité à la lumière des papyrus
grammaticaux”. LALIES = Actes des sessions de linguistique et de
littérature 15.83-101.
Swiggers, P. & A. Wouters. 1995b. “Grammar and Poetics: From Technique to
Tevcnh”. Greek Literary Theory after Aristotle. A Collection of Papers in
Honour of D.M. Schenkeveld ed. by J.G.J. Abbenes, S.R. Slings and I.
Sluiter, 17-41. Amsterdam: VU University Press.
Swiggers, P. & A. Wouters. 1995c. “The Treatment of Relational Nouns in
Ancient Grammar”. Orbis 38.149-178.
Swiggers, P. & A. Wouters. 1998. “The syndesmos in Ancient Grammar:
Grammatical Status and Subtypes”. Actes du XVIe Congrès International
des Linguistes (Paris, 16-20 juillet 1997). Paris. 1998 [CD-ROM ed.; Paper
Swiggers, P. & A. Wouters. 1999. “Les noms ad aliquid et ad aliquid qualiter
chez les grammairiens latins”. Conceptions latines du sens et de la
signification ed. by M. Baratin and Cl. Moussy, 127-142. Paris: Presses de
l’Université Paris Sorbonne.
Swiggers, P. & A. Wouters. 2000. “Grammaires grecques (et latines) sur
papyrus”. Manuscripts and Tradition of Grammatical Texts from Antiquity
to the Renaissance ed. by M. De Nonno, P. De Paolis and L. Holtz, vol. I,
59-88. Cassino: Ed. dell’Università degli Studi di Cassino.
Swiggers, P. & A. Wouters. 2002. “De adverbio: Statut et signification de
l’adverbe chez les grammairiens latins”. Grammatical Theory and
Philosophy of Language in Antiquity ed. by P. Swiggers and A. Wouters,
287-323. Leuven, Paris & Sterling: Peeters.

Taylor, D.T. 1986. “Rethinking the History of Language Science in Classical

Antiquity”. The History of Linguistics in the Classical Period ed. by D.T.
Taylor, 1-16. Amsterdam & Philadelphia: Benjamins.
Viljamaa, T. 1998. “Participium coniunctum. Syntactic definitions of the
participle in ancient grammars”. Arctos 32.265-276.
Wendel, C. 1948. “Tyrannion”. Paulys Realencyclopädie der classischen
Altertumswissenschaft, II. Reihe, Bd. VII, coll. 1811-1819.
Wouters, A. 1979. The Grammatical Papyri from Graeco-Roman Egypt.
Contributions to the Study of the ‘Ars grammatica’ in Antiquity. Brussels:
Paleis der Academiën.
Wouters, A. 1996. “Plutarch’s Comments on Plato’s ‘Grammatical’ (?)
Theories: A Few Remarks on Quaestio Platonica X”. Plutarchea
Lovaniensia. A Miscellany of Essays on Plutarch ed. by L. Van der Stockt,
309-328. Leuven: Peeters.
Wouters, A. 1999. “La grammaire grecque dans l’école antique, d’après les
papyrus”. Actes du XXXIe Congrès International de l’A.P.L.A.E.S.
(Universités Lumière-Lyon 2 et Jean Moulin-Lyon 2, Lyon 5, 6 et 7 juin
1998), 51-68. Lyon: Presses de l’Université.
Wouters, A. 2005. “Quintilian and the Roman praeceptor sermonis”.
Humanitas grecka i rzymska ed. by R. Popowski, 143-168. Lublin:
Wydawnictwo K.U. Lublin.

The University of Helsinki


The earliest references to Latin grammar in the first century BCE

associated grammar with the study of virtue and the Liberal Arts. The view of
the Liberal Arts as a complete form of education was cherished in Antiquity by
Platonists in particular, and was characterized by unity of study as well as the
idea of ascent from a lower, earthly level to higher, divine realities. These are
also the characteristic features of the most complete Platonic theory of learning
preserved from Antiquity – the one found in Augustine’s De ordine. It has
been a matter of dispute to what extent this theory depends on Varro’s
Disciplinarum libri novem. The present study has shown that the educational
theory of Late Antiquity differed from that of the first century BCE at least in
one important respect: the role of dialectic is much more prominent in
Augustine’s De ordine than it is in the earlier works.

1. Introduction
The earliest descriptions of Latin grammar are preserved in the works of
Cicero (106-43) and Varro (116-27), the two leading intellectual figures of the
first century BCE It was during this period that grammar probably became an
independent discipline, given that the Techne attributed to Dionysius Thrax
(ca. 100 BCE) is today largely regarded as inauthentic. The question of when
grammar became an independent discipline is an important one for linguistic
historiography. But it is no less important to see that grammar continued to
hold close ties with the related disciplines from which it had emancipated itself
– philosophy, rhetoric and philology – throughout Antiquity. Together with
logic and rhetoric, grammar formed the linguistic part of an educational
scheme known as the Liberal Arts, which were viewed as an ideal form of a
complete, general education.
This educational scheme, which came to be of crucial importance for
medieval intellectual life for centuries, was transmitted to the Middle Ages
through the works of the Latin encyclopaedists, such as Augustine (354-430
CE), Martianus Capella (fl. ca. 400 CE), Cassiodorus (487-583 CE) and

Isidore of Seville (c. 602-636 CE). In Isidore’s popular encyclopaedia, the

Etymologiae, grammar was defined as the first of the seven Liberal Arts and
the foundation of them all. The idea that grammar is the foundation of all
subsequent study was present at the first mention of grammar as a separate
discipline in the first century BCE. Grammar was regarded as part of the
Liberal Arts in Cicero’s rhetorical works and in Varro’s encyclopaedic
Disciplinarum libri novem. It was called the foundation of rhetoric and all
higher study in Quintilian’s description of Roman education in the first century
CE. Both Cicero and Quintilian (35?-96/100 CE) maintained that the ultimate
end of all study is philosophy and the acquisition of virtue. Within this scheme,
the role of grammar is always preparatory rather than an end in itself.
In this paper I will analyse the role of grammar within the system of the
Liberal Arts from the first century BCE to the fourth and fifth centuries CE, by
which time the seven Liberal Arts had become a fixed system. Throughout this
period, it will be emphasised, this educational scheme was an ideal inspired by
Platonism rather than an actual practice found in ancient schools.

2. The Liberal Arts or Enkyklios Paideia

The Liberal Arts are best known in the canonical form they took in Late
Antiquity. At that time, they consisted of three linguistic and logical arts –
grammar, rhetoric and dialectic – and four mathematical disciplines –
geometry, arithmetic, astronomy and music theory. 1 However, it was only in
Late Antiquity that liberal education took a fixed form. Earlier, this term and
its Greek equivalent enkyklios paideia had a much more general meaning. It
referred to a wide general education which enabled a free-born citizen to
fulfill his political tasks in public life. The Sophists, professional teachers of
rhetoric and argumentation, were famous for their claim that they were able to
teach the mastery of the entire range of human knowledge necessary for
success in political life.
It is in one of Cicero’s rhetorical works, De partitione oratoria, that
grammar (studia litterarum) was first mentioned as pertaining to liberal
education (22,76-80). 2 According to Hadot, Cicero united the linguistic and

In the Middle Ages, the terms trivium and quadrivium were generally used to denote these
two groups of disciplines. Boethius (420-526) used the term quadrivium in his works (Stahl
1971:94), but the term trivium is first attested in the ninth century (see Rajna 1928, Wagner
1983 and Luhtala 1996:289).
Sunt autem alii quidam animi habitus ad virtutem quasi praeculti et praeparati rectis studiis et
artibus, ut in suis rebus studia litterarum, ut numerorum ac sonorum, ut mensurae, ut siderum,
ut equorum, ut venandi, ut armorum, […] (De part. or. 22,80). “But there are certain other
states of mind trained and prepared for virtue by proper studies and sciences, as for instance
among personal matters the study of literature, rhythms and music, mensuration and
astronomy, riding and hunting and fencing” (tr. by Rackham). It is noteworthy that Cicero
does not use the Greek terms for the disciplines; in De Finibus the term grammatica is used

mathematical disciplines in a way that is reminiscent of their later canonical

interpretation (1984:51). 3 The Liberal Arts, which here include grammar,
music, geometry, astronomy, equitation, hunting and heraldry, are mentioned
among things that are preparatory for the acquisition of virtue and the study of
philosophy. Dialectic and rhetoric, by contrast, count among virtues
themselves and form the logical part of philosophy. Thus, Cicero’s scheme
differs from the seven Liberal Arts of Late Antiquity in at least three ways.
Firstly, the number of subjects included was open. Secondly, grammar,
rhetoric and dialectic are not on the same level; grammar is propaedeutic for
virtue while dialectic and rhetoric are virtues in themselves. Thirdly, the role of
dialectic was not as prominent as in Late Antiquity, when it was regarded as a
tool relevant for all learning.
Cicero’s De partitione oratoria represented the rhetoric of the Middle
Academy, and it described a Platonist ideal of a rhetorician’s education rather
than the actual practice found in ancient schools. In accordance with the
Platonist ideal, Cicero maintained the unity of liberal education: “[…] there is
also the truth enunciated by Plato […] that the whole of the content of the
liberal and humane sciences is comprised within a single bond of union”. 4
Cicero had been influenced by Platonism while studying in Athens with
Antiochus of Ascalon (130/120-68 BCE). 5 Antiochus was a key figure in the
process of syncretism whereby Stoic and Platonic doctrines were gradually
amalgamated. In the first century BCE, the process of synthesis even involved
Pythagorean and Peripatetic elements. Varro, who attended Antiochus’ lectures
in Rome (Dillon 1977:62), wrote an encyclopedia entitled Disciplinarum libri
novem, which probably consisted of treatises on grammar, rhetoric, dialectic,
geometry, arithmetic, music, medicine and architecture.
The subjects pertaining to liberal education continued to vary from author
to author and even within the works of one author up until the times of
Augustine and Martianus Capella. In Vitruvius’ De architectura, enkyklios

instead of studia litterarum (3,2,5). ‘Ars numerorum’ could be interpreted as arithmetic (see n.
11). For Augustine’s terminology concerning the Liberal Arts, see Burton 2005.
The four mathematical disciplines formed a unity already in the fifth century BCE. They
were mentioned by Hippias (c. 485-415), one of the Sophists, who thought that a profound
knowledge of arithmetic, geometry, astronomy and theory of music were also indispensable for
the general education, but this view was not shared by many of his contemporaries. Plato
recommended these four arts to be studied on a more advanced level (Hadot 1984:11-15).
[...] est etiam illa Platonis vera […] omnem doctrinam harum ingenuarum et humanarum
artium uno quodam societatis vinculo contineri (De oratore 3,6, 21). “[...] there is also the
truth enunciated by Plato, [...] that the whole of the content of the liberal and humane sciences
is comprised within a single bond of union” (tr. by Sutton). In Plato’s view, these arts are
intimately associated with each other through a common element, number.
Antiochus was head of the New Academy in 79-78 BCE when Cicero visited Athens. In 88
BCE Antiochus accompanied Philon of Larissa (160/59-c.80 BCE) to Rome where Cicero
attended their lectures.

paideia comprised literature, drawing, geometry, history, ethics, physics,

music theory, medicine, law and astronomy. These arts, which are described as
steps leading to the ‘temple’ of architecture, possess a lasting value, according
to Vitruvius (1st cent. BCE or CE). They belong to the gifts of Fortune, which
abide secure to the end of life, 6 and are united to form one single body. 7
In Seneca’s (c. 1-65 CE) account, the Liberal Arts were subordinated to
philosophy and included grammar, geometry, music, counting and astronomy
(Ep. 88, 3-10). Galen (129-200?), too, kept philosophy distinct from the other
arts. He regarded philosophy as the highest of the divine gifts allotted to man,
thanks to his rationality, which raises him above the other animals. Galen drew
a distinction between the manual arts and the arts that are guided by reason; the
latter, which were invented by Hermes (or Mercurius), comprise medicine,
rhetoric, music, geometry, arithmetic, astronomy, grammar and law (Protr. 3,
p. 105, 15f.). Vitruvius, Seneca and Galen failed to include dialectic in the
Liberal Arts, but it appeared in Jerome’s (348-420 CE) list of reason-based
arts, which comprised grammar, rhetoric, philosophy, geometry, dialectic,
music, astrology and medicine (Ep. 53, 6). As for grammar, some authors fail
to mention it altogether and it does not yet form a unity with the other arts of
discourse. 8

Omnia enim munera fortunae cum dantur, ab ea faciliter adimuntur; disciplinae vero
coniunctae cum animis nullo tempore deficient, sed permanent stabiliter ad summum exitum
vitae (De architectura 6, preface 3). “For all gifts of Fortune, as they are bestowed by her, so
are they easily withdrawn; but when training is conjoined with mental power, it never fails, but
abides secure to the final issue of life” (tr. by Granger).
Cum ergo tanta haec disciplina sit, condecorata et abundans eruditionibus variis ac pluribus,
non puto posse iuste repente profiteri architectos, nisi qui ab aetate puerili his gradibus
disciplinarum scandendo scientia plerarumque litterarum et artium nutriti pervenerint ad
summum templum architecturae. […] cum autem animadverterint omnes disciplinas inter se
coniunctionem rerum et communicationem habere, fieri posse faciliter credent; encyclios enim
disciplina uti corpus unum ex his membris est composite (1, 1, 11-12). “Since, therefore, so
great a profession as this is adorned by, and abounds in, varied and numerous
accomplishments, I think that only these persons can forthwith justly claim to be architects
who from boyhood have mounted by the steps of these studies and, being trained generally in
the knowledge of arts and the sciences, have reached the temple of architecture at the top [...]
When, however, it is perceived that all studies are related to one another and have points of
contact, they will easily believe it can happen. For a general education is put together like one
body from its members” (tr. by Granger).
Such are Nicomachus of Gerasa (c. 50-150 CE) and Theon of Smyrna (fl. c.115-140 CE),
who deal with the mathematical disciplines as propaedeutic for the study of Platonic
philosophy. The only subject related to grammar treated by Alkinoos (Albinus, fl. 2nd cent CE)
is etymology. In the works of Plutarch (c. 46-120 CE), Apuleius (c. 125-161/180) and
Maximus of Tyre (c. 125-185 CE), the study of poetry or grammar is becoming propaedeutic
for the study of philosophy, but it does not yet form a unity with dialectic and rhetoric (Hadot

When encyclopaedic works gained popularity in Late Antiquity, the

number of the Liberal Arts became fixed at seven, a sacred number for the
Platonists. The orientation of these encyclopaedias was definitely Platonic (or
Neoplatonic), and at least some of them were inspired by Varro. Marius
Victorinus (4th cent. CE), the chief representative of Greek Platonism in Rome,
wrote treatises on grammar, rhetoric and dialectic. Augustine intended to write
treatises on all the Liberal Arts, but only three survive, those on grammar,
dialectic and music. Augustine was ambiguous in his attitude towards the
Liberal Arts, but these works were written quite soon after his conversion,
when he was heavily influenced by Neoplatonic philosophy. 9 He also dealt
with the seven Liberal Arts in one of his educational dialogues, De ordine,
whose Platonist orientation is striking. Of Platonist inspiration is also
Martianus Capella’s allegorical De Nuptiis Philologiae et Mercurii, which
consists of brief manuals on the Liberal Arts.

3. The Roman Educational System

To some extent the evolution of grammar in Antiquity must also be viewed
in relation to the history of education. In the first century BCE, when grammar
became an autonomous discipline, grammar schools were differentiated from
the rhetorician’s schools. Thus, language and literature, which earlier had been
taught together with rhetoric, began to be taught in separate grammar schools
in which boys between eleven (or twelve) and fifteen were taught. Afterwards,
the boys entered rhetorician’s school, where they learned to master the skills of
public speaking necessary for the future career of a statesman or an advocate.
Rhetoric was the most important subject in this curriculum. However, the
orator’s education was not complete unless he knew the other Liberal Arts and
even philosophy. So Quintilian maintained in his Institutio oratoria, which
offers us the best description of ancient education.
It was the grammarian’s task to introduce pupils to the great literature of
the past and to teach the basic skills of linguistic analysis. The grammarian’s
role was important in transmitting the literary heritage that was crucial for the
nation’s collective memory and in helping to protect the language of the
ancient texts against corruption. Thus, the grammarian acted as the guardian of
tradition, historiae custos; history included fiction and mythology alongside
true history. Because literary culture was regarded, in a common metaphor, as
a ‘mystery of the Muses’ or the ancients, the texts of the poets were thought of
as being divinely inspired. Cicero, for instance, appealed to the sacred nature
of poetry when defending the citizenship of the poet Archias: “Rightly, then,
did our great Ennius call poets holy, for they seem recommended to us by the
To what extent Augustine changed his attitude towards the study of the secular arts (and
especially dialectic) is a matter of dispute, see e.g. Marrou 1958, Pépin 1976, Lienhard 1997,
Brachtendorf 2001 and Hessbruggen-Walter 2005.

benign bestowal of God. Holy then, gentlemen, in your enlightened eyes let the
name of poet be!” (Pro Arch. 8,18, tr. by Watts). Not only were the poets
called holy and their poems considered sacred, but even the grammarian´s
profession could be described as holy. 10
Yet grammar teaching had been criticised for being trivial, and Quintilian
therefore discussed this subject in an apologetic tone. According to Quintilian,
the study of literature was blamed for immersing its students in technical
minutiae, ignoring wider issues. This is not true of an ideal grammarian, he
argued. Since grammar is the foundation of rhetoric and all subsequent study,
it is important to master even the smallest technical details of language.
Language and literature are, moreover, studied for a higher aim, namely the
stable and lasting values achieved by contemplation of divine issues. I will
quote here in full a passage from Quintilian’s defence of grammar, as it at once
illustrates the propaedeutic nature of grammar and its link with the divine. By
means of these laborious and seemingly trivial studies, Quintilian claimed, the
pupil gradually approaches the inner shrine of the sacred place.

For this reason those who criticize the art of teaching literature as trivial and lacking
in substance put themselves out of court. Unless the foundations of oratory are well
and truly laid by the teaching of literature, the superstructure will collapse. The study
of literature is a necessity for boys and the delight of old age, the sweet companion of
our privacy and the sole branch of study which has more solid substance than display.
The elementary stages of the teaching of literature must not therefore be despised as
trivial. It is of course an easy task to point out the difference between vowels and
consonants, and to subdivide the latter into semivowels and mutes. But as the pupil
gradually approaches the inner shrine of the sacred place, he will come to realize the
intricacy of the subject, an intricacy calculated not merely to sharpen the wits of a
boy, but to exercise even the most profound knowledge and erudition. (I, 4, 6 tr. by
Butler, p. 65)

As regards the other Liberal Arts (or enkyklios paideia), Quintilian first
discussed music. It is the oldest art related to literature (I,10,10), he claimed,
and quoted several examples of the role that music had played in social and
political life and even in warfare in ancient times. There are practical needs for
training in music, because the orator can stir the emotions of the audience by
controlling his voice. Furthermore, one has to command metre and rhyme in
order to appreciate poetry (I,4,4). Playing an instrument was also regarded as
part of an educated man’s culture (I,10,19). Through literature, music was
related even to divine things, Quintilian explained (I,10,11). For instance,
Orpheus was regarded as uniting the roles of musician, poet and philosopher,
all being of divine origin.

Many examples are quoted in Kaster (1988:15-17).

Quintilian further pointed out that music was important for such
philosophers as the Pythagoreans and Plato; the latter regarded the knowledge
of music as necessary for the ideal statesman (I,10,12-15). It is true that Plato
thought that poets have a divine gift, comparable to that of the prophet. Plato
(428/7-349/8 BCE) recognised poetry as divine madness, inspired by the
Muses. Being possessed by the Muses was indispensable to the production of
the best poetry (Dodds 1951:80). Poets like Hesiod (fl. c. 700 BCE) received
their message directly from the gods. Although Plato himself remained critical
of poetry, which he associated with the irrational part of the soul (Dodds
1951:217), the idea of poetry as an avenue to divine knowledge was cherished
in later Platonism.
As for the mathematical arts,11 an orator must know geometry in order to
make calculations, and to follow lawsuits when the arguments involve
measurements and boundaries. Geometry is also related to rhetoric because
both make use of logical arguments (I, 10, 34-37). Knowledge of philosophy is
necessary, for there are many difficult passages in poetry that are based on
natural philosophy, and many philosophers have expounded their doctrines in
verse (I, 4, 4). Astronomy is needed because the poets often give their
indications of time by reference to the rising and setting of the stars (I, 4, 4).
All these arts are hidden in the perfect orator’s skill, but Quintilian pointed out
that he was not describing any existing orator, but an ideal one who “seeks not
the transitory gains of advocacy, but those stable and lasting rewards which his
own soul and knowledge and contemplation can give” (I, 12, 18 tr. by Butler).
Here we are in the realm of contemplation, the highest of human activities.
Quintilian dedicated a long discussion to the importance of grammar,
music and geometry for the education of the orator, and briefly mentioned even
gymnastics. The fact that he failed to mention dialectic in this discussion
suggests that dialectic was not regarded as relevant for the study of these arts
in the first century CE. It is also worth noting that we find no traces of a
connection between dialectic and grammar in Quintilian’s exposition. Thus,
for Quintilian, dialectic was hardly the pre-condition for all further acquisition
of knowledge as it was later in Augustine’s De ordine.
During the following centuries, several grammatical manuals were
compiled in Rome, yet seldom do they mention the higher goals of language
study. Most do not define the art of grammar at all, much less address
grammar’s association with the Liberal Arts. Only a few grammarians reflect
briefly upon the role of grammar in education. Phocas (5th cent. CE) lamented
that the youth of his own time engaged in literary study neither from their
desire for learning nor for love of virtue, but as though driven by necessity. He

The Greek term ‘arithmetic’ is not used by Quintilian, who mentions numerus as one of the
divisions of geometry (I, 10, 35).

referred to school as the gymnasium of wisdom, which shows the way to a

happy life. 12 The image of the gymnasium relates literary studies to physical
exercise, which requires great diligence and discipline. In the introduction to
his grammar, Diomedes (fl. 4th cent. CE) pointed out that by virtue of the
laborious process of learning, we are as superior to the uneducated as they are
to cattle. 13 This image, which extends back at least to Isocrates (436-338 BCE),
was popular in describing the distinction achieved by man through literary

4. Augustine: De ordine
An ideal orator, as described by both Cicero and Quintilian, gradually
proceeds through the study of the Liberal Arts towards higher things, the
acquisition of virtue and the contemplation of the divine. These Platonic ideas
are made much more explicit in one of Augustine’s educational dialogues, De
ordine. 14 This text, whose sources are under debate, is the fullest description
preserved of the Platonist theory of learning. It was written in 386, quite soon
after Augustine’s conversion, when he was heavily influenced by Neoplatonic
philosophy. 15 In this educational theory, learning is described as an ascent from

Adulescentes vero nostri saeculi non desiderio litterarum nec amore virtutis ad studia se
applicare, sed aut necessitate conpulsos […] et gymnasium sapientiae, quo ad beatam vitam
semita demonstratur, velut taeterrimum carcerem detestari (GL V, 411,2-7). “[...] the youth of
our times does not pursue literary studies for their love of virtue or literature, but as if driven
by necessity , [...] and they despise the gymnasium of wisdom, which shows the path to a
happy life, as the most horrible prison […]”.
Superest ut singula recolendo memoriae tenaci mandentur, ne frustra cum tempore evanescat
labor, quo tanto maxime rudibus praestare cognoscimur […] quanto ipsi a pecudibus differre
videantur (GL I, 299, 18-24). “It remains for us to renew the cultivation of these studies, and
to record them, in order that the work would not vanish without avail in the course of time –
the work which is as superior to primitive studies, as human beings appear to differ from wild
In my recent book (2005) I have accepted Hadot’s view (1984) that the scheme of the
Liberal Arts in De ordine reflects Neoplatonic influences. Hadot thinks that Augustine’s
immediate source was Porphyry. According to the earlier standard view, De ordine depended
on Varro (Ritschl 1877), or Varro’s Pythagorean sources (Dyroff 1930). Hadot’s position has
been recently challenged by Shanzer (2005), who believes that De ordine is heavily indebted
to Varro’s Disciplinarum libri. Although I find her arguments convincing, I do not think that
De ordine reflects only Varronian influences. In the present state of research, I do not want to
exclude Neoplatonic elements from Augustine’s text (see also n. 20).
In Retractationes (426/7) Augustine says that he intended to write books about the
disciplines but was able to finish only one of them, that on grammar. He described the
disciplines as steps by which one could arise from corporeal to incorporeal things: Per idem
tempus quo Mediolani fui baptismum percepturus, etiam diciplinarum libros conatus sum
scribere, [...]; per corporalia cupiens ad incorporalia quibusdam quasi passibus certis vel
pervenire vel ducere. Sed earum solum de grammatica librum absolvere potui, […] (1,6). “At
the very time that I was about to receive baptism in Milan, I also attempted to write books on
the liberal arts, [...] and desiring by definite steps, so to speak, to reach things incorporeal

the sphere of earthly, corporeal things towards incorporeal ideas and the realm
of contemplation. The ascent takes place by means of the seven Liberal Arts.
Now their number and order is fixed and they are divided into two groups:
three linguistic and logical arts and four mathematical disciplines. However, all
seven subjects form a unity, which is based on numerical relations. It is a
typically Platonic view to regard numbers and quantities as the unifying factor
of the linguistic arts and the higher mathematical disciplines.
According to Augustine, learning is based on divine order – hence his title,
De ordine. This order is in God, on the one hand, and in the soul of a wise
man, on the other (2, 8, 25). In the process of learning, the pupil must first
subject himself to authority in order to be able to receive moral teaching so that
the irrational part of the soul is subjected to reason. The first stage does not
require study; it consists of being able to follow simple, authoritative
commands whose importance will be understood only later. The second and
third stages of learning, which consist of the seven Liberal Arts, are based on
Reason. The second level, that of learning (discendo), covers the three
linguistic disciplines – grammar, rhetoric and dialectic. The third level,
delectando, points to the arts of the quadrivium – music, arithmetic, geometry
and astronomy – as they lead to the contemplation of being and God and of
philosophy itself.
Augustine then proceeded to deal with grammar, rhetoric and dialectic in
more detail. Because human language is rational, it is capable of forming a
bond between human beings so that people can talk and exchange ideas (2, 12,
35). Things had to be named by means of significant sounds (sonos
significantes), because people could not read one another’s minds without the
mediation of words. But the words of those absent could not be heard, and that
is why Reason invented letters in order to represent the spoken sounds.
Afterwards Reason distinguished consonants, vowels and semivowels and
recognised syllables as well as the eight parts of speech. Their entire evolution,
purity, and articulation were skillfully and minutely differentiated. Thereafter,
numbers and measures entered into play. Reason directed the mind to the
different quantities of sounds and syllables, and thereby it discovered the time-
intervals through which the long and the short syllables were extended, some
double and others simple. It noted these points as well, and reduced them to
fixed rules. Reason has this ability to invent meanings, sounds and letters
because linguistic units are based on mathematical relations between them.
Grammar’s association with numbers is metrical harmony.
Grammar could already have been complete in being restricted to linguistic
analysis, but it has the additional task of examining all that has been written

through things corporeal and to lead others to them. But I was able to complete only the book
on grammar, [...]” (tr. Bogan 1968:21-22). Augustine gives slightly different lists of the
Liberal Arts in his various works (see Shanzer 2005:72).

down, all that is worth remembering (2, 12, 37). This part, which is called
history, even includes mythology, which is why grammar also examines things
that are untrue. This threatens its reason-based nature. But Augustine argued
that the untrue element does not lie in grammar itself. Grammar must deal with
all kinds of literary heritage, but this does not affect its method, which is
nevertheless true. But wherein lies the truth in grammar? Augustine asked and
suggested that it inheres in the connection between grammar and dialectic,
whose definitions, divisions and syllogistics make grammar a science. This
connection is important with the other sciences, too, which are sciences insofar
as they use the dialectical method. Dialectic, by contrast, is true in itself.
Grammar thus deals with unscientific subject matter such as myth and other
literary fiction by means of a scientific method.
Dialectic occupies a unique position among the arts in this theory.
Augustine praised it as the discipline of disciplines, which teaches to teach and
to learn (2, 13, 38). Dialectic is the result of the fact that Reason reflects upon
itself and its own tools. It is the action of pure Reason without the involvement
of the senses. Dialectic renders the other disciplines scientific through its
purely rational method. Dialectic gives the rational mind the tools to know
itself and its origin, to understand its ontological value and the higher
principles. Only dialectic can render a man wise.
Rhetoric is needed to convince an unlearned public. But rhetoric is not
sufficient as such; it must also make use of dialectical arguments so that the
public will understand and be convinced. The three linguistic disciplines form
the rational part of the theory of signification, Augustine explained. Grammar,
dialectic and rhetoric are on a lower, more earthly level from which one can
ascend towards higher realities through the study of the four mathematical
disciplines of the quadrivium. When dealing with the first of them, music,
Augustine dwells at some length on numeric proportions, which are divine and
eternal: “Because whatever the mind is able to see is always present and is
acknowledged to be immortal, numeric proportions seemed to be of this
nature” (2, 14, 41). Music is connected with both grammar and poetry through
the quantity of syllables, which are based on numeric proportions.
Reason proceeds to deal with geometry, arithmetic and astronomy, all
branches of study which present their subject matter as numerically
proportioned. Then Reason ventures to prove the soul immortal. The soul
comes to feel that it possesses great power and that it owes all its power to
numerical proportions. Something wondrous urges it on. And it begins to
suspect that it itself is perhaps the very number by which all things are
numbered or, if not, that this number can be found where it is striving to arrive.
Many of the characteristics that Augustine attributed to the Liberal Arts in
De ordine had been expressed by various authors before him: the unity and the
divine nature of the arts, as well as the idea that it is possible to ascend, by

studying the Liberal Arts, from a lower, corporeal level to higher, incorporeal
realities. Yet there are definitely novel elements in the theory of De ordine: for
instance, the enormous emphasis placed on the role of dialectic in rendering all
the other disciplines scientific. As far as grammar is concerned, I believe that
this emphasis reflects a relatively new tendency rather than the situation in the
first century BCE. 16

5. Conclusion
The earliest references to Latin grammar by Cicero and Quintilian
associated grammar with the study of virtue and the Liberal Arts. Both authors
cherished the idea of a perfect orator, an idea heavily influenced by Platonism.
The Platonic view of the Liberal Arts was characterised by unity of study as
well as the idea of ascent from a lower, earthly level to higher, divine realities.
These are also the characteristic features of the most complete Platonic theory
of learning preserved from Antiquity – the one found in Augustine’s De
ordine. It has been a matter of dispute to what extent this theory depends on
Varro’s encyclopaedic work Disciplinarum libri novem.
Almost nothing definite is known of the contents and the philosophical
orientation of Varro’s encyclopaedia. It is therefore a matter of speculation to
what extent Augustine’s De ordine and Martianus Capella’s De nuptiis
depended on Varro. The present study shows, however, that the educational
theory of Late Antiquity differed from that of Cicero and Quintilian at least in
one important respect: the role of dialectic is much more prominent in
Augustine’s De ordine than in Cicero’s rhetorical works and in Quintilian’s
description of the education of a perfect orator. In Quintilian’s Institutio
oratoria, we find no traces of an interaction between dialectic and grammar.
Dialectic is associated with the study of rhetoric and is not presented as a tool
relevant for all learning as is the case in De ordine.
Only brief allusions are made to the Liberal Arts during the first centuries
CE, when the number varied from one author to another. Grammar is not even
mentioned by all authors dealing with the Liberal Arts. It is only in the fourth
and fifth centuries that the Liberal Arts took on canonical form, and their
number was fixed at seven. Then a number of encyclopaedic works were
written. To what extent the revival of encyclopaedic works was inspired by
Varro’s Disciplinarum libri novem is a matter of speculation. Yet it seems safe
to say that the encyclopaedists of Late Antiquity did not slavishly copy Varro’s

If the philosophy of the Liberal Arts in De ordine should reflect only Varronian influences,
one would expect to find such influences even in Quintilian’s treatment of the Liberal Arts. In
my recent book (2005) I have claimed that there was a new wave of philosophical influences in
grammar in Late Antiquity which involved Platonic elements. This development probably left
its traces in Priscian’s Institutiones grammaticae and could be reflected in De ordine.

work. It was in Late Antiquity that dialectic became the discipline of

disciplines – the role which it was given again in the Carolingian Renaissance.


Primary Sources
Augustine. De ordine. Ed. by W. M. Green. Corpus Christianorum Series
Latina 29. Turnhout: Brepols, 1970.
Augustine. Retractations translated by M. I. Bogan. The Fathers of the
Church: A New Translation 60. Washington, D.C.: The Catholic
University of America Press, 1968.
Cicero. De oratore with an English translation by E. W. Sutton. Loeb Classical
Library. London: Heinemann, 1959.
Cicero. De partitione oratoria with an English translation by H. Rackham.
Loeb Classical Library. London: Heinemann, 1942.
Cicero. Pro Archia poeta. The Speeches with an English translation by N. H.
Watts. Loeb Classical Library. London: Heinemann, 1961.
Diomedes. Diomedis artis grammaticae libri III. GL I.299-529.
Galen. Protreptikos. Scripta minora 4. Ed. I. Marquardt & I. Mueller & G.
Helmreich. Stuttgart: Teubner, 1884.
GL = Grammatici latini. Ed. by Heinrich Keil. Leipzig: Teubner, 1855-80.
(Repr. Hildesheim: Olms 1981, 2002.)
Martianus Capella. De nuptiis Philologiae et Mercurii. Ed. by Adolf Dick.
Stuttgart: Teubner, 1998.
Phocas. Ars de nomine et verbo. GL V.410-439.
Quintilian. Institutio oratoria with an English translation by H. E. Butler.
Loeb Classical Library. London: Heinemann, 1920.
Seneca. Ad Lucilium epistulae morales with an English translation by Richard
M. Gummere. Loeb Classical Library. London: Heinemann, 1962.
Vitruvius. De architectura. Edited with an English translation by F. Granger.
Loeb Classical Library. London: Heinemann, 1955.

Secondary Sources
Brachtendorf, J. 2001. “The Decline of Dialectic in Augustine’s Early
Dialogues”. Studia Patristica 37.25-30.
Burton, Philip. 2005. “The Vocabulary of the Liberal Arts in Augustine’s
Confessions”. Pollmann & Vessey, eds. 2005.141-164.
Dillon, John. 1977. The Middle Platonists 80 B.C. to A.D. 220. Ithaca & New
York: Cornell University Press.
Dodds, E.R. 1951. The Greeks and the Irrational. Los Angeles & London:
University of California Press.

Dyroff, Adolf. 1930. “Ueber Form und Begriffsgehalt der augustinischen

Schrift De Ordine”. Aurelius Augustinus: Festschrift der
Görresgesellschaft zum 1500. Todestage des heiligen Augustinus ed. by
Martin Grabmann & J. Mausbach, 15-62. Cologne: J.P. Bachem.
Hadot, Ilsetraut. 1984. Arts libéraux et philosophie dans la pensée antique.
Paris: Études Augustiniennes.
Hessbruggen-Walter, Stefan. 2005. “Augustine’s Critique of Dialectic:
Between Ambrose and the Arians”. Pollmann & Vessey, eds. 2005.184-
Kaster, Robert A. 1988. Guardians of Language. The Grammarian and Society
in Late Antiquity. Berkeley & Los Angeles: University of California Press.
Lienhard, Joseph. 1997. “Augustine on Dialectic: Defender and Defensive”.
Studia Patristica 33.162-166.
Luhtala, Anneli. 1996. “Grammar and Dialectic: A Topical Issue in the Ninth
Century”. Iohannes Scottus Eriugena: The Bible and Hermeneutics:
Proceedings of the Ninth International Colloquium of the Society for the
Promotion of Eriugenian Studies Held at Leuven and Louvain-la-Neuve,
June 7-10, 1995 ed. by Gerd Van Riel, Carlos Steel & James McEvoy,
279-302. Leuven: University Press.
Luhtala, Anneli. 2005. Grammar and Philosophy in Late Antiquity. A Study of
Priscian’s Sources. (= Amsterdam Studies in the Theory and History of
Linguistic Science, 107.) Amsterdam & Philadelphia: John Benjamins.
Marrou, Henri-Irénée. 1958. Saint Augustin et la fin de la culture antique. 4th
ed. Paris: Bibliothèque des Écoles francaises d’Athènes et de Rome, 145.
Pépin, Jean. 1976. Augustin et la dialectique. Paris: Villanova.
Pollmann, Karla & Mark Vessey, eds. 2005. Augustine and the Disciplines
from Cassiciacum to Confessions. Oxford: Oxford University Press.
Rajna, Pio. 1928. “Le denominazioni Trivium e Quadrivium”. Studi Medievali
Ritschl, Friedrich. 1877. “De M. Varronis disciplinarum libris commentarius”.
Opuscula Philologica 3.352-402.
Shanzer, Danuta R. 2005. “Augustine’s Disciplines: Silent diutius Musae
Varronis?” Pollmann & Vessey, eds. 2005.69-112.
Stahl, William Harris. 1971. Martianus Capella and the Seven Liberal Arts. 1.
The Quadrivium of Martianus Capella. New York & London: Columbia
University Press.
Wagner, David L., ed. 1983. “The Seven Liberal Arts and Classical
Scholarship”. The Seven Liberal Arts in the Middle Ages ed. by David L.
Wagner, 1-31. Bloomington: Indiana University Press.

Lawrence University


This article critically assesses the linguistic and pedagogical value of

Priscian’s analysis of Latin nouns and verbs in his short treatise on the noun,
pronoun, and verb. After determining that the Institutio de nomine et
pronomine et verbo differs substantially from Priscian’s other grammatical
texts, the author demonstrates that Priscian’s accounts of the noun and verb are
a “mixed bag”. Priscian’s lengthy list of nominative singular endings is less
than satisfactory, and his paragraphs on the Latin verb are often bizarre (e.g.,
the first person plural is derived from the second singular by the insertion of an
infix: ama-s thus becomes ama-mu-s). Priscian’s Institutio does, however,
provide a succinct enumeration of the five declensions and four conjugations
and an appropriately organized array of nominal and verbal paradigms that
make the teaching and learning of Latin much easier for mediaeval
schoolmasters and students.

Priscian’s monumental Institutiones Grammaticae is arguably the most

well known and most important grammatical treatise to survive from classical
antiquity, but it is not Priscian’s only linguistic endeavor. Both his Partitiones
Duodecim Versuum Aeneidos Principalium and Institutio de Nomine et
Pronomine et Verbo are grammatical as well as pedagogical texts. The latter
plays an especially significant role in the history of Latin pedagogy, but it is an
idiosyncratic text that manifests some decidedly curious linguistic features.
What follows is an attempt to characterize the Institutio and to assess critically
its rather striking analyses of Latin nouns and verbs.
The Institutiones Grammaticae is an exhaustive account—18 books and
over 900 printed pages in Keil (GL II:1-597 and III:1-377)—of Latin grammar,
especially and almost exclusively morphology, as it had developed up until
Priscian’s floruit in the early 6th century C.E. It contains a myriad of precisely
articulated definitions; it includes thousands of citations and quotations from
classical literature, which both delighted and informed the mediaeval mind; it

is laden with philosophy that inspired later philosophical speculation; and its
last two books offer us one of the most exciting surprises in all of western
intellectual history, namely, a sophisticated treatment of selected Latin
syntactic phenomena. This last is based, as Priscian readily and almost
proudly proclaims on several occasions, on the seminal work of the second
century C.E. Greek grammarian Apollonius Dyscolus, but even so,
successfully adapting Greek syntactic analyses to Latin is an extraordinary
accomplishment. Books 17 and 18 are therefore unique in the history of Latin
linguistics and were often referred to as Priscianus Minor and copied
separately from the first sixteen books, which were known as Priscianus
Major. If Plato is correct in claiming that swans sing most sweetly right before
their death, then Priscian’s magnum opus is the swan song of Greco-Roman
language science.
Priscian authored several other treatises, and they confirm what we already
know about his craft from the Institutiones. Thus definitions and literary
quotations abound even in his short treatises on numerical symbols, Terence’s
meters, and rhetorical exercises (GL III:403-440). Priscian’s Analyses of the
Twelve First Lines of the Aeneid (GL III:457-515) is in a question-and-answer
format à la Donatus’ Ars Minor. Vivien Law (2003:87) describes it as “a
lesson in progress”, and it is also a schoolteacher’s labor amoris ‘labor of
love’. Priscian deploys a deep-seated affection for Vergil in every question he
asks and every instruction he gives. Each word of each first line of each of the
twelve books of Vergil’s Aeneid receives almost a page of linguistic scrutiny.
The student who can respond correctly to Priscian’s interrogazioni has not
only mastered his parsing but has also learned to think on his feet about both
grammar and literature, for he must scan the verse, parse each word and define
selected features of its accidence, explicate the syntax, decline or conjugate—
sometimes via synopsis—either the word at issue itself or a compound or
derivative therefrom, explore features of word-formation, rationally explain
otherwise opaque grammatical behavior, relate Latin forms to their Greek
precursors, recite related lines in other poets, and so on and so forth almost ad
infinitum. The work is a gem from the point of view of any old-fashioned
Latin teacher, and it is obvious that Priscian has poured his heart and soul into
this work. Equally as obvious, the Partitiones manifests all the same features
as does the Institutiones: definitions, morphology in profusion, syntax,
philosophy, quotations.
If we now turn to Priscian’s Institutio de nomine et pronomine et verbo
(GL III:441-456; Passalacqua 1992 is the best critical text), we find ourselves
in a completely different text entirely, and no one seems to have noticed that.
What has been noticed is that the Institutio belongs to the regulae genre of
grammatical literature (Law 2003:85) and is more systematic than any other
regulae work (idem:87). That latter observation goes a long way toward

explaining why Priscian’s Institutio competed favorably with Donatus’ Ars

Minor as the most popular and most widely used Latin “textbooks” in the early
Middle Ages until they were superseded by their authors’ more grandiose
tomes, the Institutiones Grammaticae and the Ars Major respectively.
What is so different about the Institutio? It contains not a single definition,
not a word about syntax, nothing remotely philosophical, and only a few
literary citations. It is derived from the much larger Institutiones, sometimes
verbatim, compendii causā “for the sake of an abridgement” according to its
author, but it is strikingly un-Priscianic. The simple fact of the matter is that
Priscian has dumbed down his grammatical curriculum in the Institutio—he
did not just abridge it—and in so doing he made it an incredibly useful text.
The dumbing down is, at once and paradoxically, the major weakness and the
major strength of the Institutio. Let us then take a closer and more critical look
at this short but influential treatise.
Priscian’s title informs us that his topic is the inflected parts of speech—
noun, pronoun, and verb—and it is indeed exclusively morphological. In
Keil’s text the noun receives six pages of attention, the pronoun one, and the
verb seven; the Institutio is therefore evenly divided between declension and
conjugation. The title also informs us that the text is not intended for
beginners, for it presupposes an understanding of the parts of speech, i.e., of
the sort of knowledge contained in Schulgrammatik texts like Donatus’ Ars
Minor. Moreover, like all our Latin grammatical texts it is written in Latin,
that is, in the target language, and therefore presupposes readers with at least a
modicum of Latin. A quick glance at the text reveals the presence of a number
of Greek glosses and examples, demonstrating that the Institutio is intended for
Greek speakers who need to master Latin. That only makes sense since
Priscian was teaching in Constantinople. The text is designed ad instituendos
pueros “for the purpose of instructing boys”, and it is intended to be used in
tandem with the Institutiones Grammaticae, as several explicit references to
more extensive analyses in the larger work indicate. It is highly unlikely,
however, that any but a few magistri would have had access to both texts.
Any cursory glance at the text also reveals that it is not a textbook in the
usual sense. Rather, it is a grammatical manual expressly designed for
teachers, i.e., an inventory of the inflected forms of the Latin language and
therefore a classic instance of a regulae-type of grammar. The Institutio
provides the teacher with a more or less complete and a superficially well
organized list of the nominal and verbal paradigms that he needs to teach to his
students, and to that extent it resembles what we find nowadays at the end of a
beginning Latin textbook, namely, an appendix containing declensions and
conjugations. The text begins abruptly: Omnia nomina, quibus Latina utitur
eloquentia, quinque declinationibus flectuntur, quae ordinem acceperunt ab
ordine vocalium formantium genetivos. (“All nouns that the Latin language

employs are inflected in five declensions, which have acquired their order from
the [sc. alphabetical] order of the vowels forming the genitives.”) The Greek
schoolboys learning Latin in the classroom of Magister Priscianus Caesariensis
in the early sixth century C.E. must have rejoiced when they heard or read that
sentence, because in their study of their own native language they had been led
to believe that Greek has over 50 kanōnes. Life in Latin was destined to be so
much easier for them with only five declensions to master.
The students would also have been pleased with the examples which their
teacher provides, for they are for the most part semantically transparent. To
hear them is to know what they mean because they are either borrowed or
cognate Greek words like syllaba ‘syllable’ or pater ‘father’ or they are Latin
words with which the boys are familiar, like senatus ‘senate’ or meridies
‘noon’, which is when the school day ends; so they may concentrate on the
morphology, which is, after all, their main task in the course. Unfortunately,
after its first short paragraph their teacher’s manual on Latin morphology gets
complicated, needlessly complicated they probably thought, and modern
Latinists would agree with them. Priscian notes that the nominative of the first
declension has two terminal letters but three terminations and plunges
immediately into an analysis of the final letters and terminations of all types of
nominative singulars in Latin, including Greek proper names. Worse, gender
rears its ugly head for the first time and remains prominent throughout
Priscian’s account of nominal morphology. Put differently, the brilliant first
paragraph that is so practical and so quintessentially Roman is followed by
four pages of Greek emphasis on phonological and morphological matter that
is simply not germane to the classification of nouns into declensions and that is
not necessary for students to learn—not way back then and not now.
We can easily see how needlessly complicated this approach can become.
Nouns ending in -a, whether Greek or Latin, masculine or feminine or
common, are of the first declension. Exceptions are those adjectives that are
declined like pronouns and end in long -īus in the genitive. Neuter nouns
ending in -a, however, are Greek and are of the third declension. Names of
letters, half of which end in -a in Greek, are not declined. Priscian concludes
with some ringers, namely, Greek proper names that end in long -ēs and look
like first declension patronymics but are not; his examples are the names of
Thucydides and Euripides. Such names are of the third declension, as are
barbarian names with the same ending. By the time Priscian gets to the fifth
declension, three and a half pages later, he literally has nothing left to say.
“The fifth declension has a single ending in long -ēs, which we have
exemplified above when we were discussing the third declension.” In other
words he has proceeded via a process of elimination, discussing every
nominative singular ending in the Latin language and assigning it to one or the

other of the five declensions in Latin. Nothing that Priscian says about Latin
nouns in this section of his treatise is wrong, but neither is it necessary.
Priscian’s first short paragraph informs his students and us that all Latin
nouns are classified into five declensions based on the ending in the genitive
singular, and he then spends the next four and a half pages on the spelling of
the nominative. The only morphological merit in this mess is that the student
who masters this material might occasionally be able to assign a new
vocabulary item to the correct declension, but since there are often so many
options, his success may be due more to luck than to his learning. Basing a
morphological account on the spelling of the nominative is just bad linguistics,
as Theodosius’ account of the more than 50 kanōnes he posits for Greek
(Bekker 1821:975-1002) eloquently but unfortunately attests.
As for the bad linguistics, here is what the early 5th century grammarian
Consentius has to say (GL V:357-358). He notes that it is “difficult and
arduous” to use the nominative singular as the base form of the other cases.
Nouns with the same nominative ending have different genitives and therefore
different oblique cases throughout their declension. The inconstantia
“inconsistency” is such that the nominative is virtually ending-less and lacking
a well defined termination. Ergo, he says, “on account of these difficulties we
have refrained from establishing rules on the basis of the nominatives because
they are not logically unambiguous and are inexplicable by virtue of their
quantity.” He opts for the genitive, arguing that “all that variation, which most
[sc. grammarians] have tried to explain on the basis of the nominative, is
included in three or four well defined rules based on the genitive.” Consentius
is a man who knows whereof he speaks. So is Varro of course, and he says
much the same thing and in fact establishes the theoretical basis for
Consentius’ entire argument in De Lingua Latina X.52-62 (cf. Taylor
Priscian has obviously followed the confusing and linguistically
incompetent practice of “most [sc.grammarians].” Two, maybe even three,
reasons explain why he does so. First, as Law (2003:88) notes, Priscianus
Major is a fusion of Schulgrammatik and regulae genres, and the former
privileges the nominative. In an abridged treatise it is only natural that he
would retain the Schulgrammatik material, for better or worse. Second, he is
in Constantinople and is indubitably affected by the Greek grammatical
tradition, which uniformly privileges the nominative. So the weight of both
the Greek and the Roman grammatical traditions is almost forcing him to
proceed as he does. Third, some pedagogical value inheres in what is
otherwise morphological mayhem, for the lexical and phonological
information that Priscian provides is worthwhile, perhaps not priceless but
valuable nonetheless. It is the sort of information that does not fit

systematically into any grammar but that enhances and augments any student’s
understanding of Latin nouns.
We learn, for example, that ficus is masculine and fourth declension when
it refers to the vitium corporis ‘bane of the body’, i.e., ‘hemorrhoids’, but
feminine and either second or fourth declension when it means ‘ficus tree’. Vir
and its compounds are the only second declension nouns ending in -ir. Greek
nouns ending in -ros convert that ending to -er in Latin and are second
declension, like Alexandros in Greek but Alexander in Latin, unless, that is,
they are possessives. Nouns ending in -e, -l, -n, -t, or -x are all third
declension, and so are nouns ending in -r if they are not second declension.
Although nouns ending in long -ēs may be first, third, or fifth declension, if the
-ē- is preceded by an -i-, the noun belongs to the fifth declension. Nouns
ending in short -es are third declension. All nouns ending in -is are third. And
so on and so forth. None of this will tell the students how to form the genitive,
which is the base form for every declension, as Priscian has already pointed
out in his first paragraph, but any student who masters this sort of information
will be intellectually better prepared than those who have not. Perhaps then
Priscian included all this not because he was compiling a Latin
Schulgrammatik section or was heir to the Greek tradition but because he was a
good Latin teacher who wanted his students to have every advantage they
could get. The plethora of references to Greek nouns and Greek endings and
how both are “converted”, to use his term, into Latin suggest that he had his
own Greek students in mind when he composed this section of the Institutio.
At the end of this Schulgrammatik section Priscian immediately and
without warning plunges into his regulae section, and the five paragraphs
provide us with the morphological information that he promised in the very
first paragraph, namely, the five declensions. Each declension gets one
paragraph, and the entire section takes only a page and a half. It is a perfect
example of linguistic economy, and the naked simplicity of truth is a beauty to
behold. Priscian does not arrange his examples into the formulaic schemes that
are ubiquitous in modern textbooks and grammars, for that is a procedure that
does not enter intellectual history until later; such paradigms are clearly
implicit in his prose, however. Priscian is not the only grammarian to present
the five declensions in their now canonical form, but we ought remember that
not all grammatici Latini do so. Priscian is among the elite. His account of
pronouns follows, but it is relatively transparent and requires no comment.
Priscian’s first sentence on the verb is straightforward: “All regularly
inflected verbs end in -o or -or and have four conjugations.” The next four
sentences define those four conjugations on the basis of the vowel before the
final -s in the second singular. This is standard operating procedure for the
grammatici Latini and originates with Varro, who had discovered the
conjugations that way. In what is a harbinger of things to come, Priscian

derives the second singular by rules of addition, subtraction, and substitution or

conversion. Thus the first conjugation changes final -o into long -ās: amō ‘I
love’ thereby becomes amās ‘you love’. The -eō in the second conjugation
becomes long -ēs: doceo goes to docēs. Third conjugation -io verbs drop the
final -o and add an –s; other third conjugation verbs, however, change the final
-o of the first person into a short –is: facio becomes facis, lego becomes legis.
The fourth conjugation converts the -io of the first person to a long -īs in the
second: audio audīs. Priscian is cogently aware, here and elsewhere, of the
importance of vowel length. The four conjugations are therefore clearly and
unambiguously defined by their first and second persons singular, specifically,
by how the latter is derived from the former. Priscian’s initial foray into the
verb conjugations and verbal morphology is therefore familiar enough.
Nothing which follows on the next five pages on verbs could be described
as familiar, however, with the exception of general categories like tense and
mood and so forth, but even here there are surprises. The tenses of the
indicative are five in number: present, preterite imperfect, preterite perfect,
preterite pluperfect, and future. Tense clearly takes precedence over aspect,
which is not considered a grammatical category, and there is neither a present
perfect nor a future perfect, both of which Varro had long ago identified.
There are indicative and subjunctive moods, of course, but also an optative
mood. Greek has an optative, so Latin does too. We know it as the volitive
subjunctive or subjunctive of wish, but to Priscian and his Greek students it is
an optative despite being isomorphic with the subjunctive. Meaning or
signification trumps morphology. At least there is no conjunctive mood in
Priscian as there is in some other grammatici Latini. We know the conjunctive
as, inter alia, the future perfect indicative, which is a subjunctive to Priscian.
What is least familiar, however, and therefore most fascinating is the way
Priscian derives the other persons in any given tense or mood from his two
“base” forms, i.e., the first and second persons of the present indicative. No
other mainstream Latin grammarian does what Priscian does here; Sacerdos
and the Bobbio excerpt of Macrobius contain similar but not exactly the same
analyses. To the extent that Priscian’s analysis of Latin verb forms differs
from a modern one, it is nothing short of bizarre.
It begins auspiciously enough. The third singular is derived from the
second singular by simply converting the final -s to a -t and by shortening the
vowel if it is long: amās amat, docēs docet, legis legit, audīs audit. So far, so
good. If we expect Priscian to continue substituting plural endings for the final
-s of his second singular “base” form, however, we would be sorely
disappointed. Priscian instead inserts the syllable -mu- into the second singular
before the final -s. So amāmus is not amā- plus -mus, the only first plural
active voice morpheme in the Latin language, but is instead considered as amā-

plus -mu- plus -s. That is also how we get the second plural—by inserting the
syllable -ti- into the second singular before the final -s.
The third plural in the first and second conjugations is formed by inserting
an -n- in front of the final -t of the third singular: amat becomes amant and
docet docent. So much for using the second singular as a base form. In the
third and fourth conjugations, the third plural is derived from neither the
second singular nor the third singular but from the first singular by converting
the final -o to -unt: facio faciunt, lego legunt, audio audiunt. Priscian’s rules
work, but they are woefully misconceived. He simply has no concept of
morpheme and simply fails to see that the terminal morphemes for the six
persons recur throughout the several tenses and moods.
The preterit imperfect is an even better case in point. It consists of a
progressive stem, the infix -bā- for the past tense, and the personal endings -m,
-s, -t, -mus, -tis, -nt, which are virtually universal in the Latin verb system and
which are just about the first verbal morphemes that modern-day students
learn. But that is not Priscian’s approach to the imperfect. In the first and
second conjugations he drops the second singular ending from the present and
adds -bam, thereby combining and confusing tense and person morphemes, but
since he does not know what a morpheme is, we should not be too surprised.
In verbs of the third and fourth conjugations he converts the final -o of the first
person singular to a long -ē and then adds -bam. It is not necessary to belabor
the obvious: Priscian’s analyses yield some strange verbal bedfellows.
Priscian employs the first person singular as the base form for the preterite
perfect since it always ends in long -ī. By adding -stī we get the second
person; by adding -t and shortening the vowel we derive the third. Keeping the
vowel short and adding -mus and -stis make the first and second persons
respectively. The third plural requires special treatment: we change the final -
ī to a long -ē- and add -runt or -re. Pluperfects are not perfect stem plus -erā-
plus personal endings, nor are they perfect stem plus the imperfect of “to be”.
Rather, we must begin with the same base form, namely, the first singular of
the perfect, and then we change that final long -ī to a short -e- and add -ram.
Enough is enough.
Priscian’s verbal regulae will work, but they are askew and amiss,
linguistic mumbo-jumbo, pedagogical hogwash. Because he has no concept of
a morpheme and therefore no morphophonemics, he is simply juggling
syllables and letters, not indiscriminately but also not scientifically. It is a
classic instance of not being able to see the forest for the trees. Latin verbs
have stems, infixes, and terminal morphemes: the endings specify person and
number and voice; the infixes, including a zero-morpheme, specify tense and
mood; and the stems specify aspect and signification. Priscian, it would seem,
is unaware of this in general and only dimly aware of it in part. Beauty, for
better or worse, is in the eye of the beholder, and Priscian’s system is pretty

ugly. Most of his students probably just memorized paradigms. To Priscian’s

credit, he has tried to analyze the Latin verb in all its complexity as a rule-
governed system; he just got the rules wrong.
In fine, Priscian’s accounts of nouns and verbs leave much to be desired
both linguistically and pedagogically. The one gives too much extraneous
information, the other arranges its information and its misinformation un-
insightfully. Can anyone learn to decline nouns and conjugate verbs from the
Institutio? Yes, as thousands of mediaeval students undoubtedly did, but they
did so almost in spite of their textbook or their teacher’s manual. Historians of
linguistics raised on Thomas Kuhn’s The Structure of Scientific Revolutions
tend to emphasize those periods of revolutionary science when paradigms shift
and to ignore those periods of normal science when textbooks are written.
This is sometimes too bad, for textbooks and teachers’ manuals tell us about
what is being taught and, hopefully, learned (cf. Bursill-Hall 1977:1-3).
Priscian’s Institutio de nomine et pronomine et verbo is a product of normal
science that transports us into the late antique and mediaeval Latin classroom,
and despite its shortcomings it is a fascinating text that offers us otherwise
unobtainable insights into the history of Latin language science.


Bekker, Immanuel, ed. 1821. Anecdota Graeca. Vol. III. Berlin: Reimer.
(Repr., Graz: Akademische Druck- u. Verlagsanstalt, 1965.)
Bursill-Hall, G. L. 1977. “Teaching Grammars of the Middle Ages: Notes on
the Manuscript Tradition”. Historiographia Linguistica 4.1-29.
GL = Keil, Heinrich, ed. 1855-80. Grammatici Latini. 7 vols. + supplement.
Leipzig: Teubner. (Repr., Hildesheim: Olms, 1961 and 1981.)
Kuhn, Thomas. 1996[1962]. The Structure of Scientific Revolutions. 3rd ed.
Chicago: University of Chicago Press.
Law, Vivien. 2003. The History of Linguistics in Europe: From Plato to
1600. Cambridge: Cambridge University Press.
Passalacqua, Marina, ed. 1992. Prisciani Caesariensis Institutio de Nomine et
Pronomine et Verbo. Urbino: Quattro Venti.
Taylor, Daniel J. 1996. Varro De Lingua Latina X: A New Critical Text and
English Translation with Prolegomena and Commentary. (= Studies in the
History of the Language Sciences, 85.) Amsterdam & Philadelphia: John

UMR 7597, Université Paris Diderot (Paris 7)


This article is an account of a detailed study of the collection of sources

and references used by Conrad Gessner in his Mithridates, de differentiis
linguarum [...] obseruationes (1555). Although this volume is of modest
length, its author intends for it to be a comprehensive survey of the world’s
languages. He does not claim his work to be original, however: he clearly
affirms the debt he owes to his predecessors, whom he cites as systematically
as possible; quotations make up nearly 50% of the total text. The vastness of
Gessner’s scholarship yields a cumulative index – both of the direct sources of
the Mithridates and of authors quoted – in excess of 150 names, cited more
than 500 times.
Taking different criteria into consideration (chronology, frequency of
sources, quantitative importance of the cited texts), this article first covers the
forms of citation (in terms of length and precision), then gives global statistical
data, before providing a more detailed analysis organized by period: antiquity
(85 authors cited), Middle Ages (15 authors), and Renaissance (53 authors).
These three periods are thus represented in the Mithridates, with a clear
predominance of historians from antiquity, along with Renaissance
cosmographers, explorers and naturalists.

0. Introduction
Dans son Mithridate, Conrad Gessner (1516-1565) entreprend de présenter
les langues du monde, tant les langues anciennes que les modernes, à partir des
données qu’il a rassemblées dans ses ouvrages encyclopédiques, notamment sa
Bibliotheca Universalis (1545), complétée par un Appendix (1555).
Gessner ne prétend pas faire œuvre originale: il affirme nettement la dette
qu’il doit à ses prédécesseurs, qu’il cite le plus systématiquement possible,
parfois avec précision, parfois avec plus de flou. Et comme son érudition est
immense, l’index cumulé des sources directes du Mithridate et des noms
d’auteurs cités est riche de plus de cent cinquante noms.

Dans le présent article, nous envisageons d’étudier l’ensemble de ces

sources et de ces références dont la somme représente ce qu’on peut appeler
l’“horizon de rétrospection” du Mithridate. La notion d’horizon de
rétrospection, qui recouvre l’ensemble des connaissances préalables
reconnues, assumées ou même refusées par un auteur x à un moment x de
l’histoire, est particulièrement exploitable pour des travaux encyclopédiques,
comme l’Encyclopédie de Diderot (cf. Auroux & Colombat 1999). Elle l’est
aussi pour un ouvrage comme le Mithridate qui, bien que de taille modeste
(160 petites pages, moins de 185 000 signes espaces compris), entend donner
un aperçu complet des langues de l’univers.
Pour explicite que soit l’indication des sources du Mithridate, la perception
par le lecteur desdites sources et surtout de leur étendue est beaucoup moins
évidente. Cela tient tout autant aux habitudes d’alors (citation approximative
des auteurs, des textes et surtout des passages dans les textes) qu’aux
contraintes techniques de l’époque (pas de guillemets, utilisation limitée de
l’italique, etc.). Après une brève typologie de ces citations, nous tenterons d’en
établir un inventaire complet en tenant compte des éléments suivants:

— chronologie: les auteurs anciens côtoient les contemporains de

Gessner, sans que ce dernier n’établisse de degré de validité entre les
différentes époques;
— fréquence des sources: certains auteurs et certains textes sont
mentionnés très souvent dans l’ouvrage, d’autres ne sont cités qu’une
— importance quantitative des textes cités: à côté de citations
extrêmement brèves, des passages entiers sont reproduits, par exemple
pour le traitement étendu de certaines langues, comme l’ancienne
langue gauloise et le germanique.

La corrélation de ces éléments doit permettre d’affiner la perception que

peut avoir un moderne de ce texte, une des premières compilations
linguistiques, en identifiant soigneusement ce qui relève de l’emprunt et ce qui
relève de l’interprétation propre de Gessner.

1. Forme et typologie de la citation

Ayant abordé ailleurs (Colombat 2006) la question de la citation, nous ne
donnerons qu’un échantillon qui nous permettra au moins d’entrevoir comment
se présente la citation dans le texte de Gessner et quelles difficultés doit
affronter l’exégète.
vel Libyca. nam Punicam eandem Arabicae faciunt eruditi.

PVNICA lingua apud Afros mures quidam zegeries uocantur, quod in Graeca ualet
idem quod βουνοί (id est colles,) Herodotus libro 4. ¶ Accipiter Libycè uocatur
barbax, Hesychius. ¶ Libyes battos appellant reges, Scholiastes Pindari. (Mithridate,
1555:8r 1 )
ou libyque, Les savants en effet confondent langue punique et langue arabe.
En langue punique, chez les Africains, certains rats sont appelés zegeries, ce qui en
grec signifie la même chose que βουνοί [bounoi], c’est-à-dire collines (Hérodote, IV).
L’épervier est appelé en libyen Barbax (Hésychius). Les Libyens appellent les rois
Batti (scholiastes de Pindare).]

Le texte de Gessner peut être entièrement constitué de citations: tel est le

cas dans cet extrait. Les citations sont séparées par des pieds de mouche (¶) ou
des espaces plus larges. Comme dans les autres textes de cette époque, aucun
usage n’est encore fait des guillemets, mais les parenthèses—(id est colles,)
(Valla colles transfert,)—permettent d’introduire une remarque, que Gessner
propose parfois en italiques, notamment dans les textes longuement cités. La
source est donnée en fin de citation, par le nom de l’auteur, éventuellement
suivi d’une précision (le livre 4 d’Hérodote). Elle est traduite du grec en latin,
toutes les citations de ce passage étant originellement en grec. Pour bref qu’il
soit, cet échantillon est assez représentatif de la méthode de Gessner: ce dernier
apporte des témoignages, dont il prend soin de mentionner la source, mais sans
la précision qu’attend un lecteur moderne.
Quelle typologie peut-on dresser de ces citations?
— Elles peuvent être courtes, voire très courtes—comme dans l’échantillon
donné ci-dessus—ou, au contraire, très longues. De fait, des pages entières du
Mithridate sont constituées de longues citations. Par exemple, dans le passage
sur la uetus lingua Gallica, c’est-à-dire le gaulois, sont successivement
reproduits de longs passages de Glareanus (20r-22v) et d’Aventinus (23r-25r),
soit 9 pages de citation pratiquement continue. Le même Aventinus est à
nouveau longuement cité (30r-31v, puis 34v-35v) dans le chapitre consacré à
la langue allemande. Dans ce cas, Gessner s’efface complètement derrière
l’avis donné, se contentant de commenter dans des parenthèses (souvent en
italiques, notamment dans les longues citations) le texte cité et en prenant bien
soin de rappeler par des indices, parfois peu visibles pour le lecteur moderne,
qu’il s’agit bien d’une citation (un titre: De uetera lingua Gallica Hen.
Glareani iudicium, 20r; une reprise: (inquit idem Glareanus), 22v, mox,
“bientôt”, c’est-à-dire “un peu plus loin dans l’ouvrage”, 25r; des parenthèses:
(inquit Io. Aventinus), 23r; une fin de citation: Haec ille “Voilà pour lui
[Joachim Vadianus]”, 27v).

À l’avenir, nous donnons uniquement le numéro du folio (recto ou verso) dans l’édition de

— Elles sont relativement précises. Certains auteurs ne sont certes pas

mentionnés, mais ce n’est jamais volontaire, contrairement à une pratique
fréquente à l’époque. Les mentions legere memini “je me souviens d’avoir lu”
(11r, 16v), memini uidere “je me souviens d’avoir vu” (72r), si bene memini
“si je me souviens bien” (18v, 48v, 68r) sont signes d’un souvenir flou, voire
d’un doute sur ce souvenir, mais en aucun cas d’une volonté de masquer
l’origine. De même, les mentions vagues, comme aliqui... alii “quelques-uns…
d’autres”, recentiores quosdam “certains modernes” (27v), Nuper quidam…
asseruit “Récemment quelqu’un a prétendu…” (61r), Grammatici quidam
“Certains grammairiens” (61v), sont relativement peu nombreuses.
— Elles peuvent être imbriquées les unes dans les autres, soit que Gessner
intervienne dans la citation, à son propre titre ou encore pour introduire un
autre auteur (du pseudo-Caton ou du Pline l’Ancien cités en italique dans du
pseudo-Bérose, 11r), soit que l’auteur cité cite lui-même un autre auteur: ainsi
Platon cité par Clément d’Alexandrie (3v, 4r), Homère cité par Strabon (14r),
Pline l’Ancien cité par Vadianus (27r), etc. Il faut donc distinguer citation de
premier niveau et citation de second niveau (la citation dans la citation).
— Elles peuvent être adaptées au besoin de l’exposé. Citant Tacite
(Germanie, 43, 3), Gessner (30r) remplace une longue liste de peuples
germains cités par Tacite par la simple phrase Ex iis etiam Naharuoli sunt (Il y
a aussi parmi eux les Naharvales), puisqu’il ne s’intéresse alors qu’à ce peuple.

2. Données globales
Si l’on tente, malgré les difficultés et les inévitables approximations, de
discerner ce qui constitue proprement des échantillons de langues, des citations
d’auteurs (dont on a ôté les échantillons) et la part qui revient à Gessner lui-
même, on obtient la répartition suivante: les citations dominent nettement
(47% du texte total) et si on leur ajoute les échantillons (11%) avec lesquelles
elles occupent 58% du volume total, il ne reste que 42% pour le texte de
Gessner lui-même: Gessner fait parler les langues, fait parler les autorités dont
il se réclame, et derrière lesquelles il s’efface de bonne grâce.
Attachons-nous maintenant aux données suivantes: le nombre d’auteurs
mentionnés; le nombre de mentions qui en sont faites; la quantité de texte cité
attribuable à chacun d’eux, et rapportons ces données aux périodes
correspondantes: Antiquité, Moyen Âge, Renaissance et Humanisme.
153 auteurs sont mentionnés dans le Mithridate, ce qui est considérable
dans un livre aussi court (2+78 feuillets, soit 158 512 caractères, espaces non
compris). Les noms de ces 153 auteurs sont mentionnés 507 fois: autrement
dit, un auteur est utilisé, simplement mentionné ou explicitement cité, en
moyenne 3,3 fois. Mais il y a une forte disparité entre les auteurs :
Les trois auteurs dont le nom est cité plus de vingt fois totalisent 69 mentions (14%):
Tacite, César, Aventinus.

Les huit auteurs cités de 10 à 19 fois totalisent 105 mentions (21%): Hérodote,
Münster, Pline l’Ancien, Athénée, Strabon, Postel, pseudo-Bérose, pseudo-
Les dix-huit auteurs cités de 5 à 9 fois totalisent 121 mentions (23%): Jérôme, Beatus
Rhenanus, Eusthate, Homère, Ammien Marcellin, Clément d’Alexandrie, Denys
le Périégète, Tite Live, Miechowita, Pirckheimer, Plutarque, Althamer, Diodore
de Sicile, Glareanus, Aeneas Sylvius, Herberstein, Marco Polo, Platon.
Les 31 auteurs cités trois ou quatre fois totalisent 101 mentions (20%).
Les 93 auteurs cités une ou deux fois totalisent 111 mentions (22%).

Rapportons maintenant la mention de ces auteurs aux périodes délimitées:

Antiquité, Moyen Âge, Renaissance et Humanisme (Tableau 1, col. 2).
L’Antiquité domine avec 85 auteurs cités, suivi par l’Humanisme (53 auteurs),
le Moyen Âge restant loin derrière (15 auteurs).
Tableau 1

Colonne 1 : Colonne 2 : nombre Colonne 3: nombre Colonne 4: longueur

Période d’auteurs cités de mentions des textes cités
Antiquité 85 (55%) 306 (60%) 22943 c. (31%)
Moyen Âge 15 (10%) 34 (7%) 2120 c. (3%)
Renaissance et 53 (35%) 167 (33%) 48637 c. (66%)
Total 153 (100%) 507 (100%) 73700 c. (100%)

La part de l’Antiquité augmente légèrement si l’on prend en compte le

nombre de mentions des auteurs (Tableau 1, col. 3). Mais si l’on tient compte
de la longueur des textes cités pour chaque auteur (citations de premier
niveau), les proportions s’inversent entre Antiquité et Humanisme, alors que la
part du Moyen Âge est considérablement réduite (Tableau 1, col. 4).
Gessner puise donc amplement dans le stock antique, qu’il connaît
admirablement, mais c’est aux auteurs humanistes qu’il accorde sa confiance,
au point de s’effacer parfois complètement derrière eux. Les citations
humanistes représentent en effet 66% de l’ensemble des citations, mais aussi
31% du volume total du texte du Mithridate, soit presque un tiers de l’ouvrage.
Intéressons-nous maintenant aux auteurs le plus souvent mentionnés
(Tableau 2, col. 1 et 2). Les auteurs antiques dominent: les historiens (Tacite,
César, Hérodote) viennent en tête, suivis par un naturaliste (Pline l’Ancien), un
antiquaire (Athénée), un géographe (Strabon). Les humanistes ne sont
représentés que par l’historien Johannes Aventinus, Sébastien Münster et
Guillaume Postel. Une place à part est à faire à Bérose et Caton qui sont en fait
un pseudo-Caton et un pseudo-Bérose (cf. infra).
Mais si l’on tient compte de la longueur de texte cité (Tableau 2, col. 3 et
4), les humanistes reprennent le dessus, avec la domination écrasante
d’Aventinus, suivi par Münster, Glareanus et Pirckheimer (38% à eux quatre).

Hérodote vient en 5e position, suivi par le pseudo-Bérose, Strabon et Tacite,

Miechowita et Aeneas Sylvius s’intercalant en 7e et 9e position.

Tableau 2

Col. 1. Les 11 auteurs Col. 2. Nombre Col. 3. Col. 4.

mentionnés 10 fois ou de citations Les 10 auteurs le plus Pourcentage de
plus longuement cités texte cité
Tacite 25 Aventinus 15%
César 24 Münster 8%
Aventinus 20 Glareanus 8%
Hérodote 19 Pirckheimer 7%
Münster 16 Hérodote 5%
Pline l’Ancien 14 Bérose 4%
Athénée 13 Miechowita 4%
Strabon 12 Strabon 3%
Postel 11 Aeneas Sylvius 3%
Pseudo-Bérose 10 Tacite 3%
Pseudo-Caton 10 Autres auteurs 40%

3. Les périodes
3.1 Auteurs de l’Antiquité (Tableau 3)
Gessner associe étroitement auteurs latins et auteurs grecs, qu’il cite en
traduction, mais en donnant assez souvent le terme grec original significatif,
notamment pour Athénée, Hérodote, Hésychius, Jean le Grammairien,
Plutarque et Strabon. En version originale, on trouve également deux vers
d’Homère sur la diversité des langues parlées à Troie (Iliade, 4.437-438) et un
extrait d’Aristophane (Les oiseaux) sur le langage indistinct des Illyriens.
Si TACITE (ca.55-ca.116) est l’écrivain le plus cité, c’est en tant qu’auteur
de la Germanie, notamment à cause du long développement que Gessner
consacre à la langue allemande (27r-44r), mais pour la longueur des citations,
il ne vient qu’en 3e position. Les citations sont en effet nombreuses (une
douzaine), mais courtes. Le passage le plus cité (26v, 44r-v, 62v) concerne la
langue pannonienne des Oses (Germanie, 43.1) qui démontre que ces derniers
ne sont pas des Germains.
CESAR (100a.C.-44a.C.) est mentionné dans le développement sur la langue
gauloise (17v-25v), mais aussi dans celui sur l’allemand, naturellement pour le
Bellum Gallicum, dont trois passages sont reproduits: le célèbre début (Gallia
est omnis diuisa in parteis treis...), l’évocation de l’absence de druides chez les
Germains (6.21), un passage du livre 3 (22) sur les sol[i]durii.

Tableau 3

Col. 1. Les auteurs de Col. 2. Col. 3. Les 10 auteurs de Col. 4.

l’Antiquité cités 6 fois et Nombre de l’Antiquité le plus Pourcentage de
plus 2 citations longuement cités texte cité
Tacite 25 Hérodote 15%
César 24 Pseudo-Bérose 12%
Hérodote 19 Strabon 10%
Pline l’Ancien 14 Tacite 8%
Athénée 13 Plutarque 8%
Strabon 12 Clément d’Alexandrie 8%
Pseudo-Bérose 10 Athénée 6%
Pseudo-Caton 10 Diodore de Sicile 6%
Jérôme 9 Pseudo-Caton 4%
Ammien Marcellin 8 César 3%
Homère 8 Autres auteurs antiques 20%
Clément d’Alexandrie 7
Plutarque 7
Tite Live 7
Diodore de Sicile 6

HERODOTE (ca.484a.C.-425a.C.) est le 3e auteur pour le nombre de

mentions, mais le premier pour la quantité de citations. Gessner cite comme un
leitmotiv l’affirmation d’Hérodote selon laquelle, au temps de Crésus, les
Athéniens l’emportaient parmi les Ioniens, et les Lacédémoniens parmi les
Doriens. Le passage qui contient cette phrase (Hérodote 1.56-57) est cité plus
longuement dans le développement sur la langue pélagique (62r-v) et Gessner
affirme avoir utilisé la traduction de Valla corrigé par lui-même d’après le
manuscrit grec. L’autre passage développé concerne les langues des cités
d’Ionie (Hérodote 1.142, Mithridate, 57r). Est également citée, aux deux bouts
de l’ouvrage (2v et 69r), l’affirmation selon laquelle les Troglodytes
Éthiopiens sifflent comme des chauves-souris.
L’Histoire naturelle de PLINE L’ANCIEN (23-79) est utilisée pour le nombre
de langues parlées à Dioscurias (6.15), pour le don des langues du roi
Mithridate (7.88), pour les Scythes appelés Sacae ou Aramaei (6.50,
Mithridate, 11r, 63v) et pour quelques termes attribuables à telle ou telle
langue: eglecopala en gaulois (17.46), brechmasis en “indien” (12.27, mais
nous lisons aujourd’hui bragma sic), gadir en punique (4.120).
ATHENEE (déb. IIIe s.p.C.), auteur grec des remarques érudites du Banquet
des Sophistes, est lui aussi cité pour différents termes isolés, dont l’origine est
asiatique, “indienne”, laconienne, macédonienne, perse ou thébaine. On peut
lui associer CLEMENT D’ALEXANDRIE (ca.150-215), auteur des Stromates,
beaucoup cité au début du Mithridate, à propos du nombre de langues, de la
À l’exception de Denys le Périégète dont le nom apparaît toujours cité dans le commentaire

notion de dialecte, d’un prétendu langage des animaux, puis à nouveau

mentionné à la fin pour le terme Βέδευ ou Bedy qui signifierait ‘eau’ ou ‘air’
en phrygien ou en macédonien. DIODORE DE SICILE (Ier s.a.C.) est également
cité dans le développement préliminaire pour le voyage de Iambolos, témoin
des particularités linguistiques des Éthiopiens, puis à propos des hiéroglyphes
égyptiens. Le géographe STRABON (ca.63a.C-19p.C.) est cité principalement
pour un long développement sur la langue carienne, les langues barbares et le
barbarisme en général (14.2.28). Le passage est cité en traduction, mais la
dernière phrase est en outre citée en grec, sans doute pour faire apparaître sous
leur forme originale les termes καρίζειν [karizein], βαρβαρίζειν [barbarizein] et
σολοικίζειν [soloikizein].
PLUTARQUE (ca.46-120) est cité pour son De fluuiis à propos des termes
Araxa et Βριξάβα [Brixaba], pour sa Vie de Lycurgue à propos du dialecte
laconien, et surtout pour son De Homero, à propos des témoignages que la
langue homérique fournit sur l’éolien, l’attique et le dorien.
Comme on l’a déjà dit, BEROSE et CATON sont en fait un pseudo-Bérose et
un pseudo-Caton—auxquels il faut ajouter un pseudo-XENOPHON, moins
souvent cité—tous les trois imaginés par le faussaire ANNIUS DE VITERBE
(1432-1502), dont la supercherie a trompé la plupart des érudits pendant plus
d’un demi-siècle. Tous les trois sont utilisés notamment pour parler de la
langue arménienne (10v-11v) et expliquer le lien entre Noé et Janus et, plus
généralement, le lien entre la Bible et l’occident. Bérose est aussi utilisé pour
la distinction entre les Thuyscons (qui seraient les ancêtres des Germains) et
les Celtes et à propos de la langue étrusque. Caton l’est en outre pour
l’appellation du Pô Botigon par les Étrusques, Botigum par les Ligures, pour
les Osci devenus Capuani et les Volosci devenus Volsci.
TITE-LIVE (59a.C.-17p.C.) est cité pour ses témoignages sur les Galates
descendants des Celtes, sur l’origine germanique des noms de rois celtes (dans
une citation de Beatus Rhenanus) et sur le rapport entre l’étrusque, le latin et le
rhétien. AMMIEN MARCELLIN (ca.330-400) l’est pour des noms germaniques
Les noms de saint MATTHIEU et de saint JEROME apparaissent à l’occasion
des versions données du Notre Père, et le second est en outre considéré comme
inventeur (avec Cyrille) du cyrillique et traducteur de la Bible en dalmate.
Saint PAUL est mentionné pour sa lettre aux Galates.

3.2 Auteurs du Moyen Âge

En tête des auteurs médiévaux cités, on peut s’étonner de voir des auteurs
aussi différents que le commentateur byzantin Eustathe et le voyageur vénitien
Marco Polo. EUSTATHE (ca.1115-ca.1195) est cité (8 fois) pour son
commentaire du géographe Denys le Périégète: il l’est pour l’assimilation du
colchidien à l’égyptien, et de l’ionien au vieil attique, ainsi que pour la mention

de termes en phrygien, en scythe ou en thrace. Le témoignage de MARCO POLO

(ca.1250-1324; 5 mentions) apparaît essentiellement à la fin de l’ouvrage, pour
l’évocation des peuples rencontrés lors de son périple en orient, pour dire
simplement que tel ou tel peuple a une langue propre (habent regem & idioma
proprium / linguam propriam habent / habet autem linguam propriam &
SUIDAS (Xe s.; 4 mentions) est cité à propos de noms germaniques et pour
le prétendu nom du pain, bekos, en lydien ou phrygien. BEDE (ca.673-735) est
mentionné 3 fois (dont une fois dans une citation de Münster) à propos de
l’origine germanique de l’anglais.
La plus longue citation médiévale revient à ÉGINHARD (ca.770-840; 2
mentions) qui évoque les noms des mois et les noms de vents inventés par
Charlemagne (35v-36r). SAXO GRAMMATICUS (ca.1150-1220) et WITICHIND
(Xe s.) sont également mentionnés deux fois, le premier comme historien des
peuples scandinaves, le second comme historien des Saxons.

3.3 Auteurs humanistes (Tableau 4)

Tableau 4

Col. 1. Les auteurs Col. 2. Col. 3. Les 10 auteurs Col. 4.

humanistes cités 5 fois et Nombre de humanistes le plus Pourcentage de
plus citations longuement cités texte cité
Aventinus 20 Aventinus 22%
Münster 16 Münster 12%
Postel 11 Glareanus (Loriti) 12%
Beatus Rhenanus 8 Pirckheimer 11%
Miechowita 7 Miechowita 5%
Pirckheimer 7 Aeneas Sylvius 4%
Althamer 6 Francus, Fabianus 3%
Glareanus (Loriti) 6 Wernher, Georg 3%
Aeneas Sylvius 5 Crinitus, Petrus 3%
Herberstein 5 Herberstein 3%
Rhodiginus 3%
Beatus Rhenanus 2%
Postel 2%
Autres auteurs humanistes 15%

On observe une corrélation étroite entre les colonnes 2 et 4 du Tableau 4,

avec une relative exception pour Postel et Beatus Rhenanus. Postel est souvent
cité (11 fois), mais les citations sont très brèves. Beatus Rhenanus est
mentionné 8 fois, mais Gessner n’en tire que quatre citations relativement
Il est difficile d’établir une typologie des auteurs utilisés par Gessner, tant
est large la palette de compétences de la plupart de ces auteurs. Par exemple
Aeneas Sylvius, alias le Pape Pie II, est présenté dans le Larousse du XIXe

siècle comme “théologien, orateur, diplomate, canoniste, historien, géographe,

poète latin”. Pour simplifier, on peut dire que la source humaniste ciblée par
Gessner n’est pas principalement le linguiste, mais l’historien-géographe, ou
“cosmographe” pour utiliser l’appellation de l’époque, ou encore le
commentateur d’historien ou de géographe.
Johannes AVENTINUS (1477-1534), chroniqueur bavarois, auteur des
Annales des Boïens, est à la fois l’humaniste le plus souvent mentionné et
l’humaniste le plus longuement cité, notamment à propos du gaulois (23v-25r)
et de l’allemand (30r-31v, 34v, 35r-v), mais aussi, de façon plus brève, à
propos du hongrois et des langues slaves. Gessner tire de longues citations du
livre I des Annales des Boïens, en prenant donc parti pour la thèse faisant du
gaulois une langue germanique. Il utilise aussi la Nomenclatura établie par
Aventinus qui lui procure de nombreuses étymologies de mots allemands.
Aventinus est suivi d’un cosmographe, en l’occurrence Sébastien
MÜNSTER (1488-1552), dont la Cosmographie est amplement utilisée (9
citations) pour la langue anglaise, le gaulois, l’allemand, l’irlandais, le
moscovite, le sarde et l’origine du Rotwelsch ou langue des vagabonds. De
Münster, Gessner utilise en outre un long passage de la Grammaire chaldaïque
pour traiter de la langue éthiopienne. C’est aussi en tant que cosmographe
qu’est utilisé AENEAS SYLVIUS (1405-1464), pour sa Cosmographie en deux
parties (Europe et Asie), qui sert à Gessner pour la Hongrie, la Valachie (la
Roumanie) et les nomades Zigares (Tsiganes).
Pour la Hongrie, Gessner utilise également l’ouvrage de Georg WERNHER
(déb. XVIe s.) consacré avant tout aux sources thermales de ce pays
(Hypomnemation de admirandis Hungariae aquis), mais qui évoque aussi les
peuples, notamment les Juhres, ancêtres des Hongrois. L’érudit polonais
MIECHOWITA (Matthias à Michau/Michou, 1456-1523), auteur d’un Tractatus
de duabus Sarmatiis, est également mis à contribution sur le même sujet, ainsi
que sur les langues slaves et lituaniennes, pour lesquelles Gessner utilise aussi
le baron HERBERSTEIN (Sigismundus Liber Baro, 1486-1566), diplomate
autrichien connu pour avoir donné aux occidentaux, après deux voyages à
Moscou, une meilleure connaissance de la Russie dans ses Rerum
moscouitarum commentarii.
Pour l’Est de l’Europe, Gessner utilise la Germaniae ex uariis scriptoribus
perbreuis explicatio de l’érudit allemand Willibald PIRCKHEIMER (1470-1530),
qui lui fournit des informations sur les Tcherkesses, les Khazares, les
Mengrels, la Chersonèse taurique (la Crimée) et les peuples slavophones.
Gessner utilise aussi le récit, par l’historien italien Paulus JOVIUS (1483-1552),
d’une ambassade de Basile le Grand, prince de Moscovie, auprès du pape
Clément VII.
Les commentateurs occupent aussi une place importante. Le Suisse
Heinrich Loriti dit GLAREANUS (1488-1563) fait l’objet d’une longue citation

(5,5 p., 20r-22v) empruntée à ses Annotations aux Commentaires de la guerre

des Gaules de César. Glareanus y défend la thèse selon laquelle le gaulois était
identique à la langue des Helvètes et des habitants de la Rhénanie et montre
que les mots gaulois ont été profondément corrompus par les auteurs latins,
notamment César. En complément naturel de la Germanie de Tacite, BEATUS
RHENANUS (1485-1547) est mis à contribution pour ses Castigationes in
libellum Taciti de situ, moribus & populis Germaniae, à propos de la parenté
des Gaulois et des Germains, de la romanisation des Romains, d’une leçon
dans un manuscrit de Tacite et des termes Truton et Trutmann.
À côté de ces historiens, cosmographes et commentateurs, linguistes et
philologues sont moins nombreux. Outre Guillaume POSTEL (1510-1581), déjà
mentionné, qui fournit des informations sur les langues orientales, l’Italien
Ludovicus Caelius RHODIGINUS (1469-1525), auteur de notes (Antiquae
lectiones) sur les auteurs grecs et latins, est cité pour la tendance des Doriens à
ouvrir largement la bouche et à beaucoup utiliser la lettre a, pour l’interdiction
que se sont faite les Carthaginois, trahis par le grec dans leur guerre contre
Denys l’Ancien, d’apprendre l’écriture et la langue grecques, pour l’évocation
(inspirée d’Athénée) des Grecs “chasseurs de noms” et grands créateurs de
mots. Du traité conçu à la manière des Nuits attiques par PETRUS CRINITUS
(1474-1507), De honesta disciplina, Gessner tire un long développement sur
les quatre dialectes du latin, l’archaïque, le latin, le romain et le mixte,
développement qui doit beaucoup (sans que Crinitus le dise) au livre IX des
Étymologies d’Isidore. Un assez long passage de l’Orthographia Deutsch de
Fabianus FRANCUS (XVIe s.) est utilisé pour montrer les variétés de
prononciation en fonction des différents dialectes allemands.
Les naturalistes (qui n’apparaissent pas dans le Tableau 4) sont également
mis à contribution: Pierre BELON (ca.1517-1594) tant pour son Histoire de la
nature des oyseaux que pour ses Observations recueillies d’un voyage en
orient, Gisbert de LONGUEIL (Longolius, ca.1507-1543) auteur d’un Dialogus
de auibus, Guillaume TARDIF (ca.1440-ca.1500), auteur d’un Art de
faulconnerie et des chiens de chasse. Le premier donne un terme désignant un
oiseau des îles Moluques, le second dit que le nom allemand du bec de l’oiseau
est précisément bec (rostrum en latin), le dernier fournit le nom de certaines
races de faucons chez les Égyptiens, Babyloniens et Assyriens.
La dernière catégorie de textes à évoquer est celle des récits de voyageurs,
particulièrement utilisés à la fin de l’ouvrage, dans un développement intitulé
De variis linguis, praesertim in remotissimis terris imperii Tartarici, & Orbis
noui (“Des diverses langues, surtout celles parlées dans les pays les plus
reculés de l’empire Tartare et du Nouveau Monde”, 70r-71v). Outre Marco
Polo, déjà évoqué, Gessner utilise Josaphat BARBARO (XVe s.), patricien
vénitien qui fut envoyé en Tartarie, puis chez les Perses et qui tira une relation
en italien de ses voyages sous les titres Viaggio alla Tana et Viaggio in Persia.

Pour le nouveau monde, Gessner se sert de Pierre MARTYR (1457-1526),

Alvise CADAMOSTO (ca.1432-1488), et Amerigo VESPUCCI (1451-1512). Le
premier est un érudit italien, secrétaire apostolique à la cour de Ferdinand
d’Aragon et d’Isabelle de Castille, qui fit l’histoire de la découverte des
Amériques dans ses Décades du nouveau monde. Gessner l’utilise pour des
renseignements sur les langues parlées au Panama, à Cuba et à Hispaniola
(Haïti). Le second lui fournit l’étymologie de Madère (‘bois’), le troisième des
informations sur la langue des habitants de la Grande Canarie.

4. Conclusion
Sur les 153 auteurs répertoriés, nous n’avons évoqué que les plus
fréquemment cités. En fait ces mentions et citations ne représentent que
l’émergence de l’iceberg par rapport à l’immense connaissance qu’a
accumulée Gessner en élaborant ces ouvrages majeurs (au moins par la taille)
que sont notamment la Bibliotheca universalis, l’Historia plantarum et
l’Historia animalium. Ces monuments constituent l’arrière-fond d’un
Mithridate voué à traquer les “différences” entre les langues. Le bilan de
l’étude des sources est un peu déroutant pour un lecteur moderne, car les
ouvrages grammaticaux ou linguistiques ne sont pas dominants, loin de là. En
fait, ce n’est pas la structure des langues qui intéresse Gessner, mais l’indice de
leur variété: ce qu’il recherche avant tout, c’est le moindre témoignage, réduit
à un seul mot — les mots isolés ou en petite série sont légion dans le
Mithridate — ou la simple mention d’une langue différente parlée à un endroit
identifiable de la terre (cf. le prétendu turc Inte mename de Louis de Barthème
[69v] ou les indications extrêmement vagues d’une lingua propria reprises à
Marco Polo).
Le Mithridate n’est en aucun cas le bilan linguistique d’une époque. Il n’est
même pas le bilan des époques passées. Certes Festus ou Servius ne sont pas
oubliés (ils font l’objet de trois mentions chacun), mais ils sont là surtout
comme témoins de langues peu connues (l’osque pour l’un, le punique et le
thessalien pour l’autre).
Le Mithridate est plutôt un fichier ouvert où grammairiens et linguistes,
s’ils ne disposent pas de l’érudition de Gessner, peuvent puiser à leur gré des
informations difficiles d’accès. À ce titre, il constitue un document
extrêmement précieux à la fois pour le philologue à la recherche des sources
alors disponibles et pour l’historien de la linguistique qui peut s’imaginer
quelle représentation globale des langues on pouvait se forger au milieu du
XVIe siècle.

Annexe: liste des auteurs mentionnés par période et nombre de mentions 3

Antiquité: Tacite (25); César (24); Hérodote (19); Pline l’Ancien (14); Athénée (13); Strabon
(12); Bérose (pseudo-) (10); Caton (pseudo-) (10); Jérôme (saint Jérôme) (9); Homère (8);
Ammien Marcellin (8); Clément d’Alexandrie (7); Denys le Périégète (7); Plutarque (7);
Tite Live (7); Diodore de Sicile (6); Platon (5); Aristophane (4); Hésychius (4); Paul (saint
Paul) (4); Appien (3); Cicéron (3); Ennius (3); Festus (3); Hippocrate (3); Jean Ammonius
(Jean le grammairien) (3); Matthieu (saint Matthieu) (3); Pausanias (3); Ptolémée (3);
Servius (3); Suétone (3); Théocrite (3); Varron (3); Virgile (3); Xénophon (pseudo-) (3);
Dioscoride (2); Élien (2); Ephore (2); Moschos (2); Nicolaus Damascus (2); Porphyre (2);
Ruffus (2); Stobée (2); Accius (1); Alcée (1); Aristote (1); Arnobe (1); Cassiodore (1);
Daniel (1); Démosthène (1); Étienne le grammairien (1); Flavius Josèphe (1); Fulgence
(1); Galien (1); Horapollon (1); Ixion Demetrius (1); Jonathan (1); Julius Capitolin (1);
Lactance (1); Lucien (1); Marcellus Empiricus (1); Naevius (1); Nicolaus Myrepsus (1);
Onkelos (1); Pacuvius (1); Philippus (1); Philon (1); Phortius (1); Pindare (1); Plaute (1);
Polyen (1); Pomponius Mela (1); Procope (1); Proculus (1); Quadrigarius (1); Salluste (1);
scholiastes de l’Odyssée (1); scholiastes de Pindare (1); Sisenna (1); Solin (1); Térence
(1); Timée de Locres (1); Timosthène (1); Titinnius (1); Valerius Antias (1).
Moyen Âge: Eusthate (8); Marco Polo [Marcus Paulus Venetus] (5); Suidas (4); Bède (3);
Eginhard (2); Saxo Grammaticus (2); Witichindus (2); Avicenne (1); Cyrille (1);
Gotofridus Viterbensis (1); Hunibald (1); Jordanus (1); Notger de Saint-Gall (1); Paul
Diacre (1); Wasthaldus (1).
Humanisme: Aventinus, Johannes (20); Münster, Sébastien (16); Postel, Guillaume (11);
Beatus Rhenanus (8); Matthias à Michou [Miechowita] (7); Pirckheimer, Willibald (7);
Althamer, Andreas (6); Glareanus, Henricus [Loriti, Heinrich] (6); Aeneas Sylvius (5);
Herberstein [Sigismundus Liber Baro] (5); Belon, Pierre (4); Crinitus, Petrus [Ricchio,
Pietro] (4); Gyraldus [Giraldi, Giglio Gregorio] (4); Wernher, Georg (4); Bale, John (3);
Barbaro, Josaphat (3); Gelenius, Sigismond (3); Georgevitz, Bartol (3); Martyr, Pierre (3);
Rhodiginus, Ludovicus Caelius [Ricchieri, Ludovico] (3); Varinus, Phavorinus [Guarinus
Camers] (3); Francus, Fabianus (2); Luther, Martin (2); More, Thomas (2); Potken,
Johannes (2); Stumpfius, Johannes (2); Tardif, Guillaume (2); Valla, Lorenzo (2);
Agricola, Georges (1); Alunno, Francesco (1); Bellunensis, Andreas (1); Bibliander,
Théodore (1); Bifrons, Jacob (1); Breitenbachius, Bernhardus (1); Burrana, Petrus (1);
Cadamustus, Aloysius [Cadamosto, Alvise] (1); Celtis, Conrad (1); Cranzius, Albertus (1);
Dalburgio, Johannes Camararius a (1); Denisot [de Nisot], Nicolas (1); Galeotto Marzio
(1); Jovius, Paulus (1); Lodoicus Patritius [Louis de Barthème] (1); Longolius, Gybertus
[Longueil, Gilbert de] (1); Magnus, Johannes (1); Magnus, Olaus (1); Stabius, Johannes
(1); Trithème, Jean (1); Tschudi, Gilg (1); Vadianus, Joachim [Watt, Joachim von] (1);
Vespucci, Amerigo (1); Wied, Anton (1); Ziegler, Jacob (1).


Auroux, Sylvain & Bernard Colombat. 1999. “L’horizon de rétrospection des

grammairiens de l’Encyclopédie”. Recherches sur Diderot et
l’Encyclopédie 27.111-152.

Nous ne donnons que quelques variantes des patronymes, notamment pour les auteurs

Colombat, Bernard. 2006. “Citation des sources, citation des langues dans le
Mithridate de Conrad Gessner”. Hôs ephat’, dixerit quispiam, comme
disait l’autre… Mécanismes de la citation et de la mention dans les langues
de l’Antiquité, éd. par C. Nicolas, Recherches & Travaux, hors série
Colombat, Bernard & Manfred Peters. À paraître. Conrad Gessner. Mithridate
(1555). Traduction française avec introduction, annotation et indexation.
Genève: Droz.
Gessner, Conrad. 1555. Mithridates, de differentiis linguarum tum veterum tum
quae hodie apud diuersas nationes in toto orbe terrarum in usu sunt [...]
Obseruationes. Tigurini [Zürich]: C. Froschoverus. Cf. Peters 1974.
Peters, Manfred, ed. 1974. Konrad Gessner. “Mithridates”. Neudruck der
Ausgabe Zürich 1555. Aalen: Scientia Verlag.

Philosophy Department
Pontifícia Universidade Católica do Rio de Janeiro


Michel de Montaigne (1533-1592) was one of the most influential authors

in the rise of modern thought, especially in the French context, his influence
reaching philosophers as important as Descartes and Pascal in the seventeenth
century. He was also one of the main representatives of the revival of ancient
skepticism in the modern age, with his own brand of skeptical thought
contributing to the development of modern skepticism as distinct from the
ancient one. Language was discussed in the Essais (1570-1580) from a
skeptical perspective, mainly in relation to knowledge, moral and cultural
relativism, and religious experience. Montaigne contributed to the
development of a philosophical discussion of language, addressing
fundamental questions such as: the diversity of languages and cultural
relativism, the imperfection and inadequacy of ordinary language for
philosophical reflection, and consequently, the need for a special language and
a specific style such as that found in the Essais.

Il y a le nom et la chose: le nom, c’est une voi que remerque et

signifie la chose; le non, ce n’est pas une partie de la chose ny de la
substance, c’est une pièce estrangere joincte à la chose et hors d’elle.
Montaigne, “De la gloire”, Essais, II, 16.

1. Introduction
Michel de Montaigne (1533-1592) was a leading humanist and one of the
most influential authors of early modern thought, especially in the French
context, his influence reaching from nowadays obscure thinkers such as Pierre
Charron and François La Mothe Le Vayer to saints like Francis of Sales to
philosophers as eminent as Descartes and Pascal in the seventeenth century. He
was also one of the main representatives of the revival of ancient skepticism in
the modern age, with his own brand of skeptical thought - nouveau
pyrrhonisme - contributing towards the development of modern skepticism as

distinct from the ancient one. He discussed language in his Essais (1570-1580)
with a skeptical perspective, mainly in relation to knowledge, moral and
cultural relativism, and religious faith. The issue of moral and cultural
relativism is particularly relevant as Montaigne took into consideration the
anthropological problems raised by the discovery of the New World. The
anthropologist Claude Lévi-Strauss (1991), for instance, considered him a
predecessor of modern ethnography. 1
Were the inhabitants of the Americas human beings in the same sense as
the Europeans, that is, descendants of Adam, or did they come from another
origin? Did the languages they spoke develop from the same sources, or were
they different forms of signifying and communicating? Did these “savages”
have souls and could they be converted to Christianity, or was this a hopeless
task? If so, how should they be addressed in order to understand the Gospels?
These were some of the fundamental questions we find in the sixteenth century
debate as a result of the European conquest of the New World and of the
impact the contact with this until then terra incognita had on traditional
knowledge (Bernand & Gurzinski 1991).
To discuss these matters, Montaigne took issue with the prevalent
conceptions of language, culture, and human nature of his time, opening the
way to a debate involving multiculturalism and relativism. The discovery of
the New World works as a counterpoint against which European culture must
revaluate itself. It also points towards a future in which both worlds, Old and
New, will be changed:
Notre monde vient d’en trouver un autre, non moins grand, plein et membru que lui,
toutefois si nouveau et si enfant qu’on lui apprend encore son ABC; il n’y a pas
cinquante ans qu’il ne savait ni lettres, ni poids, ni mesure, ni vêtements, ni blés, ni
vignes. Il était encore tout nu au giron, et ne vivait que des moyens de sa nourrice. Si
nous concluons bien de notre fin, et ce poète de la jeunesse de son siècle, cet autre
monde ne fera qu’entrer en lumière quand le nôtre en sortira. L’univers tombera en
paralysie; l’un membre sera perclus, l’autre en vigueur. (“Des coches”, Essais, III, 6)

[Our world has lately discovered another as large, well peopled, and fruitful, as this
whereon we live; and yet so raw and childish, that we are still teaching it its ABC: ’tis
not above fifty years since it knew neither letters, weights, measures, vestments, corn,
nor vines; it was then quite naked in the mother’s lap, and only lived upon what she
gave it. If we rightly conclude of our end, and this poet of the youthfulness of that age
of his, that other world will only enter into the light when this of ours shall make its
exit; the universe will fall into paralysis; one member will be useless, the other in
vigour]. (trans.: Cotton 1952)

Montaigne realized from the very beginning the meaning of the impact of
the exchange between these radically different cultures. The European empires
would try to conquer the New World, and would almost certainly succeed, but

See chapter XVIII, “En relisant Montaigne” (Lévi-Strauss 1991).

the New World, with its different cultures, its riches, its challenges, would also
irrevocably transform European cultures and would also force them to
revaluate themselves. The contrast between a New World and Europe, the Old
World, is particularly significant.
All these matters had a deep influence on Montaigne’s contribution to a
philosophical discussion of language, as for him it consisted in addressing
fundamental questions such as: the diversity of languages and cultures, the
imperfection and inadequacy of ordinary language for philosophical reflection,
and as a consequence, the need for a special language and a specific style such
as that found in the Essais, in his own words, this “only book in the whole
world of its kind” [“seul livre au monde de son éspece”] (II, 8).

2. Skepticism in the Essais

The subject I examine here is part of a larger project on theories of
language in modern philosophy. My central hypothesis is that the revival of
ancient skepticism in modern thought contributed decisively towards the
philosophical interest in language and particularly in what could be called a
pragmatic as opposed to a logical or formal view of language. I consider
Montaigne precisely as representative of this pragmatic view as he gives
central importance to ordinary language and to concrete human experience.
The criticism of traditional conceptions of knowledge and culture,
dominant in the Middle Ages, is one of the main aspects of the skeptical
outlook that begins to take shape in the sixteenth century. Language which was
considered either from a grammatical or from a logical point of view begins
now to be seen as a cultural phenomenon (Wlaswo 1987:173-181).
There is controversy around the inclusion of Montaigne among modern
skeptics. Historians of ideas such as Richard Popkin (2003) defend this,
considering him “the most significant figure in the sixteenth-century revival of
ancient skepticism” (2003:44). Other interpreters of Montaigne’s philosophy
do not consider him a skeptic at all, but only a philosopher whose thought had
a skeptical phase, but who ended up adopting an eclectic or even a fideist
stance (Villey 1908).
The fact is that the “Apologie de Raymond Sebond” (1962), the longest of
the Essais (2002), written around 1575-76, contains many references to ancient
Greek skepticism, and certainly became one of the major sources of knowledge
of this philosophy in the sixteenth century, influencing most of the
philosophers of the seventeenth century in this respect. Popkin (2003:47)
describes Montaigne’s crise pyrrhonienne in terms of the doubts raised in him
by his reading of the ancient skeptics such as Pyrrho of Elis (4th century BCE);
and, as we shall see, Montaigne frequently referred to skeptic philosophers as
“Pyrrhonians”. In fact, the revival of ancient skepticism on the eve of modern
thought corresponds to the interest these philosophers brought to a context of
change, crisis, and conflict between traditional doctrines and the new

challenges the great transformations of those times were provoking. The

downfall of ancient physics and astronomy and the rise of the new science, as
well as the religious and political crisis of the Reformation, meant a context of
diaphonia – or conflict of doctrines – precisely in the sense discussed by
ancient skeptics. It is certainly highly significant that the term “new” was then
used both to describe the New World and the new science, in this latter case by
Galileo himself who employed the expression Scienza Nuova.
Montaigne was the first major philosopher of modern times in whom we
find an important and influential reflection on the social and cultural impact of
the major changes of his age: the discovery of the New World, the
Reformation and the Scientific Revolution. The fact that these three historical
episodes practically coincide made the sixteenth century an almost unique
period in history in which we find one of the most complete breakdowns of
traditional thought.
The nouveau pyrrhonisme that in a certain way Montaigne inaugurates has
to be seen against this background and results from the consideration of the
relativity of human beings’ intellectual and moral achievements, a relativity
that was to undermine the whole concept of human nature, of its place in the
cosmos, and of its universality (Popkin 2003:44). “Pyrrhonism”, a term
derived from Pyrrho of Elis, the father of Greek skepticism in the fourth
century BCE, amounts then to the realization of the limits of human knowledge
due to the frailty of human nature. This is a typical skeptic argument, and one
of its major sources, the 10th trope of Aenesidemus found in Book I of Sextus
Empiricus’s Pyrrhonian Hipotiposes, is precisely the trope of ethical relativity
showing that beliefs and values are culture dependent (Annas & Barnes
1985:151-171). Comparing different cultures and even different customs
among the Greeks and other peoples of the East, Sextus Empiricus shows that
we cannot establish the foundations of universal values. Montaigne sees the
impact of the discovery of the New World as giving new grounds for this same
line of argument, and in fact making it even more radical, since the difference
between European cultures and the various cultures of the New World is so
deep, we could almost say they are incommensurable (“Des cannibales”, “Des
In consequence of this diversity, the adoption of a notion of knowledge as
absolutely certain and any attempt at building science as a perfect explanation
of the essence of reality breaks down and is considered by Montaigne to be a
vain enterprise. Any attempt at establishing a universal ethics would also lack
adequate criteria. We should rather accept the relativity of knowledge and its
inconclusiveness, and the philosopher should rather pay closer attention to
practical matters and give moral advice in the sense of a strictly applied ethics,
than speculate on metaphysical and theoretical matters.
Can we say that the Essais are a work of skeptical philosophy? According
to one of Montaigne’s most recent interpreters, Sylvia Giocanti (2001),

skepticism probably constitutes the only philosophical standpoint capable of

giving coherence and intelligibility to the Essais. She agrees with the three
stages of interpretation of the Essais originating from Pierre Villey’s classic
study (1908). 2 Book I is considered as representing Montaigne’s eclecticism,
that is, his appropriation of elements of Stoic, Epicurean and Socratic
philosophies, the so-called tour de la philosophie in order to point out the
richness and complexity of traditional philosophy, but also its inconclusive
character, and finally its diaphonia, or the insoluble conflict of doctrines.
These doctrines can all contribute towards our philosophical reflections, but
ultimately do not give us any definite answers and fail to achieve their own
aims of explaining the nature of reality and defining the way towards the “good
life”. The skeptic standpoint is adopted in Book II, especially in the “Apology”
(II, 12), not to replace these traditional doctrines with a better one, but taking
ancient skepticism as an ancestor, a predecessor working as a starting point to
Montaigne’s own doubts and questions, and to the realization of the
impossibility of attaining certainty on any given subject under investigation. In
a way, anthropology, as a conception of human nature, is the foundation of
Montaigne’s skepticism, since human reason is characterized by irresolution
(Giocanti 2001) and due to its limits, certainty is an illusion. The limits of
human reason and intellectual powers are the background against which Book
III proposes a practical ethics, based on the acceptance of diversity and the
relativity of customs and therefore of values. This practical ethics, having as its
basic lesson the virtue of tolerance, is always to be put to test (the literal
meaning of essai) in our experience (Levine 1999). Detachment, tolerance,
tranquility, and humbleness are the virtues which can give us through
reflection some assurance of happiness and a life as good as possible in this
troubled, imperfect, changing world.

3. The relevance of language

Language is not a central theme in Montaigne’s Essais (if indeed there is
any), and although a brief survey of specialized journals such as Montaigne
Studies and of the main interpreters of his work reveals very few analyses of
his views on language, I propose to highlight some passages of the Essais
which according to my interpretative hypothesis allow us to consider language
one of the key notions for the understanding of Montaigne’s philosophy,
underlying the whole conception of his work. 3
My claim is that the way Montaigne understands language, indeed the way
he sees the need for a new language, is part of the context of revival of ancient
skeptic philosophy towards which he gave such a major contribution.

This interpretation is, however, contested by S. Rendall (1992) among others, defending a
less structured composition of the Essais.
Some noteworthy exceptions are Berven (1995) and Mathieu-Castellani (1988).

Although there is not a systematic treatment of language in the Essais, as in

fact the work does not deal with any subject systematically, we find throughout
the text a number of reflections on some of the main issues of philosophy of
Montaigne’s conception of meaning at word level is nominalistic, as he
Il y a le nom et la chose; le nom c’est une voix qui remarque et signifie la chose; le
nom, ce n’est pas une partie de la chose ni de la substance, c’est une pièce étrangère
jointe à la chose, et hors d’elle. (“De la gloire”, Essais, II, 16)

[There is the name and the thing, the name is a voice which denotes and signifies the
thing, the name is no part of the thing, nor of the substance, it is a foreign piece joined
to the thing and outside it]. (Trans.: Cotton 1952)

Word meaning does not give us any access into reality, and at the end of
“Des coches” (III, 6) he says, “nous n’avons aucune communication à l’être”
[we have no communication with Being]. He adopts a thoroughly
conventionalist conception of meaning, compatible with his skepticism.
Therefore we should not dispute over words as scholastic philosophers
often did and as Protestant and Catholic theologians were doing in
Montaigne’s time. He says that most of the occasions for the troubles of the
world are grammatical (II, 12) and our disputes are often merely verbal (III,
13), thus exposing the frivolous character of these disputes.
At a propositional level, Montaigne challenges the privilege or centrality of
the assertion or descriptive sentence adopted by philosophers since Plato and
Aristotle as the basic form of propositions:
Je vois les philosophes pyrrhoniens qui ne peuvent exprimer leur générale conception
en aucune manière de parler: car il leur faudrait un nouveau langage. Le notre est tout
formé de propositions affirmatives, qui leur sont du tout ennemies. (“Apologie de
Raimond Sebond”, Essais, II, 12)

[The Pyrrhonian philosophers, I see, cannot express their general conception in any
kind of speaking; for they would require a new language on purpose: ours is all
formed of affirmative propositions, which are totally hostile to them]. (Trans.: Cotton

In this new language (nouveau langage) affirmative sentences in which we

express certainty would have no privilege, would not be considered the
essential form of our relation to reality; rather the expressions of ignorance,
“J’ignore”, and of doubt, “Je doute”, are at least equally important and
guarantee the avoidance of dogmatism.
Is Montaigne one of those “Pyrrhonian philosophers”? We could say so, at
least as regards the essays of Book II, according to most interpreters, and
certainly in the “Apology of Raymond Sebond”. We could also say that his is

precisely this “new language” one should adopt, avoiding dogmatic assertion
and refusing the declarative proposition the status it had since Aristotle’s
Organon. That is, avoiding the employment of judgment as saying of reality
that it is as it is, and being true or false in relation to it.
But what kind of “new language” should this be? We find already in Book
I a defense of a “Pyrrhonian language”, for instance when Montaigne says:
Et si j’eusse eu à dresser des enfants, je leur eusse tant mis en la bouche cette façon de
répondre enquêteuse, non résolutive – “Qu’est-ce à dire?”; “Je ne l’entends pas”; “il
pourrait être”; “Est-il vrai?” (Des boiteux, III, 11)

[Had I been to instruct children, I would so often have put this manner of answering
in their mouth, enquiring and not resolving: “what means it? I understand it not. It
may well be. Is it true?”] (Trans.: Cotton 1952)

The trope of child instruction is a traditional one, later influencing the

study by linguists, especially psycholinguists and pedagogues of language
acquisition. 4 One of the best known examples is the much quoted passage,
quoted among others by Wittgenstein (Philosophical Investigations § 1), of
Saint Augustine’s Confessions (I, 8) in which children learn the meaning of
words as adults point things out to them and pronounce the name attached to
these things, the so-called “ostensive definition”. Montaigne is saying precisely
the opposite, namely that we also need a new language to teach children not
word meaning through ostensive definitions but how to use words to question
and interrogate, as children so often do. This is how we learn language; this is
how knowledge is acquired and developed. This is in fact how language works
according to our experience and contrary to the idealized model of language
favored by logicians.
Language is used in communication, as an instrument of human
interaction, and as a way through which we understand each other, negotiate,
and express our recognition of each other. “Nous ne sommes hommes et ne
nous tenons les uns les autres que par la parole” [We are only men, and only
relate to each other through language] (I, 9). But it should not be seen as a
perfect instrument, nor should understanding be idealized. It is a very
imperfect tool and we have to acknowledge and accept its limitations.
Language always says too much or too little; it is inadequate and its
inadequacy is part of the skeptic argument about the limitations of human
nature (“De l’art de conférer”, Essais, III, 8). Since we depend on it to express
ourselves, in fact to become ourselves as linguistic subjects, we have to accept
that subjectivity is not the locus of certainty, and what is unsaid can only be
expressed indirectly, can only be pointed out as something to be sought after.
“Mais quand tout est conté, on ne parle jamais de soi sans perte” [But, when all

See Essay I, 26, “On the education of children”.

is summed up, a man never speaks of himself without loss] (“De l’art de
conférer”, Essais, III, 8).
Therefore, although these two conceptions – that of communication and
that of the limits of linguistic expression – are apparently in contradiction, they
are fundamental features of the functioning of language. Language should not
be idealized, and it is not required that meaning should be perfectly
unambiguous and understanding be total for communication to take place.
These should not be seen as flaws but as part of the skeptic view of human
nature and of Montaigne’s skeptic strategy of giving emphasis to concrete
human experience (Buzon 1992). On the contrary, it is the presupposition of
definite certainty and absolute truth found in dogmatic philosophy which is
problematic. New experiences such as the natural and cultural reality of the
New World prove this and show that we need a new language, in fact a
language always capable of renewing itself as the only adequate way of
expressing the diversity of meanings of our experience.

4. Concluding remarks
The Essais’ digressive style results from Montaigne’s skeptic idea of the
impossibility of dealing conclusively with any matter. Language and style
should reflect directly the way the subject matter is discussed; the
philosopher’s discourse must be coherent with the philosophy he defends. In
this sense the language of the Essais is an essential feature of Montaigne’s
philosophy, it is an expression of how he thinks philosophy must be done.
Montaigne lived in troubled times; the great majority of the beliefs on
which tradition was anchored were beginning to collapse, France was
immersed in a civil war, and kings and political leaders were being murdered. 5
In a very Socratic way, he felt that the philosopher’s main duty was towards
life and that he should first and foremost renounce any presumption,
abandoning the vanity of certain knowledge, “la vanité de savoir”. He was at
the same time politically engaged and socially detached, a skeptic and a
believer, a hermit and a man of the court, friend of kings, queens and noblemen
as well as of the peasants of his Bordeaux region. His main lesson to us seems
to be one of tolerance towards the diversity of customs and cultures, pointing
out the relativity of all our values and beliefs - religious, political, even
scientific. “Nous sommes Chrestiens à mesme titre que nous sommes ou
Perigordins ou Alemans” [We are Christians by the same title that we are
Perigordians or Germans] (“Apologie de Raimond Sebond”, Essais, II, 12;
Trans.: Cotton 1952). He points to Europeans of his time that had they been

Namely, the political assassinations in 1572 of the Protestant leader Admiral de Coligny,
giving rise to protests which led to the massacre of Saint Bartholomew (August 24th of the
same year); later in 1588 of Henri, duke of Guise, a Catholic leader whose murder was ordered
by the king, Henri III; and in 1589 of King Henri III himself, murdered by a monk.

born in the New World, their customs, beliefs and language would have been
radically different.
Therefore, we should be open to new experiences, and language shows us
that the possibilities of meaning are innumerable. In the spirit of the ancient
skeptics with their skeptiké agogé, philosophy should be considered more as a
way of conduct than as a set of doctrines or theories. Moderation, reflection
and self-examination are thus essential to the philosopher’s attitude. The scales
in perfect balance were Montaigne’s symbol.
It is also highly significant that he chose, according to the “Apologie”, as
his motto a question: Que sais-je? [What do I know?].


Annas, Julia & Jonathan Barnes. 1985. The Modes of Scepticism. Cambridge:
Cambridge University Press.
Bernand, Carmen & Serge Gurzinski. 1991. Histoire du nouveau monde. Paris:
Berven, Dikka. 1995. Language and Meaning: Word Study in Montaigne’s
Essais. New York: Garland.
Buzon, F. de. 1992. “L’homme et le langage chez Montaigne et Descartes”.
Revue Philosophique de la France et de l’Etranger 4.451-466.
Giocanti, Sylvia. 2001. Penser l’irresolution: Montaigne, Pascal, La Mothe Le
Vayer: trois intinéraires scéptiques. Paris: H.Champion.
Lévi-Strauss, Claude. 1991. Histoire de lynx. Paris: Plon.
Levine, Alan. 1999. Early Modern Skepticism and the Origins of Toleration.
New York: Lexington Books.
Mathieu-Castellani, Gisèle. 1988. Montaigne et l’écriture de l’essai. Paris:
Presses Universitaires de France.
Montaigne, Michel de. 1952. The Essays. (=Great Books of the Western World,
25). Translated by Charles Cotton. Chicago: Encyclopaedia Brittanica.
Montaigne, Michel de. 1967. Apologie de Raymond Sebond. Paris: Gallimard.
Montaigne, Michel de. 2002. Les essais, texte en français moderne établi par
Claude Pinganaud. Paris: Arléa.
Popkin, Richard H. 2003. The History of Scepticism from Savonarola to Bayle.
Oxford: Oxford Univ. Press.
Rendall, Steve. 1992. Distinguo: Reading Montaigne Differently. Oxford:
Clarendon Press.
Villey, Pierre. 1908. Les sources et l’évolution des Essais de Montaigne. Paris:
Wlaswo, Richard. 1987. Language and Meaning in the Renaissance. Princeton
University Press.
Wittgenstein, Ludwig. 1953. Philosophical Investigations. Oxford: Blackwell.

California Institute of Integral Studies


Many 17th-century philosophers, theologians, and educators were engaged

in developing a universal language to remedy the confusion caused by the
multiplicity of languages. George Dalgarno (c.1619-1687), who published Ars
Signorum (The Art of Signs) in 1661, and John Wilkins (1614-1672), who
published An Essay Toward a Real Character, and a Philosophical Language
in 1668 emerged as the two leading theorists and practitioners of the universal
language movement in England. Early on in their work, Dalgarno and Wilkins
collaborated; but soon after, they parted ways and worked separately. In
Dalgarno’s “Treatise”, evidence emerges that educational reform may well
have been the critical issue which eventually divided Dalgarno and Wilkins.
While the scholarly debate about who was the more original language designer
is interesting, the most relevant point about the Dalgarno and Wilkins dispute
for the history of the 17th-century universal language movement is that it sheds
light on the failure of the movement. In Dalgarno’s and Wilkins’ unresolved
struggle as reflected in their critiques of each other, an explanation emerges
why a movement that drew significant scholarly attention, even a sense of
urgency, in the early to mid 17th century had dissolved, for the most part, by
the end of that century.

Many prominent 17th century philosophers, theologians, and educators

strongly advocated for a universal language that could overcome the confusion
caused by the multiplicity of languages – that is, the curse of the biblical
Tower of Babel (for an informative study of the theological issues in the
universal language movement, see Poole 2003). Among these advocates were
Francis Bacon (1561-1626), René Descartes (1596-1650), Marin Mersenne
(1588-1648), and John Amos Comenius (1592-1670). The two leading
theorists and practitioners of the universal language movement in England
were George Dalgarno (c.1619-1687) who published Ars Signorum (The Art of

Signs) in 1661 and John Wilkins (1614-1672) who published An Essay

Toward a Real Character, and a Philosophical Language in 1668. Early in the
conceptual formations of their universal language projects, Dalgarno and
Wilkins were supportive colleagues, even collaborators, until, rather abruptly,
they discontinued their working relationship.
While David Cram drew attention to Dalgarno’s “The Autobiographical
Treatise” in 1980, the “Treatise” was not published until 2001 when he and
Jaap Matt included its publication in their excellent study, George Dalgarno
on Universal Language. As Cram and Maat (2001:26) have pointed out: “Until
recently, very little was known about the joint efforts of Dalgarno and Wilkins
towards the construction of a universal language”. Before the publication of
the “Treatise”, scholars have been left to conjecture regarding Dalgarno’s and
Wilkins’ working relationship and ultimate parting of ways. Dalgarno’s
detailed account of his initial collaboration and enduring dispute with Wilkins,
albeit one-sided, provides a balanced perspective regarding their working
relationship as well as significant insight regarding the 17th century
philosophical language movement.
In this paper, I will focus on the “Treatise”, accepting the well-documented
conclusion that the contention between Dalgarno and Wilkins was due to
incompatible goals and was aggravated by the fact that each passionately
believed that his, and only his, approach should be pursued. As we shall see,
the relevance of a universal language to educational reform was a critical issue
which eventually divided Dalgarno and Wilkins. Unquestionably, each
passionately believed that he could not compromise his approach without
jeopardizing the integrity of his project. While the scholarly debate about who
was the more original language designer is interesting, I believe what is most
relevant about the Dalgarno and Wilkins dispute for our understanding of the
history of the 17th century universal language movement is the light it sheds on
the failure of the movement. In Dalgarno’s and Wilkins’ unresolved struggle as
reflected in their critiques of each other, we see an explanation of why a
movement that drew significant scholarly attention, even a sense of urgency, in
the early to mid 17th century had dissolved, for the most part, by the end of that
century (see Salmon 1974, Cram 1985, and Lewis 2001 for a history of the
flawed attempts to revive the Essay at the end of the 17th century).
We find in Dalgarno’s and Wilkins’ critiques of each other an insightful
articulation of the central dilemma of the 17th century universal language
movement. In short, what Dalgarno intended to achieve with his language –
ease of learning the language – was for Wilkins inadequate for achieving the
grand scheme of a philosophical language; and what Wilkins intended to
accomplish with his language – the classification of reality – was for Dalgarno
impossible to learn and unfeasible in its aspiration to include all knowledge. In
the final analysis, Dalgarno believed that Wilkins was attempting to do too
much, and Wilkins believed that Dalgarno was attempting to do too little.

Paradoxically, while Dalgarno and Wilkins held opposing views, both were
correct. As Cram (1980) has summed up the situation, the primary priority for
Dalgarno was the secondary priority for Wilkins and vice versa.
In the narrative of the “Treatise”, Dalgarno confirmed that his collaborative
relationship with Wilkins began when he was introduced to John Owen (1616-
1683) who was Vice Chancellor at Oxford from 1652 to 1658. After reviewing
Dalgarno’s tables, Owen acknowledged that they were the bases for a universal
language. Accordingly, he offered to arrange a meeting for Dalgarno with Seth
Ward (c.1600-1662) “whose judgment he would trust” (Cram and Maat
2001:356). In fact, it was Ward who suggested to Wilkins that he work on a
universal language. As promised, Dalgarno met Ward; Ward was impressed
with the quality of Dalgarno’s work, but he criticized the limited range of his
tables. Nonetheless, Ward offered to introduce Dalgarno to Wilkins.
Dalgarno’s first reaction to the possibility of meeting Wilkins was delight –
in fact, he described the prospect as “incredible” (ibid. 356). However, he
added: “But afterwards I knew by my own experience that it was nothing more
than deserved” (ibid. 356). Dalgarno initially met Wilkins while Wilkins was
strolling with Ward in Oxford. Wilkins told Dalgarno that he was intrigued by
his papers; he invited Dalgarno to Wadham College, where he was Warden of
the College. Dalgarno gratefully accepted Wilkins’ invitation, and he was well
received by Wilkins at Wadham.
Of their early meetings, Dalgarno asserted that he greatly appreciated
Wilkins’ approach to producing a universal language. He wrote: “But of all my
scholars Mr. Warden was the greatest proficient, for out of his singular
curiosity he took more pains than they to understand me [Note that due to the
intense nature of their disagreement, Dalgarno never mentioned Wilkins by
name in his “Treatise” nor did Wilkins mention Dalgarno by name in his
Essay]” (ibid. 357). After considerable discussion with Wilkins, Dalgarno
indicated that Wilkins, like Ward, “began to suggest that my tables were not
full, more particularly defective in the predicament of substance, having made
no provision for the species of natural bodies” (ibid. 357).
Dalgarno’s response to Wilkins’ criticism brings us to the core of their
dispute. Dalgarno wrote:
But when I had perused it and perceived that it quite destroyed the principles that I
had laid down to myself to be guided by, specially it did thwart that which was my
grand principle, to find out the shortest, easiest and distinctest medium of
communication for the common use of all mankind, I returned it, excusing myself that
I reverenced his judgment but could not go against the dictates of my own. (ibid. 358)

Notable is Dalgarno’s emphasis on “the shortest, easiest and distinctest

medium of communication for the common use of all mankind…” The
growing intensity of the disagreement is reflected in Dalgarno’s resolve to
keep “within the bounds of that which I called the body of a language and not

to meddle…which would swell my tables to such a bulk as to discourage any

to learn it” (ibid. 366). In this statement, Dalgarno clearly set his priorities:
ease of learning a language superseded its expression of all knowledge.
Wilkins countered Dalgarno’s approach by maintaining that persisting in
Dalgarno’s plan would produce a “lame design” (ibid. 366). Dalgarno
described in detail the ensuing debate between Wilkins and himself – each not
giving ground to the other. In the words of Dalgarno, Wilkins concluded:
That a philosophical language should contain a regular enumeration according to the
received theory of all the notions of nature and art, that so the lexicon of radicals
might be a synopsis or index of the whole encyclopedia of arts digested in a
predicament series. (ibid. 366)

In Dalgarno’s statement describing Wilkins’ formulation of his primary

goal, as throughout the “Treatise”, we see that Dalgarno was balanced and
accurate in his reporting events and sentiments. For example, compare
Dalgarno’s statement to that of Wilkins in the Essay:
The Second Part [of the Essay] shall contain that which is the great foundation of the
thing here designed, namely a regular enumeration and description of all things and
notions, to which marks or names ought to be assigned according to their respective
natures, which may be styled the scientifical part comprehending universal
philosophy. It being the proper end and design of the several branches of philosophy
to reduce all things and notions unto such a frame, as may express their natural order,
dependence, and relations [italics are those of Wilkins]. (Wilkins 1668:1)

Dalgarno pointed out that he was certain that the global precision that
Wilkins sought would not only be impossible for Wilkins to achieve, but also
for any person or group to achieve or even to persuade others that it could be
achieved. Dalgarno reported that he and Wilkins “argued the case often but
without conviction on either side” (Cram and Maat 2001:358). Dalgarno
Hence followed a perfect and declared breach of judgment, though not of friendship,
for I conversed with him a long time after this, sometimes personally and sometimes
by letters, as familiar as ever before. (ibid. 358)

Lest we think this breach was amiable, let us look at a letter that Dalgarno
wrote to Samuel Hartlib in 1659:
[…] Dr Wilkins, who, if he publish anything he deals neither ingenuously nor justly
with me and I fear if he attempt anything on this subject [philosophical language] he
shall have small credit of it for besides that all he will do will be to discover another
man’s labors, this chiefly will reflect that (as I am very confident) he will do it with
several material defects for albeit I did for almost a year’s time by daily converse with
him open up the whole mystery so far as the vein of my invention has then gone yet

since the time of my retirement I have discovered advantages which I persuade

myself have not fallen under his consideration. (ibid. 426)

After summarizing his work with Wilkins leading up to the “breach,”

Dalgarno moved his attention “to compare the method of Ars Signorum with
that of the Essay” (ibid. 364). He noted that Wilkins wanted radical characters
for each species, and Dalgarno planned to “keep within the bounds” (ibid. 366)
of what he thought appropriate for a natural language. As already quoted,
Dalgarno bluntly stated that he did not want to “swell [his] tables to such a
bulk as to discourage any to learn it” (ibid. 366). Cram and Maat insightfully
assessed that “[…] Wilkins’s fundamental mistake, in Dalgarno’s opinion, was
to attempt to force two incompatible designs into one – that is, to serve the
purposes of scientific classification and those of linguistic practice at the same
time” (ibid. 29).
While Cram and Maat acknowledged the dilemma of the universal
language movement, a dilemma that eventually led to its demise, they excluded
discussion of a significant goal of the universal language movement: the
reform of education in 17th century England, particularly university education.
Dalgarno, a schoolmaster, referred to this reform of education as he wrote that
a benefit of universal language would be “perfecting and propagating humane
learning” (ibid. 364); however, he never elaborated on this possibility. In Cram
and Maat’s extensive study, “education” does not appear in the index. On the
other hand, for Wilkins, education was the raison d’être for his philosophical
language as he stated in the opening pages of his Essay:
[…] if these marks or notes [of his universal language] could be so contrived, as to
have such a dependence upon, and relation to one another, as might be suitable to the
nature of things and notions which they represented […] The best way of helping the
memory by natural method, the understanding likewise would be highly improved;
and we should, by learning the character and the names of things, be instructed
likewise in their natures, the knowledge of both which ought to be conjoined.
(Wilkins 1668:21)

Wilkins’ philosophical language was based on the notion that “knowledge

of [language and nature]…ought to be conjoined”, and this “conjoining” was
the defining intention and goal of his universal language. He insisted that as
one learned the words of his language, one also learned the nature of the
referents of these words. To appreciate the role that Wilkins’ philosophical
language played in advancing educational reform, it is necessary to consider
his aspiration for the Essay. Wilkins was among those in the science revolution
in 17th century England who held a controversial view that science, education,
and philosophical language were interrelated. Starting with Bacon, through
Comenius, and culminating in the debate between John Webster (1610-1682)
and Ward in 1654, a new model of education was growing out of the expansion
of 17th century empiricism. Wilkins challenged not only the content of the

traditional university curriculum, but also the prevailing styles of teaching and
In 17th century England, the challenge presented by Bacon, Comenius, and
the leaders of the new science to those holding traditional assumptions
regarding knowledge and learning is well documented in the various calls for
reform in higher education (see Subbiondo 2001). For the most part, those
most supportive of integrating the new science into the university’s curriculum
were the same people who collaborated with Wilkins in establishing the Royal
Society and in responding to the Society’s first major initiative, the
development of a philosophical language. The founding members of the Royal
Society held that the primary goal of the Society would be the promotion of
science and science education, and they believed the Society’s philosophical
language project was inextricably linked to the advancement of the new
science and education reform. While they recognized that a philosophical
language would facilitate international communication, they asserted that it
needed to be more than a 17th century Esperanto. The founders of the Royal
Society desired to develop a scientific language in which the elements of each
word were analogous to the terms of an algebraic formula. In short, in learning
the word, one learned the constituent parts of its referent.
In no other 17th century text has the argument against the interrelationship
of the new science, educational reform, and philosophical language been better
documented than in Webster’s critique of science in the universities,
Academiarum Examen, or the Examination of the Academies (1654). In
Vindiciae Academiarum (1654), Wilkins and Ward rebutted Webster by
arguing about several issues pertaining to higher education; however, their
views that philosophical language could improve scientific education are most
relevant to this study.
In the preliminary pages of the Essay, Wilkins expressed his gratitude to
Ward for providing him with the structural framework for a philosophical
language that would be “useful” and “feasible”. He acknowledged the efforts
in Italy and France to compile a dictionary, but he considered them “far
enough from being finished”. He wrote:
But for all such attempts to this purpose, which he [Ward] had either seen or heard of,
the authors of them did generally mistake in their first foundations; whilst they did
propose to themselves the framing of such a character, from a dictionary of words,
according to some particular language, without reference to the nature of things, and
that common notion of them, wherein mankind agree […]. (Wilkins 1668: “To the

Wilkins maintained that the natural tables that formed the basis of his
language served the following objectives: they provided a comprehensive
inventory of extant knowledge, offered a design to “promote and facilitate the
knowledge of nature”; presented an order and classification of knowledge;

served as “the most useful repository in the world” of scientific knowledge;

advanced the goals of the Royal Society; and reflected a scientific method.
Also, Wilkins emphasized that his language echoed the motto of the Royal
Society – nullius in verba (nothing in words) as he stressed. “As things are
better than words, as real knowledge is beyond elegancy of speech” (Wilkins
1668: “To the Reader”). Thus, In Wilkins’ and Ward’s view, Dalgarno’s tables
and language did not meet the requirements of a language that could advance
learning. Educational reform was so important to Wilkins that he could not
work with anyone who did not share his goal.
In the “Treatise”, Dalgarno reported the depressing consequences of the
“breach” on his career: he noted that “both scholars and benefactors fell off”
(Cram and Maat 2001:358) from their support of him and his project. He
wrote: “After this breach was noised abroad I found that all my former and
great hopes and expectations for a reward of my labors were almost quite
defeated” (ibid. 358). Unfairly, Dalgarno found himself dismissed as a credible
inventor of a universal language. For example, Owen, who was the first
significant person at Oxford to pledge his support of Dalgarno, wrote about
Wilkins’ work without even mentioning the efforts of Dalgarno.
Despite the lifelong struggle and the many disappointments, Dalgarno
never wavered in his commitment to his project as the following statement will
attest: “I cannot discern any considerable blemish in Ars Signorum, but remain
still fixed in the same judgment, whereof I was then” (ibid. 364).
Maat, in his excellent Philosophical Languages in the Seventeenth
Century: Dalgarno, Wilkins, Leibniz, published in 2004, judiciously cautions
[…] much of the literature on seventeenth century artificial languages has focused on
Wilkins on the assumption that his scheme, besides being the largest and the most
complete one of all, was exemplary for the whole intellectual movement it was a part
of. But in fact, the Essay represented one of several alternatives which were
considered and debated within that movement rather than the culmination of a
unidirectional development. The contrary assumption has not only led to
misrepresentations of Dalgarno’s views, but also to a flawed picture of the movement
as a whole, which overlooks the diversity of the schemes involved and underestimates
the sophistication of the views underlying them. (Maat 2004:144)

In order to appreciate the Dalgarno and Wilkins debate, it is important to

realize that Dalgarno and Wilkins were presenting two competing universal
languages – it is not so much a case that one language was better than the other
as each had its particular strengths and weaknesses. In the end, the weaknesses
of both languages reflected the weaknesses of the 17th century universal
language movement – the impossibility of including all knowledge and the
difficulty of learning an artificial language – and help explain the demise of
this enterprise. Also, the fact that Wilkins’ aspiration of improving education

through universal language never materialized was a loss for the 17th century
educational reformers.


Cram, David. 1980. “George Dalgarno on Ars Signorum and Wilkins’ Essay”.
Progress in Linguistic Historiography: Papers from the International
Conference on the History of Linguistic Sciences ed. by E. Koerner, 113-
121. Amsterdam: John Benjamins.
Cram, David. 1985. “Universal Language Schemes in Seventeenth-Century
Britain”. Histoire Epistemologie Langage 7.35-44.
Cram, David and Jaap Maat. 2001. George Dalgarno on Universal Language.
Oxford: Oxford University Press.
Lewis, Rhodrie. 2001. “The Efforts of the Aubrey Correspondence Group to
Revise John Wilkins’ Essay (1668) and their Context”. Historiographia
Linguistica 28.331-364.
Maat, Jaap. 2004. Philosophical Languages in the Seventeenth Century:
Dalgarno, Wilkins, Leibniz. (=New Synthese Historical Library, 54.)
Dordrecht etc.: Kluwer.
Poole, William. 2003. “The Divine and the Grammarian: Theological Disputes
in the 17th Century Universal Language Movement”. Historiographia
Linguistica 30.273-300.
Salmon, Vivian. 1974. “John Wilkins’ Essay (1668): Critics and
Continuators”. Historiographia Linguistica 1.147-163.
Subbiondo, Joseph L. 2001. “Educational Reform in Seventeenth-Century
England and John Wilkins’ Philosophical Language”. Language and
Communication 21.273-284.
Ward, Samuel & John Wilkins. 1654. Vindiciae Academiarum. Oxford:
Thomas Robinson. Facsimile reprint in A. Debus, Science and Education
in the Seventeenth Century: The Webster-Ward Debate. London:
Webster, J. 1654. Academiarum Examen, or the Examination of the
Academies. London: Giles Calvers. Facsimile reprint in A. Debus, Science
and Education in the Seventeenth Century: The Webster-Ward Debate.
London: Macdonald.
Wilkins, J., 1668. An Essay Towards a Real Character, and a Philosophical
Language. London: Gelibrand and Martyn.

Université de la Sorbonne nouvelle / Paris III


The object of this article is to present one aspect of the development of the
theory of sounds during the 17th and 18th centuries, after a discussion of the
hypotheses put forth by i) Sylvain Auroux, who believes that during this
period, the letter played a regulating role analogous to that of the phoneme
among modern linguists, and ii) Daniel Droixhe, who suggests that certain
authors (such as those of the Port-Royal grammar) were guided by a “sense of
the phoneme”. The important issue in this debate is the way in which the
authors of this period conceived the units of sound in the language, their status,
and the criteria for defining them. The article examines these questions using a
corpus of texts from the 17th and 18th centuries, tackling three problems
characteristic of phonetic descriptions during that time: the “discovery” of
nasal vowels, that of semivowels, and the identification of the different values
of the letter e in French.

Cet article se veut une contribution à l’histoire de la notion d’unité sonore

telle qu’elle se bâtit progressivement dans les grammaires françaises de l’âge
classique. Les grammairiens se sont livrés à partir du 16e siècle à un travail de
description phonétique qui a enrichi progressivement l’inventaire des sons du
français et les a conduits à reconnaître un certain nombre d’unités restées
inaperçues. Ce travail se poursuit tout au long du 17e et pendant une partie du
18e siècle. De nombreux textes de cette série présentent donc des
développements hautement intéressants d’un point de vue historique et
épistémologique: ils argumentent en faveur de la découverte d’un son, ou
discutent la pertinence des arguments allégués en faveur de découvertes
supposées. L’instabilité des listes de sons proposés par les auteurs apparaît en
effet comme un trait caractéristique des descriptions phonétiques de l’âge
classique. Un des cas les plus discutés est celui des voyelles nasales, mais il en
est d’autres, comme celui des semi-voyelles [j], [w], [], ou du nombre des e
du français. Nous allons examiner successivement ces trois exemples de

découverte, en portant essentiellement notre attention sur les arguments

proposés en vue d’établir l’existence empirique d’une unité sonore que la
perception immédiate n’a pas suffi à rendre manifeste.
Deux thèses ont été proposées pour rendre compte des difficultés que les
auteurs rencontrent, du fait qu’ils ne disposent pas du concept de phonème
pour réguler le nombre des unités, et distinguer entre ce qui constitue le
matériel stable mis en œuvre par la langue et l’infinie variété des
prononciations. Nous les rappelons brièvement:

• Celle que Sylvain Auroux développe dans plusieurs textes (1973, 1979,
1992) consiste à montrer que la phonétique de l’âge classique se développe
indépendamment des autres domaines d’étude du langage, et que n’aura jamais
lieu, au cours de la période, l’articulation de la description phonétique et d’une
sémiotique. Ainsi, en dépit de la mise en œuvre fréquente dans les descriptions
de paires minimales, jeune / jeûne; côte / cote etc., les analyses n’aboutissent
jamais à la notion d’unité distinctive. Par ailleurs, au cours du 18e siècle, la
description du système vocalique s’appuie sur ce qui cesse d’être une simple
métaphore et devient un véritable modèle acoustique, supposé rendre compte
de l’émission de la voix humaine: l’identification de l’appareil vocal à un
instrument à vent, comme une flûte (Cordemoy 1668, Boulliette 1760, Boindin
1753, De Brosses 1765...), ou à une corde en vibration. Ce modèle constitue en
lui-même un obstacle à l’analyse des systèmes linguistiques. Il conduit en effet
à distinguer les voyelles qui sont de véritables sons, constituant un continuum
unidimensionnel au sein duquel on peut potentiellement identifier une infinité
de positions, et les consonnes, qui ne sont que des articulations ne pouvant être
réalisées sans les premières. Les arguments et les faits en faveur de cette thèse
sur les représentations du système vocalique à l’âge classique sont nombreux.
La description des voyelles sous la forme d’une série, ordonnée selon le plus
ou le moins à partir d’un seul critère (degré d’ouverture, ou hauteur) apparaît
très tôt. On en trouve une première formulation chez Meigret (1550) qui donne
le premier classement linéaire des voyelles du français, et l’idée que la série
dans son ensemble est produite par la plus ou moins grande ouverture de la
bouche est formulée explicitement par la grammaire de Port-Royal. La
conséquence produite par la mise en œuvre de ce modèle d’analyse du
vocalisme selon Sylvain Auroux est dans l’impossibilité de penser la voyelle
comme une unité. Dès lors, c’est en quelque sorte la lettre qui joue ce rôle. Les
lettres voyelles pour les linguistes de l’âge classique sont des êtres naturels,
légués par la tradition, susceptibles de variations ou prononciations
nombreuses, et jouant un rôle fonctionnel analogue au phonème dans les
descriptions modernes.
• Daniel Droixhe (1977) a proposé une autre interprétation de l’approche
de la notion d’unité sonore dans les textes de l’époque classique: les auteurs

seraient parfois guidés par un “sentiment du phonème”, une conscience en

quelque sorte épiphonologique, analogue à celle qui est à l’œuvre dans
l’invention des alphabets. C’est ce “sentiment du phonème” qui conduit ainsi
les Messieurs de Port-Royal à ne pas parler des voyelles nasales dans le
chapitre qu’ils consacrent aux voyelles, et à décrire l’opposition des deux o du
français comme une opposition d’ouverture, une opposition de timbre donc, et
non une opposition de longueur, anticipant sur une évolution qui n’est pas
encore fonctionnelle dans le français du milieu du 17e siècle selon André
Martinet (1946).

Nous allons discuter ces deux thèses à partir de l’histoire de trois

problèmes de description: la découverte des voyelles nasales, celle des semi-
consonnes, et le problème de la description des différents e. 1

1. La découverte des voyelles nasales

Si donc les Messieurs omettent de parler des voyelles nasales, selon D.
Droixhe, c’est que celles-ci n’ont pas acquis encore en 1660 un statut
phonématique. Ce ne serait le cas que lorsque Dangeau identifie la série
complète dans son Mémoire sur les voyelles de 1694 contenu dans ses Essais
de grammaire. Pour bien mesurer la portée de cette thèse, il faut avoir à l’esprit
que de nombreux textes grammaticaux du 16e siècle proposent déjà une
description des voyelles nasales. Dangeau ne découvre donc pas de nouveaux
sons, il consacre seulement leur statut d’unité dans le système vocalique du
français. C’est donc que pendant la période précédente tel n’était pas le cas, et
qu’il s’agissait de simples variantes combinatoires de la séquence voyelle +
consonne nasale. Autrement dit, pour que les voyelles nasales constituent des
phonèmes, il faut, comme l’écrit Martinet, “que la consonne nasale qui est la
source du phénomène (de nasalisation) ait complètement disparu” (Martinet
1965:145). L’affirmation de la thèse d’un sentiment du phonème qui guiderait
les analyses de Port-Royal conduit donc Daniel Droixhe a en proposer une
autre dans le domaine de l’histoire du français et à modifier la chronologie
généralement admise du phénomène de la nasalisation des voyelles: en 1660,
les voyelles nasales ne sont pas encore des phonèmes, donc la double
articulation des voyelles nasales n’a pas encore disparu. Le mécanisme dans
son ensemble est décrit alors selon le synopsis suivant:

Nous prenons le parti ici d’utiliser la lettre comme moyen de notation, ce qui correspond au
fond à la façon dont la question est conçue au cours de cette période. Le problème posé par les
grammairiens est en effet celui de l’identification des différents sons que note la lettre. Un peu
plus loin, il nous faudra distinguer parmi eux entre la série des é (ouvert, fermé, long, bref,
etc...), que nous noterons de cette façon, et le e muet.

15e siècle chute du ə final => peizãnə —> peizãn

la différence morphologique entre le masculin et le féminin est menacée:
(masc) peizãn vs (fem) peizãn

16e siècle thérapeutique quantitative en vu de revitaliser l’opposition qui tend à se

masc: peizã:n vs fem: peizãn

17e siècle diphtongaison: (masc) peizã:n —> peizãun

fin 17e nasalisation phonologique vs dénasalisation de la voyelle devant consonne


début 18e s. peizã vs peizan

Nous voudrions faire ici deux observations, de portées très différentes.

La première est factuelle. Les plus anciennes descriptions de la nasalisation
des voyelles en français remontent aux premières grammaires du français
(Palsgrave 1530); mais il est sans doute raisonnable de penser que la
systématicité de la description de Dangeau a pour préalable leur accession à un
statut pleinement phonématique. Néanmoins, on dispose de descriptions assez
précoces de l’abandon de l’articulation de la consonne suivant une voyelle
nasale. C’est indiscutablement le cas dans ce passage de la grammaire de
Maupas (1607): m “à la fin des syllabes ne s’exprime que foiblement & sans
serrer les lèvres l’une contre l’autre, ainsi Nom, renom, faim”, de même “N,
finissant la syllabe ne s’exprime non plus, quoy que ce soit sans faire toucher
le bout de la langue contre le palais de la bouche, ton bon conseil Henri”
(Maupas 1607:18). De même, Chifflet semble reconnaître l’existence de
voyelles comportant un trait nasal en tant que telles. Il les identifie comme des
sons simples, quoiqu’il se serve de la notion de diphtongue pour les
appréhender, et les définit comme des “Diphtongues qui se forment d’un seul
son, comme, EU, OU, AI”. Ainsi on réalise “les sons de An, En, In, On, Un,
sans prononcer l’n comme ferraient de vraies diphtongues” (Chifflet
1659:172). 2
La seconde de mes remarques concerne le type d’argumentation utilisé par
les découvreurs des voyelles nasales. Celles que nous venons de citer
proposent une analyse (très fine et très juste au fond) de l’articulation, et ne
visent pas la description acoustique d’un son vocalique nasalisé (ce que tente à
l’inverse Palsgrave (1530), qui décrit le passage du flux d’air par le nez, et
l’impression produite d’un son qui comporte “some thyng in the noose”), mais
s’efforcent plutôt de distinguer la voyelle nasale et la réalisation d’une
séquence voyelle + consonne nasale. C’est ce qui rend d’autant plus fiable le
témoignage de Maupas sur la chute de la consonne après la voyelle nasalisée

Voir sur ce point Yves-Charles Morin (1994), et E. Coseriu (1994).

en syllabe fermée dès le début du 17e siècle. Mais ce qui fait sans doute tout le
poids de l’analyse de Dangeau, en qui les contemporains ont vu le véritable
inventeur des voyelles nasales, est qu’il argumente sur un autre terrain. Il quitte
en effet le domaine de l’analyse articulatoire ou acoustique et donne trois
arguments reposant sur la prise en compte du contexte:

• L’argument du port de voix:

Il est impossible de fredonner sur les deux voyelles d’une diphtongue. Si vous voulez
fredonner sur le mot depuis, tout votre port de voix se fera sur l’i, et point du tout sur
l’u […]. Quand vous voudrez fredonner sur tyrans, biens, profonds, communs, tout
votre port de voix se fera sur an, en, on, un. Si le son de an était composé de la
voyelle a, et de la consonne n, votre port de voix se ferait sur l’a, et vous
commenceriez à prononcer l’n, quand votre port de voix viendrait à finir: c’est ce qui
n’arrive pas quand vous fredonnez sur tyrans, et votre port de voix est tout sur le son
de an, preuve certaine que ce son n’est pas composé. (Dangeau 1694:18)

• S’il s’agit de véritables voyelles (simples), alors elles sont susceptibles de

se trouver en poésie en position d’hiatus comme les autres. Celles-ci étant
ignorées jusque là en tant que telles, de tels hiatus se trouvent nécessairement
chez tous les auteurs, qui, composant généralement “la plume à la main”, et
non en se fiant à l’oreille, n’évitent pas ces sortes de défauts, parce qu’ils
prennent les voyelles nasales pour des diphtongues voyelles + consonnes
nasales. Si ce raisonnement est juste, on doit en trouver la confirmation chez
trois sortes de poètes:

i) les poètes d’origine normande. Le parler normand, en effet, dénasalise

les voyelles nasales et restitue un N consonne après la voyelle. Dans ce cas,
l’hiatus disparaît de lui-même. Donc, un poète normand comme Corneille
devrait commettre plus fréquemment cette faute de prosodie qu’un autre. Le
dépouillement des tragédies de Corneille comparé à celui des ouvrages de
Racine confirme la plus grande fréquence chez le premier des hiatus provoqués
par la rencontre de deux voyelles dont la première est une voyelle nasale.
Dangeau peut écrire qu’il est “assez content de voir son raisonnement confirmé
par l’expérience” (Dangeau 1694:24).
ii) les poètes qui sont en même temps des acteurs, et qui ont donc
instinctivement un plus grand souci des contraintes de la diction. La rareté des
hiatus avec voyelle nasale dans le Misanthrope de Molière confirme également
ce point.
iii) les poètes qui composent pour l’opéra doivent éviter plus
soigneusement encore ces rencontres (quoique instinctivement toujours) du fait
de la nécessité d’éviter ces “baillements” dans le chant. C’est encore ce que
confirme la lecture des pièces de Quinault où elles sont exceptionnelles.

• Enfin Dangeau se réfère à ce qui lui paraît une règle générale relative à la
prononciation des consonnes finales: celle-ci est supprimée, “au moins dans la
conversation” (1694:15). Mais elle est restituée dès que se présente le risque
d’un hiatus. On peut ainsi distinguer nous parlons et nous appelons. C’est
exactement ce qui se passe avec le n de on: on parle / on appelle. Ce qui
“prouve” a contrario que devant consonne le n de on n’est pas prononcé (c’est-
à-dire, prononcé comme une consonne). Autrement dit, ‘on’ n’a pas la
prononciation d’une diphtongue, mais d’une voyelle simple.

2. La découverte des semi-consonnes: [j], [w], []

Buffier est le premier à définir le i de païen, ou de naïade comme une
consonne. Cette opinion est reprise par Boindin dans ses Remarques sur les
sons de la langue française (1753), par Duclos dans ses Remarques sur la
grammaire de Port-Royal (1754), et par Wailly (1754). Harduin combat cette
opinion dès 1757 dans ses Remarques diverses sur la prononciation et sur
l’orthographe, et fait le point avec une lucidité remarquable sur les enjeux
théoriques de cette discussion. La question touche au fondement des analyses
phonétiques de cette période, savoir la conceptualisation de la différence entre
consonne et voyelle. Nous l’avons rappelé plus haut: pour l’ensemble des
auteurs, seule la voyelle est un son véritable. La consonne n’est qu’une
articulation, ou mieux selon Harduin un “mouvement modificatif” de la
voyelle avec laquelle elle forme une syllabe, agissant comme, dit-il, en
reprenant et en aménageant la comparaison topique dont se servent presque
tous les auteurs, les coups de langue et non les doigts du flûtiste: “les voyelles
répondent aux tons divers causés par la diverse application des doigts sur les
trous de la flûte; et les consonnes répondent aux coups de langue qui précèdent
ces tons” (1757:20). Si l’on examine les suites de voyelles, on constate une
différence entre le cas où la première appartient à la série i, u, ou, et tous les
autres cas:

[…] quand il faut passer des sons a, e, eu, o, à quelque autre son, ce changement ne
peut s’opérer sans une certaine contrainte qui arrête la voix, et empêche le premier de
se joindre assez rapidement au second pour que leur union produise un son double.
(Harduin 1757:21)

Et un peu plus loin:

Les sons doubles ne pouvant donc commencer par a par eu ni par o, il n’y a de
véritables que ceux qui commencent par i, par u, & par ou. (1757:23)

Il précise dans un texte publié un peu plus tard l’analyse articulatoire de ces
diphtongues dans ces termes:

La célérité de la prononciation fait qu’on passe au son de la seconde voyelle (de la

diphtongue) lorsque celui de la première n’est, pour ainsi parler, qu’ébauché; en sorte
que le mécanisme d’où naît ce premier son semble se réduire un simple mouvement
modificatif [souligné par nous] du deuxième, et n’opérer qu’une simple articulation.
(Harduin 1760a:29)

L’argumentation tend ainsi à prouver que le premier élément de ces

diphtongues est bien une consonne, comme l’ont proposé Buffier et Boindin
(1709, 1753). Pourtant telle n’est pas la conclusion à laquelle se range Harduin
dans ce même texte et dans sa Dissertation sur les voyelles et sur les
consonnes: il faudrait aller contre la tradition qui reconnaît des diphtongues en
français, et abandonner cette notion elle-même.

Malgré tout ce que j’avais observé moi-même sur la manière de prononcer i, ou, u
dans les mots que je viens de citer, j’ai dit que ces lettres ne laissaient pas d’y garder
leur nature de voyelles, parce que j’avais pour moi quantité d’habiles grammairiens,
et que je ne pouvais pas me ranger au parti de Wallis, ni même adopter la maxime du
Père Buffier, et de MM. Boindin et Duclos, sans rejeter toutes les diphtongues.
(Harduin 1760b:18)

Enfin, cette étrange argumentation d’une lucidité paradoxale, est complétée

un peu plus loin par une observation sur la distribution contextuelle de ces
unités que Harduin refuse d’admettre tout en les définissant avec une plus
grande précision que ses adversaires:

Je pense qu’il faut nécessairement de deux choses l’une: ou que l’i tréma de naïade
soit maintenu dans son ancienne qualification de voyelle, ou que l’on ne reconnaisse
plus aucune diphtongue, et que l’alphabet [...] soit augmenté de trois consonnes;
savoir i, u et ou lorsque chacun de ces caractères est suivi d’une voyelle qui fait partie
de la même syllabe. (1760b:19)

Rassemblons les faits qui se dégagent de cette brève histoire. Le processus

de découverte de la série des semi-consonnes en tant qu’unités du système du
français s’appuie d’abord sur l’observation fine d’un processus articulatoire
jusque là mal perçu, la lettre constituant sans doute un obstacle à la
reconnaissance d’une consonne sous la graphie d’une voyelle. Mais la netteté
des analyses suffit à montrer que les auteurs ne paraissent pas outre mesure
limités dans leur investigation par le fait graphique en lui-même. La précision
de l’analyse articulatoire est assez efficace pour permettre de poser ou de
postuler des unités en quelque sorte contre la lettre. Il est remarquable, pour
finir, que les unités soient définies au bout du compte par deux types d’analyse
combinées: une analyse articulatoire. Elle est première et garantit l’existence
empirique du fait que le grammairien tente d’isoler. Elle est articulée in fine à
une analyse fonctionnelle, qui délimite la place de l’unité en système, et

pourrait être idéalement sanctionnée par une modification de l’alphabet. En ce

sens, la lettre est bien le concept régulateur de la phonétique de l’âge classique.

3. Le problème de la diversité des e du français

C’est un problème formulé dès les plus anciennes grammaires du français,
auquel les grammairiens s’intéressent pendant longtemps, et qui amène par
exemple Buffier à composer un traité exclusivement consacré à cette question
en 1714. Une des particularités du français est de donner à la lettre e, donnée
pour être toujours prononcée [e] en latin, diverses valeurs, typiquement
françaises donc, et dont l’inventaire ne laisse pas d’embarrasser les
grammairiens de l’époque classique. Certains en distinguent trois (Pillot 1550,
Masset 1606, Maupas 1607…), d’autres quatre (Oudin 1632, Chifflet 1659, et
Irson 1656), cinq (voir l’article consacré à la lettre E dans l’Encyclopédie qui
identifie outre le e muet, un e fermé et un e ouvert commun, un e plus ouvert et
un e très ouvert), ou même six (Boulliette, 1760, compte quatre nuances du e
ouvert, qui s’ajoutent au e fermé et au e muet). Les auteurs distinguent ces
différents e selon des critères articulatoires (plus ou moins grande ouverture)
ou acoustiques (caractère aigu ou grave). Je ne creuse pas davantage ce point
faute de place, l’histoire de ces distinctions est bien documentée (voir les
travaux de Liselotte Biedermann-Pasques) et nécessiterait un long
Deux faits sont ici particulièrement intéressants pour mon propos.
• On peut montrer que le raffinement des analyses successives tend à
produire une inflation des unités reconnues, de trois e (e muet compris), à cinq
voire six (Biedermann-Pasques 1992). Les problèmes discutés peuvent être
ramenés à la question de l’existence, entre le e ouvert et le e fermé, d’un e
moyen. C’est une thèse défendue dès le 16e siècle pour la première fois, selon
Adrien Millet (1933:33), par Etienne Pasquier dans une lettre adressée à
Ramus (tII; col. 58 C des Œuvres, Paris), puis par Harduin, Beauzée (1767),
Buffier, Duclos, d’Olivet, tandis que nombre d’auteurs du 17e siècle ne
l’identifient pas (Port-Royal, Masset, Maupas...). Pour étrange que cette
affirmation puisse paraître à un lecteur moderne (il n’est guère question de e
moyen dans les descriptions du français contemporain), il faut rappeler que
c’est un fait pris en compte par Martinet (1945) dans son étude de la
prononciation du français contemporain réalisée à partir d’une enquête
conduite dans un camp de prisonniers en 1941; il lui accorde d’ailleurs un
statut phonématique: les trois e du français permettraient dans certaines
variétés de distinguer piquait, piquet, et piqué.
La confusion et la diversité des analyses de ce champ de problèmes au
cours de la période que nous observons paraît donc ici à son comble: les unités
se distinguent-elles par le degré d’ouverture de la bouche (Buffier), par la
tonalité (Boindin, Duclos, Beauzée,), ou par la longueur (Chifflet), voire par

une combinaison de ces critères? On peut penser avec Adrien Millet que cette
histoire un peu enchevêtrée voit émerger progressivement la conscience de la
différence acoustique qui permet de distinguer les trois e de tête (ouvert grave),
tète (ouvert aigu), et vérité (fermé aigu). On retrouve en effet ces intuitions
acoustiques très imprécises notées avec une certaine stabilité tout au long du
18e siècle (Boulliette 1760, Féraud 1787, Fromant 1768) jusqu’au système des
voyelles décrit par Demandre (1769), dont Millet note l’étonnante proximité
avec celui auquel Rousselot aboutit à partir des premières analyses
expérimentales. Les exemples donnés sont fréquemment les mêmes: tempête /
trompette; tête / tète... On aurait donc, à partir du consensus des impressions
auditives établi dans la communauté des grammairiens au cours du 18e siècle,
la découverte de la tonalité, précédant de près d’un siècle l’invention de
l’appareillage expérimental adéquat.
•Mais tout aussi remarquable me paraît la contribution de Buffier à ce
débat compliqué. Dans son Traité de 1714, l’auteur de la Grammaire sur un
plan nouveau déplace le problème du terrain de l’analyse des perceptions
acoustiques ou articulatoires, sur celui de la distribution des unités, et formule
une série de lois de position, gouvernant la distribution des réalisations. La
conséquence immédiate et naturelle de cette analyse est de réduire le nombre
des unités: il n’y a plus qu’un seul e en français, dont on peut observer un
certain nombre de réalisations par l’effet d’une “mécanique naturelle”. En
l’occurrence, chaque fois que e est suivi dans la même syllabe d’une consonne,
les mouvements préparatoires à son articulation provoque l’ouverture de la
voyelle. Celle-ci est plus grande lorsque la consonne est un l ou un r, mais elle
existe également, à un moindre degré avec les autres consonnes. Dans les
syllabes ouvertes (en finale notamment) le e reste fermé. Il va sans dire que
Buffier ne traite pas tous les faits, ce qui est un peu gênant. Néanmoins, il me
paraît très remarquable que l’on ait là une analyse qui dispose des moyens
épistémologiques de sortir de l’impasse que constitue l’étude des impressions
auditives et articulatoires. Le résultat est le retour à une unité qui n’est autre
que la lettre (il reste un seul e, diversement accentué) qui, comme le remarquait
Sylvain Auroux, joue alors un rôle fonctionnel équivalent au phonème.
On voit donc que le basculement de l’étude des sons du plan phonétique à
ce que nous appelons phonologie, ne survient peut-être pas brutalement avec
l’invention du phonème, qui n’apparaît lui-même que lorsque se fait
l’articulation des descriptions phonétiques avec une sémiotique, mais se trouve
préparé tout au long de l’âge classique par ces études fonctionnelles, dans
lesquelles tout se passe comme si la notion de variante combinatoire avait
précédé celle de phonème.


Arnauld, A. & Lancelot, C. 1660. Grammaire générale et raisonnée, Paris: le

Petit; 1754, avec remarques de Duclos, Paris: Prault fils aîné; 1768,
contenant les Reflexions sur les fondements de l’art de parler pour servir
d’éclaircissement et de supplément à la Grammaire générale et raisonnée,
de l’abbé Fromant, Paris: Prault père.
Auroux, S. & Calvet, L.-J. 1973. “De la phonétique à l’apprentissage de la
lecture; la théorie des sons du langage au XVIIIe siècle”. La Linguistique
Auroux, S. 1979. La sémiotique des Encyclopédistes. Paris: Payot.
Auroux, S. 1992. “Note sur les progrès de la phonétique au 18e siècle”.
Histoire des idées linguistiques ed. par S. Auroux, tome 2, 598-606. Liège:
Biedermann-Pasques, Liselotte. 1992. Les grands courants orthographiques au
XVIIe siècle et la formation de l’orthographe moderne. Tübingen: Max
Niemeyer Verlag.
Boindin, Nicolas. 1753. Remarques sur les sons de la langue française. Paris:
Boulliette, abbé. 1760. Traité des sons de la langue française. Paris: Mérissant.
Beauzée, Nicolas. 1767. Grammaire Générale. 2 vols. Paris: Barbou.
Buffier, Le Père Claude. 1709. Grammaire française sur un plan nouveau.
Paris: N. Le Clerc et al.
Buffier, Le Père Claude. 1714. Grammaire française sur un plan nouveau...
nouvelle édition augmentée d’un Traité sur la prononciation des e différens
de la langue française... Paris: P. Witte.
Chifflet, R. P. Laurent. 1659. Essay d’une parfaite Grammaire de la langue
française. Paris: P. Mauger [Slatkine reprints, Genève, 1973].
Cordemoy, Géraud de. 1668. Discours physique de la parole. Paris: F.
Coseriu, Eugène. 1994. “La ‘découverte’ des voyelles nasales”.
Communication et cognition 27.27-110.
Dangeau, Abbé Louis de Courcillon. 1694. Essais de Grammaire contenus
dans trois lettres d’un académicien à un autre académicien. Paris: J.-B.
De Brosses, Charles. 1765. Traité de la formation méchanique des langues et
des principes physiques de l’étymologie. Paris: Saillant.
Demandre, A. 1769. Dictionnaire de l’élocution françoise. Paris: Lacombe.
Droixhe, Daniel. 1977. “Sentiment du phonème et omission des voyelles
nasales chez Port-Royal”. La Grammaire Générale des Modistes aux
Idéologues ed. par A. Joly et J. Stéfanini, 157-164. Lille: PUL.
Féraud, Jean-François, abbé. 1787. Dictionnaire critique de la langue
française. Marseille: J. Moss.

Harduin, Alexandre-Xavier. 1757. Remarques diverses sur la prononciation et

sur l’orthographe. Paris: Prault.
Harduin, Alexandre-Xavier. 1760a. Lettre à l’auteur du traité des sons de la
langue française par l’auteur des Remarques diverses sur la prononciation
et sur l’orthographe. Paris: Imprimé par l’auteur.
Harduin, Alexandre-Xavier. 1760b. Dissertation sur les voyelles et sur les
consonnes. Paris: Lambert.
Irson, Claude. 1656. Nouvelle méthode pour apprendre facilement les
principes et la pureté de la langue français. Paris: l’auteur [Slatkine
reprints, Genève, 1973].
Martinet, André. 1945. La prononciation du français contemporain. Genève-
Paris: Droz.
Martinet, André. 1946. “Note sur la phonologie du français vers 1700”. BSL
XLIII.13-23. Repris dans Le français sans fard, 155-167. Paris: Presses
universitaires de France, 1969.
Martinet, André. 1965. “Les voyelles nasales du français”. La Linguistique 2.
Repris dans Le français sans fard, 144-154. Paris: Presses universitaires de
France, 1969.
Masset, Jean. 1606. L’Exact et Tres-Facile Acheminement à la Langue
Françoise, Par Iean Masset. Mis en latin par le mesme autheur, pour le
soulagement des estrangers. Paris: David Douceur.
Maupas, Charles. 1607. Grammaire et syntaxe française. Orléans: O. Boynard
[Slatkine reprints, Genève, 1973].
Meigret, Louis. 1550. Le Tretté de la Grammere françoeze. Paris: C. Wechel
[Slatkine reprints, Genève, 1972] .
Millet, Adrien. 1933. Les grammairiens et la phonétique ou l’enseignement
des sons du français depuis le XVIe siècle jusqu’à nos jours. Paris: J.
Morin, Yves-Charles. 1994. “Quelques réflexions sur la formation des voyelles
nasales en français”. Communication et cognition 27.27-110.
Oudin, Antoine. 1632. Grammaire française rapportée au langage du temps.
Paris: A. de Sommaville [Slatkine reprints, Genève, 1972].
Palsgrave, John. 1530. Lesclaircissement de la langue française. Londres: R.
Pynson et J. Haukins.
Pasquier, Etienne. 1971. Œuvres. Genève: Slatkine.
Pillot, Jean. 1550. Gallicae linguae institutio, latine sermone conscripta. Paris:
Stephanus Grouleau [Slatkine reprints, Genève, 1972].
Wailly, Noël François. 1754. Principes généraux et particuliers de la langue
française. Cité d’après la 2e édition. Paris: Barbou.


Laboratoire d’Histoire des Théories Linguistiques
CNRS / Université Paris 7 / École Normale Supérieure Lettres et Sciences


The discovery of the works of Jean Macé has pushed back the publication
of the first known French “general and rational grammar” to 1651 for certain,
and perhaps as early as about 1635. The simplest way of realizing the
historian’s goal of accounting for certain “phenomena” is to organize them into
“series”. The Port-Royal Grammaire Générale et Raisonnée (1660), then, is
the point of convergence of three distinct series:
-The grammatization of vernaculars: the European movement to i)
rationalize vernaculars and ii) produce universal grammars. Macé is
situated in this movement, probably being the first French grammarian to
develop the subject.
-The movement to analyze Latin in search of “causes”: a “theoretical”
approach towards languages (Lancelot).
-The evolution of logic: the logic of ideas (Arnauld and Nicole)
The grammatization of the vernaculars is the empirical cause that brings about
the general grammar, not only as a problem, but as an intellectual and
pedagogical project. The causalist tradition supplies it with a descriptive plan;
the logic of ideas provides its theoretical foundation and its limits. Macé
belongs to only one of these three series; his text could not have had the same
impact as Lancelot’s.

Nous tenons à remercier B. Colombat et A. McKenna pour leur aide dans la recherche de
plusieurs documents.

Pendant très longtemps les historiens des sciences du langage 2 ont

considéré que la Grammaire générale et raisonnée de Port-Royal (désormais la
GGR-PR, 1660; Privilège de 1659) était l’origine du mouvement même de la
“grammaire générale”. 3 Ils ont également entériné l’idée selon laquelle la
recherche scientifique concernant ce domaine s’était rapidement éteinte au
début du XIXe siècle pour faire place à la grammaire comparée (Schwartz
1981), véritable avènement de la science des langues ou linguistique. Les
critiques contre la méthodologie de la grammaire générale, menée notamment
par les disciples de Humboldt (le célèbre texte de A.-F. Pott “Zur Geschichte
und Kritik dersogenannten Allgemeinen Grammatik”, 1863 4 ), et qui visaient
sans doute davantage les travaux idéalistes d’obédience kantienne, comme
ceux d’A.-F. Bernhardi (Schlieben-Lange & Weydt 1988) ou de J. S. Vater, 5
ont durablement lié la grammaire générale avec le rationalisme. Chez les
historiens, la vulgate prend donc la forme suivante: la grammaire générale, née
avec Port-Royal, est une discipline rationaliste (née du cartésianisme), a priori
et de peu de valeur empirique, par manque d’intérêt pour la facticité (ou
l’historicité) des langues; sous cette forme, elle a été le paradigme dominant
aux XVIIe et XVIIIe siècles 6 (pour un récapitulatif et une défense de cette
position, voir Droixhe 1978).
Or, la croissance des connaissances et l’affinement des modèles de
représentation historique 7 ont conduit:

Le “mythe port-royaliste” a son origine chez les premiers historiens français de la science
grammaticale: F. Thurot (“Discours préliminaire” à sa traduction française de l’Hermes de
J. Harris, 1794; réd. sous le titre Tableau des progrès de la science grammaticale, par A. Joly,
Bordeaux, 1970) et Dieudonné Thiébault (“Lettre à Monsieur Pinglin sur l’histoire de la
Science Grammaticale”, dans sa Grammaire philosophique, 1802; rééd. par D. Droixhe,
Stuttgart-Bad Cannsatt, 1977). Ce dernier n’hésite pas à périodiser la discipline, distinguant
l’état dans lequel cette science est restée “depuis son origine” et le développement qu’elle a
connu depuis “le premier ouvrage <la GGR-PR> […] dans lequel on ait donné à cette science
la forme qui pouvait le plus l’élever à la perfection” (II, p. 174).
Il ne faut pas confondre cette thèse avec celle de Dominicy 1984 qui voit dans les travaux de
Port-Royal (où la Logique de 1661 tient une part essentielle) la “naissance” de la grammaire
moderne. Par ailleurs, on notera la prudence de Donzé (1967) qui remarque que “l’histoire de
la grammaire générale reste à faire” et qu’au “début du XVIIe siècle on entendait par ce terme
tantôt l’étude de ce qui est commun à toutes les langues, tantôt celle des notions grammaticales
dans leurs rapports avec les catégories logiques” (l.c., p. 35). Donzé cependant soutient que PR
est l’origine du genre en France.
Zeitschrift für Philosophie und philosophische Kritik, N.F. XLIII, p. 102-41 et p. 185-245.
Versuch einer allgemeinen Sprachlehre, 1801.
Foucault, dès 1966, par la notion d’ “épistèmè de l’âge classique”, avait donné une
consistance théorique à cette périodisation: le langage pour l’âge classique est élidé dans son
existence, “seul subsiste son fonctionnement dans la représentation” (l.c.:95).
Nous laissons de côté ce qui concerne la prétendue “fin” de la grammaire générale, dont la
critique est déjà argumentée par Aarsleff 1967 et 1982. Voir également Bourquin 2005. On
oublie tout simplement des auteurs aussi importants que Marty, Husserl ou Hjelmslev.

• à tenir compte de travaux antérieurs, chez les Allemands par exemple, 8 qui
abordent la question de la construction d’une grammaire (universelle ou
générale) valable pour introduire à plusieurs langues (Auroux (dir. publ.)
• à noter le foisonnement de la grammatisation (constitution de dictionnaires
et de grammaires 9 ) d’une langue comme le français, notamment sous forme de
traités “partiels” (nous voulons dire voués à un aspect particulier d’une langue)
tant au XVIe siècle que dans la première moitié du XVIIe siècle. Ce ne sont ni
exactement des grammaires ni exactement des dictionnaires, mais ils
développent des descriptions locales d’une langue particulière, et participent
largement à la formation des métalangages et des méthodes de la
grammatisation (Mazière 2006). Pour le pédagogue comme pour le théoricien,
ils finissent par constituer un fouillis inextricable, face auquel les auteurs
réclament des règles ou des principes généraux d’unification;
• à constater la relative rareté des ouvrages de grammaire générale; en
France, notamment, il y a une sorte de vide de production entre l’œuvre de PR
et celle de la génération des encyclopédistes (à partir de 1750) qui aboutira à la
Grammaire Générale de N. Beauzée (1769) (Auroux 1982b);
• à noter que si la grammaire générale est liée avec le rationalisme chez des
auteurs comme Beauzée ou les kantiens, il l’est avec l’empirisme chez
• à faire l’hypothèse que la grammaire générale était l’une des réponses
possibles face à la dispersion produite par le vaste mouvement de
grammatisation des langues du monde (Auroux 1994:138, 142).

Une découverte empirique, due à F. Mazière, vient confirmer avec éclat

cette dernière hypothèse et changer la chronologie admise jusqu’ici. Il s’agit
des travaux de Macé (ou Du Tertre, ou Père Léon — c’est son nom de carme
— ou Noël François, ou François Irénée; rien n’est signé Macé, mais, à la
Bibliothèque Nationale de France, c’est sous Macé ou plus rarement Du Tertre
qu’on trouve tous les ouvrages; voir notre Annexe) qui font remonter,
sûrement à 1651 et probablement aux alentours de 1635, la publication de la
Elle n’était pas ignorée de M. H. Jellinek: “Die allgemeine Grammatik will das allen
Sprachen Gemeinsame darstellen, ihre Absicht ist bei Hellwig und Ratichius noch rein
didaktisch.” (1913, 1:30).
Sur ce concept voir Auroux 1994; initialement la tradition occidentale ne comporte de
grammaire que du grec (par innovation autochtone) et du latin (par transfert technologique); ce
n’est qu’à partir de la Renaissance (on peut noter la grammaire castillane de Nebrija en 1492)
que les langues européennes vont être massivement “grammatisées” (un Tableau abrégé dans
Auroux 1994:74-75), simultanément, au reste, avec quantité d’autres langues, notamment
amérindiennes. Nous avons nommé ce phénomène massif “révolution technologique de la
grammatisation”. Sa prise en considération, qui aurait été impossible sans la multiplication des
travaux sur l’histoire des sciences du langage, constitue l’un des principaux résultats récents de
cette discipline.

première “grammaire générale et raisonnée” française dont nous avons

connaissance à ce jour. Dans cet article, nous nous proposons, sur la base
d’une analyse du contenu, du contexte et des conditions de cette publication,
d’en tirer les conséquences pour notre conception de l’histoire des sciences du

1. Macé et la première “grammaire générale et raisonnée” connue

Si nous laissons de côté l’œuvre proprement apologétique du Père Léon de
Saint Jean (qui à ce titre était connu d’Arnauld, lequel eut un échange
polémique avec lui; voir notre Annexe), on note rapidement que le contexte
d’apparition du concept de “grammaire générale et raisonnée” est celui d’un
travail encyclopédique de compilation référent à la “sagesse <science>
universelle”, à la “science générale”, à des “méthodes abrégées”, à des “tables”
présentant systématiquement tel ou tel type de savoir. On possède par le
biographe P. J. Levot (voir Annexe) et par la Bibliothèque Nationale de
France, entrée Macé, Jean (en religion le P. Léon de Saint-Jean, pseud. Du
Tertre, Noël François et François Irénée), les titres suivants:
Fr . Leonis Encyclopaediae praemessum, seu sapientiae universalis delineatio, adumbrans
generalis eloquentiae atrium, templum, sacrarium; quibus praemittitur de virtutis,
scientiarum et eloquentiae corruptelis deque earum restauratio accurata disquisitio.
Lutetiae Paris. ex typis J. Guillemot, 1635.
Manière abrégée de conférer et disputer sur les matières de controverse. Poitiers, 1647.
Méthode universelle pour apprandre facilement les langues, pour parler purement et écrire
nettement en françois, recueillie par le sieur Du Tertre, seconde édition, Paris, I. 1651.
L’image de la sagesse, avec une idée générale des sciences. Paris 1654.
Méthode de la sagesse et de la connaissance universelle. Paris 1654.
Neuf sciences générales, divisées en neuf tables. Paris 1654.
Le Portrait de la Sagesse Universelle avec l’Idée Générale des Sciances; Et leur Plan
représenté en Cent Tables, par le R. P. F. L. Paris 1655, 1656, 1657.
La politesse de la langue françoise, pour parler purement et écrire nettement. Par N. Fr.,
prédicateur et aumônier du Roi. Paris, 1656, 1663 (Bruxelle), 1664,1674.
Studium sapientiae universalis. Paris, 1657. Fol., 3 part.
L’Académie des sciences et des arts, pour raisonner de toutes choses et parvenir à la sagesse
universelle, par le R.P.L.P., 1979 et 1680. 3 vol.

Dans ce corpus, la première apparition de la grammaire générale et

raisonnée se trouve dans la seconde édition de la Méthode (1651), dont le
privilège date du 4 décembre 1649, sous un autre titre (Méthode Abregée pour
apprandre facilement la Langue Latine, pour parler purement et escrire
nettemant en François, par le Sieur Du Tertre). Si l’on suit l’auteur, ce ne
serait qu’une esquisse, reproduisant un texte imprimé auparavant:
J’ai mis à l’entrée, le crayon d’une GRAMMAIRE GENERALE ET RAISONNEE,
permission de donner au jour la Pièce entière qui est toute prête entre mes mains; avec
le Dictionnaire Vulgaire et Critique, qui l’accompagne.

Après avoir un peu médité et compris la Grammaire générale et raisonnée, il faut se

prescrire (par exemple en la Latine) peu de règles générales, propres, pures et
intelligibles. Elles doivent être choisies parmi celles de la Grammaire que nous avons
dressée, il y a plus de quinze ans: ou en d’autres meilleures, notre siècle en ayant
produit un grand nombre. (p. 37-38, à propos de la méthode latine)

Nous aurions donc une première publication vers 1635. Nous n’avons
malheureusement pu en retrouver la trace jusqu’à ce jour. 10 Comme document
attesté nous disposons donc de la Méthode de 1651 (p. 1-58, sous le titre
Méthode générale et raisonnée pour apprendre la langue latine contiennent,
p. 39-58, La GRAMMAIRE générale, et raisonnée); du Portrait de la Sagesse
universelle de 1655 (elle occupe les p. 5-18, sous le titre La grammaire
générale et raisonnée); et, enfin, de la publication posthume de L’académie
des sciences et des arts en 1780 (elle y occupe les p. 9-29).
On ne peut pas dire que la critique moderne ait totalement ignoré Macé.
F. Brunot le mentionne au tome 3 de l’Histoire de la langue française pour sa
Méthode Universelle (qu’il date de 1650) comme un de ceux qui ont fait passer
dans leurs livres la substance des Remarques de Vaugelas. Il a visiblement
ouvert l’ouvrage et note qu’il reprend les remarques de La Mothe Le Vayer et
d’un auteur inconnu (p. 63). Padley (1985:409) signale que “The third part of
the work simply repeats material from Vaugelas’ Remarques”: il n’a pas ouvert
le livre et a lu Brunot en diagonale.
Pour l’instant, nous n’avons pas retrouvé d’écho des travaux de Macé dans
la littérature grammaticale de l’époque. Toutefois, la récurrence de la
publication (y compris posthume) et la notoriété de son auteur (proche de
Richelieu qu’il assiste à son agonie, prédicateur du roi et polémiste contre les
protestants et les jansénistes, partisan modéré de la réforme catholique) nous
conduisent à conclure qu’il est impossible qu’ils soient demeurés inconnus.
Dans ces conditions, l’absence d’écho (pour peu qu’elle se confirme) peut
provenir d’abord d’un manque de reconnaissance chez les grammairiens (Macé
ne sera jamais considéré comme une autorité pour le français ou le latin; il ne
sera pas académicien), ensuite, du fait que dans le contexte encyclopédique qui
est le sien, la formulation de l’idée de “grammaire générale et raisonnée”
apparaît comme une dépendance naturelle du projet méthodique, un “truc” de
pédagogue, rien de notable, en quelque sorte une banalité. Il en résulte, par
ricochet, que la raison du succès de Lancelot et Arnauld proviendrait d’une
autre source que l’invention soudaine d’un concept de “grammaire générale et

La F.L. encyclopaediae praemessum, dont la publication correspond pour la date, ne la
contient pas.

2. L’universel et le général
Pour se faire une idée de ce que Macé entend sous le concept de “général”,
le plus simple est de partir de son ouvrage principal, Le Portrait de la Sagesse
Universelle avec l’Idée Générale des Sciances; Et leur Plan représenté en
Cent Tables (1655).
L’universalité de la sagesse est celle de l’encyclopédie (cf. A celui qui lit,
p. iij) qui procède par tables et se réfère à R. Lulle; il s’agit donc d’une totalité
englobante, ce que l’on peut nommer une universalité extensive. Il ne s’agit
pas d’une rhapsodie, la rationalité est introduite par “l’ordre du Traité”: “Je
rapporte tout à l’HOMME […] sur le modèle de la structure de l’Homme j’ai
formé l’idée d’une Science Générale, qui se développe et multiplie en Trois,
puis en Neuf. La Science Sensible, la Science Raisonnable et la Science
Spéculative. La Première conduit les sens dans l’usage de la Parole Propre,
Poétique, et Historique […]”. Une “science générale” est donc une science
extensivement universelle.
Toutefois, le paysage se complique lorsque est présentée “L’IDÉE
Générale DES SCIANCES DISTRIBUÉES Dans les divers Cercles DE
L’ENCYCLOPÉDIE”. L’idée générale d’une science est-elle la même chose
que l’idée d’une science générale? Une “idée générale” ne saurait être une
totalité englobante; c’est quelque chose d’abstrait. L’abstraction consiste à ôter
des éléments; il y a, toutefois, deux modèles d’abstraction. L’un consiste dans
l’extraction de la “communauté” entre différents éléments; l’idée générale est
“comprise” dans toutes les idées particulières (universalité distributive, au sens
de la logique médiévale); le rapport est de l’un (l’idée générale) au multiple
(les éléments particuliers); pour aller du “général” au “particulier”, on peut
utiliser la déduction. L’autre consiste à ôter des “détails” sur un ensemble
d’éléments donnés (je dessine une maison sans les ouvertures; je fais une carte
de l’Afrique sans les fleuves, ni les lacs, ni les montagnes). L’idée générale
demeure en relation univoque avec une totalité à laquelle elle renvoie et dont
on peut dire qu’elle est une représentation abrégée. Les parties de l’idée
générale n’ont pas besoin d’être présentes dans toutes les parties de l’ensemble
représenté, la généralité n’est pas “distribuée”, elle n’offre pas de voie
“déductive” (dans une “géographie générale”, je dois entrer la définition d’un
lac, cela n’implique pas qu’il y ait des lacs dans le Sahara, puisqu’au contraire
l’absence de lac entre dans la définition d’un désert).
Allons plus loin. Pour Macé “La science est une connaissance certaine et
évidente de la vérité. Ce qui fait que les mêmes raisons nous persuadent qu’il
n’y a qu’une vérité en toutes choses, au moins par proportion et ressemblance:
nous obligent de reconnaître qu’en la même manière, il n’y a aussi qu’UNE
SCIENCE GENERALE ET UNIVERSELLE” (ibid.). Les termes “généralité”
et “universalité” sont donc synonymes (susceptibles d’être employés l’un pour

l’autre 11 ), mais évidemment distincts comme l’impose l’utilisation rhétorique

du doublet synonymique. Nous avons donc trois éléments qui s’enchâssent.
D’abord la “science universelle” qui est une totalité englobante de l’ensemble
des “connaissances certaines et évidentes de la vérité”; ensuite la “science
générale” qui est une partie de la précédente; enfin, l’“idée générale de la
science” qui est, de nouveau, une partie de la précédente. Une contrainte forte
relie ces trois éléments: l’idée générale de la science doit être une
représentation (“portrait”) de la science générale et la science générale une
représentation de la science universelle, à première vue comme l’abrégé l’est à
Il semble exclu qu’une quelconque voie déductive nous mène du général à
l’universel ou aux éléments dans lesquels il est englobé; pareillement du
général ou de l’élément à l’universel, puisque comme totalité englobante celui-
ci est atteint par sommation. Macé défend, au reste, une épistémologie
empiriste quant à l’origine et à l’ordre des connaissances (celle que l’on
trouvait dans le dernier chapitre des Seconds Analytiques d’Aristote):
Suivant le style de la Nature et de la Doctrine, il faut commencer par les plus basses
<i.e. les sciences> qui sont les plus voisines de nos sens. (ibid. Aij)

Parce que la connaissance animale, matérielle et grossière marche toujours devant la

spirituelle d’abord les espèces des choses corporelles se présentent aux sens qui sont
les premières vedettes ou sentinelles avancées pour découvrir les objets, qui
s’approchent principalement de la vue. (ibid. 4)

Si l’on comprend le concept encyclopédique de l’universalité, reste à

comprendre la place que peut tenir, dans une telle épistémologie, la généralité
de la grammaire.

3. La grammaire générale
Macé appuie son analyse sur une théorie générale du langage. Les choses
rendent des bruits “non par aucune action de vie ou mouvement intérieur, qui
soit en ces choses mêmes”. Les animaux ne parlent pas (rugissements,
beuglements, hennissement “sont à la vérité des voix ou actions, qui sortent du
principe de leur vie naturelle et animale”, “ce ne sont que des bruits vagues
mêlés et confus”, M.:2). 12 Le psittacisme (perroquets, pies et sansonnets)

Macé désigne parfois sa grammaire générale comme “grammaire universelle” (“comme je
diray dans la Grammaire Vniverfelle”, M.:23), vocable qui sera peu fréquent en français, où
l’on a tendance à le réserver à une totalité extensive exhaustive; Beauzée notera, pour cette
raison, qu’une grammaire universelle est impossible; seul Court de Gébelin — peut-être sous
l’influence de l’Hermès de Harris — utilisera le vocable (sur tous ces problèmes voir Auroux
M. désigne la première partie de l’ouvrage Macé 1651, intitulée Méthode Générale et
Raisonnée. Voir l’Annexe pour la composition d’ensemble.

manque d’étendue, d’invention et de la capacité de dépasser “la leçon étudiée”

(c’est l’argument classique que l’on retrouve dans la cinquième partie du
Discours de la Méthode, 1637).
Or, l’Homme […] est doué d’une voix naturelle, vitale, articulée et raisonnable. Et
c’est ce qu’on appelle Parole. Par le moyen de laquelle chacun forme en soi-même ses
pensées, et le communique aux autres. Comme donc dans la supposition des plus
sages Philosophes, il n’y a qu’une vérité: aussi ne puis-je reconnaître qu’une Parole,
revêtue néanmoins de plusieurs couleurs, livrées et apparences. Si l’Homme par la
méditation de l’étude s’entretient soi-même, unifiant ou diversifiant la vérité de
quelque sujet; c’est une parole intérieure, spirituelle ou mentale. S’il se communique
aux Personnes présentes, c’est une parole extérieure, corporelle et orale. Si finalement
il veut transmettre la connaissance de cette vérité aux absents […]: alors, c’est la
même parole, mais couchée, écrite ou imprimée sur le papier.
Cependant parce que la vérité et la pensée, c’est-à-dire la connaissance de la vérité
sont comme naturelles à l’homme: elles sont toujours conformes, et semblables à
elles-mêmes; gardant partout également, la naïveté et la pureté de leur première
innocence. Le Juif, le Gentil, le Chrétien et le Turc, prononçant Jehova, Jupiter, Dieu
et Alla: ne veulent exprimer qu’une pensée, et donner une même vérité. (M.:4-5)

Toutefois les Pensées “s’altèrent” quand elles sortent du lieu de leur

origine; elles “changent d’habits, selon les théâtres sur lesquels on les fait
paraître. De là procède la diversité des noms, des paroles et des langages”,
selon la discrétion (i.e. l’“arbitraire”) des sages, le hasard et la coutume. 13 Cette
diversité est telle que “ces paroles sont vraiment comme des contrats moitié
naturels, moitié artificiels: et comme des concordats solennels que la Nature a
premièrement enseignés, que l’industrie a achevés et que l’usage a signés et
scellés […]” (M.:6). Il y a donc de la facticité et de l’irrationnel dans le
langage, ce qui conduit l’auteur à une hypothèse de sous-détermination
[…] tout ainsi que la parole de l’Homme fuyarde et changeante, ne peut être
représentée par les traits d’aucun pinceau: de même le bien dire, ne saurait être
dépeint de ses couleurs plus naturelles, ni enfermé tout entier dans le détroit des
règles et des préceptes. (Macé 1651, Préface au Lecteur:7)

Tous les préceptes que l’on peut donner au sujet des paroles tiennent à trois
choses: la Raison “qui suivant les lois de la Grammaire, de la Dialectique, et
de la Rhétorique: prescrit certaines règles dans le discours, dont il ne faut pas
aisément se départir”; l’Usage, qui semble être le Roi, ou le Tiran, tant de la
Parole que de l’Écriture; l’Analogie que notre langue française, par exemple,
peut avoir avec les étrangères (celle d’où elle tire son origine, comme le grec

Il ne s’agit pas là d’une situation optimale: “Les paroles, au fonds ne sont instituées, que
pour servir aux pensées. Et c’est une étrange injustice, de rendre les méditations de l’esprit,
esclaves d’une chose si faible comme est le langage, puisque la bonté, principalement en cet
endroit doit être préférée à la beauté” (Macé 1651, Préface au Lecteur:5).

ou le latin; celles qui lui sont voisines — comme l’espagnol ou l’italien (Macé
1651, Préface au Lecteur:9-10). On ne voit aucune langue qui n’abandonne
souvent la raison; celle-ci n’est pas toujours ou connue ou suivie; l’usage n’est
pas moins douteux et incertain qu’inconstant et bizarre; les autres langues se
trouvent souvent dans la même peine que la nôtre.
La parole et les langues sont établies, afin d’entretenir le commerce, des
Compatriotes (Maisons, Familles et Républiques), des Nations (Voyages,
Trafics, Alliances et Confédérations), des “Studieux” (Sciences, Disciplines,
Arts). Tout cela serait plus aisé s’il n’y avait qu’une sorte de parler dans le
monde ou “comme quelques uns promettent de le faire voir, s’il se trouvait une
Langue matrice et primitive, générale et universelle: ainsi qu’on le soutient à
l’égard de la Science” (M.:7). Mais peut-être que “la nature même de la Parole,
à raison qu’elle est étrangère et corporelle, est aussi cause de cette diversité”.
De là résulte, non seulement l’inutilité de la langue universelle, mais encore la
nécessité de la pluralité des langues:
[…] la diversité des noms et des idiomes, est non seulement nécessaire pour
s’entendre les uns les autres: mais encore profitable, pour atteindre à la connaissance
de la vérité qu’on veut exprimer. Vous diriez, à proprement parler, que ce sont divers
rayons de lumière qui éclairent davantage un même objet: ou comme les effets
différents, qui déclarent les propriétés d’une même cause. (M.:8)

La grammaire générale ne peut être qu’un préambule; on ne peut rester

dans des généralités et il faut encore que les généralités aient quelque
pertinence face à la diversité. Par son but même la pédagogie de Macé doit
affronter l’aporie de la contingence: comment concilier l’universel (c’est-à-
dire en fin de compte l’unicité de ce qui est signifié) et la diversité dont on
reconnaît les avantages? Dans ces conditions la Methode Universelle pour
Apprandre Facilement les Langves […] (1651), adopte un point de vue
pragmatique 14 pour déterminer ce qui doit être appris dans l’ensemble des
langues afin de déterminer quelles Langues chacun doit s’efforcer d’acquérir:
Un Homme qui ne doit jamais sortir de Paris, et qui n’a affaire qu’avec les citoyens
de ce Petit-Monde: travaillerait assez inutilement, s’il voulait apprendre l’anglais,
l’espagnol, l’allemand, ou l’Esclavon. (M.:8-9)

A mon avis, un savant en France doit savoir la perfection du français et du latin: avoir
l’intelligence du grec, n’être pas ignorant dans l’hébreu; et savoir se servir, au moins
se tirer de l’italien et de l’espagnol, encore si on veut de l’anglais et de l’allemand.
Car au reste de posséder les Langues Orientales et du Septentrion […] ce n’est pas
l’œuvre d’un jour. 15 (M.:10)

“Les langues du pays, de commerce, et d’usage: s’apprennent pour le besoin, et pour la
nécessité” (M.:14).
Et c’est même l’affaire de “monstres d’esprit”.

La connaissance des langues n’a pas de valeur en soi (“j’aimerais mieux

savoir cent mille vérités en un mot qu’une vérité en cent mille paroles”, M.:11):
“l’esprit de l’Homme étant borné et limité, la vie courte et les occupations
grandes: je crois qu’il est bien malaisé d’apprendre parfaitement les Langues et
les Sciences”. Dans le cercle des sciences les trois langues primitives,
principales et universelles sont l’hébreu, le grec et le latin. Il y a trois niveaux
de connaissance des langues: le bas (lorsqu’on entend, et qu’on parle une
langue grossièrement, “seulement, comme on dit, pour passer pays”), le
médiocre (on l’entend et on la parle avec facilité et moins de défaut; “cela est
passable et hors de reproche, surtout si la Langue est étrangère”) et l’exquis
(“lorsqu’on veut atteindre la perfection d’une Langue, et y acquérir de la
réputation; sachant ses propriétés, ses beautés, élégances, ornements,
délicatesses, et les autres perfections”). D’où une approche de la diversité par
une réduction, non seulement de la multiplicité à atteindre, mais encore du
niveau à acquérir:
Et je ne vois en ce Royaume, la nécessité de ce degré exquis: qu’au regard du français,
et du latin. A mon jugement le médiocre suffit, dans l’hébreu et dans le grec; excepté
au regard de quelques Professions. Et le plus bas, dans les Langues étrangères et de
commerce; excepté à l’égard de ceux qui agissent dans les Traités publiques, comme
sont les Ambassadeurs, les Secrétaires, les Interprètes et autres. (M.:12-13)

Ce niveau correspond à une autre restriction, celle du vocabulaire à

[…] je demande au mieux disant Latin, qui soit en France; que sert de savoir en cette
Langue Romaine, tout le menu équipage d’un Navire? tous les ustensiles d’une
cuisine? 16 Tous les outils d’un Cordonnier? avec tous ces mots contraints ou inventés
dans les Colloques de Cordier, de Vivès, de Pontan, d’Erasme, de Robert Estienne, et
de semblables Maîtres d’Ecole? (M.:13-14)

La visée pédagogique, qui tient compte de la contingence de la diversité,

place directement le préambule de l’accès à cette diversité dans l’optique de la
généralité extensionnelle: on coupe, on restreint. Il n’est pas question de mettre
au jour une généralité inhérente à un sujet raisonnable universel, mais de
travailler dans une rationalité “générale” restreinte à l’utilité de sujets réels
engagés dans des activités diversifiées. La contingence est peut-être une aporie
théorique; pour Macé, elle est d’abord le lieu d’ancrage de l’apprentissage des
langues, raisonné dans les limites du raisonnable.

Il n’est pas exclu que Macé vise directement la pédagogie de Comenius qui passe par les
choses; voir, par exemple, dans l’édition Elzévir de 1643 de la Janua linguarum reserata, le
chap. XI , “De la cuisine”.

4. Le contexte européen de la GGR-M

Ni la GGR-PR, ni la grammaire générale de Macé (désormais la GGR-M)
ne sont des hapax dans le contexte européen. On trouve des exemples plus
précoces de ce type de projet en allemand et en anglais. En 1619, la
Grammatica Universalis/Allgemeine Sprachlehr de Wolfgang Ratke (1571-
1635) donne dans une première partie (25 pages) les définitions des concepts
grammaticaux sans donner d’exemples et présente dans une seconde partie les
paradigmes grammaticaux allemands. La même année, Christopher Hellwig
(1581-1617) rassemble différentes grammaires sous le titre: Libri didactici
grammaticae universalis, Latinae, Graecae, Chaldaicae/Sprachkünste: I.
Allgemeine. II. Lateinische. III. Hebraïsche. La même année encore, le
Portugais Amaro de Roboredo (fl. XVIIe s.) publie à Lisbonne sa Methodo
grammatical para todas as linguas; quelques années plus tard, il choisira
d’accroître la Janua linguarum de Bathe (1564-1614) (1611), qui ne
comportait initialement que l’espagnol et le latin (voir plus loin). En 1648, C.
Ravius (1613-1677) publie A Generall Grammar for the ready attaining of the
Ebrew, Samaritan, Calde, Syriac, Arabic and the Ethiopic languages, with a
pertinent Discourse of the Orientall Tongues, dont le titre résume bien le
Cet ensemble est à mettre en relation avec un vaste mouvement
pédagogique, dont les implications concernent le rassemblement des
connaissances sous forme d’encyclopédie permettant un apprentissage aisé de
toutes les sciences, et l’accès aux langues par le biais de nouvelles grammaires.
En 1630, Johann Heinrich Alsted (1588-1638), dans sa Scientiarum omnium
encyclopaedia septem tomis distincta, l’une des encyclopédies les plus célèbres
jusqu’à l’ouvrage de Diderot et d’Alembert (1751), n’hésitait pas à reprendre
la distinction médiévale entre la “grammatica universalis”, ou plus simplement
“grammatica speculativa”, et les “grammaticae speciales” 17 (sur ces concepts
voir Corvino et al. 1983; Rosier 1984): la première est une science, et, comme
telle, nécessaire; 18 les secondes sont accidentelles. C’est par lui, sans doute,
que le célèbre pédagogue Comenius (1592-1670) confortera son projet
encyclopédique, jamais mené à bien. 19 C’est de Ratke (comme lui protégé par
l’évêque Lanecius-Lanecky), cependant, que le Tchèque tient son projet de
pédagogie linguistique, qui inspira, notamment, ses Grammaticae Facilioris

Voir t. 1, p. 271: “Praecipuum Grammaticae generalis officium in eo est, ut Grammaticas
notiones (seu entia Grammatica) componat cum notionibus seu entibus Logicis”.
Voir, par exemple, Incipiunt quaestiones disputae super Prisciano Minori de Gentili de
Cingulo (ca 1290-1318).
Le Lingua Bohemicae Thesaurus sera perdu dans l’incendie de Leszno en 1656; il devait
faire partie d’un ensemble plus vaste comprenant le Theatrum sanctae scripturae et le
Theatrum universitatis rerum to jest Divaldo sveta conçu comme une grammaire générale des
choses existantes (veci v byti jsoucich) et dont il ne subsiste que des fragments (Cauly

Praecepta (Prague, 1616; cet ouvrage semble aujourd’hui perdu, cf. Cauly
1995:62) et l’un des ouvrages les plus cités de l’époque, 20 la célèbre Janua
Linguarum reserata sive seminarium linguarum et scientiarum omnium, hoc
est, compendiosam Latinam et quamlibetaliam linguam […] methodus sub
titulis centum, periodis mille comprehensa (Leszno 1631; édition tchèque
1633; édition en français et en latin 1642; en 1637, l’édition de Londres
comporte des abrégés des grammaires latine, anglaise et française). Celui qui
fut le dernier chef spirituel de l’Église de l’Unité des frères tchèques y critique
notamment la Janua Linguarum (1611) du jésuite William Bathe destinée à
l’apprentissage du latin à partir de l’espagnol (il était à l’époque réfugié au
Collège irlandais de Salamanque); cet ouvrage, avec une trentaine d’éditions
au XVIIe siècle, a été un incontestable succès qui a précédé celui du Morave.
Parmi les rééditions ou traductions, il importe de noter que plusieurs d’entre
elles ajoutent d’autres langues au latin: A messe of tongues (Londres 1617, par
J. Barbier); Porta de linguas (Lisbonne 1623, par A. de Roberedo, déjà cité);
Janua linguarum silinguis (Strasbourg 1629, par I. Habrecht); Mercurius
quadrilinguis (Padoue et Bâle 1637, par C. Shoppe). La clé de la pédagogie
linguistique de Comenius, contrairement à celle des jésuites qui privilégient le
latin, est le passage par l’exhibition visuelle des choses. Comme le note Cauly:
“Le monde des mots doit demeurer l’image du monde des choses et le langage
rendre sensible et intelligible à la fois l’ordre universel créé et voulu par Dieu”
Macé connaît au moins l’un de ses prédécesseurs. Dans la Méthode,
lorsqu’il fait la liste des livres que l’élève doit mettre dans son programme de
lecture, il note: “[…] lorsqu’on sera un peu fort, Ianua Linguarum qui est sans
doute un Oeuure de grand prix en cette matière, & tout conforme à notre
dessein” (M.:28).
Un peu plus loin, il intègre l’ouvrage lui-même à sa pratique pédagogique
en l’utilisant comme manuel; l’élève progressera dans le vocabulaire en
“réduisant par ordre alphabétique tous les Mots qu’il lira chaque jour en Ianus
Linguarum, ou ailleurs” (M.:30). Compte tenu de la pratique de l’époque qui
consiste à ne pas donner de nom d’auteur, on pourrait se demander si la Ianua
Linguarum citée est celle de Comenius ou celle de Bathe. On imagine mal un
proche de Richelieu favoriser ouvertement les jésuites. 21 La date qui nous place
à une époque où Comenius est plus répandu que Bathe, comme l’existence
d’une version française dont la parution est encore proche (1642), nous font
pencher en faveur du Morave. On imagine mal, en effet, Macé recommander

Il fait évidemment partie des ouvrages recommandés par Macé, voir plus loin.
Il en fait cependant un éloge ambigu à propos des Collèges dont on devrait suivre la “routine
accoutumée” dans les “Académies et les Vniversités”, “Quoy qu’il soit auantageus, d’y
apporter toute la modération possible; comme fait tous les jours avec fuccés cette illuftre
Compagnie qui tient l’empire des lettres, aussi bien que de la piété” (M.:35).

pour le débutant un ouvrage qui n’est pas dans sa langue, puisque cela serait
contraire à tout son projet. Dans ces conditions, nous avons une filiation Macé
Å Comenius Å Ratke, qui nous ramène au premier terme de notre série.
On voit nettement que ce vaste mouvement répond à des problèmes
pratiques. D’abord le problème posé par l’apprentissage du latin dans un
contexte où non seulement il n’est plus la langue maternelle de quiconque,
mais où de plus sa pratique orale — encore recommandée par la pédagogie
jésuite — est en nette diminution. 22 De manière générale, l’enjeu est de
conduire au latin par la voie du vernaculaire national; il en va de même pour le
grec: Lancelot dans la Préface du Jardin des racines grecques (qu’il signe avec
de Sacy) note que l’une des deux “entrées” dans une langue est le vernaculaire
du locuteur. Le problème, ensuite, de la grammatisation des vernaculaires eux-
mêmes et de la description précise de leur fonctionnement: les différentes
remarques appellent des “méthodes”, c’est-à-dire des “chemins” (odos)
susceptibles de guider l’apprenant. Celui, enfin, posé par la diversité des
langues nationales qui sont en voie de grammatisation, 23 et dont la division de
l’Europe en nations se recommandant de langues différentes, rend nécessaire
un apprentissage. 24 Dans un contexte, où la grammaire est conçue comme la
principale méthode d’apprentissage des langues, 25 puisque pour dominer les
langues il faut dominer les grammaires, la question devient: comment dominer
les grammaires de toutes les langues, c’est-à-dire toutes les grammaires des
langues? Comment ramener le multiple à l’unité? 26 Il est clair que l’on peut,
Et très critiquée; voir Macé, M. p. 32 sur les “Enfans que la crainte du figne ou de la ferule,
empèche de parler François”.
La grammatisation des vernaculaires européens tient largement ses motivations de la volonté
d’affirmation des identités nationales.
Ce n’est pas simplement un problème réservé aux diplomates, aux militaires et aux
voyageurs: au gymnase de Leszno, Comenius dut faire face à la cohabitation dans la même
classe d’enfants allemands, polonais et tchèques.
Lorsque les grammaires sont monolingues et destinées à des locuteurs natifs (Denys, Donat)
ou, simplement, trop compliquées pour des débutants (Apollonyos, Priscien, Sibawayhi,
Panini) on peut difficilement penser qu’elles sont des techniques d’apprentissage initial de la
langue. Nous avons formulé l’hypothèse selon laquelle c’est le fait qu’une langue bien
grammatisée (le latin) soit devenue une langue étrangère pour des locuteurs qui avaient à
l’apprendre, parce qu’ils ne connaissaient que leur vernaculaire, qui a fait passer la grammaire
(du latin d’abord) au statut d’instrument pédagogique pour débutants étrangers (Auroux
1994:82). L’idée qu’il faille commencer par une grammaire pour apprendre une langue a
toutefois toujours été l’objet de contestations.
L’orientation pratique explique à nos yeux la restriction aux principales langues
européennes; nous avons, en effet, noté (Auroux (éd) 1990; Auroux 1994) que la
grammatisation des langues du monde, notamment amérindiennes, est rigoureusement
contemporaine de celle des vernaculaires européens. Le mouvement de grammaire générale a
une origine pédagogique; il faudra attendre Beauzée pour voir intégrer véritablement la
diversité des langues et il restera relativement isolé dans cette voie; elle sera explorée sur un
autre registre par les grandes compilations du dernier tiers du XVIIIe siècle et du début du
XIXe (Monbodo, Court de Gébelin, Hervas, Adelung/Vater, Balbi). Ces compilations ont

comme le rappelle Alsted, recourir à la catégorie médiévale de la grammaire

universelle. Il y a toutefois une différence essentielle entre les savants de la
grammatica speculativa et les Européens de l’époque de la grammatisation des
vernaculaires. Les premiers sont dans un univers intellectuel latin, ils peuvent
faire abstraction de la diversité et rester dans la “science” qu’ils n’exemplifient
que sur une seule langue de référence. Les seconds ont l’obligation pratique de
donner accès à la diversité: la grammatisation des vernaculaires leur a donné
un nouvel objet et ils doivent affronter l’aporie de la contingence.
On aura remarqué au vu de nos références précédentes que la grande
tentation pour surmonter l’aporie est la généralité extensive: après des
linéaments introductifs, on entasse les abrégés de grammaire. C’est faire
comme si les langues étaient de simples nomenclatures. 27 Au reste, la
nomenclature est souvent la solution choisie par ceux qui ont inventé une autre
voie pour réduire l’un au multiple: construire une langue universelle, projet qui
intéressa la Royal Society. On notera que John Wilkins (Essay towards a real
character, and a philosophical language, 1668) invoque la grammaire
universelle ou philosophique, au sens où le faisait Francis Bacon en 1605, dans
le second volume de The Advancement of Learning, c’est-à-dire en proposant
de réduire les règles grammaticales au suivi de l’analogie des choses. Les
projets de langue universelle se conçoivent dans le même contexte que celui de
la grammaire générale 28 mais ils abandonnent la question de l’apprentissage
des vernaculaires et éludent l’aporie de la contingence: plutôt que de chercher
un accès à la diversité, ils proposent de construire l’unité qui la remplacera
pour le plus grand bien de l’humanité et de la science. En s’efforçant de créer,
on introduit le futur, c’est-à-dire l’utopie. C’est une voie que Macé refuse
GGR-M et GGR-PR sont donc bien en résonance avec le contexte
européen. On notera la concomitance de la parution de la GGR-PR et celle des
grammaires italiennes et espagnoles de Lancelot (auxquelles la première sert
en quelque sorte d’introduction), ainsi que le travail considérable du même
Lancelot sur le grec et le latin. Dès les premières lignes de sa Préface à GGR-

d’abord un but descriptif, “historique”, voire généalogique (Vater est l’inventeur du mot
“linguistique”, au sens de discipline qui classe les langues et remonte à leur origine). Elles sont
distinctes du programme de la GGR, même si Vater et Court de Gébelin écriront également des
grammaires générales. On doit faire remonter le premier élément de cette série au Mithridates
(1555) de K. Gessner, qui, quoique très en amont de son apparition, en est une référence
explicite chez Adelung et Vater.
Un nomenclateur universel propose un listage des mots par classes; en donnant un nom (ou
un chiffre) à chaque classe, on espère constituer un vocabulaire qui vaut pour tous les mots (de
toutes les langues) de la classe en question. L’exemple de Macé, cité supra, du nom de Dieu en
différentes langues, n’est donc qu’un effet de nomenclature universelle.
Avec évidemment une emphase plus prononcée (la question prend la première place) sur le
développement des sciences modernes.

PR, il note la liaison de son projet avec la diversité des langues sur lesquelles il
a travaillé:
L’engagement où je me suis trouvé […] de travailler aux Grammaires de diverses
langues, m’a souvent porté à rechercher les raisons de plusieurs choses qui sont ou
communes à toutes les langues, ou particulières à quelques unes.

L’originalité de GGR-PR tient essentiellement dans sa référence (dès le

titre) aux “fondements de l’art de parler”, à l’idée qu’il faut les expliquer et
chercher les “raisons” des différences. En France même, le pédagogue Irson,
proche des jansénistes, note dès 1660 l’intérêt de la grammaire universelle:
On ne peut bien comprendre les diverses sortes de significations, qui sont enfermées
dans les mots que l’on n’ait bien compris auparavant ce qui se passe dans nos
pensées; puis que les mots n’ont été inventés que pour les faire connaître. C’est en ce
sens que la grammaire est universelle et qu’elle suppose la Logique, au moins la
Naturelle, pour la définition de ses termes et pour l’explication de ses Règles
générales qui sont infaillibles”. (Irson, 2e éd. 1660, Préface:1)

Le passage que nous avons noté en italique est repris textuellement de la

GGR-PR (II.I; voir Delesalle & Mazière 2006).
Cette orientation n’est pas nécessairement suivie d’effets chez les auteurs
qui sont les plus proches chronologiquement. Le polygraphe Jacques du
Roure 29 publie en 1661 un abrégé d’encyclopédie qui contient une grammaire
Dessein d’une institution UNIVERSELLE, Avéque le dénombrement des Arts, des
Sciences & des Livres neceffaires à ce Dessein. / GRAMMAIRE GENERALE. /
Nouveaux Rudimens: Et nouvelles Regles de la langue Latine. / A Paris, / Chez
l’Auteur: Avec Privilège du Roy. / M. DC LXI.

PR n’y est cité que pour la Nouvelle Méthode latine. Dans ce dernier cas, le
contexte est donc à peu près le même que celui de Macé; le projet est
semblable et l’on retrouve le même constat de l’impossibilité d’apprendre de
nombreuses langues, comme de la vanité de la connaissance des mots face à la
connaissance des choses. 30 La grammaire générale n’occupe que quatre pages
(p. 9-12). Elle contient des définitions rapides des différents sons et lettres,

Nous devons cette référence à B. Colombat (Université Paris 7, Laboratoire d’Histoire des
Théories Linguistiques).
“[…] les Anciens avoient fur nous cet incomparable avantage qu’ils apprenoient de leurs
nourrices les paroles, comme de leurs Maîtres, ils apprenoient feulement les chofes. Pour
apporter quelque remede à nôtre malheur, ou du moins pour ne l’accroître point, nous ne
devons ni embaraffer nôtre efprit par une multitude de langues, ni comme il arrive
ordinairement, favoir d’autant moins de chofes que nous aurons appris plus de mots, & et que
par exemple nous pourrons dire de la bouë & des fétus en plus de Langues Orientales ou
Occidentales” (du Roure 1661:12).

quelques éléments de morphologie et de classification des mots, dont la

classification en huit parties du discours qui est expliquée, les différents cas,
les temps et les modes du verbe, la syntaxe de convenance ou de régime 31 (“un
mot est la cause de la variation d’un autre mot” [du Roure 1661:11]).
On remarquera que la dernière édition de GGR-M (dans L’Académie des
sciences et des Arts), publiée pourtant vingt ans après GGR-PR, ne fait aucune
référence à Port-Royal.
Un ouvrage un peu plus tardif présentera une vision originale de la GGR-
PR. En 1689 le Britannique George Hickes fait paraître ses Institutiones
gramaticae anglo-saxonicae et moesogothicae dont le titre s’inspire
ouvertement de Priscien. Dans sa Préface (p. III), il fait référence à Port-Royal
et insiste sur le rôle de la GGR-PR comme introduction aux langues romanes;
il entend faire une œuvre comparable pour les langues germaniques. L’ouvrage
suit le plan traditionnel: exposition des lettres, puis des parties du discours. La
particularité de l’ouvrage tient, d’une part à ce que Hickes propose une
grammaire de la langue utilisée par la Bible d’Ulphila, donc d’une langue
morte qu’il nomme le moeso-gothique; d’autre part, à ce qu’il met
systématiquement en relation les éléments de cette langue avec les éléments
plus modernes de l’anglo-saxon. Ainsi, il propose une “lettre” du moeso-
gothique et la met en correspondance avec l’élément correspondant de l’anglo-
saxon; de la même façon, sous la catégorie d’une partie du discours, il présente
une forme moeso-gothique et la (les) formes correspondantes de l’anglo-saxon.
Autrement dit, Hickes, en grammatisant une langue morte pour donner accès à
des langues vivantes, invente quelque chose qu’il faut bien classer dans le
genre “grammaire historique et comparée”, quand bien même la distance est
considérable entre Bopp et lui.
Il faut en conclure que grammaire historique et comparée et grammaire
générale répondent au même problème de ramener l’un au multiple: l’une
espère le faire en rangeant toutes les formes de toutes les langues sous les
mêmes catégories; l’autre en rangeant toutes les formes d’un groupe de langues
sous les formes d’une langue plus ancienne. 32

Comme la GGR-PR, du Roure note l’arbitraire de ce type de syntaxe qui doit plus à l’usage
qu’à la raison. Ce fait serait cause que “l’on a souhaité que, les Philofophes trouvassent une
Langue Vniverfelle […]; […] à l’exemple des Arithméticiens qu’ils choififfent […] des
Caractères signifians à chacun les chofes et les rapports des chofes” (1661:11-12).
Il convient de prendre la mesure de ce que signifie “multiplicité des langues” à l’époque;
Macé en reste à la tradition des pères de l’Église et en envisage soixante-douze. De fait,
comme on l’a vu, il note celles qui l’intéressent: latin, grec, hébreu, français, italien, espagnol
(en ajoutant à la rigueur l’allemand et l’anglais, langues qui sont des vernaculaires de la série
des grammaires universelles; voir Ratke et Ravius), mais il en reste au latin et au français.
Port-Royal va plus loin: latin, grec, hébreu, espagnol, italien, français. Le rôle essentiel du
français se conçoit, puisque, comme le remarquait Lancelot dès la Préface du Jardin des
racines grecques, le vernaculaire est la première “entrée” dans les langues. On voit bien
l’erreur de perspective de M. Foucault, lorsque dans sa Préface à une réédition de la GGR-PR

5. GGR-M et GGR-PR
Que contient une grammaire générale et qu’est-ce qui fait d’elle une
“grammaire raisonnée”? Si l’on suit Macé, on atteint vite le contenu de la
grammaire générale qui consiste à définir les éléments suivants: lettres,
distinction en consonnes et voyelles; syllabes; les mots accessoires ou
principaux (neuf espèces); le verbe et ses modes; les personnes; les dérivations;
déclinaisons, genres, cas, conjugaisons. Les définitions sont extrêmement
succinctes: “Le Mode ou le moeu, marque l’air & la façon dont la Perfonne
agift, ou patift”. “Ces spéculations […] font vraiment ouurages de
Metaphysique & de Logique, aufquelles appartient de dreffer les Arts, les
Difciplines & les Methodes” (M.:47). Elles sont complétées par des exemples,
lesquels sont exclusivement en français ou en latin. À la fin de sa grammaire
générale et raisonnée, le carme indique comment celle-ci s’articule à la
Mais afin de mieux entandre toutes ces Regles, & les autres particulieres: il faut
s’arréter à vne des Langues plus les 33 illuftres, & neceffaires; donnant en fuite,
Grammaire Latine.
C’est vn Art, ou vne Methode; qui enfeigne à lire, efcrire, parler, &c.

On doit en conclure que la rationalité de la grammaire réside dans la

généralité de la catégorisation et des rapports stables qui peuvent exister entre
les catégories; ce sont ces rapports qui constituent les règles, comme par
exemple “L’infinitif marquant l’action, ou la passion, fans indiquer aucune
Perfonne: est moins déterminé, & neanmoins eft tres neceffaire pour la
formation de tous les Verbes” (M.:57).

(Paris, Republications Paulet 1969), il note: “À aucun moment Arnault et Lancelot ne

cherchent à dominer un ensemble de langues qu’ils pourraient connaître; le domaine auxquels
ils s’adressent est relativement étroit.” (p. IX). La particularité de PR c’est, tout au contraire,
de faire référence à des langues que Lancelot domine; lorsque la GG élargira son champ
linguistique (avec Beauzée, puis avec la Grammaire Universelle et comparative de Court de
Gébelin 1772) cette compétence deviendra impossible et on devra passer par des données
contenues dans d’autres grammaires. On saisit immédiatement le danger d’affirmations comme
celle de Foucault: “On comprend pourquoi le projet d’une grammaire générale n’a jamais
engendré de méthode comparative […]. La grammaire générale ne définissait un espace
commun à toutes les langues que dans la mesure où elle ouvrait une dimension intérieure à
chacune; c’est seulement là qu’on devait la chercher” (l.c., p. XI). On ne peut, bien entendu,
faire grief au philosophe d’ignorer les connaissances qu’ont produites quelque trente-cinq
années de recherches spécialisées en histoire des sciences du langage. Mais on doit interroger
une méthode qui consiste, pour l’historien, à définir abstraitement des conditions de
possibilités qui délimitent ce qui est formulable dans une période envisagée à partir d’un
échantillonnage restreint de textes. Il s’agit d’un des derniers avatars du transcendantalisme.
Cette inversion est dans le texte, il s’agit probablement d’une faute d’impression.

La “règle” contient une explication: l’infinitif est moins déterminé parce

qu’il n’a pas de marque de personne. Dans ces conditions, il semble que la
généralité de la grammaire soit distributive: on retrouve les règles dans toutes
les langues; la diversité linguistique ne fait pas question; on traite, par
exemple, la déclinaison comme une catégorie universelle. La particularité ne
fait qu’exemplifier la généralité ou ajouter d’autres paramètres à ces principes.
La diversification des paramètres peut même déjà apparaître dans le corps de la
grammaire générale: “En quelques Langues joignant le gérondif, on exprime
vne action que l’on veut faire avec vne autre” (M.:57).
Pour un linguiste, le traité de Macé n’est pas vraiment quelque chose de
bouleversant. Les faits qu’il rapporte sont bien connus; il n’invente ni catégorie
nouvelle, ni explication particulièrement subtile; les banalités sur l’universalité
de la pensée justifient la généralité de quelques catégories et règles
linguistiques et la contingence de l’histoire la diversité des langues. Dans le
projet pédagogique, si la généralité est la porte des langues, elle ne dispense
pas de travailler chacune d’entre elles et l’unique recourt pratique est l’astuce
des méthodes et l’adéquation des “extraits”. Toutefois, il ne faut pas sous-
estimer l’importance de deux éléments qui apparaissent comme des acquis
dans la trame de l’œuvre de Macé: d’abord, la nécessité de passer par la langue
de l’apprenant et, ensuite, celle de commencer par les éléments les plus
généraux (les moins spécifiques) que l’on puisse trouver.
À l’inverse de Macé, Lancelot est un véritable “professionnel” de l’analyse
linguistique et des techniques d’apprentissage: il domine au moins quatre
langues étrangères (latin, grec, italien, espagnol) pour lesquelles il a écrit des
manuels, il possède des rudiments d’hébreu, et, bien évidemment, s’intéresse
au français. Comme grammairien, il travaille dans un tout autre contexte. On a,
en effet, remarqué combien chacune des rééditions de la Nouvelle Méthode
Latine (unanimement citée) s’inspirait davantage de la Minerve de Sanctius,
c’est-à-dire d’une vision causaliste (Minerva seu de Latinae Linguae causis &
Elegantia, Lyon, 1587; Clérico 1982). Les causes des phénomènes d’une
langue sont constituées par une explication rationnelle en termes de catégories
et d’harmonie des règles (on postule une organisation “rationnelle” de la
langue elle-même, donc on repousse l’aporie de la contingence). Il suffit de lire
dans la GGR-PR (II, XXI) la discussion pour refuser au gérondif la qualité
“d’adjectif passif” et lui conférer celle de “substantif actif”, pour comprendre
la méthode et la sophistication de cette grammaire théorique; c’est sur elle que
repose l’insistance sur la recherche des “principes et des raisons”.
Le contenu de la GGR-PR ne diffère guère de la GGR-M et du mouvement
européen dans lequel Lancelot semble se placer, si l’on s’en tient aux matières
traitées (lettres, parties du discours, etc.), que sur deux points, il est vrai
considérables: d’abord, un plus large approfondissement de la discussion et un
recours plus net à la diversité des langues (le texte est dix fois plus long que
celui de Macé), ensuite l’introduction d’éléments supplémentaires comme les

figures 34 ou la syntaxe (il est vrai très succincte). Parmi ces éléments
supplémentaires, il faut mettre au premier plan le premier chapitre de la
seconde partie: “Que la connaissance de ce qui se passe dans notre esprit est
nécessaire pour comprendre les fondements de la grammaire; et que c’est de
là que dépend la diversité des mots qui composent le discours.” À première
vue, il s’agit d’une banalité que l’on trouve comme telle chez Macé. Mais,
d’une part, la liaison de la pensée au langage est devenue celle du principe à
ses conséquences (“dépend”); d’autre part, l’exposé des opérations de l’esprit
(concevoir, juger, raisonner) ne se trouve généralement pas dans les
grammaires; 35 enfin, la réduction de ces opérations à l’activité de l’esprit
(concevoir, c’est-à-dire former une idée, et, juger, c’est-à-dire lier des idées),
nous fait pénétrer dans une nouvelle forme de la logique, cette “logique des
idées” (Auroux 1993) qu’exposeront Arnauld et Nicole (La Logique ou l’Art
de penser, 1662), mais qui a des sources chez un Clauberg, par exemple. C’est
probablement dans ce chapitre qu’il faut voir la principale contribution du
philosophe Arnauld; elle a pour conséquence d’introduire dans la grammaire la
notion de “proposition” (qu’on ne trouve, bien entendu, pas chez Macé) et un
modèle théorique pour cette proposition (composition d’idées). C’est à partir
de là que les grammairiens du siècle suivant (comme Beauzée ou Condillac)
s’efforceront d’expliquer les phénomènes linguistiques par des opérations
sous-jacentes sur les idées. 36
L’historien a pour but d’expliquer des “phénomènes”; pour cela, le plus
simple est de les mettre en “série”. La GGR-PR est le point de convergence de
trois séries distinctes:
• Grammatisation des vernaculaires Æ mouvement européen i) de
rationalisation des vernaculaires (élaboration et simplification des règles); ii)
de grammaire universelle. C’est dans ce vaste mouvement, dont la nature
pédagogique est indéniable, que Macé a strictement sa place, parmi une foule
d’autres, tout en étant probablement le premier Français à thématiser la
• Mouvement “causaliste” d’analyse du latin Æ approche “théorique” des
langues; Lancelot en est l’éminent représentant français.
Comme chez Sanctius, la figure sert à traiter les exceptions aux règles: “Que s’il se rencontre
quelque chose de contraire en apparence à ces règles, c’est par figure […]” (GGR-PR II,
XXIV). Il s’agit là d’une approche “dure” de l’aporie de la contingence.
Ils constituent le contenu du traité aristotélicien De l’interprétation, où l’on s’accorde à
reconnaître les premiers linéaments de la théorie des parties du discours développée par les
Ainsi la concordance (l’accord) de deux mots a pour but de marquer que l’on additionne
deux idées en restreignant l’étendue de la première et en augmentant sa compréhension. Les
mots qui désignent des idées restreignant l’étendue d’une autre idée (désignée par un
substantif) sont des “adjectifs”. Beauzée distinguera les adjectifs qui restreignent l’étendue
sans toucher à la compréhension (nos déterminants) et ceux qui le font en changeant la
compréhension (les adjectifs qualificatifs).

• Evolution de la Logique Æ Logique des idées, 37 Arnauld et Nicole en

donnent la première version simple et cohérente.
La grammatisation des vernaculaires est la cause empirique qui fait naître
la grammaire générale, comme problème, aussi bien que comme projet
intellectuel et pédagogique; la tradition causaliste lui fournit son projet
explicatif; la logique des idées est ce qui lui donne son assise théorique et ses
limites. 38 Incontestablement, Macé n’appartient qu’à l’une de nos trois séries;
son texte, honorable, en parfait accord avec la modernité de son temps, outre
son manque d’innovations théoriques, ne pouvait donc avoir le même
retentissement que celui de Lancelot.

Sur la bio-bibliographie de “Jean Macé”

La Biographie universelle de Feller (1851) donne Jean Macé à l’article Léon de Saint
Jean, son nom de Carme. Hoefer dans la Nouvelle biographie générale en 46 volumes de 1852
fait de même. Aucun ne mentionne ses écrits profanes, mais ils s’accordent à le faire naître à
Rennes en 1600 et mourir au couvent des Billettes en 1671 et à le créditer d’une œuvre
considérable et de grande influence (prédicateur de deux rois, familier de trois papes). C’est
dans la Biographie bretonne de Levot (1852-1857) qu’on trouve, listée, l’ensemble de son
œuvre (63 titres), l’entrée, là encore, se faisant par le nom de Léon avec, entre parenthèses:
“Jean Macé, connu en religion sous le nom de Léon de Saint-Jean”.
Toujours sous Léon de Saint-Jean, au tome 7 de la somme parue à Lille en 1975
Catholicisme, hier, aujourd’hui, demain (Letouzey), l’auteur est qualifié de “théologien,
philosophe, apologiste, prédicateur, auteur spirituel, artisan de la Réforme catholique en
France” et l’on insiste sur l’étendue de ses publications, sur ses qualités de prédicateur (devant
Louis XIII puis le jeune Louis XIV), de controversiste contre les calvinistes (il obtient la
conversion d’Henriette de Coligny) et les libertins, sur son apologétique résolument rationnelle
(ouverte au “nouvel âge de la raison”). Cet aspect de l’engagement de Macé, ami personnel de

Chez Kant, cette logique des idées deviendra une “logique transcendantale”, théorie des
formes a priori de l’entendement; dans cette version, elle permettra aux grammairiens
d’obédience kantienne de contourner l’aporie de la contingence, en déduisant les catégories
grammaticales des formes de l’entendement et en écartant la discipline de toute contagion de la
contingence et de la diversité des langues. Kant lui-même considérait la relation du langage à
la pensée comme arbitraire: “Tel lie la représentation d’un certain mot avec telle chose, tel
autre avec telle autre chose; l’unité de la conscience dans ce qui est empirique, n’est, par
rapport à ce qui est donné, ni nécessairement, ni généralement valable” (Critique de la Raison
pure, deuxième version de la Déduction transcendantale, fin du paragraphe 18).
Les grammairiens ultérieurs vont s’efforcer de réduire l’aporie de la contingence en
appréhendant la grammaire générale comme une science, c’est-à-dire une discipline exprimant
des lois nécessaires. Ces dernières reposent sur les catégories grammaticales. Pour un
rationaliste comme Beauzée, il faut s’efforcer de trouver les catégories nécessairement
présentes dans toutes les langues et sous lesquelles se brancheront les catégories contingentes
des autres langues. Pour un empiriste comme Condillac, il s’agit de déterminer les catégories
suffisantes à partir desquelles on peut construire les catégories de toutes les langues. Quand
bien même les dites catégories ne se trouveraient pas dans certaines langues ou qu’une langue
en contiendrait de nouvelles, toutes celles que l’on rencontre devront se décomposer en une
suite de catégories connues (Auroux 1988).

Richelieu qu’il assiste à sa mort, rend peu probable que l’ensemble de ses écrits n’aient pas été
connus d’Arnauld. Il publie en 1643 un texte court (24 pages) mais pressant: Sentiments
sincères et charitables sur les questions de la prédestination et de la fréquente communion,
qu’il signe François Irénée et dédie à la Reyne Régente: “Il importe sans doute, Madame, à
ceux que Dieu destine au gouvernail, de cognoistre les vents, qui forment les tempêtes”. Ce
texte fait part de sa vive inquiétude devant la possibilité d’une “guerre” entre catholiques.
“Deux principales disputes troublent en nos jours la paix de l’église et la menacent sans doute
de quelque funeste accident. [...] La première dispute est spéculative, de la Prédestination; la
seconde, de la Communion, regarde la pratique”. Il propose une conciliation par convergence
d’approches pour la première dispute, et prêche la paix des pratiques par abandon des censures
trop violentes et le retour aux mœurs de l’Eglise primitive. Il a entraîné une réponse (anonyme)
d’Arnauld: Lettre d’un Docteur en Théologie a un de ses amis sur un livre intitulé:
Sentiments sincères et charitables sur les questions de la prédestination et de la fréquente
communion, par François Irénée (1644). 39 Le janséniste est parfois très dur dans sa
polémique: son adversaire aurait attaqué “avec de si mauvaises preuves et des raisonnements si
faux, qu’il ressemble à une personne qui s’égare de son chemin, et qui, dans ses égarements
mêmes, se laisse encore tomber par des chutes continuelles” (ibid.).
Les écrits qui nous intéressent, en particulier la Méthode Universelle, ne sont pas signés
Père Léon, ni Noël François, ni François Irénée, ses autres noms de plume, mais “sieur du
Tertre” avec cette précision “recueillie par le sieur du Tertre”. Levot signale l’ouvrage en 1650
avec une seconde édition en 1652. Il se pourrait que ce qu’il qualifie de première édition soit
une Méthode abrégée pour apprendre facilement le latin par le sieur du Tertre. Cependant
l’édition de la BNF, qui classe toutes les œuvres de du Tertre-Léon sous le nom de Macé, est
de 1651 et porte la mention “seconde édition”. Méthode universelle pour apprendre facilement
les langues, pour parler purement et écrire nettement en français, recueillie par le sieur Du
Tertre, Revue, corrigée et augmentée en cette seconde édition. L’ouvrage conjoint: Epitre, 40
Préface: Au lecteur; 41 puis, p. 1 à 58: Méthode générale et raisonnée pour apprendre
facilement les Langues, principalement la Latine par laquelle “on apprendrait la Science et la
Langue par LA GRAMMAIRE Générale et raisonnée” p. 39 à 58; puis Discours sur les
difficultés de l’orthographe française, p. 59 à 106; puis: Recueil alphabétique des remarques
sur la langue française, p. 107 à 246; puis: Alphabet pour l’orthographe, non paginé.
Sous la signature Noël François, certaines parties de la Méthode sont reprises dans La
politesse de la langue française pour parler proprement et écrire nettement par N. Fr,
Prédicateur et Aumônier du Roi; 1663, 2e éd., Bruxelles; 1664, 3e éd.; 1672, 4e éd.: Paris, chez
Michel Bobin. La BNF ne garde pas trace d’une première édition. Cependant, le privilège est
de 1655 et le premier “achevé d’imprimer” de 1657. Levot signale trois dates: 1656, 1664,
1668. L’entrée en matière “au curieux de la langue française” reprend la Préface de la Méthode
signée du Tertre. Dans cet ouvrage, cependant, l’auteur a supprimé la GGR pour ne garder que
le Recueil des Remarques avec les Censures et le Discours sur les Difficultés de
La GGR sera reprise en 1679 et 1680 sous la signature R. P. Léon dans L’Académie des
sciences et des arts, pour raisonner de toutes choses, et parvenir à la sagesse universelle, par

Nous devons ces références à Antony McKenna.
Dans cette épître, adressée à l’évêque et comte de Dol, du Tertre parle de “celuy sous lequel
i’ay eu l’honneur de travailler quelques années: et dont je ne fais ici que ramasser les
C’est dans cette préface que nous rencontrons le passage déjà signalé: “J’ai mis à l’entrée, le
crayon d’une Grammaire générale et raisonnée, que j’ai vue imprimée il y a plus de quinze ans.
Et j’espère bientôt la permission de donner au jour la pièce entière, qui est toute prête entre
mes mains, avec le Dictionnaire Vulgaire et Critique qui l’accompagne”.

le R .P. Léon, prédicateur de leurs Majestés très chrétiennes, publié Chez Charles Osmont,
dans la grand’salle du Palais, du côté de la Cour des Aides, à l’Écu de France; le privilège est
de 1654. Il n’y a pas de préface, seulement une table détaillée des points traités dans les
Sciences. Dans le corps de l’ouvrage apparaît un nouveau titre: L’académie des arts et des
sciences [inversion des termes], distribuée dans les divers Cercles de l’encyclopédie. P. 9 à 29:
La Grammaire Générale et Raisonnée est textuellement celle de la Méthode, mais ce qui est dit
“recueilli” dans la Méthode, par Du Tertre, est assumé ici. Après une annonce sur les sciences
humaines à présenter dans une “encyclopédie” des sciences, on a une “Grammaire Générale et
raisonnée”. La part dévolue à l’orthographe est très réduite (2 pages) qui précèdent, p. 32 à 46,
les “Tables de la grammaire”, présentées avec accolades et tableaux fermés. Le texte est
pratiquement le même que celui de la Méthode, avec une plus grande insistance sur le lien
entre Dieu et la parole. Suit l’histoire, la chronologie (avec la géographie et l’œil de l’histoire),
etc, jusqu’à l’arithmétique (400 pages). A rapprocher du Portrait de la sagesse universelle,
publié en 1655 et repris en 1656 et 1657, également sous la signature du Père Léon de Saint-

Sources primaires
Alsted, Johann Heinrich. 1630. Johannis Henrici Alstedii, Encyclopaedia
septem tomis distincta […]. Herbornae Nassoviorum: typis G. Corvini.
Arnauld, Antoine. 1644. “Lettre d’un Docteur en théologie à un de ses amis”.
Sentiments sincères et charitables sur les querelles de la prédestination et
de la fréquente communion (s.l., 1644, 4°, 40 p. in Œuvres d’Arnauld, vol.
XXVIII, p.461-492).
(=GGR-PR) Arnauld, Antoine et C. Lancelot. 1660. Grammaire générale et
raisonnée. Suivie de La Logique ou l’Art de penser, 1662, A. Arnauld et P.
Nicole. Genève: Slatkine; Paris: diffusion Champion, 1972.
Du Roure, Jacques. 1661. Dessein d’une institution universelle. Grammaire
générale. Nouveaux Rudimens: Et nouvelles Règles de la Langue Latine.
Paris: chez l’Auteur.
Hickes, George. 1688-1689. Institutiones grammaticae anglo-saxonicae et
moeso-gothicae [Texte imprimé], auctore Georgio Hickesio,...
Grammatica islandica Runolphi Jonae, catalogus librorum
septentrionalium. Accedit Eduardi Bernardi Etymologicon britannicum.
Oxoniae : e theatro Sheldoniano.

Sources secondaires
Aarsleff, H. 1967. The Study of Language in England, 1780-1860. Princeton:
Princeton University Press.
Aarsleff, H. 1982. From Locke to Saussure. Essays on the History of
Linguistics and Intellectual History. Minneapolis: University of Minnesota
Amacker, R. 1990. “Sept thèses sur l’origine de la grammaire générale”.
Sprachtheorie und Theorie der Sprachwissenschaft. Geschichte und

Perspektiven, éd. par R. Liver, I. Werlen et P. Wunderli, 16-36. Tübingen:

Gunter Narr Verlag.
Asworth, E. J. 1974. Language and Logic in the Post-Medieval Period.
Dordrecht: Reidel.
Auroux, S. 1982a. “Empirisme et théorie linguistique chez Condillac”. Condillac
et les problèmes du langage, J. Sgard (dir. publ.), 177-219. Genève: Slatkine.
Auroux, S. 1982b. “General Grammar and Universal Grammar in
Enlightenment France”. General Linguistics 23.1-18.
Auroux, S. 1988. “La grammaire générale et les fondements philosophiques
des classements de mots”. Langages 92.79-91.
Auroux, S. 1993. La logique des idées. (=Collection Analytiques, 6.) Montréal:
Bellarmin; Paris: Vrin.
Auroux, S. 1994. La révolution technologique de la grammatisation. Liège:
Auroux, S. 1998. La raison, le langage et les normes. Paris: PUF.
Auroux, S. (dir. publ.). 1989-2000. Histoire des idées linguistiques. 3 vol.
(=Philosophie et langage.) Bruxelles: P. Mardaga.
Auroux, S. (dir. publ.). 1992. Histoire des idées linguistiques. Tome 2: Le
développement de la grammaire occidentale. Liège: Mardaga.
Auroux, S. & I. Rosier. 1987. “Les sources historiques de la conception des
deux types de relatives”. Langages 88.9-29.
Bartlett, B. E. 1975. Beauzée’s Grammaire Générale. Theory and
Methodology. La Haye: Mouton.
Baum, R. 1982. “La grammaire idéologique et sa place dans l’histoire de la
grammaire philosophique”. Histoire, Épistémologie, Langage IV:1.23-33.
Bourquin, J. (dir. publ.). 2005. Les prolongements de la Grammaire Générale
en France au XIXe siècle. Besançon: Presses Universitaires de Franche-
Brunot, Ferdinand. 1966. Histoire de la langue française, des origines jusqu’à
1900. Tome III: La formation de la langue classique (1600-1660).
Première partie 1ère éd., 1909. Paris: Armand Colin.
Cauly, O. 1995. Comenius. Paris: Éditions du Félin.
Clerico, G. 1982. Franciscus Sanctius. Minerve. Villeneuve d’Ascq: Presses
Universitaires de Lille.
Colombat, B. (dir. publ.). 1992. “Les parties du discours”. Langages 92.
Colombat, B. 2005. “Changement d’objectif et/ou changement de méthode
dans l’apprentissage du latin au XVIIe siècle? La Nouvelle méthode latine
[...] de Port Royal” Tous vos gens a latin. Le latin, langue savante, langue
mondaine (XIVe-XVIIe siècles), E. Burry (dir. publ.), 177-202. Genève:
Corvino, F., L. Miccoli, A. Ponzio, P. Calefato, G. Mininni, & M.
A. Bonfantini. 1983. Linguistica Medievale. Bari: Adriatica Editrice.

Cram, D. 1985. “Universal Language Schemes in Seventeenth-Century

Britain”. Histoire, Épistémologie, Langage VII:2.35-44.
Delesalle, S. & F. Mazière. 2007. “Grammaire générale et grammaire
particulière: Les Méthodes de Claude Irson”. Actes du dixième colloque
ICHoLS, São Paulo/Campinas, août 2002 ed. par Eduardo Guimarães et
Diana Luz Pessoa de Barros, 41-62. Amsterdam & Philadelphia: John
Dominicy, M. 1984. La naissance de la grammaire moderne. Langage,
logique et philosophie à Port-Royal. Liège: Mardaga.
Donzé, R. 1967. La grammaire générale et raisonnée de Port-Royal. Berne:
Droixhe, D. 1978. La linguistique et l’appel de l’histoire (1600-1800).
Rationalisme et révolutions positivistes. Genève: Droz.
Foucault, M. 1966. Les mots et les choses. Paris: Gallimard.
Harnois, G. 1929. Les théories du langage en France de 1660 à 1821. Paris:
Belles Lettres.
Hassler, G. 1984. Sprachtheorien der Aufklärung. Zur Rolle der Sprache in
Erkenntnissprozess. Berlin: Akademie Verlag.
Jacquemet, G. (dir. publ.). 1947. Catholicisme : hier, aujourd’hui, demain.
Tome 7: Latran-Lutte des classes. 1975. Paris: Letouzey et Ané.
Jellinek, M. H. 1913. Zur Geschichte der neuhochdeutschen Grammatik von
den Anfängen bis auf Adelung, 2 tomes. Heidelberg: Winter.
Julien, J. 1988. “La terminologie française des parties du discours et de leurs
sous-classes au XVIe siècle”. Langages 92.65-78.
Knowlson, J. 1975. Universal Language Schemes in England and France.
Toronto: University of Toronto Press.
Lecointre, C. 1989. “Le traitement de l’énoncé selon la grammaire générale et
raisonnée”. Sens et Être: Mélanges en l’honneur de Jean-Marie Zemb,
E. Faucher, F. Hartweg et J. Janitza (dir. publ.), 139-148. Nancy: Presses
Universitaires de Nancy.
Levot, Prosper. 1852-1857. Biographie bretonne, recueil de notices sur tous
les Bretons qui se sont fait un nom [...] depuis le commencement de l’ère
chrétienne jusqu’à nos jours, par P. Levot, [...] avec la collaboration de
MM. Bizeul, Aymar de Blois, Aug. Burnel, [...] etc., [..]. Vannes :
Marin, L. 1975. La critique du discours. Sur la “Logique de Port-Royal” et les
“Pensées” de Pascal. Paris: Éditions de Minuit.
Mazière, F. 2006. “Synonymie et définition. Ajustement des sens lexicaux en
français”. Synonymie et ‘differentiae’: théories et méthodologies de
l’époque classique à l’époque moderne, Actes du colloque international,
Université de Messine, 6-8 octobre 2003, M. G. Adamo, P. Radici Colace
(dir. publ.), 95-111. Messine: Accademia Peloritana-Napoli, ESI.

Mazière, F. 2007. “Emergence de ‘la langue française’”. Les Français en

émergence éd. par Enrica Galazzi et Chiara Molinari, 9-21. Bern: Peter
Michael, I. 1970. English Grammatical Categories and the Tradition to 1800.
Cambridge: Cambridge University Press.
Nuchelmans, G. 1982. Judgement and Proposition. From Descartes to Kant.
Amsterdam: North Holland Publishing Company.
Padley, G.A. 1985. Grammatical Theory in Western Europe, 1500-1700.
Trends in Vernacular Grammar I. Cambridge: Cambridge University
Rickard, P. 1978. The Embarassments of Irregularity. The French Language in
the Eighteenth Century. Cambridge: Cambridge University Press.
Rosier, I. 1984. La grammaire spéculative des Modistes. Villeneuve d’Ascq:
Presses Universitaires de France.
Schlieben-Lange, B. & H. Weydt. 1988. “August Ferdinand Bernardhi (1770-
1810)”. Histoire, Épistémologie, Langage X:1.81-100.
Schwartz, E. 1981. Les idéologues et la fin des grammaires générales
raisonnées. 2 tomes. Lille: Atelier national de reproduction des thèses,
Université de Lille 3.

University of Potsdam


Lat. analogia, Fr. analogie, Engl. analogy, Germ. Analogie is a concept

widely used in the history of linguistics. But the content of the term and its use
show considerable variation and change over time. It first designated the
relation between linguistic elements and represented a kind of opposition to
arbitrariness. Analogy was regarded as a principle of the formation of words.
In some linguistic theories, analogy is used to explain the development of
language, for example in the description of all words out of analogous roots. In
eighteenth-century linguistic theories, the concept of analogy is mainly applied
to relations between referents in the formation of new words, but similarities
between sounds in syllables also become important. Analogy is regarded as an
important factor of sound change and becomes specialised as a concept of
historical comparative linguistics.

The starting point of my paper is a concept-historical approach that we use

in the context of the onomasiological lexicon of language-theoretical concepts
of the 17th and 18th centuries. I would like to illustrate the intention of this lexi-
con by using as an example the concept of analogy. Due to the character of the
historical-comparative method, analogy is seen to be closely connected with
the laws of phonetics, as Hans Helmut Christmann (1980:520) has already
noted. Research regarding the term- and concept history is primarily directed
towards phonetic law and, as far as analogy is concerned, most studies begin
right with the 19th century. Other studies deal first with antiquity only to then
more or less directly fast-forward to the 19th century. This paper, however, will
focus on the period that is usually omitted and, for comparison, the transition
that occurred in the 19th century.

1. On the history of analogy as a term and concept prior to the 17th century
Already back in antiquity, analogy, meaning similarity or agreement of
proportion, was contrasted with anomaly, i.e. irregularity. In ancient times,

ever since the emergence of grammar as one of the three liberal arts of the
trivium, it was fiercely disputed whether analogy could be found in language
formations, or whether these only showed irregularities (compare Lallot 1995).
Aristarchus of Samothrace (ca. 170 BCE) argued in favor of analogy, while the
whole group of Stoics and in particular Crates of Mallus (2nd century BCE)
argued in favor of anomaly.
Also Quintilian (M. Fabius Quintilianus ca. 35 – ca. 96 CE), who as a
teacher contrasted his clear and unembellished language with the contempo-
rary, contrived language of the so-called asianism style, is an important refer-
ence author for statements on analogy. In grammar, however, it was Marcus
Terentius Varro (115 – 27 BCE) who treated the concept of analogy together
with its counter-concept, anomaly. In his De lingua latina libri XXV, Varro dis-
cussed the dispute between the analogists and the anomalists. In volumes VIII-
XIII, written in 47-45 BCE, he extensively discussed word formation, conjuga-
tion and declination. Herein he first explored the conflict between irregularity
originating from daily usage (consuetudo) and analogy. After presenting the
arguments for both anomaly and analogy, he proposed his solution in the sense
of the latter and illustrated this using derivation and the expression of time
relationships through verbs. Analogy is defined by Varro as following the prin-
ciple of similarity; anomaly, on the other hand, would be the same as dis-
similarity (Varro 1993:453). Both have their justification in language.
However, while anomaly is based on people’s usage of the language, analogy
follows reasoning.
Basically, the Renaissance philologists adhered to the ancient grammari-
ans’ understanding of analogy, which can be reduced to three aspects (Christ-
mann 1980:520):

1) Criterion of regular form formation in inflection and word derivation

2) A certain correspondence of meaning and form
3) The practice based on these principles to declare a form as rule-adherent in
the individual case by means of proportion.

Furthermore, Julius Caesar Scaliger’s (1484-1558) treatise De causis lin-

guae Latinae had a great influence through the 17th century. In volume XIII, a
discussion is devoted to the relationship between analogy and anomaly. Sca-
liger’s preceding list of earlier grammarians’ errors shows that the established
categories of Latin grammar based on the Aristotelian system of Cause and
Effect were to be revised.
Ever since the French philologist Henri Estienne (1528?-1598), analogy
has finally attained the status of a linguistic term and is used as a criterion for
the formation and evaluation of new words. As ‘correspondence’, ‘agreement’
or ‘parallelism’, analogy hereby refers to Greek and Latin; on the other hand, it
refers to elements already present in French. As an expert on Varro and editor

of his De lingua Latina (1573), Henri Estienne refers to Varro’s concept of

analogy, but with a different emphasis. While for Varro the focus was more on
declination and conjugation, Estienne was primarily concerned with naming
and word formation. The great problems confronting the French language at
his time were not to be solved by the people, in Estienne’s opinion, but by
philologists who were experts in the ancient languages.
The linguistic-grammatical concept of analogy could easily be connected to
the philosophical and theological concept that creation may be diverse, but it is
not chaotic. The task of language ensues automatically, and it consists of
faithfully representing the given order of things.

2. Analogy in the language standardization of the 17th century

Initially in the 17th century, the meaning of analogy was again limited to a
criterion for judging – or condemning – language forms already in existence.
In particular, this is expressed in the work that was the authority on French
grammar of the 17th century: Remarques sur la langue françoise (1647) by
Claude Favre de Vaugelas (1585-1650), who also refers to Varro. Vaugelas,
however, is against the formation of new words. When new word formation
does not come into question and when usage is the determining authority for
language, then analogy only gets the cases that have not been regulated yet by
usage. These cases must be decided according to similarity (ressemblance) and
agreement with the usage. Here analogy behaves toward the exemplary court
usage like a copy to the original. Most of the French grammarians of the 17th
century adhered to Vaugelas’ concept of analogy. This is also true for Antoine
Arnauld (1612-1694) and Claude Lancelot (1615/16-1695), who stressed
regulations and rational explanations, but who never doubted the precedence of
usage. A clear divergence from Vaugelas first takes place through Charles
Pinot Duclos (1704-1772), who, in the annotations to his edition of Grammaire
générale et raisonnéee (1754), equates analogie and raison.
In Germany, two alternative principles of language standardization whose
tradition dated back to antiquity also rivaled each other (compare Gardt
1999:128). For the analogists, the point was to derive structural principles from
the grammatical characteristics of a language. These were to serve as guide-
lines for standardization. For instance, it was suggested that the forms of the
plural marker be adjusted to the forms that were most often used. Analogous
standardizations thus broke with the established language usage. By contrast,
the anomalous view prohibited such interventions and set a certain usage as
leader-variant. When Justus Georg Schottel (1612-1676) decided for the analo-
gous position, this was due primarily to the fact that he rejected East Middle
German as a leader-variant. Moreover, the analogous principle would provide
a more unambiguous criterion of explanation, which would at the same time
correspond to the inner nature of language.

3. Analogy as a grammatical term in lexicography

Due to the widespread dissemination of the term ‘analogy’, the correspond-
ing lemma already existed in the lexicography of the 17th century. However, in
the dictionary of the Accademia della Crusca (1612) analogia does not appear
as a lemma. This is linked to the fact that the word does not appear in the
literary language usage of the great authors of the 14th century, which
comprises the body of the Crusca. But it appears six times as a descriptive term
to denote word formation relationships that came into being through derivation
(cuffia-scuffiare, stendare-tende) as well as for inflection relationships
(Accademia della Crusca 1612: Introduzione).
The dictionary of the French Academy (1694) first defines analogy as a
dogmatic term and then as similarity that forms the basis for logical conclu-
sions. The formation of metaphors is also traced back to comparison and
similarity: thus, the lower part of a mountain was named pied ‘foot’ in analogy
to the human foot. Word formation relationships are also named as
grammatical analogy. Through analogy, for instance, ambitionner was formed
from ambition, and likewise passionné evolved from passion.
In the 1718 edition of the Academy dictionary, analogy is first introduced
as a mathematical concept referring to relationships in numerical series.
Alongside that, however, analogy is explicitly defined as a grammatical term.
On the one hand, it determines the relationship of different meanings of a word
to each other; on the other hand, it determines relationships in word formation.
With its entry as a philosophical term in the 1762 edition, the similarity
between things and word meanings takes center stage, a development that was
also able to emerge from the background of the popularity of metaphorism (Du
Marsais, Des Tropes, 1730).
The 1798 edition finally differentiates – still using the examples from the
1694 edition – between a mathematical, a philosophical, a historical, a moral
and a grammatical concept of analogy. The grammatical concept is reduced to
the relationships that exist between words, due to their formation.
In its first edition from 1726, the dictionary of the Real Academia
Española, however, only gives a general relationship based on similarity as the
meaning of analogy. This similarity serves as the basis for drawing
conclusions. A specifically grammatical concept of analogy first appears in the
1803 edition: analogy is characterized as the second part of a grammar,
determining how individual parts of speech, their characteristics and their
flexion are to be treated. This definition is retained verbatim in the subsequent
editions until 1852.

4. Analogy from a language-theoretical viewpoint

In language-theoretical texts, analogy is explained as the result of human
practice, out of which – despite all arbitrariness – similarities in language

nomenclature and a great uniformity of usage arise. In this way, for example,
the genus of nouns, the conjugation forms of verbs and the number of sub-
stantives can be derived from rules (Lamy 1688:75). Analogy is seen here in
the regularity of certain inflectional endings, on the basis of which conclusions
can be made as to meanings. At the same time, analogy is imputed to have a
stabilizing effect on language. It was due to analogy that grammarians were
first actually able to formulate rules. The usefulness of these grammatical rules
stands beyond question, for without it the people by themselves would have to
tediously derive the rules from the real existing analogy. Quite similarly, this is
how analogy is defined in the grammars of the 18th century. Analogy is thus
seen as a real given characteristic in a language which can become the object
of grammatical description, but which also exists without this.
Another use of the term analogy is close to the philosophical language
usage and pertains to the denotation of similar phenomena with the same word
which has been extended in its meaning, as a result of name transference. The
basis for that are similarity relationships between the world that can be per-
ceived with the senses and abstract phenomena that are not accessible to the
human senses. For instance, it is possible on this basis to apply the word
masque ‘mask’ to people who hide their real intentions and opinions even if
they are not wearing masks. César Chesneau Du Marsais (1676-1756) even
goes as far as to assume an originally objective meaning for prepositions,
which was drawn through imitation or also through improper application to
analogous facts and ultimately led to an extension of the meaning. For the
process that lies at the basis of such developments, he uses the term cata-
chresis, whereby it remains open whether by this really ‘bad imitation’ or
‘faded representativeness’ is meant (Du Marsais 1730:51-52).
Analogy, which has been discussed since antiquity in the context of word
formation, is seen in the 18th century – just like the formation of metaphors
based on similarity – as a force leading to the extension of language. Through
analogy, new words can easily be formed for terms that are to be denoted.
Moreover, as a result of the effect of analogy, systematic relationships are
established through which analogous words lead to analogous ideas and at the
same time lighten the burden on our memory (Condillac 1746:147).
In the Encyclopedia article Analogy, the usual definition appears, to begin
with, as the similarity of otherwise qualitatively different objects and the possi-
bility of naming them with the same word on the basis of this commonality.
New words are coined on the basis of analogy, i.e., new terms are attached to
them according to already existing words. Ambiguities due to language
resulted from that, which can only be remedied with the aid of analogy. In
grammar, however, analogy is defined on different levels. First, on the sound
level, minimal differences in articulation are established, thus the sound P
merely consists of a more tense articulation in comparison to B, just as analogy
exists between B and V. In contrast, there is no analogy between the French on

dit and the Latin dicitur or the Italian si dice; each of these expressions was
considered language specific. On the contrary, there is definitely an analogy
between on dit and the German man sagt, since both of them go back to the
denotation for ‘man’ and the third person of the verb ‘to say’. Analogy is also
assigned great weight for conclusions in the field of declination and for further
morphological facts. Encompassing phonetics, words, expressions, idioms and
sentences, analogy pertains to all areas of language.
The quality of being able to speak analogously is also ultimately linked to
mastery of the special character of a language (Encyclopedia article Analogue).
According to the article, a shibboleth for identifying foreigners was their
production of non-analogous expressions. In the same way, almost all
contemporary authors who wrote in Greek or Latin produced sentences which
were analogous to their native languages, but not to the intended target lan-
The conceptual closeness of analogy to the special character of a language
could go even further, in that it was referred to as the genius of language.
Thus, Johann Heinrich Lambert (1728-1777) characterizes the genius of
language as the force that gives word formations their drive based on
similarity, but which also sets limits where there is not yet any pattern on
which one could formulate similar words (Lambert 1764, II:624).
At the same time, comparisons increase between different languages in
which analogy, i.e. the similarity of the forms of one language with the forms
of another, is determined. While such comparisons with reference to expres-
sions and individual forms can be considered meaningful, Nicolas Beauzée
(1717-1789) rejects it rigorously for conclusions about the relationship of
languages. Analogous words can at best shed light on the degree of
communication between peoples; not, however, on the relationship of lan-
In contrast, analogies are considered a sustainable means to pursue the path
of a language back to its origins. On a multitude of direct, transversal ways,
dependent on different conditions, the languages had developed from analogy
to analogy starting from the common root words (De Brosses 1765: xvi-xvii).
The designation of abstract terms that are not further accessible to the senses
can thus be explained by producing comparisons and by naming with words
analogous to real and physical objects.

5. Analogy as an instrument of cognition

With rationalist language theories as background, the rule adherence of
analogy was reinterpreted to mean a force that guided people back to a sen-
tence construction corresponding to correct logical thinking. For Gabriel
Girard (1677/78-1748) the languages that have a subject-verb-object word
order are called analogous languages (langues analogues, Girard 1747:23-24):

According to Du Marsais (Du Marsais, “Construction” in Encycl. 1754,

vol. 4) the logically given order is not always kept. Rather, it is disturbed by
emotional factors and a search for greater force of expression. Irregularities
arising through this are, however, rectified by the listener, even when the pro-
duced speech opposes this. Here analogy helps us by allowing us to draw con-
clusions, going from the familiar to the unfamiliar. We acquire analogy via
imitation and through a long learning process.
Analogy is also deemed conducive to learning a language, to the extent that
conjugation forms, for example, could be easily identified and formed through
comparison with similar forms (Radonvilliers 1768:xv-xvi, Mayans y Siscar
In the 18th century, analogy becomes an epistemological instrument.
Analogy first describes a correspondence of the relationship that exits between
the linguistic signs on the one hand and ideas on the other hand. On the basis
of this, one sign evokes another sign. Since this correspondence can also be
viewed under a genetic aspect, analogy also becomes the explanatory principle
for the development of language. Corresponding to the first naturally given
signs, i.e. in analogy to the first signs, more and more signs were developed.
Language rules are nothing but continued analogies. Where analogy is only
weakly developed, people have fewer possibilities to develop the language and
to adapt it to the requirements of knowledge. The blending of languages
reduces analogy. A language with perfect analogy, as Étienne Bonnot de
Condillac (1714-1780) regarded algebra to be, was considered the episte-
mological ideal in the 18th century.
With the reference of the analogy of signs to the linking of ideas (liaison
des idées) and to the explanation of the origin of language, the concept of
analogy also undergoes an extension into the historical sphere. Even if
Condillac’s attempt to explain the origin of language, like later attempts in the
18th century, is rather hypothetical and directed toward an interpretation of the
essence of language, analogy here is used as a principle to explain the further
language development after the emergence of the first signs. Analogy is
disturbed when a language mixes with another. The rules of analogy have to
develop anew in these mixed languages.
Condillac discusses the principle of analogy in the emergence and develop-
ment of languages in yet more detail in the Grammaire, the second part of the
Cours d’études pour l’instruction du Prince de Parme (1775). He distances
himself here from the principle of arbitrary selection of designations for things.
He replaces it with the guiding function of nature in inventing the first signs
and with analogy in the further development of languages (Condillac 1947-
51[1775], I:431).
Condillac’s historical treatment of analogy influenced numerous authors,
among them also rationalist language theorists. When Beauzée wrote a lengthy
addendum to the analogy article for the Encyclopédie méthodique (Grammaire

et littérature) (Vol. I, 1782:176-180; compare Christmann 1980:528, also

Christmann 1979 and 1984), he repeated parts of his article Usage, but also
discussed the role of analogy in the origin of language. Unlike before, he no
longer purported Vaugelas’ usage as an explanatory principle, but referred to
analogy as a “lumière des langues”. For Beauzée, however, analogy is the
result of raison; it existed already before the very first language and fell from
heaven during mankind’s creation in order to determine the form of language.
Regarding analogy as the decisive factor for determining rules of a lan-
guage suggested that this characteristic was seen as positive. It was determined
to develop it further with the aim of language extension. In this sense, Joseph
Priestley (1733-1804) thought that with a meaningful arrangement, all
analogies existing in a language would be recognizable all at once, and
redundant or ambiguous linguistic phenomena would have to stand out. He
characterized the most important aim of language improvement to be the pro-
duction of analogy, which would lead to a more convenient and more meaning-
ful language usage (Priestley 1762:182-185).
Analogy was also popular with the ideologues, both as a goal of extending
the national language and structural principle of artificial languages as well as
an explanatory principle of linguistic phenomena. The ability to solve equivo-
cal cases in syntax and orthography was attributed to it. Finally, analogy
seemed to bring order to the incalculable number of linguistic signs and to
reduce their variety (Thurot 1837:14).
After the example of chemical terminology, the national language was to
be changed in such a way that by varying the suffixes, accurate and above all
comprehensible terms could be attained. Even in the transition to philological
studies, the former ideologists adhered to analogy as an explanatory principle.
Towards the end of the 18th century, however, there were increasing
warnings about having too much trust in analogy. These were particularly
directed against drawing conclusions about areas that could not be perceived
with the senses. Though analogy was based on universal laws, these were
undergoing so much change that they could easily be the reason for
misjudgments and only superficially took interrelationships into account.
(Compare e.g. Dieudonné Thiébault (1733-1807).) This once again
underscored the characterization of analogy as merely providing some clues
but no real evidence to draw conclusions.

6. Analogy in the 19th century

With the development of historical-comparative linguistics, a paradigm
shift in the concept of analogy took place. Analogy was no longer understood
as merely the image of the given rules in languages. Rather, it also became the
determining principle for deciding whether languages were related or not. This
development is also connected with the focus on the sub-word level.

Wilhelm von Humboldt (1767-1835) regards analogy as a principle of

language extension and also uses the designation organic structure for it. In his
Announcement of a Treatise on the Basque Language and Nation, he wrote in

It can be assumed as a firm principle that everything in a language is based on

analogy, and that the structure of language is organic down to its finest parts. (Hum-
boldt 1904:295)

In stating his position that this analogy, i.e. the organic structure, was
disrupted by the mélange of languages, he agrees with Condillac and other
language theorists of the 18th century. The disruption of the existing analogy
ultimately leads to a new formation of analogy. The language organism is
defined by Humboldt as a connected tissue of analogies (Di Cesare 1989:69).
Analogy is for Humboldt “an inner relationship of the structure of language
which ensures the cohesiveness of its parts” (Di Cesare 1989:68).
The term analogy is also used in German historical-comparative linguistics.
Franz Bopp (1791-1867) speaks of analogically formed series (analogisch
gebildete Reihen), within which the same sounds characterize the same forms.
Moreover, he also applies the analogy term to cases where a form changes by
following the pattern of other forms within their series. Furthermore, the tran-
sition of a form from one analogy series into the other is a central theme. Such
transitions, which are found in historical comparative grammar for e.g. strong
and weak declinations, are referred to by Jacob Grimm (1785-1863) as
inorganic (unorganisch), which corresponds to the expression false analogy
(falsche Analogie), coined by August Friedrich Pott (1802-1887) (compare
Christmann 1980:530; Gao 2000:7) that was commonly used later. The desig-
nation inorganic or false indicates a violation of the order of the original form
system or a deviation from the original phonetic system.
The term analogy is used in the 19th century in three ways. In the first stage,
the usage of Franz Bopp dominates. He regards both the analogously formed
series as well as the transitions into another series, i.e. formations of series and
false analogy, as analogy formations. In the second stage, the term analogy is
used exclusively in the sense of false analogy for such morphological trans-
formations that break the phonetic law. Phonetic law and analogy are regarded,
for instance, by August Friedrich Schleicher (1821-1868) as two forces that
determine the modifications of language, whereby the phonetic law was effec-
tive without exception and analogy as disruption of the phonetic law was false.
This strict separation between phonetic law and analogy makes the attribute
false superfluous. Finally, this separation is canceled in the third stage, where
the subsequent Neogrammarians such as Karl Friedrich Brugmann (1849-
1919) and Hermann Paul (1846-1921) reject a subjective evaluation of
language phenomena. Phonetic law and analogy are now considered to be

equal. Formations of forms that cannot be explained by the laws of phonetics

are recognized as analogy formations and are considered to be just as correct
as phonetic law formations. The equal standing of phonetic law and analogy
means that, in contrast to Humboldt’s analogy concept, the Neogrammarians
do not regard analogy as having any intellectual quality about it at all. Analogy
is regarded just as mechanically as the phonetic law (Christmann 1980, Gao
2000:8, Pater 1996).
Paul’s analogy theory assumes that the morphological forms of a language
are not isolated, but rather that they combine in a certain way in groups. He
differentiates between two kinds of such groups. Substantial groups are
comprised of inflection, formed from the derivation forms of a stem, but also
from words that correspond to each other in meaning (Paul 1909:106). Formal
groups, by contrast, are made up of forms that each have the same category,
such as all comparatives, all nominatives, or all first persons of the verbs.
Paul’s most general principle of morphological transformation states that
innovations come into being through analogy formation. New morphological
forms have always had models in the system (Wurzel 1988:537-539).
The use of the analogy term in contemporary descriptions of word forma-
tions has been definitively influenced by Hermann Paul, whose language-inter-
nal analogy concept is a basis that makes the two opposing word formation
models, namely the syntactical and the lexical, or rather the compositional-
regular and the analogous-holistic, equally legitimate. A language-internal
analogy concept will, however, only inadequately do justice to the creative,
language-modifying aspect. Language-externally, the term analogy is dis-
tinguished by its cognition-functional implications, which can be found in
Wilhelm von Humboldt’s concept of analogy.
In the 20th century there were attempts to resolve the dualism of analogy
and phonetic law. Already Hugo Schuchardt (1842-1927) had developed the
theory that an inner, not antithetic relationship exists between phonetic law and
analogy. The need for a differentiation between phonetic and conceptual
analogy was emphasized in the 20th century. Likewise, the mechanical charac-
ter of the Neogrammarians’ analogy concept was criticized. Heinrich Lausberg
(1912-1992) goes even further in Humboldt’s direction by subsuming the
phonetic law under an even broader analogy concept. He criticizes that the
terminology of the Neogrammarians limited analogy to the form structure,
word body structure and sentence structure. For him, the traditional term pho-
netic law only represents the phonological analogy, while analogy as an
encompassing concept is efficacious in all areas of language.


Académie française. 1694. Le dictionnaire de l’Académie française. Paris: J.

B. Coignard.
Académie française. 1718. Nouveau dictionnaire de l’Académie françoise.
Éditeur scientifique Regnier-Desmarais, François-Séraphin (Abbé). Paris:
Académie française. 1762. Dictionnaire de l’Académie françoise. Quatrième
édition. Paris: chez la Vve B. Brunet.
Académie française. 1798. Dictionnaire de l’Académie françoise. Cinquième
édition. Paris: chez J. J. Smits, an VII [1798].
Accademia della Crusca. 1612. Vocabulario degli accademici della Crusca.
Venezia: Giovanni Alberti.
Brosses, Charles de. 1765. Traité de la formation méchanique des langues, et
des principes physiques de l’étymologie. Paris: Saillant, Vincent, Desaint.
Christmann, Hans Helmut. 1979. “Zum Begriff der Analogie in der Sprach-
wissenschaft des 19. Jahrhunderts”. Festschrift für Kurt Baldinger zum 60.
Geburtstag: 17. Nov. 1979, hrsg. von Manfred Höfler, Henri Vernay &
Lothar Wolf, 102-115. Tübingen: Niemeyer.
Christmann, Hans Helmut. 1980. “Zum Begriff der Analogie in der
Sprachbetrachtung des 16. bis 19. Jahrhunderts”. Stimmen der Romania.
Festschrift für W. Theodor Elwert zum 70. Geburtstag hrsg. von Gerhard
Schmidt & Manfred Tietz, 519-535. Wiesbaden: Heymann.
Christmann, Hans Helmut. 1984. “Wilhelm von Humboldts Schriften zum
Baskischen und sein Begriff der Analogie”. Navicula Tubingensis hrsg.
von Francisco J. Oroz Arizcuren, 85-88. Tübingen: Narr.
Condillac, Étienne Bonnot de. 1746. Essai sur l’origine des connoissances
humaines : ouvrage où l’on réduit à un seul principe tout ce qui concerne
l’entendement humain. Amsterdam: P. Mortier.
Condillac, Étienne Bonnot de. 1947-51[1775]. “Grammaire”. Oeuvres
philosophiques. Texte établi et présenté par Georges Le Roy, vol I, 425-
513. Paris: Presses Universitaires de France. 3 vols. (Corpus Général des
Philosophes Français.) (Parme: Impr. royale 11775.)
Di Cesare, Donatella. 1989. “Wilhelm von Humboldt: Die analogische
Struktur der Sprache”. Wilhelm von Humboldts Sprachdenken hrsg. von
Hans-Werner Scharf, 67-80. Essen: Hobbing.
Diderot, Denis & Jean Le Rond d’Alembert. 1751-1765. Encyclopédie ou
dictionnaire raisonné des sciences, des arts et des métiers. Paris: chez
Briasson etc.; [puis] Neufchastel: chez Samuel Faulche.
Duclos, Charles, éd. 1754. Grammaire générale et raisonnée. Paris: Prault fils

Du Marsais, César Chesneau de. 1730. Des tropes ou des differens sens dans
lesquels on peut prendre un mème mot dans une mème langue. Paris: chez
la Veuve de Jean-Batiste Brocas.
Encyclopédie méthodique. 1782-1786. Encyclopédie méthodique. Grammaire
et littérature. Paris: chez Panckoucke, libraire; Liège: chez Plomteux,
Imprimeur des États.
Estienne, Henri. 1972. Traicté de la conformité du language françois avec le
grec. De Latinitate falso suspecta. Project du livre intitulé De la
Precellence du langage françois. Genève: Slatkine reprints.
Gao, Yali. 2000. Analogie und Wortbildung. Eine wortbildungstheoretische
Anwendung des Analogiebegriffs Wilhelm v. Humboldts. Universität Pas-
sau, Dissertation.
Gardt, Andreas. 1999. Geschichte der Sprachwissenschaft in Deutschland.
Vom Mittelalter bis ins 20. Jahrhundert. Berlin: de Gruyter.
Girard, Gabriel. 1747. Les vrais principes de la langue françoise ou La parole
réduite en méthode, conformément aux loix de l’usage. Paris: Le Breton.
Humboldt, Wilhelm von. 1904. Gesammelte Schriften hrsg. v. der Kgl. Preuss.
Akademie der Wissenschaften, von Albert Leitzmann. Bd. III. Berlin: Behr
1904 (Nachdruck 1968).
Lallot, Jean. 1995. “Analogie et pathologie dans la grammaire alexandrine”.
LALIES 15.109-123.
Lambert, Johann Heinrich. 1764. Neues Organon, oder Gedanken über die
Erforschung und Bezeichnung des Wahren und dessen Unterscheidung
vom Irrthum und Schein. Leipzig: J. Wendler.
Lamy, Bernard. 1688. La rhetorique ou L’Art de parler. Troisième Edition,
reveuë & augmentée. Paris: Chez André Pralard, ruë S. Jacques.
Mayans y Siscar, Gregorio. 1768. Idea de la Gramatica de la lengua latina.
Valencia: Por la Viuda Joseph de Orga.
Pater, Wim A. 1996. “Analogie”. Sprachphilosophie: ein internationales
Handbuch zeitgenössischer Forschung = Philosophy of Language: An
International Handbook of Contemporary Research = La philosophie du
langage: manuel international des recherches contemporaines hrsg. von
Marcelo Dascal, Dietfried Gerhardus, Kuno Lorenz & Georg Meggle, vol.
2, 1237-1250. Berlin & New York: De Gruyter.
Paul, Hermann. 1909. Prinzipien der Sprachgeschichte. 4th edition.
Halle/Saale: Niemeyer.
Priestley, Joseph. 1762. A Course of Lectures on the Theory of Language and
Universal Grammar. Warrington: W. Eyres.
Radonvilliers, Claude-François Lizarde de. 1768. De la manière de’apprendre
les langues. Paris: Saillant.

Real Academia Española. 1726-1739. Diccionario de la lengua castellana, en

que se explica el verdadero sentido de las voces, su naturaleza y calidad,
con las phrases o modos de hablar, los proverbios o refranes, y otras cosas
convenientes al uso de la lengua [...]. Compuesto por la Real Academia
Española. Tomo primero: Que contiene las letras A.B. Madrid: Imprenta de
Francisco del Hierro.
Real Academia Española. 1803. Diccionario de la lengua castellana
compuesto por la Real Academia Española, reducido a un tomo para su
más fácil uso. Quarta edición. Madrid: Viuda de Ibarra.
Scaliger, Jules César. (1540). De causis linguae latinae libri tredecim.
Lugduni: apud S. Gryphium.
Thurot, François. 1837. Oeuvres posthumes de M. Fr. Thurot, membre de
l’Institut, professeur au collège royal de France. Leçons de grammaire et de
Logique. Vie de Reid. Paris: chez L. Hachette.
Varro, Marcus Terentius. 1993. On the Latin Language. With an English
Translation by Roland G. Kent. Cambridge, Massachusetts; London,
England: Harvard University Press.
Wurzel, Wolfgang Ullrich. 1988. “Analogie. Hermann Paul und die natürliche
Morphologie”. Festschrift für Germanistik N.F. 9.537-44.

Université de Reims-Champagne-Ardenne
UMR CNRS 7597 “Histoire des théories linguistiques”


Historians agree that the first treatises about the history of linguistics
emerged during the last decade of the 19th century, with the works of D.
Thiébault (1793, “Lettre à M. Pinglin sur l’histoire de la science
grammaticale”) and F. Thurot (1796, “Tableau des progrès de la science
grammaticale”). Each of these works takes a different perspective on past
grammatical theories. This difference is heightened by the comparison with the
historiographical essays of Lanjuinais (1816, Preface to the republication of
L’Histoire naturelle de la Parole de Court de Gébelin), and Volney (1819,
“Discours sur l’étude philosophique des langues”). The aim of this paper is to
examine the specificity of Thiébault’s work, by focusing on the way the
representation and selection of linguistic theories of the past leads to a kind of
celebration of the ‘Grammaire philosophique’ as a discipline – celebration
which serves a didactic purpose.

La Lettre à M. Pinglin sur l’histoire de la Science Grammaticale de

Thiébault (1793) a été identifiée par l’historiographie de la linguistique, il y a
une vingtaine d’années, comme participant d’une série de textes marquant,
précisément parce qu’ils forment série, la naissance de l’histoire moderne des
théories linguistiques en France (cf. Références primaires, 1). L’idée qu’une
discipline ne peut se constituer que par la production de travaux faisant série
n’est pas contestable, mais elle peut laisser penser qu’ont existé,
antérieurement à la série, des ouvrages isolés assurant la même fonction ou du
moins relevant aussi de cette forme d’épistémologie descriptive de la
linguistique qui prend en compte une dimension temporelle. 1 Auroux (1986) et
De Clercq & Swiggers (1993) ont ainsi pu exhiber des textes plus anciens (cf.
Références primaires, 2) et, tout récemment, De Jonge (2005) a proposé de

Nous suivons ici la définition de l’histoire de la linguistique donnée par Auroux 1980.

voir dans l’histoire de la théorie des parties du discours rapportée par Denys
d’Halicarnasse le prototype de l’approche occidentale traditionnelle de
l’histoire de la linguistique.
Il est évident que la liste des ouvrages proposant, pour un contenu de
connaissance de la théorie linguistique ou pour une de ses réalisations
disciplinaires, un ‘horizon de rétrospection’, peut encore être étendue et n’est
pas limitée à une période donnée. 2 Cependant, il est tout aussi évident que l’on
ne peut considérer comme équivalents un état des lieux de la réflexion
grammaticale sur les parties du discours ou les classes de verbes, une
bibliographie commentée, et un discours prenant explicitement pour objet
l’histoire des savoirs linguistiques depuis leur origine. Ce dernier type de
discours, qui naît en France à la fin du 18e siècle, présente, malgré des
distorsions importantes dans la sanction des œuvres du passé, une forte
homogénéité méthodologique et stylistique.
Nous nous proposons ici, à travers l’examen de la Lettre à M. Pinglin, de
compléter l’interprétation avancée par Auroux 1986 selon laquelle le
développement des études historiques à cette époque répond à une fonction de
régulation appelée par une forte croissance de la production grammaticale.
Nous essaierons ainsi de montrer que la Lettre de Thiébault s’inscrit dans une
stratégie à la fois disciplinaire et didactique qui impose une modalité
particulière du recours à l’histoire des savoirs linguistiques.

1. La Lettre à M. Pinglin et la grammaire ‘philosophique’

Située comme en annexe, bien qu’incluse dans la quatrième partie de la
Grammaire philosophique de Thiébault publiée en 1802, la Lettre à M. Pinglin
se présente comme une réponse de l’auteur à la demande de son collègue
Pinglin. 3 Elle s’ouvre ainsi:
Vous me proposez, mon cher et digne ami, de rédiger l’histoire de la Science
grammaticale, non pour les savants, à qui je serais le premier à la demander, mais
pour mes élèves, auxquels il vous semble qu’elle serait utile. (1793:161)

On pourrait ainsi ajouter au deuxième ensemble Du Roure (1661), qui présente p. 7 des listes
de grammairiens et de lexicographes, et au premier ensemble le second volume de Henry
(1812). Ces deux ouvrages sont référencés sous le point 3 des Références primaires.
F. Pinglin (1752?-1814), professeur de logique au Palais national des Sciences et des Arts, est
l’auteur de deux cours de logique parus en l’an VI (cf. Droixhe 1977:61). La lettre est datée du
22 vendémiaire an II (13 oct. 1793). Elle ne peut donc s’inspirer du Discours préliminaire de
Thurot, paru trois ans plus tard. C’est pourtant ce qu’affirment Joly 1970, et Désirat & Hordé
1982. Il est vrai qu’on ne connaît pas de publication de la Lettre à M. Pinglin antérieure à
1802, mais le contenu même de la lettre ne présente pas d’indices certains d’une lecture du
Discours préliminaire.

La vocation didactique de l’entreprise est d’emblée affirmée: c’est en qualité

de professeur de grammaire que Thiébault est sollicité, même si les troubles de
la période révolutionnaire l’ont pour un temps éloigné de l’enseignement. 4
Après s’être récrié devant l’ampleur de la tâche qui lui est proposée,
Thiébault réclame le droit de ne donner de l’histoire de la science
grammaticale qu’une esquisse, genre qu’il distingue de l’abrégé en ce que
l’abrégé doit “conserver la forme d’un ouvrage régulier” et contenir des
“détails sommaires très fréquents” et “des listes faites selon l’ordre
chronologique le plus sévère” (1793:163). 5 Moins que par une limitation de
volume, l’esquisse se caractérise par une visée prospective, c’est une sorte
d’état des lieux d’un domaine permettant de définir des programmes d’études,
ou d’infléchir les décisions à prendre. Il reste qu’il s’agit ici d’une forme brève,
ce qui justifie l’ajout à la Lettre de deux textes complémentaires: “De l’histoire
des langues” et “Dissertation concernant l’autorité de l’usage sur les langues”,
initialement publiée en 1781 dans les Mémoires de l’Académie de Berlin, dont
l’auteur était membre.
Le premier volet de cet ensemble, intitulé “De l’histoire de la science
grammaticale” (26 pages sur un ensemble de 86), figure ainsi comme l’une des
trois formes de prise en compte de la dimension temporelle dans le champ de
la grammaire: histoire d’une discipline, histoire d’un objet, réflexion sur le
statut de la norme, du changement et de la variation. L’histoire de la ‘science
ne doit être ni l’histoire des langues, ni l’histoire des Grammaires particulières et
spéciales; […] elle ne doit s’attacher qu’à la découverte et admission successive des
principes qui constituent véritablement le corps de la science. (1793:165)

Les principes de cette science - la grammaire ‘philosophique’, 6 qui unit

métaphysique, grammaire et logique - n’auraient été découverts que
récemment. Le projet historiographique consiste alors à retrouver le fil des

De 1765 à 1784, Thiébault a occupé la chaire de ‘style et grammaire raisonnée’ de la
prestigieuse Ecole militaire de Berlin. Il est à partir de 1799 professeur de grammaire générale
à l’Ecole centrale de la rue Saint-Antoine. Sur la vie de Thiébault, cf. Droixhe 1977.
Trois ans plus tard, Thurot (1796:xiv) présentera son Tableau des progrès de la science
grammaticale comme une “notice abrégée des hommes qui ont le plus contribué à
perfectionner cette science, et de leurs ouvrages les plus remarquables”. Sur Thurot 1796, voir
Andresen 1978 et De Clerc & Swiggers 1994.
La distinction posée par Thiébault entre grammaire générale et grammaire philosophique,
pour intéressante qu’elle soit, importe peu ici. Disons rapidement qu’elle revient à séparer
comme relevant de deux disciplines différentes les généralisations linguistiques obtenues par
induction (grammaire générale) ou par déduction (grammaire philosophique) - séparation pour
le moins périlleuse. Malgré la nouveauté de l’appellation, le contenu de la grammaire
philosophique correspond fidèlement au plan d’études préconisé par les Idéologues: analyse de
l’entendement, grammaire générale, logique.

prémisses de la ‘grammaire philosophique’, prémisses nécessairement

imparfaites puisque développées antérieurement à la découverte des véritables
principes de la discipline. D’emblée donc, le regard rétrospectif est biaisé,
orienté vers une forme d’auto-célébration militante de la ‘grammaire
philosophique’, voire de l’ouvrage même qu’il accompagne.
L’histoire de la ‘science grammaticale’ telle qu’elle s’écrit ici opère un
découpage de cette histoire en trois ‘périodes’ principales: de l’origine à Port-
Royal; de Port-Royal à nos jours; la période “où nous entrons”. On notera que,
comme dans l’Esquisse de Condorcet, l’avenir a sa place dans l’histoire,
comme figurant la ‘dernière période’ (la ‘Xe époque’ de Condorcet). Et
l’avenir n’est en quelque sorte qu’une expansion du présent, puisqu’il
appartient à la même ‘période’. Ainsi la Grammaire philosophique se donne-t-
elle implicitement pour un ouvrage ‘d’avenir’.

2. Les trois périodes de la science grammaticale

Toute représentation périodisée d’une discipline pose le problème du statut
accordé aux événements marquant le changement de période. Dans la Lettre,
deux ‘révolutions’ de nature hétérogène viennent perturber et réorienter le
développement du savoir grammatical: la parution de la Grammaire générale
et raisonnée d’une part, la radicale nouveauté de l’époque présente, de
l’‘aujourd’hui pour demain’, d’autre part. Chacune des trois périodes ainsi
délimitée fait l’objet d’un mode particulier de présentation de ses productions
en matière de science grammaticale, le contraste de ces modes de présentation
venant renforcer l’impression de quasi-autonomie de chacune des périodes.

2.1 Première période

Après avoir écarté les grammaires des traditions autres qu’occidentales au
motif que ce sont des “grammaires spéciales et usuelles, étrangères aux
principes essentiels de la Science, aussi bien qu’aux études de [s]es élèves”
(1793:166), Thiébault distingue trois localisations de la réflexion
grammaticale, qui servent à opérer une sous-périodisation: “d’abord chez les
Grecs, ensuite chez les Romains, et enfin chez les peuples modernes, surtout
depuis l’invention de l’imprimerie” (ibid. 167). Sans surprise, le Moyen-Age
est passé sous silence, la période qui sépare la chute de l’empire romain de la
prise de Constantinople étant qualifiée de “trop longue barbarie”. On notera
que dans le Discours de Thurot, bien que la périodisation reste implicite, le
découpage est le même. 7

Thurot (1796:xxxiij-xxxiv) divise l’histoire des sciences et des arts en deux périodes: 1) des
premières républiques grecques à la décadence de l’empire romain; 2) de la renaissance des
lettres en Italie (datée du milieu du 15e siècle) à ‘nos jours’. Cette vision discontinue de
l’histoire des savoirs est encore plus nette dans la préface de Destutt de Tracy à sa grammaire:
“Les longues annales du genre humain ne nous présentent que deux intervalles de lumière que

Pour chacune des trois sous-périodes, il s’agit d’évaluer leur contribution à la

science grammaticale. Sans analyser ici les problèmes soulevés par la sélection
des auteurs et des œuvres (relevés en Annexe), nous retiendrons les verdicts
formulés: des Grecs, “il ne nous est parvenu […] aucun ouvrage qui traite
directement, formellement, et complètement, de la Grammaire philosophique”
(ibid. 168). Chez les Romains, seuls quelques textes (ceux cités en Annexe, à
l’exception de Varron) “méritent une attention sérieuse” (ibid. 169). Quant aux
peuples modernes enfin, Thiébault loue les Humanistes pour la densité et de la
diversité de leurs travaux sur les langues et les textes, mais considère que ceux-ci
restent étrangers à la grammaire philosophique, dont ils ne forment qu’un socle
préparatoire. Leur grand mérite est d’avoir fait naître l’idée que chaque langue a
“ses formes, son génie, ses règles, et sa Grammaire particulière” (ibid. 172).
Cependant, si l’on considère les grammaires des vernaculaires proprement dites,
“on n’aperçoit guères hors de la France, à cet égard, qu’un silence affligeant et
qu’une immense solitude” (ibid. 173), et ce jusqu’au milieu du siècle de Louis
Cette rétrospection cavalière n’établit pas de lien d’explication causale entre
le travail de grammatisation des vernaculaires et le développement de la
grammaire générale, au motif que les Humanistes n’auraient pas compris la
nécessité d’une grammaire unique valant “pour tous les peuples”. Au contraire,
Thurot (1796) affirmera qu’une véritable grammaire générale ou philosophique
ne pouvait être rédigée qu’en langue vulgaire, parce que seule la réhabilitation
des langues vulgaires pouvait permettre à l’étude des langues anciennes de sortir
du “torrent de l’érudition” (ibid. lxj). Son interprétation du rôle historique joué
par la Grammaire de Port-Royal s’écarte notablement de celle de Thiébault: il
reconnaît à la grammaire des Messieurs la vertu d’avoir ouvert un double
programme de recherche: celui de la grammaire générale bien sûr, mais aussi
celui des grammaires particulières ‘raisonnées’. Cette appréciation d’un partage
disciplinaire complémentaire nous paraît aujourd’hui autrement fondée que la
vision cloisonnée des successions disciplinaires proposée par Thiébault, qui ne
peut expliquer la continuité d’un certain type de production grammaticale. La
revue de la première période s’achève en effet sur un parti-pris significatif: la
grammaire de Régnier-Desmarais est regardée comme le dernier des ouvrages
relevant de cette période:
Je place la Grammaire de Regnier Desmarets dans un ordre antérieur à celles de
Messieurs d’Arnaud et Lancelot, quoique celle-ci ait été imprimée avant celle-là: mais
en ce moment, l’antériorité et la postériorité m’ont paru devoir se décider d’après les

nous connaissions assez en détail pour en bien juger: l’un est celui où brillèrent les Grecs et les
Romains, et l’autre comprend les trois ou quatre derniers siècles qui viennent de s’écouler, et
qu’ont illustrés les recherches des différents nations européennes. Ce qui les précède ou ce qui
les sépare se perd dans la nuit des temps, ou dans les ténèbres de l’ignorance” (1803:1-2).

progrès de la science, et la marche des principes, plutôt que d’après la succession des
années. (Thiébault 1793:174)

Cette liberté prise par l’auteur d’une esquisse vis-à-vis de l’ordre

chronologique ‘sévère’ de l’abrégé signe une posture paradoxale: la
coexistence temporelle de modèles théoriques et d’activités scientifiques
différents – et, pour Thiébault, divergents - est à la fois reconnue et niée.
L’ordre chronologique est négligé au bénéfice de l’ordre strictement successif
des régimes scientifiques.

2.2 Deuxième période

En faisant commencer la deuxième période avec la Grammaire générale et
raisonnée, Thiébault se conforme au discours d’auto-représentation de la
discipline épars mais dominant dans le texte même des grammaires générales
du 18e siècle. Une fois assigné à la GGR le double statut de texte fondateur et
de terminus a quo de la ‘grammaire philosophique’, le travail de l’historien
consiste à opérer le partage des ouvrages grammaticaux postérieurs: d’un côté
ceux qui poursuivent l’œuvre des Messieurs, de l’autre ceux qui ignorent cette
discipline nouvelle. Seuls les premiers méritent d’être présentés dans cette
esquisse de l’‘histoire de la science grammaticale’. La sélection s’effectue en
deux temps: sont d’abord écartés comme ne relevant pas du champ
disciplinaire sanctionné les grammaires particulières, les traités de lecture,
prononciation, orthographe, les traités sur les participes ou la prosodie. Ayant
ainsi retenu par défaut les ouvrages de grammaire générale, Thiébault en exclut
encore les grammaires de Court de Gébelin et de Condillac, sous des prétextes
pour le moins fallacieux: la partie de l’œuvre de Gébelin relevant proprement
de la grammaire philosophique ne serait qu’une synthèse des grammaires
précédentes, où il n’entrerait rien “de son propre fonds” (ibid. 179); quant à
Condillac, il n’aurait traité de la science grammaticale que superficiellement
(“il semble n’avoir songé à en parler, que pour compléter le cercle de ses
œuvres”, ibid. 179). Condamnations sommaires, qui marquent nettement les
réserves de Thiébault vis-à-vis des principes sensualistes assumés par les
Idéologues (Staum 1992). A l’opposé, Thurot fait l’éloge de ces deux auteurs
et intègre à son horizon de rétrospection deux ouvrages majeurs de Condillac
négligés par Thiébault: l’Essai sur l’origine des connaissances humaines et la
Une fois ces évictions prononcées, il ne reste plus aux yeux de Thiébault
que quatre auteurs ayant su, à la suite des Messieurs, rendre de “grands
services” à la Grammaire philosophique: l’abbé Girard, Du Marsais, Beauzée,
et le Président de Brosses. Chacun fait l’objet d’une présentation particulière,
dans le but de “saisir l’idée qu’il convient de donner de chacun d’eux à la
jeunesse” (1793:180). La revue suit un ordre à la fois chronologique et
axiologique, allant du moins au plus utile ou savant. Le détail de cette revue

permet de saisir le mode d’évaluation d’œuvres instituées à la fois en lignage et

en héritage.

a) L’abbé Girard
A la différence de Thurot, qui regarde les Vrais principes comme un
ouvrage au style plein d’affectation, proposant des dénominations nouvelles
“multipliées sans sujet” et “rarement heureuses” (1796:xcj-xcij), Thiébault
défend Girard contre ses accusateurs, et souligne même l’intérêt de sa nouvelle
nomenclature. Le verdict est cependant sans appel: cette grammaire “peut
infiniment servir à celui qui voudra nous créer une Grammaire philosophique,
telle que nous l’entendons aujourd’hui: mais elle n’est point cette Grammaire”

b) Du Marsais
Outre le caractère incomplet des travaux de l’encyclopédiste, Thiébault lui
reproche de ne pas avoir dépassé le cercle de la grammaire particulière,
française ou latine. Pour ces raisons, son ouvrage “n’est point encore la
Grammaire philosophique que nous desirons” (ibid. 182).

c) Beauzée
Sa Grammaire générale est très estimée de Thiébault, qui lui a consacré
quatre mémoires (cf. Thiébault 1771 et 1773):
[La grammaire de Beauzée] est un ouvrage entier, dans lequel il seroit long ou
difficile de calculer de combien de divisions exactes, de définitions heureuses ou
justes, et d’observations importantes et variées, il a enrichi la Grammaire
philosophique. C’est d’après tous ces faits, que Beauzée est, à mes yeux, un des
premiers Grammairiens que nous ayons à citer ici. (ibid. 183)

L’appréciation de Thurot sur le même ouvrage est tout à fait contraire: la

Grammaire générale contient “des divisions oiseuses et beaucoup trop
multipliées, des analyses peu exactes, et des définitions quelquefois fausses, un
style lourd et extrêmement diffus”. Bref, c’est un ouvrage dont la lecture est
“pénible et fatigante” (1796:xcvij). L’aveuglement idéologique de Thurot vaut
bien ici celui de Thiébault à propos de Condillac.

d) Le Président de Brosses
C’est l’auteur recueillant les éloges les plus vifs: bien que son ouvrage ne
soit pas une grammaire, il faut suivre ses vues, car “par-tout il pose les vrais
fondements de la Science que tant d’autres ont voulu nous donner” (1793:184).
Si les travaux du Président sont ainsi célébrés par la Lettre, c’est qu’ils
semblent proposer une méthode d’investigation pour l’histoire des langues,

domaine que Thiébault regarde comme le pendant de l’histoire de la science

La sélection d’un horizon de rétrospection positif aussi restreint s’explique
aussi bien par la vocation didactique de la Lettre (le professeur oriente et
prépare ainsi les lectures de ses élèves) que par les présupposés théoriques de
la Grammaire philosophique.

2.3 Troisième période

Cette section ne mentionne aucun nom d’auteur ni d’ouvrage, Thiébault se
refusant à juger ses contemporains, de même que les ‘étrangers’. 8
Programmatique, elle expose la marche à suivre pour accomplir enfin, avec
l’établissement d’une véritable Grammaire philosophique, la fusion de la
grammaire avec la métaphysique et la logique: il faut soumettre à examen et
discussion le contenu des ouvrages précédemment sélectionnés, afin d’en
combler les lacunes et d’en corriger les erreurs. Le souhait exprimé d’un effort
concerté vers l’établissement des vérités scientifiques s’achève par une longue
évocation, aux accents prophétiques, de l’avenir proche qui verra la perfection
de la science grammaticale:
Quand on aura donné aux nations civilisées, la Grammaire philosophique dont nous
venons de tracer l’esquisse, rien ne sera plus facile ensuite que de rédiger nos
Grammaires particulières et usuelles, de manière à les rendre également utiles, faciles,
régulières et complettes: les regles qu’elles auront à prescrire, les exceptions qui y
sont inévitables, les usages qu’elles doivent faire connoître, et même les idiotismes,
les anomalies, les licences, et les formes singulières par où les langues se distinguent
et se caractérisent, viendront sans peine et comme d’elles-mêmes, se placer à la suite
des branches auxquelles elles peuvent appartenir. Ainsi tout sera clairement expliqué
et facilement saisi et connu: il n’y aura point ni doutes, ni obscurité sur aucun point:
rien ne sera équivoque ou arbitraire: dans le parallele que l’on aura lieu de faire de
diverses langues, nulle d’entr’elles ne pourra se soustraire aux reproches qu’elle
méritera: nulle n’aura à craindre qu’on lui conteste ses véritables avantages: justice
sera faite de toutes parts; et ce ne sera qu’avec connoissance de cause, que l’on pourra
préférer l’une de ces langues aux autres. (ibid. 188-189)

La grammaire philosophique est ainsi fantasmatiquement assimilée à un corps

de doctrine sûr et pur, un outil explicatif et évaluatif infaillible, qui
remplacerait définitivement, au tribunal des langues, les allégations intuitives
ou idéologiques par les preuves scientifiques. C’est sur cette ‘vision’ que
s’achève le premier discours de la Lettre à M. Pinglin.

C’est cette lacune, commune à la Lettre à Monsieur Pinglin et au Discours préliminaire de
Thurot, que Lanjuinais (1816) s’efforcera de combler.

3. Style et enjeux de l’écriture de l’histoire

Pour l’historien de la linguistique, la Lettre à M. Pinglin est singulière à
trois titres principaux:
1) les contenus de connaissance véhiculés par les ouvrages cités ne font
l’objet d’aucune exposition, à peine y est-il fait allusion (en ce sens, la
Lettre se distingue nettement de la tradition des états des lieux).
2) la perspective historique est fortement instrumentalisée, orientée vers la
justification des recherches présentes et à venir.
3) les jugements formulés sur les œuvres sont aussi brefs que péremptoires.
Il en résulte, pour le lecteur d’aujourd’hui, une impression de brouillage
épistémologique, la Lettre ne respectant pas les normes de correction de
l’argumentation scientifique. Les jugements de valeur portés sur les théories
passées ne sont jamais fondés sur une restitution préalable des contenus de
connaissance véhiculés: Thiébault ne décrit pas ces théories, les contextualise
rarement, et fait peu de cas de l’exactitude et de l’exhaustivité de ses
recensions. S’il construit bien un modèle d’évolution, les théories
grammaticales passées ne sont jamais considérées comme susceptibles
d’apporter des éléments intéressant une problématique contemporaine.
L’exception faite pour les cinq ouvrages ‘précurseurs’ de la grammaire
‘philosophique’ ne remet pas en cause le principe selon lequel,
fondamentalement, le passé est toujours dépassé. En cela, cette écriture de
l’histoire participe pleinement du mouvement d’exaltation des progrès de la
raison, qui a produit un type d’historiographie justement décrit par F. Hartog
(2003:23) comme relevant moins de l’histoire que d’une “temporalisation de
l’idéal de perfection”. La grammaire philosophique doit ainsi réaliser le point
d’aboutissement de recherches antérieures tâtonnantes, et le “Discours sur
l’histoire de la science grammaticale” vise à permettre à cette nouvelle
discipline d’accéder à la pleine conscience de son rôle historique.
Pour servir cet objectif, la Lettre construit une évaluation axiologique des
ouvrages grammaticaux retenus relativement simple: il s’agit d’assigner aux
travaux des grammairiens passés leur degré d’éloignement ou de proximité
(temporel et conceptuel) avec la grammaire ‘véritable’. Celle-ci étant encore à
venir, les grammaires passées sont toujours caractérisées par le manque, le ‘pas
assez’ et le ‘pas encore’. Ainsi, des Humanistes:
[…] tout ce qu’ils ont fait en particulier pour la Science Grammaticale […] se réduit
au service, non de nous y avoir amenés, mais de nous avoir mis en état d’y parvenir
ensuite: ils ont creusé autour de ce monde encore inconnu; mais en le laissant à
défricher, et à cultiver. (1793:171)

Et des grammairiens du 16e siècle et de leurs émules:


[…] en suivant tous ces auteurs dans la série interminable de leurs travaux, pouvons-
nous ne pas reconnaître qu’ils ne sont point parvenus au but qu’ils étaient si dignes
d’atteindre? Ne sommes-nous pas forcés d’avouer que trop souvent chez eux, le génie
reste confondu avec le limon impur et fétide des temps qui les ont précédés? En effet,
quel mélange de préjugés et de philosophie, d’erreurs et d’observations précieuses, de
crédulité moutonnière et de sagacité! (ibid. 172-173)

Au nom de la “secousse des préjugés” - mot d’ordre d’époque plusieurs fois

répété - le discours historique prend l’accent exalté du discours judiciaire, qui
joue du glissement des catégories axiologiques: le juste est ici une notion
épistémique aussi bien que morale. Appliquée aux œuvres du passé, elle devra
l’être aux œuvres à venir, dont les auteurs sont en quelque sorte avertis par ce
travail de balisage du passé de la discipline. Un horizon de rétrospection n’est
ainsi exhibé que pour être repoussé, au nom du progrès et de la perfectibilité.
Le point de vue surplombant dont Thiébault s’autorise, et l’assurance dont
il fait preuve dans l’énonciation des ‘verdicts’ ne doivent cependant pas être
mis au compte d’une simple forme d’aveuglement militant ou d’auto-
légitimation. Nous avons rappelé en introduction de cette étude l’analyse
inscrivant la Lettre dans une série d’écrits visant à réguler la production
grammaticale (Auroux 1986): la prolifération de grammaires, françaises ou
générales, dans la seconde moitié du 18e siècle nécessite l’établissement de
bilans, de classements, de hiérarchisations. Si l’on restreint cette série aux
textes prenant explicitement pour objet l’histoire des savoirs linguistiques
depuis leur origine, la Lettre apparaît comme le premier et le plus bref de ces
textes, et partant le plus réducteur. La fonction de régulation qu’il assume, par
une nécessité en quelque sorte interne au champ des études grammaticales, se
réalise conformément au ‘régime d’historicité’ propre aux premières années de
la République. Dès 1802, Thiébault est parfaitement conscient de ce
déterminisme et regarde sa Grammaire comme l’émanation de cet
enthousiasme qui porta les projets de réforme de l’instruction publique. Les
hommes alors, dit-il,
recherchoient eux-mêmes comme de concert [le mieux qu’on ne cessoit de leur
promettre], dans les lumières précédemment acquises, et dans les vues hardies que
tous les jours la fermentation des esprits produisoit avec une confiance toujours
nouvelle; et pour assurer et hâter leur brillante destinée, ne permettoient plus ni
l’hésitation aux pusillanimes, ni le doute aux circonspects; […] tout annonçoit que le
zele n’avoit plus de ménagements à garder, plus de bornes à respecter, plus d’obstacle
à reconnoître. (1802:Préface, j-ij)

L’accroissement des connaissances disponibles coïncide en effet avec une

injonction de rupture avec le passé. Paradoxalement, la ruine du passé
provoque un besoin d’histoire, ne serait-ce que pour imaginer et s’assurer un
avenir de la raison définitivement ‘juste’. L’histoire inscrite au programme des
Ecoles centrales est pensée, selon les vœux de Condorcet, comme une ‘histoire

philosophique des peuples’, qui doit “débusquer les mythes” et rechercher la

vérité des causes en vue du progrès intellectuel et matériel (cf Staum 1992).
Enfin, comme Thiébault le suggère lui-même, la Lettre doit aussi être
interprétée comme participant de l’entreprise de disciplinarisation et de
didactisation 9 de la Grammaire générale, dans la mesure où elle est motivée
par la volonté de défendre et justifier la place de la grammaire générale dans
les nouvelles institutions d’enseignement.

4. Conclusion
L’usage du passé pratiqué par la Lettre à M. Pinglin nous permet-il d’y
reconnaître à proprement parler une étude relevant de l’histoire de la
linguistique? La trame de la Lettre est bien la succession temporelle dans le
long terme, mais la norme épistémologique de la discipline, qui impose que la
description et l’analyse des théories passées justifie leur évaluation, n’est
évidemment pas respectée. On ne peut conclure de là, en mimant le geste de
Thiébault, que la Lettre doit être exclue de notre propre horizon de
rétrospection. Dès 1816 d’ailleurs, Lanjuinais (1816:ix) mentionne la Lettre
comme l’une de ses sources d’information, à côté de Henry 1812 et Thurot
1796, construisant ainsi une première représentation de la série qui nous
intéresse. Il resterait à mener la comparaison précise des textes de cette série,
d’abord pour analyser l’évolution de la répartition entre ouvrages ‘sanctionnés’
et ouvrages ‘périmés’, ensuite pour vérifier l’hypothèse d’une balance entre
ambition didactique d’une part et projet de recension et collation des œuvres
du passé d’autre part. Enfin, il faudrait interroger la fonction de ces textes
rétrospectifs vis-à-vis de la grammaire qu’ils accompagnent la plupart du
temps. Si la Lettre est le seul de ces premiers aperçus historiques à s’adjoindre
à une grammaire du même auteur, cette configuration se retrouve bien sûr,
pour peu que l’on prolonge la série jusqu’à nos jours.


Références primaires:
1. Bullet, Jean-Baptiste. 1754. Liste commentée des ouvrages consultés,
Mémoires sur la langue celtique, vol. I. Besançon: C.-J. Daclin.

Ces notions sont empruntées à Chiss & Puech 1999. Bien qu’appliquée aux théories
linguistiques du XXe siècle, l’analyse suivante nous semble convenir aussi à la série de textes
dont la Lettre fait partie: “ [Les procédures de didactisation des savoirs] ne pourraient se mettre
en place si, en amont, un groupe professionnel n’avait assumé la charge et tiré les bénéfices du
balisage à la fois axiologique, rétrospectif et projectif d’un champ intellectuel en voie de
disciplinarisation” (Chiss & Puech 1999:18-19).

Court de Gébelin, Antoine. 1772. Prospectus du Monde primitif, annonçant

pour le tome 10 une “Notice des livres que l’auteur a lus sur ces objets”.
Ephémérides du Citoyen. Paris.
Lanjuinais, Jean Denis. 1816. “Discours préliminaire”. Histoire naturelle de la
Parole éd. par Antoine Court de Gebelin, i-lvij. Paris: Plancher, Eymery,
Thiébault, Dieudonné. 1793. “Lettre à Monsieur Pinglin sur l’histoire de la
Science Grammaticale”. Thiébault 1802, vol. II.161-247.
Thurot, François. 1796. “Discours préliminaire” à la traduction de Harris,
James. Hermès, ou Recherches philosophiques sur la Grammaire
universelle, i-cxx. Paris: L’Imprimerie de la République.
Volney (Constantin-François de Chasseboeuf). 1819. Discours sur l’étude
philosophique des langues. Vol. VIII: Œuvres complètes, 1820-1822.
Paris: Bossange.
2. Anonyme. 1715. Projet d’une Bibliothèque universelle. Journal de Trévoux,
t. XV, août 1715, 1361-1390. Trévoux/Paris: E. Ganeau..
Leibniz. Epistolaris de Historia Etymologica Dissertatio [Aarsleff, Hans.
1982. 87, “The study and use of etymology in Leibniz”. From Locke to
Saussure. Essays on the study of language and intellectual history, 84-100.
Minneapolis: University of Minnesota Press.]
Suétone. De Illustribus grammaticis. [1828. C. Suetoni Tranquilli Duodecim
Caesares et minora quae supersunt opera, 421-474. Paris: N.E. Lemaire.]
3. Henry, Gabriel. 1812. Histoire de la langue française. Paris: Leblanc.
Roure, Jean-Baptiste du. 1661. Dessein d’une institution universelle, avéque le
dénombrement des Arts, des Siences & des Livres necessaires à ce dessein.
Paris: L’auteur.
4. Destutt de Tracy, Antoine. 1803. Eléments d’idéologie II: Grammaire. Repr.
1970 de l’édition de 1817. Paris: Vrin.
Thiébault, Dieudonné. 1771. Précis de la Grammaire Générale de M. Beauzée
(3 mémoires). Nouveaux mémoires de l’Académie royale des Sciences et
Belles-Lettres de la Prusse, année 1771, paru en 1773.
Thiébault, Dieudonné. 1773. Suite de l’examen analytique de la Grammaire
générale de M. Beauzée. Ibid., année 1773, paru en 1775.
Thiébault, Dieudonné. 1802. Grammaire philosophique. Stuttgart-Bad
Cannstatt: Friedrich Fromann Verlag.

Références secondaires:
Andresen, Julie. 1978. “François Thurot and the First History of Grammar”.
Historiographia Linguistica V-1/2.45-47.
Auroux, Sylvain. 1980. “L’histoire de la linguistique”. Langue française 48.7-

Auroux, Sylvain. 1986. “Histoire des sciences et entropie des systèmes

scientifiques: les horizons de rétrospection”. Archives et Documents de la
SHESL 7.1-26.
Chiss, Jean-Louis & Christian Puech. 1999. Le langage et ses disciplines, XIXe
– XXe siècles. Bruxelles: Editions Duculot.
Clercq, Jan de & Pierre Swiggers. 1993. “Aux sources de l’historiographie de
la grammaire: les Mémoires de Trévoux et un projet de bibliothèque
universelle”. Orbis 36.221-231.
Clercq, Jan de & Pierre Swiggers. 1994. “François et Charles Thurot
historiographes de la linguistique: une histoire de famille”. Florilegium
historiographiae linguisticae éd. par Jan de Clerq et Piet Desmet, 277-294.
Louvain: Peeters.
Désirat, Claude & Tristan Hordé. 1982. “Introduction”. Histoire Epistémologie
Langage IV-1.5-20.
Droixhe, Daniel. 1977. “Introduction”. Thiébault 1802.7-65.
Hartog, François. 2003. Régimes d’historicité. Présentisme et expériences du
temps. Paris: Le Seuil.
Joly, André. 1970. Introduction et notes de Thurot, Tableau des progrès de la
science grammaticale. Bordeaux: Editions Ducros.
Jonge, Casper de. 2005. “Dyonisius of Halicarnassus as a Historian of
Linguistics”. Henry Sweet Society Bulletin 44.5-18.
Staum, Martin. 1992. “L’Idéologie dans les Ecoles centrales”. L’institution de
la raison éd. par F. Azouvi, 163-196. Paris: Editions de l’EHESS.
Auteurs et ouvrages mentionnés dans “De l’histoire de la science grammaticale” 10

Première période: de l’origine à Port-Royal:

Grecs: “le traité de Longin sur le sublime, quelques morceaux de Lucien, quelques phrases de
Platon, quelques fragments de quelques autres auteurs”; Poétique, De interpretatione
et Organum d’Aristote; leçons d’Isocrate
Romains: Auctores latinae linguae in unum redacti corpus, Varron, “quelques observations de
Cicéron”, Institutions oratoires de Quintilien, “quelques passages de Saint-Jérôme, et
d’un petit nombre d’autres auteurs” (+ Isidore de Séville)
Peuples modernes: Scaliger père et fils, Budée, Turnebe, Erasme, Valla, Casaubon, Vossius
père, fils et petit-fils, Saumaise, Baudius, Abram, Gronovius, Dacier, Muratori, etc. /
Calepin, Lacerda, Passerat, Screvelius, Etienne, Boudot, etc. / Sanctius, Scioppius,
Despautere, Clénard, Giraudot, Advocat, etc. / Decolonia, Jouvency, Batteux,
Marmontel, etc./ Ramus, Chifflet, Vaugelas, Thomas Corneille, Reignier-Desmarets

Deuxième période: de Port-Royal à nos jours:

Port-Royal: GGR et LAP, + suppléments de Fromant et Duclos
Abbé Girard, Du Marsais, Beauzée, de Brosses

L’orthographe des noms propres est ici celle du texte original.

Ecartés: Buffier, Latouche, Restaud, Wailly, traités partiels, Dangeau, d’Olivet / Condillac,
Court de Gébelin

University of Illinois at Urbana-Champaign


In this article, we show how the arguments for French agreement rules
change in the history of French grammatical thought. We examine three
syntactic structures chosen from the Journal de la langue française (1784-
1792) and the factors - “historical” or “linguistic” - that grammarians invoke
for recommending agreement, or no agreement. In one structure, the agreement
is between a participle and a preceding noun, in another between an adjective
and a noun, and in the third between a collective noun subject and the
following verb. Starting from the analysis of the arguments of Urbain
Domergue, expressed in his Journal de la langue française, and continuing
through those of today’s linguists, we see both “historical” and “linguistic”
factors evoked to explain agreement rules. In the eighteenth century, we find
that a formal (mathematical) analysis true to the ideals of Enlightenment was
favored. Current linguists take a more lenient attitude toward variation than did

1. Introduction
L’accord est difficile à généraliser. D’une manière générale, on peut dire
toutefois que l’accord—un phénomène de transfert d’une ou de plusieurs
catégories morphologiques sur d’autres classes (le déterminant, l’adjectif, le
pronom représentant, le verbe) 1 —est un phénomène linguistique qui est propre
à chaque langue et qui affecte diverses classes grammaticales, pas toujours
d’une nature homogène. Dans son ouvrage l’Accord en français contemporain,

Dans la Grammaire d’aujourd’hui, Michel Arrivé et al. ajoutent à cette définition que le
phénomène linguistique de l’accord en français est jugé plus étendu qu’en allemand (où
l’adjectif ne s’accorde que lorsqu’il est épithète) et qu’en anglais (où l’article et l’adjectif sont
invariables), mais moins que dans le bantou, où tous les éléments de la phrase répètent
certaines marques du sujet (1986:20).

Høybye (1944) indique l’absence d’une étude systématique sur l’accord, qui
peut poser des problèmes aux enseignants:

Ce sont surtout des raisons pratiques qui ont orienté nos études dans cette direction,
nous étant constamment heurté, dans notre enseignement, à des problèmes d’accord
qui étaient insuffisamment traités dans la littérature grammaticale. Les
renseignements dont nous avions besoin étaient souvent dispersés çà et là dans une
vaste littérature, et souvent ils étaient nettement introuvables. (1944:9)

D’autre part, l’analyse suivante de Pellat, Riegel et Rioul (2002) nous fait
supposer une difficulté que les apprenants du français peuvent rencontrer dans
leur apprentissage de l’accord:

Fonctionnellement, les marques écrites de l’accord sont largement redondantes par

rapport aux indications positionnelles qui se déduisent de la structure syntaxique de la
phrase. Leur effacement ne nuirait guère au repérage des rapports de dépendance
syntaxique dans la phrase écrite […]. D’ailleurs, une grande partie des morphèmes
grammaticaux impliqués dans les différents mécanismes de l’accord ne se réalisent
pas dans la forme orale des énoncés. (2002:539)

On prévoit ainsi que l’apprentissage de la règle d’accord, qui est

essentiellement écrite, peut poser un problème non seulement aux locuteurs
non-natifs mais aussi aux locuteurs natifs.
Dans cette analyse des arguments d’Urbain Domergue (1745-1810),
grammairien à la fin du XVIIIe siècle, et des linguistes d’aujourd’hui, on
examinera quels facteurs se mettent en jeu lorsqu’un locuteur fait l’accord.
Cette question nous permettra de mieux comprendre le fonctionnement de
l’accord dans le français contemporain et ainsi de repérer des changements
linguistiques en cours dans l’histoire de la langue française. Dans la grammaire
scolaire, les fondements de nombreuses règles scolaires actuelles datent du
XVIIIe siècle. Ainsi, la comparaison du français actuel avec le français du
XVIIIe nous permettra de mieux tracer à travers le temps les idées
grammaticales ainsi que la norme linguistique.
A cette fin, trois structures sont choisies dans le Journal de la langue
française (1784-1792): l’accord de l’adjectif qui suit l’expression avoir l’air,
l’accord du verbe avec le syntagme nominal la plupart de et l’accord du
participe passé conjugué avec avoir. La raison pour laquelle on a choisi ces
structures particulières, plutôt que d’autres, est que les questions posées sur ces
structures par les lecteurs du Journal de la langue française intéressent
toujours les locuteurs du français d’aujourd’hui: dans les écoles, on apprend les
règles établies, souvent sans poser de questions sur ses fondements.
En adoptant la terminologie de Jean-Christophe Pellat (2003), on analysera
les facteurs qui déterminent la grammaticalité de l’accord en termes de
“principe historique” et de “principe linguistique”. Dans son article, Pellat

explique que le système d’orthographe du français du XVIIe est la mise en

œuvre du “principe phonographique”, principe qui s’appuie sur le son réel, et
du “principe historique”, principe qui recourt à l’étymologie grecque ou latine
ou à l’évolution du mot dans l’histoire du français. Plus précisément, dans
notre étude, par le “principe historique”, 2 on entend l’argument favorisant
la raison étymologique ou historique de l’expression en question; par le
“principe linguistique”, on entend l’argument favorisant la description du
fonctionnement de la langue à un moment donné du point de vue
synchronique, sans recours à l’étymologie ou à des langues classiques, que cet
argument soit formellement analysable ou non d’après la logique des
grammairiens ou des linguistes. Dans cette étude, on suppose qu’en français,
“l’accord” est une notion parallèle au “principe de l’orthographe” dans le sens
que l’accord et le principe de l’orthographe concernent tous les deux “la
formation correcte” d’une unité linguistique.
Avant de commencer notre étude sur les trois structures précises, il nous
faudrait présenter le Journal de la langue française que l’on examinera dans
cette étude. Ce journal sert de lieu de débat par excellence à la fin du XVIIIe
siècle où Domergue, rédacteur de ce périodique, invite les abonnés, les savants,
notamment les grammairiens et les philosophes à réfléchir aux questions de
langue française de l’époque. Dans son premier article qui date du
1er septembre 1784, Domergue précise que le but de ce périodique est de
“fonder sur une métaphysique claire un système de grammaire absolument
neuf, sans prétendre en exclure d’autre et de répondre aux différentes questions
sur la langue écrite ou parlée” dans la langue française (I, 7).

2. Système grammatical de Domergue

Au début de la publication du journal, Domergue propose son système
grammatical: il divise toutes les catégories grammaticales traditionnelles en
deux groupes au niveau philosophique: substantif et attribut. Ainsi le nom et le
pronom, qui représentent une première idée signifiant un être ou une chose (I,
34), entrent dans la catégorie du substantif; l’adjectif, l’article, l’adverbe, le
participe, le verbe sont classifiés comme attribut, attribut qui, à un mot “à une

Évidemment, le latin et le français ne se comportent pas de la même manière dans l’accord.
Néanmoins, les grammairiens du XVIIIe siècle recourent souvent au latin quand il y a des
doutes sur l’usage en français. Afin de comprendre les différentes réalisations de l’accord en
français et en latin, il serait utile de rappeler la notion essentielle dans la grammaire latine, que
Bernard Colombat a résumée ainsi: [a] une règle imposant au nominatif de s’accorder avec le
verbe en personne et en nombre; [b] une règle d’accord de l’adjectif avec le substantif en
genre, en nombre, en cas et parfois en personne; [c] une règle d’accord du relatif et de
l’antécédent en genre, nombre et personne (1996:11-12).

première idée, en attache une autre” (I, 43). L’attribut 3 se divise en trois
catégories: attribut commun (être), attribut particulier (article, adjectif, pronom
possessif, pronom démonstratif, participe et adverbe) et attribut combiné
(verbes à part être) (I, 44). Pour justifier le phénomène de l’accord entre le
substantif et l’attribut, dans son article du 15 septembre 1784, Domergue fait
référence à la “puissance d’identité” de Roussel de Breville, 4 auteur d’un Essai
sur les convenances grammaticales de la langue française (1784): le substantif
et l’adjectif s’accordent en genre et en nombre l’un à l’autre à cause de la
puissance d’identité (I, 19). Ainsi Domergue suit la tradition des
grammairiens-philosophes du XVIIIe siècle tels que Duclos et Morel dans sa
justification de l’accord.

3. Comparaison des trois structures

3.1 Avoir + participe passé
L’accord du participe passé conjugué avec avoir affirme cette analyse
formelle de Domergue et son système grammatical. Dans le Journal de la
langue française, depuis ses débuts en 1784 jusqu’en 1792, Domergue a publié
au total vingt articles à ce sujet où il présente les arguments des grammairiens
tels que Condillac, Duclos, Boinvilliers et Pierre Morel, et des lettres des
Dans cet accord, il y a deux niveaux à considérer: dans le fondement, c’est-
à-dire, la légitimité de la règle elle-même, et dans le fonctionnement de la règle
à l’intérieur des phrases. Cette règle en français, qui remonte à la formulation
“officielle” 5 de Clément Marot en 1538, s’appuie sur la grammaire du latin,
selon laquelle le participe est un adjectif et s’accorde en genre et en nombre
avec le nom auquel il se rapporte. Ainsi, bien que la structure “avoir +
participe passé”, issue du latin, soit devenue un temps grammatical composé en

Le terme “attribut” dans la grammaire du français contemporain a une définition différente de
celle de Domergue. Aujourd’hui, étant une notion opposée à “épithète”, l’attribut est un
élément prédicatif par excellence: “les attributs sont des constituants nominaux ou adjectivaux
du GV essentiels à la prédication, en ceci qu’ils apportent à un référent (un GN ou
l’équivalent) une caractérisation par la médiation du verbe” (Monneret & Rioul 1999:111).
Pourtant, Domergue ne partage pas toutes les vues de de Breville au sujet de l’accord dans
l’ensemble. Ce dernier présente six convenances pour expliquer l’accord en français:
“puissance du genre dans les noms”, “puissance du nombre dans les noms”, “puissance
d’attraction” (i.e. relation étroite du nom avec l’adjectif), “puissance d’identité” (i.e. l’accord
entre le substantif et l’adjectif), “accord des personnes, ou du verbe avec le sujet” et “accord
des modes & des temps”. Domergue oppose ces six genres d’accord à un seul qui constitue
“l’accord entre le substantif et l’attribut”, notion importante chez lui pour expliquer les trois
genres d’accord que l’on examinera dans cette étude.
D’après la Grammaire d’aujourd’hui de Michel Arrivé et al., Marot a formulé cette règle
sous forme d’Épître à la demande de François Ier qui voulait attribuer un statut officiel au
français (1986:27 ).

français, elle est perçue comme une structure attributive du latin, selon laquelle
le participe (“attribut particulier”) modifie le nom (“substantif”) auquel il se
rapporte et ainsi s’accorde en genre et en nombre avec lui. Dans la formation
de la règle au XVIe siècle, la raison historique l’a en somme emporté sur la
raison linguistique. Aujourd’hui, l’argument fondé sur la grammaticalisation
éclaire l’état de cette structure qui perdure. D’après Jürgen Klausenberger
(2000:155), cette structure “avoir + participe passé” est incomplètement
grammaticalisée par rapport à d’autres temps grammaticaux français tels que le
futur ou le conditionnel, dans la mesure où le passé composé n’a pas atteint un
stade d’affixation, en gardant l’auxiliaire avoir comme une partie autonome,
indépendante du participe passé. 6 De ce point de vue, on peut comprendre plus
rationnellement pourquoi cette structure garde un fort souvenir de la “structure
attributive” d’origine latine (Pellat et al. 2002:349).
Par ailleurs, dans son article du 30 avril 1791, Domergue évoque
l’utilité de l’accord dans cette structure du point de vue linguistique: “les latins
déclinent leurs participes, et c’est fort bien; la règle de l’accord en est plus
générale. Tantôt nous [les Français] les déclinons, tantôt nous ne les déclinons
pas, et cette variété que l’oreille a conseillée, que l’usage consacre, a produit
quelques nuances heureuses” (II, 162). Cet argument qui met en valeur la
précision du sens rejoint celui de Vaugelas dans ses Remarques sur la langue
française. Vaugelas précise que “le plus grand de tous les vices contre la
netteté, ce sont les equiuoques, dont la plus-part se forment par les pronoms
relatifs, démonstratifs, et possessifs” (1880[1647]: II, 367). L’accord du
participe passé conjugué avec avoir peut être compris dans cette perspective:
éviter l’équivoque, ce qui était le souci de la clarté de langue française de l’âge
classique. Ainsi, on peut comprendre que l’accord du participe détermine le
sens, par exemple dans la phrase le fils de cette femme que j’ai vue.
Aujourd’hui, selon John Charles Smith (1996:115), la justification
fonctionnelle de l’accord du participe passé conjugué avec avoir se situe dans
cette perspective, mais à un niveau plus abstrait. Ainsi dans son article
“Surfonctionnalité et hyperanalyse: l’accord du participe passé dans les
langues romanes à la lumière de deux théories récentes”, il explique que
l’accord du participe passé conjugué avec avoir supprime “l’ambiguïté
potentielle” chez les locuteurs.
Dans le fonctionnement concret de la règle, dont le concept important est
d’identifier la transitivité du verbe, Domergue recourt au principe historique ou
linguistique selon le cas. Pour tester la transitivité du verbe en question, il

Le texte original de Klausenburger est le suivant: “Right-branching HABEO CANTUM and
HABEBAM CANTUM delivered the Romance compound past and pluperfect with the
auxiliary ‘to have’ remaining pre-posed, as a proclitic, incompletely grammaticalized in not
reaching affix (i.e., prefix) status”.

recourt au critère linguistique tel que la transformation de la phrase à la voix

passive. Ainsi, la phrase la lettre que j’ai écrite (où l’accord a lieu) est jugée
correcte étant donné que l’on peut dire la lettre est écrite à la voix passive.
Mais pour certains cas où ce test de passivation ne fonctionne pas et où le
bon usage n’a pas été toujours respecté par les grands écrivains,7 il recourt au
principe historique, en s’appuyant sur l’étymologie latine, par exemple pour
expliquer les phrases dans lesquelles le verbe en question est utilisé avec le
pronom “en” ou les phrases avec des verbes tels que “coûter”, “valoir” et
“vivre” qui prennent un “complément indirect”. Ainsi “en” qui pronominalise
des syntagmes en “de + nom”—d’origine latine, portant “de” qui est un signe
du complément d’objet indirect—ne s’accorde en aucun cas avec le verbe
transitif (voir appendice 1, nº 9); pour le verbe “coûter” par exemple,
Domergue recourt à la structure syntaxique du verbe latin équivalent, constare
(voir appendice 1, nº 18). Comme ce verbe latin ne prend pas d’accusatif,
équivalent du complément d’objet direct en français, coûter, qui n’est pas un
verbe transitif, ne doit s’accorder en aucun cas. Ainsi, Domergue recourt au
principe historique pour expliquer les usages de “en” et des verbes du type
Aujourd’hui, dans le fonctionnement de la règle d’accord, spécifiquement
dans l’identification de la transitivité du verbe, en principe les linguistes
tiennent compte du principe linguistique, une analyse formelle, c’est-à-dire la
transformation de la phrase à la voix passive. Pour les cas de “en” et des types
du verbe “coûter”, les linguistes d’aujourd’hui se penchent davantage sur le
principe linguistique, reconnaissant la pluralité de l’usage, en tolérant l’accord
et le non-accord. Ainsi lorsque l’usage hésite, l’arrêté du 28-12-1976 (Haby
1976; voir appendice 3), qui est en vigueur aujourd’hui en France, tolère
l’accord et le non-accord. Ainsi, le pronom en qui peut être un complément
d’objet direct selon l’intention du locuteur (contrairement à l’argument de
Domergue) peut entraîner l’accord, par exemple dans une phrase j’en ai
vus (en se référant à plusieurs personnes); pour les verbes coûter, valoir et
vivre, l’accord et le non-accord avec leur complément circonstanciel sont tous
les deux admis.

3.2 Avoir l’air suivi d’un adjectif

Pourtant, le système grammatical de Domergue qui divise toutes les
composantes de la phrase en substantifs et en attributs ne résout pas le
problème de l’accord pour avoir l’air. Ainsi il ne mentionne pas son système
grammatical, comme il l’a fait dans l’accord du participe passé conjugué avec

Domergue fait référence à la faute commise par Racine concernant l’usage de “coûter”:
“Racine a pourtant donné à coûté les inflexions adjectives”. (I,117) (1er mars 1785).

avoir. Cette question apparaît deux fois dans le Journal de la langue française
(1784-1788) dans l’article du 15 avril 1785 et dans l’article du 1er décembre
1786. Ces questions posées nous font constater qu’il existe déjà une variation
dans l’usage de cette expression à l’époque. Dans ces deux articles, la question
posée par le lecteur est de savoir dans la phrase, Elle a l’air spirituelle ou
spirituel, avec quelle partie du discours le complément de l’expression avoir
l’air, c’est-à-dire, l’adjectif spirituel, doit s’accorder en genre et en nombre.
Pour répondre à cette question, Domergue recourt à une analyse formelle
de la phrase en évoquant l’usage des écrivains et ainsi soutient l’accord de
l’adjectif avec air, c’est-à-dire privilégiant l’accord sémantique interne au
syntagme nominal à l’accord commandé par la syntaxe avec le sujet elle:

Aucun écrivain, que je sache, n’a osé dire: Julie dans sa parure n’avoit pas l’air
décente: Les habitants de cette province ont l’air brutaux. Il s’agit dans ces deux
phrases, non du fond du caractère, mais de l’apparence. C’est l’apparence, l’air qu’on
veut qualifier; c’est donc à air qu’il faut faire correspondre le qualificatif. On dit d’un
homme, d’une femme, de plusieurs hommes, de plusieurs femmes: il a, elle a ils ont,
elles ont l’air spirituel, comme on dit, il a, elle a, ils ont, elles ont la physionomie
spirituelle. (I, 426-27)

Par exemple, on peut voir les “grands écrivains” tels que Voltaire et Diderot
suivre cette règle (voir appendice 2.1). Ainsi, pour expliquer la grammaticalité
de ce cas, Domergue recourt au principe linguistique en décrivant l’usage des
Aujourd’hui, la conscience de la grammaticalité a changé et les linguistes
considèrent l’accord de l’adjectif avec le sujet comme grammatical, tout en
laissant la possibilité de l’accord avec l’air quand locuteur veut mettre l’accent
sur l’aspect extérieur. On peut en fait constater ces deux usages dans le
français moderne selon la nuance (voir appendice 2.2). Ainsi, d’après Pellat,
Riegel et Rioul, dans la grammaire actuelle l’expression avoir l’air,
réinterprétée comme sembler ou paraître, exige que l’adjectif qui le suit
devient un attribut du sujet, non un épithète du nom air (2002:242). La notion
de lexicalisation explique bien ce changement de la perception de l’expression
chez les locuteurs dans le sens que l’expression avoir l’air est aujourd’hui
devenue une expression figée, qui n’obéit plus à la règle de l’accord typique. 8

Marie-Françoise Mortureux (1997:123) explique bien le mécanisme de la lexicalisation:
“formellement, la lexicalisation fige un signifiant en une suite sonore ou graphique déterminée,
qui ne varie plus qu’en application stricte des règles d’accord au sein d’une phrase (nombre
des noms, genre et nombre des adjectifs, conjugaison des verbes)”.

3.3 La plupart de suivi d’un syntagme nominal

Comme dans le cas d’avoir l’air, le système grammatical de Domergue
n’explique pas la grammaticalité de l’accord dans le cas de la plupart de, par
exemple dans la phrase la plupart des hommes sont bons (I, 347). La plupart
de, dans la position du sujet, combiné avec un nom au pluriel, prend toujours le
verbe au pluriel, contrairement à l’analyse formelle de Domergue: d’après son
système grammatical, la plupart, qui est un “substantif”, doit s’accorder en
genre et en nombre avec les “attributs”, le verbe être et l’adjectif bon dans ce
cas. Pourtant, l’usage veut que le verbe et l’adjectif s’accordent avec des
hommes, qui est un complément de la plupart de. Ainsi Domergue se montre
sceptique et perplexe dans son commentaire du 15 juillet 1786 sur la
grammaticalité de l’expression la plupart de. La syllepse appliquée dans
l’expression la plupart se révèle constante au XVIIIe siècle. D’ailleurs,
Vaugelas 9 a soutenu cet usage de la plupart dans ses Remarques sur la langue
française, pourtant sans donner de justification.
Aujourd’hui cet usage est toujours prédominant et admis dans la
grammaire scolaire. On peut bien expliquer ce phénomène de syllepse, en
recourant à la notion de lexicalisation dans le sens que cette locution, devenue
une structure autonome de la structure d’origine latine (plerique), ne suit pas la
règle typique de l’accord. Par ailleurs, on peut comprendre la syllepse comme
un processus mental qui résiste à une analyse syntaxique, si on suit l’argument
de Daniel Luzzati et Reza Mir-Samii:

Le problème de l’accord nous plonge enfin dans la linguistique cognitive, c’est-à-dire

dans une linguistique des processus, par opposition à une linguistique des états. En
termes rhétoriques, cela s’exprime par la syllepse qui caractérise les accords qui se
font par la pensée au lieu de se faire par la syntaxe. Dans un processus énonciatif, un
énoncé doit non seulement répondre aux contraintes syntaxiques dictées par son
contexte antérieur, mais également correspondre à une situation mentale en constante
évolution. (1996:194)

Les auteurs de cet article éclaircissent le problème de l’accord autour de

l’usage des pronoms personnels et des pronoms relatifs en français de ce
double point de vue. Désormais, on peut comprendre pourquoi les

Voici les remarques de Vaugelas sur l’accord avec la plupart: “la pluspart regit tousjours le
pluriel, comme, la pluspart se laissent emporter à la coustume, et la plus grand’part, regit
tousiours le singulier, comme, la plus grand’part se laisse emporter. Mais pour montrer ce qui
a esté dit en la remarque precedente, que le genitif donne la loy au verbe, et non pas le
nominatif (ce qui est bien extraordinaire et à remarquer) on dit, la pluspart du monde fait, quoy
que l’on die tousiours, la plupart font, parce que ce genitif singulier, du monde, donne le
regime au nombre singulier du verbe; Et si vous dites, la pluspart des hommes, vous direz
aussi, font et non pas fait” (1880[1647]:I, 109-10).

grammairiens tels que Vaugelas et Domergue, et les linguistes d’aujourd’hui

n’ont pas rationnalisé le fondement de l’usage de la plupart de.
Dans son argument sur la plupart de, il est intéressant de constater que
Domergue ne perçoit plus le recours au latin comme désirable comme il l’a fait
à propos de l’accord du participe passé conjugué avec avoir: bien que, pour
justifier un tel usage, il puisse recourir au latin, en prenant par exemple
l’équivalent latin plerique, qui prend le verbe au pluriel, Domergue omet de le
faire avec la plupart de. L’usage des écrivains de l’époque, qui ne permet pas
son analyse formelle, n’est pas non plus évoqué par Domergue, bien qu’il y ait
eu recours dans les deux cas que l’on a observés:

On ne peut nier que la construction de la plupart avec un pluriel & un masculin, ne

soit une double inconséquence: je crois qu’on peut en attribuer la cause à l’attention
scrupuleuse qu’avoient nos premiers écrivains de calquer notre syntaxe sur la syntaxe
latine. 10

Ils traduisirent plerique par la plupart, & donnèrent à la plupart le nombre & le genre
de plerique. Toutes les langues offrent de pareilles bizarreries, parce qu’elles sont
l’ouvrage, non des grammairiens philosophes, mais du peuple ou de savants
absolument étrangers à la logique grammaticale. (I, 347)

Ainsi, Domergue reconnaît les difficultés d’appliquer sa propre théorie dans ce

cas particulier et évoque finalement le côté “illogique” de l’usage.

4. Conclusion
Dans le raisonnement à propos des trois structures évoqués ci-dessus,
Domergue se révèle comme grammairien qui voit dans le langage la mise en
œuvre de schémas logiques sous-jacents (la division des mots en substantifs et
attributs) et qui, par conséquent, s’efforce parfois, jusqu’à “forcer” la réalité du
fonctionnement de la langue, de justifier une pratique en fonction de ses vertus
À cette fin, Domergue concilie selon les cas le principe historique et le
principe linguistique. Il donne clairement la priorité à une analyse formelle,
fidèle à la mentalité du XVIIIe siècle où la mise en scène de la raison est
importante. Domergue soutient tantôt l’analyse formelle appuyée sur le
principe historique (dans le fondement de l’accord du participe passé conjugué
avec avoir et dans le fonctionnement de cette règle à l’intérieur des phrases qui
contiennent en ou des verbes du type coûter), tantôt l’analyse formelle appuyée
sur le principe linguistique (l’accord de l’adjectif avec l’expression avoir l’air,
le fonctionnement de la règle de l’accord du participe passé conjugué avec
avoir à l’intérieur de la plupart des phrases, excepté en, ainsi qu’avec les

En latin, plerique au sujet prend le verbe au pluriel: “pleraque illa Solonis sunt CIC. Leg. 2,
64, la plupart de ces détails sont de Solon” (Gaffiot 1934:1190).

verbes du type coûter). Seulement, Domergue ne justifie pas le bon usage du

cas de la plupart de: étant donné que ce cas n’entre pas dans son système
grammatical, il ne souscrit pas à une analyse formelle fondée sur le principe
historique (grammaire du latin), qui justifierait fort bien l’usage en français.
Aujourd’hui, les théories de grammaticalisation (Klausenburger) et de
lexicalisation (Mortureux), qui expliquent la mise en œuvre du principe
linguistique et du principe historique dans l’accord en français, permettent de
mieux évaluer l’état respectif de ces trois structures dans l’histoire de la langue
française. Pour résumer, nous constatons qu’au XVIIIe siècle, selon Domergue
dans le Journal de la langue française, l’analyse formelle soutenue par le
principe historique ou linguistique est considérée la plus adéquate pour
expliquer le phénomène de l’accord. Aujourd’hui, la norme linguistique
représentée dans la grammaire scolaire garde cette même tradition, tout en
tenant compte de l’usage varié des locuteurs et de leur nuance.


Arrivé, Michel, Françoise Gadet & Michel Galmiche. 1986. La Grammaire

d’aujourd’hui. Paris: Flammarion.
Colombat, Bernard. 1996. “L’accord dans la grammaire latine de la
Renaissance à Beauzée”. Faits de Langues 8.11-24.
Domergue, François-Urbain. 1978. Journal de la langue française, soit exacte
soit ornée. Tome I (1784-1788). Tome II (1791, 1792). Genève: Slatkine
Gaffiot, Félix. 1934. Dictionnaire illustré latin-français. Paris: Hachette.
Haby, René. 1976. Tolérances grammaticales ou orthographiques (l’arrêté du
28-12-1976: ANNEXE F, Ministère de l’éducation, fait à Paris), le 28
décembre. Le 11 janvier 2006.
Høybye, Poul. 1944. L’Accord en français contemporain. Copenhague: Andr.
Fred. Høst & Søns Forlag.
Klausenburger, Jürgen. 2000. Grammaticalization. Amsterdam &
Philadelphia: John Benjamins.
Luzzati, Daniel & Reza Mir-Samii. 1996. “La transition de l’accord par le
relatif: vers une théorie de ‘fenêtrage’”. Faits de Langues 8.193-202.
Monneret, Philippe & René Rioul. 1999. Questions de syntaxe française. Paris:
Presses Universitaires de France.
Mortureux, Marie-Françoise. 1997. La lexicologie entre langue et discours.
Paris: SEDES.
Pellat, Jean-Christophe. 2003. “Variation et plurisystème graphique au XVIIe
siècle”. Faits de Langues 22.139-150.

Pellat, Jean-Christophe, Martin Riegel & René Rioul. 2002. Grammaire

méthodique du français. Paris: Presses Universitaires de France.
Smith, John Charles. 1996. “Surfonctionnalité et hyperanalyse: L’accord du
participe passé dans les langues romanes à la lumière de deux théories
récentes”. Faits de Langues 8.113-120.
Vaugelas, Claude Favre. 1880 [1647]. Remarques sur la langue françoise. 2
Tomes. Ed. A. Chassang. Versailles: Cerf et Fils, Imprimeurs.

Appendice 1
Exemples de Domergue dans le Journal de la langue française (1784-1792)

A: accord
N: pas d’accord
( ): exemple non cité par le grammairien, mais qui constitue mon hypothèse d’après sa théorie
en général.
Blanc: cas non pris en compte.

Le numéro qui précède chaque exemple figure pour respecter l’ordre des exemples défini par

Les exemples de Domergue Vaugelas Domergue Aujourd’hui

(1647) (1784- (grammaire
1792) scolaire)
Phrases non- 1. la lettre que j’ai écrite A A A
contestées 4. Quels héros la vertu n’a-t-elle pas A A A
formés ?
17. Cette femme n’est pas aussi savante N N
que je l’avois imaginé.
14. Je lui ai rendu tous les services que (N) N N
j’ai pu/vu/dû.
16. Les troupes que j’ai fait marcher. N N N
2. La maison que j’ai commencé de N N N
3. La résolution que j’ai prise de A A A
8. La chaleur qu’il a fait N N

Les exemples de Domergue (dans ces exemples, Vaugelas Domergue Aujourd’hui

l’accord est marqué selon le principe de Domergue)
Phrases 5.Les lettres qu’ont écrites Cicéron & N A A
contestées Pline
6. Pauvre Didon, où t’a “réduite” N A A
de tes maris le triste sort ?
7. Cette femme que j’ai trouvée N A A

13. La femme que j’ai vue peindre (en N A A ou N

parlant d’une femme peintre)
15. Les livres que j’ai laissés tomber N A A ou N
9. César a plus gagné de batailles que N NÆAÆ N A ou N
les autres n’en ont lu. (changements
18. La somme que cette affaire m’a N A ou N
19. Les honneurs que mon habit m’a N A ou N
20. Les jours que j’aurois vécu N
A ou N
21. Le peu d’exactitude que j’ai N N
trouvé dans cet ouvrage
22. Le peu de femmes que j’ai vues (N) A
23. Combien de gloire il a acquise A
24. Que de valeur il a montrée A
25. Que d’hommes on a sacrifiés A
26. Quelle quantité de pierres il a A (entassées)

NB: Parmi les phrases contestées dans le tableau ci-dessus, les douze dernières phrases (13, 15,
9, 18, 19, 20, 21, 22, 23, 24, 25, 26) témoignent de la difficulté d’expliquer la règle de manière
systématique, comme le montrent les divergences entre grammairiens. Pour toutes les phrases
ci-dessus (1-26), les réponses de Domergue sont valables encore aujourd’hui. Toutefois, dans
la grammaire du français actuel, l’arrêté de 1976 en vigueur tolère que l’accord se fasse ou ne
se fasse pas pour les cas n° 13, 9, 18, 19 et 20. Pour les cas n° 21, 22, 23, 24, 25 et 26, l’arrêté
ne précise rien. Quant au cas n° 15, l’arrêté ne précise rien de spécifique, mais l’Académie
française suggère de ne pas accorder le participe laissé lorsqu’il est suivi d’un infinitif
(“Documents administratifs” du Journal Officiel, 6-12-1990).

Appendice 2

Source: American and French Research on the Treasury of the French Language (ARTFL)

avoir l’air

• Elle avait l’air timide, embarrassé.
• Elle a l’air bien coquet.
• Ces maximes qui ont l’air un peu libertin.
• Peut-elle avoir l’air si nouveau ?
• Mais dit Candide à Paquette, vous avez l’air si gai, si content, quand je vous ai
rencontrée, vous chantiez, vous caressiez le théâtrin avec une complaisance naturelle;
vous m’avez paru aussi heureuse que vous prétendez être infortunée.
• Moïse, Aaron, Vous avez l’air un peu gascon.
• Elle a l’air bien furibond

• Cette proposition, qu’un des deux la tue, et que l’autre la venge, n’a pas l’air sérieux.
• Justement, c’est l’ombre de Samuel; elle doit avoir l’air bien méchant ?
• Mais aussi comment une production enfouie quinze pieds en terre pendant tant de
siècles peut-elle avoir l’air si nouveau ?

• Ces cygnes ont l’air fier, bête et méchant.
• Elle a l’air mannequiné.
• Et puis nous avons l’air si penetré, si vrai !
• Jacques monta, et un moment après Jacques, l’hôtesse, qui avait vraiment l’air abattu.
• Elle avaient l’air égaré et furieux.

• Elle avait toujours l’air profonde.
• Pourtant les habitants ont l’air sain et heureux. (pas d’accord)
• Ces pauvres femmes, aux lèvres toujours ruisselantes, ont l’air stupides, mais
nullement malheureuses.
• Les expressions parfois heureuses ont l’air cherchées.

Mirbeau O.
• Oui, elle a l’air assez bien faite.

Bataille H.
• Nous avons vaguement l’air rastas.

Rolland R.
• Elle avait l’air soucieuse et ne levait pas les yeux.

• Pourquoi Nadine avait-elle l’air si réjouie ?
• Elle avait l’air élégante et presque féminine.
• Elle n’avait pas l’air intelligent. (pas d’accord)
• Ils ont l’air sain et sauf. (pas d’accord)

Appendice 3

Liste des tolérances grammaticales ou orthographiques (l’arrêté du 28-12-1976)

I. — Le verbe

11. Accord du participe passé conjugué avec avoir dans une forme verbale précédée de en
complément de cette forme verbale:

J’ai laissé sur l’arbre plus de cerises que je n’en ai cueilli.

J’ai laissé sur l’arbre plus de cerises que je n’en ai cueillies.
L’usage admet l’un et l’autre accord.
12. Participe passé des verbes tels que: coûter, valoir, courir, vivre, etc., lorsque ce participe
est placé après un complément:

Je ne parle pas des sommes que ces travaux m’ont coûté (coûtées).
J’oublierai vite les peines que ce travail m’a coûtées (coûté).
L’usage admet que ces verbes normalement intransitifs (sans accord du
participe passé) puissent s’employer transitivement (avec accord) dans
certains cas.
On admettra l’un et l’autre emploi dans tous les cas.

Université de Picardie Jules Verne


The 18th century, a critical period in the evolution of human knowledge,

played a major part in the attainment of knowledge relating to the sounds of
language. In fact, an articulatory description of French of an unparalleled
closeness to our own would be provided by César Chesneau Du Marsais’
successor in the drafting of the ‘Grammar’ articles in Diderot et d’Alembert’s
Encyclopédie ou Dictionnaire raisonné (1751-1765).
From his contribution to the Encyclopédie, to his Grammaire générale
published in 1767, to his articles in the dictionary Grammaire & Littérature
(1782-1786) as part of the Encyclopédie Méthodique (1782-1832), the
grammarian Nicolas Beauzée sketched out what we may consider to be
veritable “strokes” of modern phonetics.

1. Introduction
Le siècle des Lumières est présenté (Auroux & Calvet 1973) comme le
siècle ayant apporté, à l’exception du XXe siècle, les progrès les plus
spectaculaires en matière de connaissances des aspects phoniques de la langue
française. Le rôle des grammairiens-philosophes de l’Encyclopédie ou
Dictionnaire raisonné de Diderot et d’Alembert (1751-1777) dans le
développement de ce savoir est bien évidemment prépondérant mais les
développements ultérieurs poursuivis par l’un d’entre eux méritent d’être pris
davantage en considération: ceux de Nicolas Beauzée.
Au delà de son apport décisif dans les théories syntaxiques, le successeur
de Du Marsais dans la rédaction des articles de Grammaire du Dictionnaire
raisonné constitue en effet un témoin incontournable et pourtant mal connu
pour ses réflexions sur les sons de la langue.
C’est afin de valoriser l’apport de ce personnage clé de la phonétique
française, que nous allons nous attacher ici à présenter dans leurs grandes
lignes les théories qu’il développe sur les sons depuis sa participation au
Dictionnaire raisonné, en passant par sa Grammaire générale (1767), jusqu’au

dictionnaire Grammaire et Littérature (1782-1784-1786) de l’Encyclopédie

Méthodique (1782-1832). Cet itinéraire sera l’occasion de brosser le portrait de
Nicolas Beauzée en tant que grammairien à la fois ‘héritier’ des connaissances
antérieures mais aussi et surtout comme ‘novateur’ d’exception dans
l’acquisition et la diffusion des savoirs sur les sons de la langue française.

2. Note préliminaire
Une description détaillée de la nature des connaissances développées par
Nicolas Beauzée sur les sons de la langue au XVIIIe siècle ne peut se faire sans
avoir au préalable évoqué la genèse de ces théories majeures pour l’histoire de
la phonétique française. L’évocation récurrente des ‘connaissances’ du
grammairien sur les sons que nous allons faire tout au long de cette
communication, est un renvoi à l’aspect le plus abouti des théories que ce
dernier a développées au sein de trois de ses ouvrages principaux, à savoir
l’Encyclopédie ou Dictionnaire raisonné de Diderot et d’Alembert (désormais
DD), la Grammaire Générale (désormais GG) et le dictionnaire Grammaire &
Littérature de l’Encyclopédie Méthodique (désormais EM).
Sans vouloir nous attarder sur ce triptyque théorique,1 nous pouvons de
manière rapide et quelque peu schématique, avancer que les réflexions que
Beauzée développe sur les sons au sein de la DD semblent ne constituer que
l’embryon de l’importante réflexion qui va naître dans les pages de la GG et
qui va se poursuivre dans l’EM.
Successeur de César Chesneau Du Marsais dans la rédaction des articles de
Grammaire, Beauzée va fournir pour la DD pas moins de 147 articles, au
travers desquels va transparaître la première ‘strate’ de sa théorie sur les sons
de la langue.
Composée de trois ‘livres’ divisés en de nombreux ‘chapitres’, eux-mêmes
structurés en ‘articles’, la GG offre pour sa part une place centrale
incontestable à l’étude des sons, puisqu’elle y consacre un livre entier.
Précédant les livres Des Eléments de l’Oraison et Des Eléments de la Syntaxe,
le livre I, intitulé Des Eléments de la Parole, illustre toute l’importance que le
grammairien accorde aux sons.
La contribution du grammairien-philosophe dans l’EM se fait
exclusivement au sein du dictionnaire Grammaire & Littérature, ouvrage dont
il a en charge la partie ‘Grammaire’ et dont le volet ‘Littérature’ a été confié à
Jean-François Marmontel.
Libéré du statut de simple ‘remplaçant’ de Du Marsais qu’il avait endossé
dans la DD, dans cet ouvrage Beauzée peut laisser libre cours à l’expression de
ses théories grammaticales. En tirant au maximum profit de l’agencement

Sur ce point précis nous renvoyons au chapitre IV de notre travail de thèse.

thématique de cette encyclopédie dirigée par Charles-Joseph Panckoucke,2 ce

dernier thématise d’un point de vue terminologique les avancées apportées par
la GG et permet ainsi l’émergence d’un vocabulaire propre aux sons.

3. Une description articulatoire avant-gardiste

Ce petit retour sur la genèse des connaissances de Beauzée à propos de la
partie phonique de la langue étant fait, nous allons à présent mettre en exergue
les particularités les plus saillantes de sa théorie, celles qui nous permettent
d’en faire l’une des ‘clés inexploitées de la phonétique française’.
La richesse et le particularisme des théories de Beauzée sur les sons ne se
situent pas au niveau de l’identification des unités phoniques, 3 mais au niveau
de leur description articulatoire. C’est en effet sur ce point précis que l’auteur
de la GG formule selon nous des propos décisifs proches de nos connaissances
modernes, et pourtant inexploités.

3.1 Une mise en système originale

Afin d’introduire notre développement sur la nature des progrès apportés
par Beauzée dans la description articulatoire des sons du français, il nous faut
au préalable insister sur ce qui a permis l’émergence de ces progrès. Nous
devons pour cela faire référence à la fois à la ‘systématicité’ et à la grande
‘technicité’ qui caractérisent l’ensemble de ses théories grammaticales.
Beauzée est en effet un grammairien qui de manière très systématique fait
toujours référence aux théories des autres savants - que cela soit pour formuler
des louanges ou des critiques - avant d’avancer ses propres réflexions. En ce
qui concerne les théories sur les sons, ce dernier apporte une plus grande
“technicité” dans sa description, qui se traduit notamment par l’élaboration
d’une mise en système jamais atteinte jusqu’alors et qui plus est, illustrée par
des schémas particulièrement précis.
Nous avons reproduit ci-dessous les deux schémas proposés par Beauzée
dans sa GG et dans l’EM pour décrire les sons vocaliques (Figure 1) - qu’il
appelle VOIX - et les sons consonantiques (Figure 2) - qu’il appelle
ARTICULATIONS - du français.

Pour une présentation plus ample de cette encyclopédie, nous renvoyons à Darnton (1982),
Doig (1990, 1992), Ehrard (1991), Rey Roselyne (1992), Rey Christophe (2004), Teysseire
(1991) et Watts (1958).
Cf. Rey 2004.

Retentissantes labiales

orale grave â…pâte grave eu…jeûneur

aiguë a…pate orale aiguë eu…jeunesse
A EU muette e…je
nasale an…pante nasale eun…jeun
orale grave ê…tête orale grave ô…côte
aiguë è…tette aiguë o…cote
nasale ein…teinte nasale on…conte

E é…bâté U u…sujet
II ii…bâti OU ou…soumis
Figure 1: Le système des voix dans la GG de Beauzée

constantes variables

Faibles fortes
nasales........................................ M (mut)
labiales muettes...................... B (baquet) P (paquet)
sifflantes.................... V (vendre) F (fendre)
nasales………………………... N (nut)
dentales... D (dome) T (tome)
linguales muettes gutturales G (galle) K (calle)
orales liquides... L (loi) R (roi)

dentales... sifflantes Z (zèle) S (scelle)

palatales.. J (Japon) CH (chapon)
aspirées………………………………………………….... H (héros)
Figure 2: Le système des articulations dans la GG de Beauzée

Alors que jusqu’à présent peu de grammairiens s’étaient livrés à une

organisation rigoureuse des différents sons qu’ils identifiaient, 4 Beauzée est
certainement le premier à présenter son système sous la forme de tableaux à
plusieurs entrées. Cette particularité offre incontestablement une dimension
nouvelle à l’étude des sons, une dimension qui à défaut de marquer une
révolution totale, stigmatise la maturation des connaissances antérieures et fait
entrer de plein pied ce champ de connaissances dans un des répertoires
essentiels de l’analyse grammaticale.

Citons en exemples Dangeau et Duclos.

3.2 La description articulatoire des sons vocaliques

Une étude contrastive associant certains des savants les plus illustres des
XVIe (Meigret et Ramus), XVIIe (les grammairiens de Port-Royal et Dangeau)
et XVIIIe (Duclos, Du Marsais et Beauzée) siècles, nous a permis d’illustrer la
qualité de la description articulatoire des sons chez Beauzée. La Figure 3 ci-
dessous schématise les résultats obtenus pour les unités vocaliques:

ARTICULATOIRES Meigret Ramus Port- Dangeau Duclos Du Beauzée
(1545) (1572) Royal (1694) (1754) Marsais (1767)
(1660) (1754-
Aperture + + + + + + +
Nasalité/Oralité - - - (+) (+) (+) +
Etirée/Arrondie - (+) - (+) - - (+)
Antériorité/Postériorité - - - - - - (+)
Figure 3: Connaissances des modes articulatoires vocaliques

Les symboles ‘+’ et ‘-’ indiquent que le trait articulatoire spécifié est soit
retenu (‘+’), soit ignoré (‘-’), dans la description proposée par les différents
L’opposition en fonction du degré d’aperture de la bouche étant le seul
mode articulatoire unanimement identifié par les grammairiens de notre
échantillon, nous ne nous attarderons pas sur celui-ci. Nous ne nous attarderons
d’ailleurs pas plus sur l’opposition ‘nasalité/oralité’ puisque cette opposition
qui fait son apparition avec Dangeau au XVIIe siècle, est systématiquement
reprise dans les descriptions faites par les grammairiens-philosophes.
Soulignons néanmoins que Beauzée est le premier à opposer d’un point de vue
terminologique les unités “orales” et les unités “nasales”. Dangeau, Duclos et
Du Marsais évoquent certes le concept de nasalité, mais n’ont pas recours,
d’un point de vue terminologique, au concept d’oralité; cette différence
fondamentale est explicitée dans notre tableau par le recours au symbole ‘(+)’
pour ces grammairiens et au symbole ‘+’ pour Beauzée.
L’opposition “Étirée/arrondie”, est la première opposition articulatoire
marquant une différence notoire entre Beauzée et les autres grammairiens.
En effet, alors que pour les deux traits articulatoires précédents nous avions
pu constater qu’il s’agissait d’un savoir maîtrisé par l’ensemble des
grammairiens de notre échantillon, soit d’un savoir acquis à partir des théories
d’un grammairien en particulier et ensuite repris par ses successeurs,
l’opposition sur la labialité apparaît comme un savoir discontinu. En d’autres
termes il s’agit d’une opposition connue et mise en valeur à une certaine
époque, qui n’est pas reprise ensuite chez certains grammairiens, et qui fait
finalement sa réapparition chez d’autres.

Le premier à mettre en valeur ce trait distinctif est Ramus. Par la suite,

Dangeau et Beauzée le soulignent dans leur description articulatoire, alors que
Port-Royal, Duclos et Du Marsais ne l’évoquent pas du tout.
Faute de place, nous ne nous détaillerons pas la nature des propos
développés par Ramus et Dangeau, mais insisterons plutôt sur les réflexions de
Beauzée lui-même. Ce dernier semble lui aussi avoir conscience du fait que la
position des lèvres constitue un critère de différenciation des sons vocaliques:

Les lèvres forment autour de la bouche une espèce de cercle pour produire EU; elles
se serrent davantage & se portent en avant pour O; encore plus pour U; mais pour le
son OU, elles se serrent & s’avancent plus que pour aucun autre. (Beauzée 1767:8)

La citation ci-dessus montre effectivement d’une manière claire que le

grammairien a conscience du fait que la position des lèvres a une incidence
majeure dans la formation des sons vocaliques. Néanmoins, les seules
illustrations confirmant ce que nous avançons concernent uniquement l’aspect
arrondi des lèvres. L’aspect étiré n’est lui jamais explicitement évoqué. Ainsi,
à l’image de ce qu’avait pu faire Dangeau, l’auteur de la GG caractérise les
sons étirés en fonction de la position de la langue dans la bouche:

L’ouverture de la bouche nécessaire à la prononciation de cette voix, est de toutes la

plus aisée & celle qui laisse le cours le plus libre à l’air intérieur. Le canal semble se
rétrécir de plus en plus pour les autres: la langue s’élève & se porte en avant pour Ê;
un peu plus pour É; & les machoires se rapprochent encore un peu d’avantage pour I.
(Beauzée 1767:7)

En notant que ‘les machoires se rapprochent un peu d’avantage pour I’,

Beauzée traduit la position physiologique de production de ce son, mais sans
référer à la position des lèvres. Comme pour Ramus et Dangeau, nous avons
donc eu recours au symbole ‘(+)’ spécifiant que l’opposition ‘étirée/arrondie’
n’est pas véritablement marquée chez Beauzée.
L’opposition entre les voyelles ‘antérieures’ et les voyelles ‘postérieures’,
donc entre les voyelles prononcées sur l’avant de la bouche et celles
prononcées sur l’arrière de la bouche, apparaît quant à elle comme l’opposition
la moins maîtrisée par les grammairiens de notre échantillon.
Bien que ne l’explicitant pas aussi bien que l’opposition entre voyelles
orales et nasales, Beauzée inclut cette dernière opposition dans son système
descriptif et est le seul à le faire. Voici comment il la systématise: 5

Ce commentaire a déjà été proposé lors de notre description de l’opposition ‘étirée/arrondie’,
mais nous le reproduisons afin de confirmer l’idée selon laquelle les traits ‘étirée/arrondie’ et
‘antérieure/postérieure’ sont étroitement associés l’un à l’autre dans la théorie de Beauzée.

Le canal semble se rétrécir de plus en plus pour les autres: la langue s’élève & se
porte en avant pour Ê; un peu plus pour É; & les machoires se rapprochent encore un
peu d’avantage pour I. (Beauzée 1767:7)

Le grammairien utilise le positionnement de la langue dans la bouche

comme un critère articulatoire permettant de distinguer certains sons
vocaliques. A l’intérieur de ce groupe de sons qui constituent aujourd’hui tous
des sons antérieurs, Beauzée mentionne l’existence de voyelles qui sont plus
antérieures que d’autres. Pour ces dernières, en plus de s’élever plus ou moins -
pour marquer une ouverture plus ou moins grande de la bouche - la langue
avance également plus ou moins pour les former. Les positions plus ou moins
avancées de la langue servent ici à souligner des degrés parmi ces voyelles
Beauzée a en fait recours à deux traits distinctifs pour opposer les groupes
vocaliques [o], [ø], [y], [u] et [e], [], [i], [a]. Le premier groupe, constitué
d’un ensemble de voyelles qu’il appelle labiales, se caractérise par le trait plus
ou moins arrondi des voyelles. Le second groupe, constitué de voyelles
linguales, se caractérise par le trait plus ou moins antérieur, puisque la langue
avance plus ou moins pour produire ces sons.
Lorsque pour justifier sa catégorie des voix linguales il cite Dangeau, il fait
certainement allusion au commentaire suivant:

Si la voix qui formeroit un a, est un peu resserrée par la langue, qui s’approche du
palais, au lieu d’un a, nous entendrons un e. Si la langue est un peu éloignée du palais,
ce sera un è ouvert: si elle en est plus proche ce sera un é fermé: & si elle s’en
approche encore davantage, & que la voix ne sorte que comme par une petite
ouverture entre la langue & les dents, cela formera le son de i. (Dangeau 1754:62)

En se plaçant dans la lignée de Dangeau, à qui il emprunte visiblement le

concept de ‘linguales’, nous pouvons émettre l’hypothèse que Beauzée ne
s’aperçoit pas en fait que ses propos diffèrent profondément de ceux de son
En effet, lorsque Dangeau explique la différence articulatoire de ‘é’, ‘è’, et
‘i’, il fait essentiellement allusion aux degrés de différence que possède la
langue dans le sens de la ‘hauteur’ de la bouche. Chez lui, la différence
articulatoire entre ces sons s’explique donc par la distance qui sépare la langue
du palais. Chez Beauzée, cette différence articulatoire s’explique à la fois par
l’écart variant entre la langue et le palais - donc dans le sens de la ‘hauteur’- et
par la position plus ou moins avancée de la langue dans la bouche dans le sens
de la ‘longueur’, ainsi que peut le laisser croire le passage suivant:

[...] la langue s’élève & se porte en avant pour Ê; un peu plus pour É; & les machoires
se rapprochent encore un peu d’avantage pour I. (Beauzée 1767:7)

Nous n’avons malheureusement pas trouvé, même en parcourant les

articles de la DD et ceux de l’EM, d’autres illustrations de cette opposition
concernant la position plus ou moins avancée de la langue dans la bouche lors
de la production de certains sons vocaliques. Cette rareté pourrait - et nous en
sommes conscient - constituer un démenti à l’hypothèse séduisante que nous
venons d’émettre et qui ferait de Beauzée le seul grammairien de notre
échantillon à véritablement prendre en compte ce trait distinctif.
Dans le cas où Beauzée prendrait donc en compte cette opposition, il nous
reste à préciser que ce trait distinctif n’est pas clairement mis en évidence, ce
qui nous pousse une fois de plus à recourir au symbole ‘(+)’.
Notons pour finir ce développement à propos de cette opposition chez
Beauzée, que même si ce que nous relevons peut s’apparenter à une réflexion
sur des degrés plus ou moins importants sur des voyelles antérieures, nous ne
relevons aucune illustration de l’opposition véritable entre ce que nous
appelons aujourd’hui les voyelles antérieures et les voyelles postérieures.

3.3 La description articulatoire des sons consonantiques

A l’image de ce que nous avons fait pour les sons vocaliques, nous avons
reproduit ci-dessous un tableau (Figure 4) récapitulant l’ensemble des traits
articulatoires identifiés par chacun des grammairiens retenus dans notre

ARTICULATOIRES Meigret Ramus Port- Dangeau Duclos Du Beauzée
(1545) (1572) Royal (1694) (1754) Marsais (1767)
(1660) (1754-
Nasalité/Oralité - - - (+) (+) (+) +
Voisée/Non-voisée - - - (+) (+) (+) (+)
Liquides - - - (+) (+) (+) (+)
Occlusives/Fricatives - - - - - - (+)
Figure 4: Connaissances des modes articulatoires consonantiques

Le premier constat qui se dégage de notre tableau est qu’aucun mode

articulatoire des sons consonantiques n’est systématisé avant Dangeau.
Dans un second temps, ainsi que nous l’avons fait pour les sons vocaliques,
notons que l’opposition ‘nasalité/oralité’ est la seule à être systématisée d’un
point de vue terminologique, et elle l’est par Beauzée.
Enfin, et il s’agit de notre propos principal, il semblerait que Beauzée soit
le seul à avoir une intuition assez claire de l’opposition
‘Occlusives/Fricatives’. 6

Notons, à la différence de ce que montre notre tableau, que Ramus semble avoir eu dès le
XVIe siècle une intuition sérieuse de cette opposition articulatoire (Clérico 1995:293-294).

L’opposition ‘muettes/sifflantes’ que substitue Beauzée à l’opposition

‘muettes/semi-voyelles’ des grammairiens antérieurs prend une dimension
rencontrée chez aucun autre grammairien:

Les articulations orales muettes sont celles qui naissent d’une interception totale de
l’air sonore; de manière que, si la partie organique qui est mise en mouvement restoit
dans l’état où ce mouvement la met d’abord, il ne pourroit s’échapper aucune partie
de l’air sonore & l’on ne pourroit rien faire entendre de distinct. (Beauzée 1767:52)

Les articulations orales sifflantes sont celles qui naissent d’une interception imparfaite
de l’air sonore; de manière que, quand la partie organique qui est mise en mouvement
resteroit dans l’état où ce mouvement la met d’abord, il s’échapperoit pourtant assez
d’air sonore pour faire entendre l’articulation même dont il s’agit, et même pour la
faire durer longtemps comme une sorte de sifflement, de même que l’on fait durer les
voix simples aussi longtemps que les poumons peuvent fournir de l’air: d’où vient
que plusieurs grammairiens ont donné à ces articulations le nom de demi-voyelles
(semivocales). (Beauzée 1767:52-53)

Parmi les grammairiens de notre échantillon, aucun ne se livre à une

distinction aussi pertinente. Certes Dangeau a recours à l’appellation
‘Sifflantes’, mais il ne classe dans cette rubrique que quatre sons
consonantiques, à savoir, Z, S, J et CH, et classe les sons F et V dans les
consonnes Labiales. De plus, ce classement n’est à aucun moment justifié. A
l’inverse de ce qu’il avait pu faire pour ses labiales et palatales, Dangeau
n’explicite pas le choix du nom de sifflantes. Il fournit seulement des
explications relatives à la position de la langue lors de la production de ces
La justification des catégories labiales et palatales peut peut-être
s’expliquer par le fait que ces deux catégories sont des catégories relevant du
lieu d’articulation, alors que la catégorie sifflantes est une catégorie relevant du
mode d’articulation. Dangeau introduit certes le concept de sifflantes mais
semble véritablement embarrassé avec celui-ci.
En identifiant aussi finement l’opposition entre consonnes occlusives et
consonnes fricatives, Beauzée apparaît donc comme le grammairien aux
connaissances les plus abouties.

4. Vers notre phonologie moderne

Comme le signale Sylvain Auroux, “le concept de phonème n’est pas lui-
même thématisé” (Auroux 1992:599) au XVIIIe siècle, mais les travaux de
certains grammairiens, et plus particulièrement ceux de Beauzée, nous
permettent de mettre en évidence des connaissances situées “au niveau d’une
épiphonologie” (Auroux 1992:599).
Les théories énoncées durant le siècle des Lumières par les grammairiens-
philosophes introduisent en effet selon nous une évolution vers l’avènement de

la phonologie, une évolution en faveur de l’étude des sons seulement en

fonction de leurs propriétés phoniques ou acoustiques. En d’autres termes,
dans cette période historique se joue le passage d’une morphophonologie à une
phonologie véritable; et Beauzée est l’un des artisans essentiels de cette

4.1 L’opposition foibles/fortes

L’opposition entre consonnes ‘foibles’ et consonnes ‘fortes’ constitue l’un
des points de réflexion essentiel en faveur du glissement progressif qui semble
s’être opéré au siècle des Lumières entre la morphophonologie et la phonologie
Le créateur de cette opposition entre consonnes ‘fortes’ et consonnes
‘foibles’ est le grammairien Dangeau (Louis de Courcillon, Abbé de):

J Ch
La première colonne est des lettres qu’on peut nommer foibles, & l’autre de
celles qu’on peut nommer fortes: la première est de celles qui sont précédées par une
petite émission de voix, & l’autre est de celles qui n’en ont point. (Dangeau 1754:47)

Bien que la notion de vibration des cordes vocales ne soit à aucun moment
évoquée dans les processus de production 7 et de distinction des sons que
Dangeau identifie, la répartition des sons proposée par ce dernier s’avère tout à
fait troublante puisqu’elle correspond exactement au dégroupement que nous
opérons aujourd’hui entre nos sons voisés et non-voisés.
En évoquant la présence d’une petite ‘émission de voix’ au début de la
production des consonnes ‘foibles’ - émission qui pourrait avoir été suggérée
par l’impression laissée lors de l’adduction des cordes vocales - Dangeau
semble avoir ainsi pris conscience d’une différence articulatoire fondamentale
entre les consonnes ‘foibles’ et les consonnes ‘fortes’.
Beauzée apporte pour sa part - dans sa GG - une définition de l’opposition
‘foible/forte’ qui n’est pas calquée sur celle de Dangeau mais qui nous laisse
également croire qu’il pourrait s’agir de notre opposition entre sons voisés et
non voisés. Sa définition semble plus basée sur celle de son prédécesseur dans
Ayant rédigé ses théories avant le célèbre Mémoire de Ferrein, Dangeau n’a pu bénéficier des
lumières de ce dernier à propos des données physiologiques de la parole et notamment sur le
rôle des lèvres de la glotte - lèvres auxquelles il donne le nom de ‘cordes vocales’ - et
n’évoque donc à aucun moment le rôle prépondérant joué par cet organe dans la production de
la parole.

la rédaction des articles de Grammaire de la DD, Du Marsais, dans la mesure

où elle s’appuie sur l’idée de différence de “force” dans la production des sons.

Les articulations organiques peuvent se diviser encore en deux espèces générales, les
constantes & les variables: & cette division est relative au degré de force avec lequel
se fait l’explosion, soit que ce degré dépende de la quantité de la force expulsive, ou
qu’il soit proportionné à la résistance de la partie organique qui intercepte la voix.
(Beauzée 1767:57)

Cette définition laisse très nettement percevoir l’aspect très ‘systématique’

qui caractérise la totalité de la réflexion grammaticale de Beauzée, puisque ce
dernier ne se contente pas de mettre en opposition les consonnes ‘foibles’ et les
consonnes ‘fortes’, mais oppose les consonnes qui sont susceptibles d’être
‘foibles’ ou ‘fortes’ (articulations ‘variables’) aux consonnes qui ne sont pas
susceptibles de l’être (articulations ‘constantes’).
Chaque son semble nécessiter pour sa production une situation
physiologique particulière: occlusion complète ou incomplète de la bouche
(‘muettes’ vs ‘sifflantes’), rôle de la langue (caractéristiques des linguales,
qu’elles soient dentales, gutturales, ou liquides). Or, Beauzée identifie des
paires d’articulations présentant les mêmes caractéristiques articulatoires. Le
seul élément permettant de les distinguer est alors le degré de force de
l’explosion nécessaire à leur production. Ce degré de force peut en effet être
plus ou moins important et créer soit des articulations ‘foibles’, soit des
articulations ‘fortes’. La prise en considération de l’intensité de l’explosion
nécessaire à la production d’une articulation, constitue ce que nous appelons
aujourd’hui un ‘trait distinctif’.
Au sein des articles du dictionnaire Grammaire & Littérature de l’EM,
Beauzée se montre plus loquace sur cette opposition entre articulations
‘foibles’ et articulations ‘fortes’, puisqu’il formule une entrée dictionnairique
pour chacune de ces notions:

(N.) FOIBLE, adj. Qui n’a pas toute la vigueur dont il est capable. Les articulations
variables sont foibles ou fortes. Voyez VARIABLE.
On appelle foibles celles qui n’interceptent pas la voix avec toute la vigueur dont est
capable la résistance de la partie organique qui en est le principe. B, V, D, G, Z, J,
sont des articulations variables foibles. Voyez ARTICULATION & FORT. (Beauzée

(N. ) FORT, E, adj. Qui a toute la vigueur dont il est susceptible. Les articulations
variables sont foibles ou fortes. Voyez VARIABLE. On appelle fortes, celles qui
interceptent la voix avec toute la vigueur dont est capable la résistance de la partie
organique qui en est le principe. P, F, T, K, S, CH, sont des articulations fortes. Voyez
ARTICULATION & FOIBLE. (Beauzée 1782-1786:FORT)

Ces deux définitions font très largement écho aux commentaires déjà
formulés dans la GG puisqu’elles soulignent la variation du degré d’opposition
des organes nécessaires à la production des articulations. Ces définitions
pouvant correspondre à celles de nos unités voisées et non voisées ne suffisent
toutefois pas à lever l’ambiguïté que peut faire naître une telle division aux
yeux des savants du XXIe siècle (Auroux & Calvet 1973:79).

4.2 L’utilisation des paires minimales

Dans la classification des sons qu’il propose, que cela soit dans la DD, la
GG, ou l’EM, Beauzée a systématiquement recours à une illustration sous
forme lexicale. Chaque son est mis en contexte au sein d’un lexème. Ceci
permet ainsi au lecteur de disposer d’une représentation phonique de l’élément
Si nous regardons dans un premier temps le système des unités vocaliques,
le recours aux paires minimales semble reposer sur la distinction des voix
issues d’une même voix fondamentale. C’est ainsi que sont opposés les sons
[], [a], et [ã], dans les mots ‘pâte’, ‘pate’, et ‘pante’ (ou ‘plante’ dans l’EM),
[], [], 8 et [ẽ], dans les mots ‘tête’, ‘tète’, et ‘teinte’, [œ], [ø], [] et [œ], dans
les mots ‘jeûneur’, ‘jeunesse’, ‘âge’, ‘jeun’, et [], [o], et [õ], dans les mots
‘côte’, ‘cote’, et ‘conte’, respectivement issus des voix fondamentales A, Ê,
EU, et O.
A la vue de ces exemples, le recours aux paires minimales semble donc
caractériser les sons dont les propriétés articulatoires sont les plus proches.
Nous pouvons néanmoins noter que Beauzée semble également avoir voulu
recourir aux paires minimales pour les voix fondamentales constantes, puisque
dans la GG et l’EM il utilise les exemples ‘bâté’ et ‘bâti’ pour illustrer les sons
[e] et [i]. Si nous tenons compte du fait que dans la DD ces sons étaient
illustrés par les mots ‘présent’ et ‘prison’ - pas aussi proches d’un point de vue
phonique que les exemples de la GG et de l’EM - il semblerait que Beauzée ait
fait preuve d’une volonté supplémentaire de mettre en évidence la valeur
phonologique de ces unités. Il reste toutefois regrettable que les voix U et OU,
illustrées par ‘sujet’ et ‘soumis’, n’aient pas également bénéficié de cette
même maturation.
Intéressons-nous à présent à l’usage des paires minimales que fait le
grammairien dans ses classifications des sons consonantiques. Dans ces
dernières, le recours aux paires minimales repose sur la distinction de deux
catégories de sons: les articulations ‘variables’ et les articulations ‘constantes’.
Au sein des articulations variables, nous avons vu que Beauzée opposait les
articulations ‘foibles’ et les articulations ‘fortes’, et c’est justement pour

Nous avons reproduit le son [] pour traduire le è présent dans tète, dans la mesure où nous
ne distinguons pas dans notre alphabet phonétique la différence entre tète et tête.

illustrer la différence entre ces sons très proches qu’il utilise des exemples sous
forme de paires minimales. Ces derniers restituent au mieux l’unicité du son
Un peu à l’image de ce que nous avions souligné pour les paires minimales
dans le système vocalique, Beauzée ne limite pas l’usage des paires minimales
aux seules articulations variables. Il fait appel à elles pour les articulations
constantes, à savoir les articulations nasales M et N, et les articulations liquides
L et R.
A aucun moment mises en parallèle dans la classification, les articulations
M et N sont illustrées par des exemples quasi-identiques: ‘Mort/Nort’ (‘Nord’
dans l’EM). Il en est de même pour L et R, illustrées par ‘Loi/Roi’. Beauzée
emploie clairement ici une exemplification visant à marquer la valeur
phonologique de ces différentes unités.
Soulignons enfin que dans la DD et l’EM Beauzée n’illustre pas les sons [j]
et [] pour la simple et bonne raison qu’il ne les retient pas dans sa
classification. Il ne les considère pas comme de véritables articulations.
Résolument ancré dans les théories de Beauzée, le recours aux paires
minimales n’est certes pas une nouveauté, puisque nous avons vu que pour
systématiser l’opposition entre les consonnes ‘foibles’ et les consonnes
‘fortes’, Duclos et Du Marsais 9 avaient déjà employé ce procédé, mais ce
recours est très régulier chez l’auteur de la GG et, qui plus est, à un endroit
aussi stratégique que dans ses classements des sons.

5. Conclusion
A défaut d’avoir révolutionné les connaissances sur les sons de la langue
française, Nicolas Beauzée s’impose à nous comme l’un des grammairiens qui
a le mieux su exploiter les connaissances de ses prédécesseurs et de ses
contemporains pour faire de l’étude des sons une discipline à part entière: la
La mise en avant, pour la première fois, d’un système entièrement structuré
de l’aspect phonique de notre langue, marque indubitablement l’aboutissement
d’une maturation construite sur plusieurs siècles. Héritier des Meigret, Ramus,
Dangeau, Duclos, Du Marsais et autres, l’auteur de la GG, notamment à travers
sa thématisation lexicale de l’opposition ‘nasalité/oralité’, mais aussi à travers
sa prise en considération de l’opposition ‘muettes/sifflantes’, apporte une
dimension technique jamais atteinte aux études sur les sons.
Par ailleurs, par le biais de son opposition entre articulations ‘foibles’ et
articulations ‘fortes’, opposition déjà systématisée par d’autres grammairiens,
mais surtout par le recours systématique et stratégique qu’il fait des paires
Du Marsais avait d’ailleurs brillamment illustré le recours aux paires minimales à l’article
CONSONNE de la DD, en fournissant un classement très séduisant pour opposer les
consonnes ‘fortes’ aux ‘foibles’.

minimales, Beauzée nous fait basculer d’une morphophonologie assez

répandue au XVIIIe siècle vers une épiphonologie qui préfigure la phonologie
à venir.

Sources primaires
Arnauld, Antoine & Lancelot, Claude. 1969 [1660]. Grammaire générale et
raisonnée, contenant les fondements de l’art de parler, expliqués d’une
manière claire et naturelle, les raisons de ce qui est commun à toutes les
langues, et des principales différences qui s’y rencontrent, etc. Paris:
Republications Paulet.
Beauzée, Nicolas & Marmontel, Jean-François. 1782-1784-1786. Encyclopédie
Méthodique. Grammaire & Littérature. 3 vol. Paris: Panckoucke; Liège:
Beauzée, Nicolas. 1974 [1767]. Grammaire générale ou exposition raisonnée
des éléments nécessaires du langage, pour servir de fondement à l’étude de
toutes les langues. Paris: J. Barbou, réédité en fac-similé; Stuttgart-Bad
Cannstatt: Friedrich Fromann Verlag.
Dangeau, de Courcillon Louis de. 1969 [1754]. Essais de grammaire (1694),
repris dans Opuscules sur la langue françoise (1754). Genève: Slatkine
Diderot, Denis, & Alembert, Jean Le Rond d’. 1751-1766. Encyclopédie, ou
Dictionnaire raisonné des sciences, des arts et des métiers, par une société
de gens de Lettres. Stuttgart: F. Frommann Verlag – G. Holzboog, 1990.
Duclos, Charles-Pinot. 1993 [1754]. Remarques sur la Grammaire de Port-
Royal. Paris [Genève: Slatkine Reprints].
Encyclopédie méthodique ou par ordre de matières par une société de gens de
lettres, de savants et d’artistes; précédée d’un Vocabulaire universel,
servant de Table pour tout l’Ouvrage, ornée des Portraits de MM. Diderot
et d’Alembert, premiers Editeurs de l’Encyclopédie. 1782-1832. 210 vol.
Paris: Panckoucke; Liège: Plomteux.
Meigret, Louis. 1545. Traité touchant le commun usage de l’escriture
francoise, auquel est débattu des fautes & abus en la vraye & ancienne
puissance des lettres. Paris: Marnef & Janot.
Ramée, Pierre de la (dit Ramus). 1972 [1572]. Grammaire. Paris [Genève:
Slatkine Reprints].

Sources secondaires
Auroux, Sylvain. 1992. “Note sur les progrès de la phonétique au XVIIIe
siècle”. Histoire des idées linguistiques, Tome 2, Philosophie et Langage,
598-606. Bruxelles: Mardaga.
Auroux, Sylvain & Louis-Jean Calvet. 1973. “De la phonétique à
l’apprentissage de la lecture”. La Linguistique 9.71-81.

Clérico, Geneviève. 1995. Analyses phoniques et prosodiques au XVIème

siècle. Origine et préhistoire d’une discipline. Thèse de doctorat d’Etat,
Université de Paris VIII Saint-Denis.
Darnton, Robert. 1982 [1979]. L’Aventure de l’Encyclopédie. 1775-1800. Un
best-seller au siècle des Lumières. Traduction de Marie-Alyx Revellat.
Préface d’Emmanuel Le Roy Ladurie. Paris: Perrin.
Doig, Kathleen H. 1990. “Notices sur les auteurs des quatre volumes de
“Discours” du Supplément à l’Encyclopédie”. Recherches sur Diderot et
sur l’Encyclopédie 9.157-159.
Doig, Kathleen H. 1992. “L’Encyclopédie méthodique et l’organisation des
connaissances”. Recherches sur Diderot et sur l’Encyclopédie 12.59-69.
Ehrard, Jean. 1991. “De Diderot à Panckoucke: Deux pratiques de l’alphabet”.
L’Encyclopédisme: actes du Colloque de Caen, 12-16 janvier 1987 ed. par
Annie Beck, 234-252. Paris: Klincksieck.
Rey, Christophe. 2004. Analyse et informatisation des articles traitant de
l’étude des sons dans le dictionnaire Grammaire & Littérature de Nicolas
Beauzée et Jean-François Marmontel, issu de l’Encyclopédie Méthodique.
Thèse de doctorat. Aix-en-Provence.
Rey, Roselyne. 1992. “Le cas des sciences de la vie”. Recherches sur Diderot
et sur l’Encyclopédie 12.41-57.
Teysseire, Daniel. 1991. “A propos de l’Encyclopédie Méthodique”.
Recherches sur Diderot et sur l’Encyclopédie 11.142-149.
Watts, Georges-B. 1958. “The Encyclopédie méthodique”. Publications of the
Modern Language Association of America 73, 348-366.

University of São Paulo, Brazil


The time and place under analysis, 19th century Brazil, constitutes a
privileged moment for the historiographer of linguistics attentive to the
methodological requirements of his/her craft. It is an important point in time of
reaffirmation of a Brazilian identity and of the increase of interest in the
indigenous languages of Brazil, manifested by the re-edition of their classical
texts, grammars and dictionaries, followed by a wave of scientific expeditions
that produced original materials. In this paper, an attempt will be made to
establish a link between the first efforts to set up a typological classification of
the native languages and to illustrate the individual biases that entered into
these scientist-travelers’ analysis. Particular attention will be paid to an
investigation of the status ascribed to the so-called Brazilian and Paraguayan
‘general languages’; respectively, the Tupí and the Guaraní.

1. Introduction
This paper concerns certain relations between linguistics and Brazilian
cultural diversity. The central period of reflection is the first half of the 19th
century and the scene is the recently independent Brazilian nation. It is an
important moment for the institutionalization of Brazilian identity and
everything it involves: the delineation of the geographical and political borders
of the country; the revival of interest in its colonial history; the collection and
the establishment of a representative literature in its official language,
Brazilian Portuguese, and, hopefully for this generation, the solution once and
for all of the ‘indigenous issue’; i.e., that of their definitive integration into
civilized society.
Although quite heterogeneous in its constitution, one of the long-lasting
myths of Brazilian culture is its celebrated linguistic unity. The belief that we

I am grateful to Aline Cruz and Luciana Gimenes for their help in locating various items of
the bibliography. My special thanks to Stéphane Goyette for his careful revision of my English
manuscript. Errors that still remain are my own responsibility.

speak one single language, from the north to the south, has become a cliché
since Joseph of Anchieta (1534–1597) wrote his grammar on “the most spoken
language along the Brazilian coast” (Anchieta 1990[1595]). The designations
associated with this language, as well as the functions and values aggregated to
it during the subsequent centuries may differ (Rosa 2003, Altman 2003a,
Gimenes 1999, Rodrigues 1996), but the strategy of electing one single variety,
to the detriment of hundreds of others, as the supra-regional means of
communication lasted at least up to the end of the 19th century, if not later,
even after the indigenous peoples ‘ascended’ to the category of free citizens,
and the Portuguese language was decreed the official language of the country
(Directory of 1787, of the Marquis of Pombal). This being the case, in the
search for the specificity of the Brazilian nation, the persisting existence of
dozens of other languages, still spoken in the interior of the country, was
perceived as an important obstacle to overcome.
The activities that led to the implementation of these policies were in great
part organized by the Instituto Histórico e Geográfico Brasileiro (IHGB),
created in 1838 by the Brazilian Emperor D. Pedro II (1825−1891), with the
task of supporting systematic research on Brazilian history, the country’s
inhabitants and its territory, and to promote regular meetings about its
associates’ findings. In the wake of the creation of the IHGB, there followed an
immediate spurt of interest in Brazilian languages and in the reedition of their
classical texts, grammars and dictionaries — among others, Anchieta’s 2nd
(1874) and 3rd (1876) editions were published in quick succession; also
Montoya (1876[1640]) and Figueira (1878[1687]) were printed as new
editions, besides various colonial chroniclers that were also reedited and
published in the journal of the Institute — and a wave of scientific expeditions
that produced original materials, such as those which Karl Friedrich Philipp
von Martius (1794−1868), Batista Caetano (1826−1882), Couto de Magalhães
(1837−1898), Charles Hartt (1840−1878), Karl von den Steinen (1855−1914)
and Paul Ehrenreich (1855−1914) took part in.
Not by chance, the IHGB was soon to become a pivotal institution not only
for the implementation of the strategies of the government, but also for the
organization of those domains perceived as related: geography, history,
archeology and ethnography. A Brazilian field linguistics emerged at this
moment, not as an autonomous discipline, but as a research area related to
historical studies on the indigenous peoples. This linguistics of sorts was to
expand in the following years, quite apart from the more prestigious field of
Portuguese philology, taking root in the Colégio Pedro II (also created in Rio
de Janeiro) in 1837-1838. As a consequence, this recently inaugurated field of
language studies was soon to split into two types of practices: one, perceived
as more practical, not requiring specific education, and related to field work on
the native languages; the other, perceived as more theoretical ⎯ and ‘more

scientific’ ⎯ related to the work of philologists on the (Indo-)European


2. The Brazilian ‘race’

The claim that the indigenous issue in Brazil could be solved through the
study and the teaching of their native languages did not originate in the 19th
century. One should keep in mind the hard work of Catholic missionaries in
the 16th and 17th centuries in recording and describing the ‘general languages’
(Zwartjes & Hovdhaugen, eds. 2004; Zwartjes & Altman, eds. 2005). But
while one of the main motivations of the missionaries was the catechization of
the native people for the service of the colonial empire, in the 19th century, the
focus was on the building of “Brazilian integrity and independence as a
civilized nation.”
In this context, the living indigenous languages — rather than the ‘dead
languages’, Latin, Hebrew, Greek — were said to be the ones to be studied,
preserved, and classified, in order to reveal the history of their speakers, their
origins and routes of migration (Varnhagen 1841:57). The ‘Black’ issue,
although an important part of the equation of racial integration, was put aside
by the intellectuals of the time who gravitated around the IHGB. As a rule,
there was no particular interest in knowing the African languages that entered
the country through the (still active) slave trade; on the contrary, there was a
stony silence about matters relating to slavery in general.
In 1840, two fundamental papers circulated in the IHGB. The first, by
Francisco Adolfo de Varnhagen (1816−1878) ⎯ an important figure in 19th
century Brazil: monarchist, diplomat, and the celebrated author of the first
general history of Brazil ⎯ was “Sobre a necessidade do estudo e ensino das
línguas indígenas do Brasil” [On the necessity of studying and teaching
Brazilian indigenous languages] (Varnhagen 1841), in which he proposed to
the members of the Institute three urgent measures regarding the indigenous
issue: 1) the creation of (religious) schools where the indigenous languages
could be studied and taught; 1 2) the reprinting of the documents, manuscripts,
grammars and dictionaries on these languages made by 16th and 17th century
missionaries; and 3) the foundation of an ethnographical division within the
Institute, whose responsibility would be the establishment of the names of the
various autochthonous nations existing within Brazilian territory; as well as the
names and terms to be given to their languages and dialects, locations,
migrations, beliefs, archeology, social customs and habits, and also to propose
strategies to educate them.
The second essay, by Martius (1845[1840]), was his prize-winning
monograph on “Como se deve escrever a história do Brasil” [How to Write the

It is perhaps not superfluous to remind the reader that, after the expulsion of the Jesuits in
1759, the Brazilian educational system had completely collapsed.

History of Brazil], in which he emphasized that the three ‘racial elements’ ⎯

the Indian, i.e., the American native, the Caucasian, i.e, the European
Portuguese and the Black, referred to by him as the Ethiopian ⎯ had to be
taken into account in the configuration of a Brazilian history, not in the same
proportion, though, since they should not have the same weight in the equation.
The European, or ‘civilized’ element, should prevail over the other two,
inferior to it.
What is relevant in Martius’ text, from the point of view of the observation
being made here, is his statement that American natives were not primitive
people; on the contrary, they were the decadent remnants of a once-great
Brazilian civilization (not unlike the Incas in Peru, or the Aztecs in Mexico)
whose language had degenerated into the various corrupted dialects and other
existing varieties of language. For Martius (1845[1840]; 1905), the existence
of so many different languages among the ‘barbarian peoples’ was evidence of
the continuous process of decay that had befallen that ancient civilization. For
the sake of the Brazilian nation, it was urgent, according to him, that its ancient
unity be restored by creating ways to unify its original language. This language
should be Tupi, called by him the ‘Brazilian general language’. As he notes:
A lingua principal fallada outr’ora pelos indios do Brazil em vastíssima extensão,
e entendida ainda em muitas partes, é a língua Geral ou Tupi. É sem dúvida
muito significativo que um grande complexo de raças Brazileiras entendam este
idioma. [...] e não podemos duvidar que todas as tribus, que n’ella sabem fazer-se
intelligiveis, pertençam a um único e grande Povo, que sem duvida possuiu a sua
historia própria, e que de um estado florescente de civilisação, decaiu para o actual
estado de degradação e dissolução [...]. (Martius 1845[1840]:387, my emphasis)

[The main language formerly spoken by Brazilian natives, over a vast territory,
and still understood in many regions, is the General language, or Tupi. It is no
doubt very telling that a great many Brazilian races understand this idiom. […] and
we cannot doubt that all tribes, which can make themselves understood by its means
[i.e. by means of this language], belong to a single and great People, which no doubt
had its own history, and which, from a flourishing state of civilization, collapsed to its
present state of degradation and dissolution […].]

The theory of the existence of an ancient Brazilian civilization had an

important methodological consequence as far as the study, in the 19th century,
of Brazilian languages are concerned: as the barbarians had no history (sic),
nor writing, the only way to reconstruct their past was to observe their present
languages and look, in their forms and vocabulary, for the remains of their
former grandeur. Linguistic data were, though, the observable dimension of
another phenomenon to be investigated, namely, the origin of homo
americanus, and the reasons for his moral and civil dissolution (cf. also
Varnhagen 1874[1849]:368 “Cada vez me convenço mais de que para o estudo
das raças indígenas nada nos póde ser de mais socorro do que o conhecimento

das suas línguas” [I am more and more convinced that for the study of the
indigenous races nothing can help us more than knowledge of their
On the one hand, this policy made room for pragmatically oriented
linguistic work: the building of a Brazilian history depended upon the
collection and establishment of linguistic data along with ethnographic data.
On the other, however, it had a perverse effect. Instead of stimulating the study
of the dozens of languages which still survived in the interior of the country,
what happened was that the study and codification of Tupi (to which all other
languages were to be compared and ‘reduced’) prevailed once again. In
practice, this meant concentrating the investigation on Tupian radicals
(including those of its dialects) and vocabulary. It was the emergence of the
movement which the following generation would pejoratively refer to as
Tupimania (Ehrenreich 1892).

3. The general languages, the Tupis, the Guaranis and the Tapuias
It is important to remark that the term general language, as well as the
terms Tupi and Guarani, expressed different realities through time.
Nowadays, it is agreed that Tupi was one of the varieties ‘of the most
spoken language along the Brazilian coast’ in the 16th and 17th centuries, and it
is associated by Rodrigues (1994, 1996) with the Língua Geral Paulista, a
language derived from this ancient Tupi language, used in the colonization of
the south and southeast regions of the country, i.e., São Paulo, Minas Gerais,
South Goiás, Mato Grosso and North Paraná (see also Prezia 2000). The
‘Paulista general language’ is wholly extinct today and, still according to
Rodrigues (1994, 1996), the only important document of it we have today is a
dictionary of verbs, by an unknown writer, published in Martius’ Glossaria
What we call Tupinambá, also spoken along the Brazilian coast during the
16th and 17th centuries ⎯ Bahia, Rio de Janeiro and Espírito Santo ⎯ was the
variety recorded in Anchieta’s (1990[1595]) and Figueira’s (1878[1687])
grammars. Its northern variety yielded the Língua Geral Amazônica, the
language of the Portuguese colonizers of the Amazon region, known today as
Nheengatú (Couto de Magalhães 1876, Rodrigues 1996, Bessa Freire 2004).
From today’s vantage point, therefore, the ancient Tupi (16th-17th centuries),
the Tupinambá (16th-17th centuries), the Paulista general language (17th-18th
centuries) and the Nheengatú (17th-20th centuries) languages consist of
geographical and historical varieties of the same subset of the Tupi-Guarani
family, the one disseminated along the Atlantic coast, from the south to the
Amazon region.
Regarding the ancient Guarani (many times referred to in the colonial
literature as the Carijós, cf. Cardim 1939[1625]:208, as well as the Jesuits’
extensive correspondence), in the beginning of the 17th century, their Jesuitical

reductions occupied vast territories of what is today southern Brazil, from the
region of Guairá (between the Paranapanema river and its tributaries, at the
border of the captaincy of São Vicente, today’s state of São Paulo), as far as
the region of Tape (Rio Grande do Sul, today’s south Brazil). In the north, the
Guarani missions spread as far as the Pantanal region and the state of Mato
Grosso (central Brazil), in the region of Itatim. Not unlike what had happened
in São Paulo among the Tupi and the Portuguese, where the mestizos spoke the
Paulista general language at least until the middle of the 18th century, contact
between Guarani and Spanish speakers favored the development of a mestizo
population whose language of contact was Guarani. As a consequence, we can
speak of a general language in Paraguay also, the Guarani Criollo, the
‘Paraguayan Guarani’ (Rodrigues 1996), that today makes up, together with
other modern varieties of Guarani ⎯ Kaiwá, Nhandéva, Mbyá, Xetá, Guayakí,
Chiriguano ⎯ a second subset of the Tupi-Guarani family, located in the
Paraná and Paraguai basin.
The rest of the country, up to the 19th century, was a great terra incognita
where the little-known group of the Tapuias lived (Cardim 1939[1625]:236).
The acceptance of the colonial project, the cooperation in the wars against the
Spanish and the disposition to convert to Christianity had been the main
criteria separating the Tupis (friends) from the Tapuias (enemies), from the
Portuguese perspective, of course. Although highly stereotypical, this two-way
division between Tupis and Tapuias can be considered the first classification of
Brazilian Indians and of Brazilian languages, an early organizational principle
of a reality perceived as extremely heterogeneous and chaotic.
The colonial sociolinguistic reality was, of course, much more complex
than that, especially if we remember the repeated incursions of the paulistas
during the 16th and 17th centuries in the Jesuits’ reductions of the south in
search of Guarani slaves who, once captured, were brought to the São Paulo
region (Monteiro 1994). In this region, besides the Tupi, there were at least
two groups (the Maromomi and the Guaianá) whose language type and
ethnicity were different, to say nothing of the period between 1580 and 1640,
during which Portugal and Spain, and their colonies, were joined under a
single crown. This made the linguistic exchanges between the Portuguese and
Spanish colonizers and the Tupian and Guarani population even more
In general terms, however, we are not wrong in considering the ancient
Tupi and ancient Guarani as the autochthonous basis on which two
neighboring general languages — the Paulista and the Paraguayan,
respectively — developed in the region which corresponds to today’s southern
Brazil, and the Tupinambá as the basis from which the Amazonian general
language derives.

4. The classification of the Brazilian languages

The botanist Martius had been to Brazil long before the creation of the
IHGB, together with the zoologist Johann Batist Spix (1781−1826), in the
Austrian Commission that followed the Archduchess Leopoldina, who came to
Brazil to marry D. Pedro I (1798−1834), the first Brazilian Emperor. Between
1817 and 1820, both scientists studied and described Brazilian flora and fauna,
but also collected samples of some Brazilian languages, with the help of
interpreters. The materials Martius picked up on that occasion, assembled with
data he transferred from ancient books and manuscripts, allowed him to
organize the first comprehensive glossary of the Brazilian languages known up
to that time: the Glossaria linguarum brasiliensium, published for the first time
in 1863 (see Cruz’ 2005 monographic study for details). Despite being in
Europe, where he had returned after his stay in Brazil, Martius became an
active corresponding associate of the IHGB, participating in the discussions
held by the Institute and assiduously publishing in its journal.
In the sections that follow, I will take as exemplars of the policies of the
period two proposals formulated by Martius regarding linguistic issues:
namely, the coinage of the term Tupi, as the genuine designation of the original
Brazilian language, and the proposal that southern Brazil was the point of
departure of the migrations of the Tupi people ⎯ indeed, according to him, the
cultural matrix of all other South American nations, including the Guaranis.

4.1 Tupi, Brazilian language

In his Glossaria (1863, 1969[1867]), Martius presented two sets of data:
those belonging to the language he referred to as Tupi, or Língua Geral
Brasílica, namely: dialectus vulgaris (Pará), 2 Apiacás, Cayowâs, Bororôs,
Omáguas, Campévas, Araquajú, Uaraguaçu, Mundrucûs, Muras, and Tupi
austral (1969[1867]:5-122), and those, c. 90 languages and dialects, belonging
to the non-Tupi groups, organized as follows (1969[1867]:123-286):
Guaycurus, Guanás [Mataco], Guachis; the Gentis Gês (=Cayapós, Chavantes,
Cherentes, Chicriabás, Geicó, Masacará, Acroa mirim, Apinagés,
Aponegicrans, Carahôs, Camacan, Meniens, Cotoxó, Tecuna, Catoquina,
Coretú); the Gentis Goyatacas (=Coropó, Machacali, Capoxó, Patachô,
Macuni); the Gentis Cren (Gueren) (=Botocudos, Puri, Coroado, Malali,
Guato, Patagon, Camé); the Gentis Guck: (=Cayriri, Sabujá, Pimenteira,
Manao, Marauha, Macusi, Paravilhana, Uirina, Bare, Cariay, Araicu,
Canamirim, Maxuruna, Caripuna, Culino, Uainuma, Jumana, Jucúna, Passe,
Cauixana, Tariana, Baniva, Carajás, Mariaté, Júri); and finally, what Martius
considered the gentis incertae affinitatis: Coeruna, Jupuá, Miranha, Jaúna,
Cobeu, Tucano, and Curetú.

I am following Martius’ own designations and orthography.

The other parts of the book were dedicated to the languages spoken in
Brazilian frontier territories: Kechua, Yagua, Oregones, Panos, Cocamas,
Pebas, Iquitos, Zapara, and Aruac, followed by a ‘vocabula comparata
Guyanae britanicae’ (1969[1867]:289-324). In addition, a small dictionary of
the Galibi language (1969[1867]:327-412) and a dictionary of Tupi flora and
fauna (1969[1867]: 373-544) accompanied the vocabularies.
In the preface of his Glossaria (1863:xii), before strongly recommending
“a difusão da lingua geral Brazilica entre todos os Indios” [the spread of the
Brazilian general language among all Indians], a ‘minor’ task to be performed
by the philanthropic researcher, Martius reiterated how preliminary in nature
his book was. The various vocabularies assembled were not yet comparable to
one another, as Martius thought they should be. Even so, they were, he hoped,
a good point of departure for other scientists and associates of the Institute
working in the same direction. In the end, it would gradually become possible
to compare these languages and establish their grammar on a more modern
The contribution of the Glossaria to the philosophy of languages, as put in
the terminology of the time, would be to allow future comparison of similar
linguistic forms and the discovery of the ‘organic laws’ that had caused their
phonetic changes (Martius 1863:xvii). Besides that, Martius highlighted the
high potential of applicability of this instrument. Equivalent lists of terms from
different languages would not only allow mutual intelligibility among those
who had to deal with Indians from different tribes, but it would also contribute
to the improvement of the existing general language, i.e. Tupi, through the
addition and adaptation of terms and expressions originating in the other
languages. As a result, the Tupi language would increasingly spread and
become, over the medium term, the lingua franca of all Brazilians: this was a
precondition, in his view, for their ‘ascendance’ to the civilized world.
[..] seria utilíssimo e summamente no interesse do Brazil, que a língua geral, que dois
séculos antes foi falada quazi em todo o império por muitos brancos, seja ainda agora
fomentada por todos os meios e estendida em todos os territórios, aonde vivem
Índios. Façam-se todos os esforços para que ella, como lingua verdadeiramente geral
e franca, seja substituída a todas as geringonças, e que estas, meio e rezultado das
divizões e inimizades perpetuas dos autochtones, mais e mais desvanesçam. (Martius

[it would be very useful and very much in the interest of Brazil, that the general
language, which two centuries ago was spoken in most of the Empire by many whites,
be continuously cultivated by all means and extended in all territories where Indians
live. Let every effort be made in order that it, as a truly general lingua franca, replaces
all jargons, and that they, being both the instrument and the result of the divisions and
perpetual hatred of the natives, increasingly disappear.]

It is easy to imagine how opportune the pioneering work of Martius would

become in a context where the control of the cultural and linguistic
heterogeneity of Portuguese America and the demarcation of its frontiers were
primordial tasks. The integration of what was perceived to be an extremely
heterogeneous society, in which there were people living in a deplorable state
of pre-civilization, was proposed as an urgent ⎯ and humanitarian ⎯ task to
be accomplished by the present generation. The linguistic issue was crucial in
this context: to speak the language which was assigned the status of ‘general’
meant to overcome the category of savagery and become civilized, in the sense
of becoming a participant of the working system of the missions and the
Even though it reproduced in a certain way the old colonial dichotomy of
Tupi vs. Tapuias, Martius’ Glossaria, in emphasizing the wide range of Tupian
dialects, is considered the first scientific classification of the Brazilian
languages (i.e., the first non-missionary one)—the starting point of 19th and
even 20th century scientists and historians. As a matter of fact, Martius’ index
would often be referred to and reformulated by later generations of
Americanists, but what is remarkable at this moment is the absence of any
reference to Guarani in all glossaries. I will come back to this issue in the next

4.2 Establishing borders

The interest in establishing the origins of the Tupian population dates from
this period, when naturalists like D’Orbigny 1944[1839], Ehrenreich 1892, in
addition to Martius himself, advanced hypotheses regarding their origin and
the reasons for their great territorial expansion. Since then, there has been a
consensus among linguists, ethnohistorians and archaelogists that there was a
common center of Tupian dispersion, although there is no agreement on the
exact location of this center, nor about the routes followed by the various
waves of migration (Noelli 1996, Urban 1996, Rodrigues 2000). Be that as it
may, in the period being examined here, we observe an interesting dispute
among historians and naturalists 3 about the real designation of this mother-
nation (Altman 2004). Not by chance, the discussions were polarized between
those defending the Tupi denomination and those defending the Guarani
denomination, languages metonymically related to Brazil and Paraguay, two
nations at war at the time (1864-1869).

The romantic literati and musicians were to deal differently with the mystique of the Indian;
cf. José de Alencar, Gonçalves Dias, and Carlos Gomes.

Although Martius was not quite the first to raise the hypothesis of the
precedence of the Guarani over the similar languages and dialects, 4 this
hypothesis is mainly associated with him. In his first incursions on the theme,
he in fact proposed that a Tupi civilization had its origin in the south of the
continent, more exactly in Paraguay and the south of Bolivia and that Guarani
was its original language. In his own words:
[...] não resta, pois, dúvida de que o berço dos tupis não é o extenso trecho ao longo
do litoral, habitado aquí e ali por outras hordas, mas deve ser procurado alhures. A
língua guarani, que em muitos sentidos se mostra ser a pura raiz de todos os dialetos,
indica-nos, por isso, a pátria de origem dos tupís, no Paraguai, o território entre esse
rio e o Paraná. Ali, ainda é falada atualmente essa língua, embora a já maior parte da
tribu tenha desaparecido, assim como os tapes da província do Rio-Grande-do-Sul e
de Montevidéu. (Spix & Martius 1938[1823]:202).

[[...] there is no doubt remaining, then, that the place of birth of the Tupis is not the
extensive area along the [Brazilian] coast, inhabited here and there by other hordes [of
natives], but that it must be looked for elsewhere. The Guarani language, which in
many senses seems to be the pure root of all dialects, shows us, for this reason, [that]
the original homeland of the Tupis is in Paraguay, the territory between this river and
the Paraná [river]. There, this language is spoken nowadays, although the greater part
of the tribe has disappeared, as well as the tapes from the province of Rio Grande do
Sul and Montevideo.]

His following publications, however, were gradually to oscillate between

the two designations, Tupi and Guarani, up to the complete omission of the
Guaranis in the edition of his Glossaria that, as we observed above, shows no
reference to them ⎯ Cruz (2005) interestingly associates this to the fact that
Martius’ work was being sponsored by the Brazilian Emperor at that time. The
fact is that the gradual replacement of the term Guarani by the designation
Tupi seems to be a clear policy of the time. Nothing illustrates its official
nature better than the subtitle Varnhagen gives to his 1876 reedition of
Montoya’s 1640 grammar: Arte de la lengua guarani, o más bien Tupi [Art of
the Guarani language, or better the Tupi].
History varies according to the point of view of the historians, as we know,
and the tone this issue assumed in the period under consideration is a good
example of this. To admit that the American native was originally from the
south, i.e., from the region of the Paraguay River ⎯ as Martius first proposed
⎯ also meant assuming that the linguistic variety spoken in Paraguayan
territory, i.e. Guarani, was the basis from which all the other dialects, Tupi
included, derived. From a different perspective, to postulate that Tupi (and not
Guarani) was the ‘mother tongue’ of all dialects, i.e., the true original language

There is a tendency in the Spanish literature of the 18th and 19th centuries to point to the
Guarani as the matrix of all others of the same group. Cf. Felippo Salvatore Gilij’s Saggio di
Storia Americana from 1782, for example, and Lorenzo Hervás’ Catalogue, 1800.

of the inhabitants of the territory, meant placing its point of dispersion

elsewhere, in the north of the country, as Varnhagen (1876) clearly states
El nombre guarani o mas bien guaryni, segun el propio Montoya, no quiere decir sino
guerra, o por ventura guerrero; y como la lengua era hablada tambien por individuos
de la nacion que no eran guerreros, como niños y sacerdotes (pajés) y los nefandos
tebiros y hasta por las propias mujeres, bien que con sus modificaciones, sigue-se que
ese nombre es impropio para aplicarse a la lengua; al paso que el de tupi, que era el
que correspondia a esa raza, que de norte a sur (y no del sur al norte, sugun imaginó
Martius) habia invadido casi toda la America Meridional acá de los Andes, no ofrece
tales contradicciones. Etymológicamente, segun temos demonstrado, este nombre tupi
viene de t’ypi, y significa ‘los de la generacion primera’. (Varnhagen 1876:iv-v)

[The name guarani, or better, guaryni, according to Montoya himself, means nothing
else but war, or maybe warrior; and as the language was also spoken by members of
the nation who were not warriors, like children and priests (pajés) and the abominable
tebiros and even by the women themselves, although with modifications, it follows
that this name does not properly apply to the language, while the [name] tupi, which
was the one that corresponded to that race, that which from the north to the south (and
not from the south towards the north, as imagined by Martius) had invaded almost all
of South America from this side of the Andes, does not have these contradictions.
Etymologically, according to what we have demonstrated, the name tupi originates
from t’ypi, which means ‘those of the first generation’.]

If we accept that this quotation illustrates the climate of opinion of the time, it
becomes easy to notice how the differences between the two languages were
far from being only linguistic differences. In the 19th century, the Tupi and the
Guarani languages became an official criterion of organizing the lasting
differences between the Spanish and Portuguese territories, i.e., concrete
evidence of the political and cultural borders between the two recently born
nations: Brazil and Paraguay.
So it is that the no less celebrated author of the Guarani Civilization,
Moisés Bertoni (1857−1929), a Swiss naturalist who spent a great deal of his
life in Paraguay, states the opposite of what Varnhagen had claimed:
El nombre ‘Tupi’: Creo haber demostrado ya, en mis trabajos anteriores, la
conveniencia de abandonarlo. De qualquer manera que se le emplee, cualquiera sea la
extensión que se le dé, hoy día ya no puede sino engendrar confusión. Como
sinónimo de ‘guarani’ está de sobra; como substituto, expone a desagradables
equivocaciones; como nombre histórico es de significado artificial; como nombre
indígena es de significado contraditorio; como verdadero nombre de pueblo, no
existió nunca; etimológicamente, es absurdo: por fin, los Guaraníes modernos lo
rechazan en absoluto y lo dan a sus enemigos. (Bertoni 1916:5)

[The name ‘Tupi’: I believe I have already demonstrated in my previous works that it
is convenient to abandon it. However one employs it, whatever be the extension one
gives it, nowadays it can only cause confusion. As a synonym of ‘guarani’ it is too
broad; as a substitute, it opens the way to unpleasant mistakes; as a historical name it

has an artificial meaning; as an indigenous name it has contradictory meanings; as the

true name of a people it has never existed, etymologically it is an absurdity: finally,
the modern Guaranis refuse it absolutely and give it to their enemies.]

5. Conclusion
Some of the consequences of this recurring policy of linguistic
standardization are unsurprising: of the several hundred languages spoken
within what corresponds today to Brazilian territory, only 180 or thereabouts
are left. What is perhaps more striking is the fact that, of these languages, only
34 are recognized by contemporary Brazilian linguists as well documented;
114 are considered to have some documentation, while the 23 remaining are
regarded as unknown (Franchetto 2001).
As a matter of fact, some time ago, in a comprehensive study on modern
Brazilian linguistics (Altman 2003b), I examined more than two hundred
research articles published in the most representative Brazilian linguistics
journals, from the point of view, among other parameters, of the language they
described — Portuguese, indigenous or some other foreign language — and of
the orientation the authors gave to their research —grammatical, use/variation,
historical, (meta)theoretical, or applied. The results, in absolute values, are
represented in the table below:

Language data in research articles published in Brazilian linguistic journals

Grammatical 61 8 9 78
Use/Variation 54 3 13 70
Historical 5 3 3 11
(Meta)Theoretical 24 3 1 28
Applied 16 — 5 21
TOTALS 160 17 31 208
Sources: Estudos Lingüísticos. Revista Brasileira de Lingüística Teórica e Aplicada
(1966−1968), Revista Brasileira de Lingüística (1974−1984), Cadernos de Estudos
Lingüísticos (1978−1988, current), Cadernos de Lingüística e Teoria da Literatura - Ensaios
de Lingüística (1978−1988, current), Documentação de Estudos em Lingüística Teórica e
Aplicada (1985−1988, current), in Altman 2003b.

We can draw many pictures of Brazilian linguistics of the period from these
numbers. By all means it certainly was predominantly synchronic and strongly
oriented to the grammatical description of the Portuguese language. What is
even more remarkable, however, is the relatively low number of articles
dedicated to the description of the indigenous languages ⎯ only 17 ⎯ to say
nothing about studies related to the African languages that entered the country
during the colonial period, totally absent, and studies on other minority
languages brought by 20th century immigrants, such as Germans, Italians and

Japanese. This does not automatically imply, of course, that nothing was done
in these fields during this period, but it certainly suggests that, if it were done,
it was not done by the Brazilian linguists who published in Brazilian
professional journals or, at least, by linguists who were part of the official
circuit of Brazilian linguistics.
More recently, on another level of reflection — the level of the history of
the Brazilian language sciences — I observed something similar regarding the
proportion of studies dedicated to the history of works devoted to the
Portuguese language and those focusing upon other Brazilian linguistic
traditions (Altman 2001). Out of 238 historical essays of different natures ⎯
chronicles, biographies, bibliographies, historiographies, summing-up accounts
⎯ only a few were retrospections of studies in Brazilian indigenous languages.
Apparently, linguistic science in Brazil, and the historiography that
legitimates it, by putting aside previous studies on Brazilian indigenous
languages and investing with privilege in the study of the Portuguese language,
indirectly reinforced the myth of Brazilian linguistic homogeneity at the same
time as it established the limits of retrospection for the discipline. Four
centuries after the pioneering descriptions of the Jesuit missionaries, renewed
by the work of the scientific expeditions of the 19th century, Brazilian linguists
lined up with other traditions that recognized the start of the discipline as a
science with the study of Indo-European languages (cf. Mattoso Câmara 1975
One of the challenges of the contemporary linguistic historiographer is to
understand, and, if possible, to explain, how and why in the process of
selection of the linguistic ideas and practices of our past, Brazilians seem to
have systematically left aside the study of languages other than Portuguese,
and obliterated their inheritance of studies developed under other traditions,
from previous centuries. Meanwhile, while there still is a great deal of
theoretical and empirical work to be done, we can observe that if the
missionaries are to be blamed, rightly so, for collaborating with a
homogenizing policy of cultural and linguistic annihilation, we, scientists from
the 19th and 20th centuries, seem, so far, not to have done much better.


Altman, Cristina. 2001. Historiografias Lingüísticas. Três Questões em

Produção Lingüística Brasileira. 2 vols. Thesis of Livre-Docência.
Faculdade de Filosofia, Letras e Ciências Humanas, Universidade de São
Paulo. Unpublished Ms.
Altman, Cristina. 2003a. “As Línguas Gerais Sul-Americanas e a Empresa
Missionária: linguagem e representação nos séculos XVI e XVII”. Bessa
Freire & Rosa, eds. 2003.57-84.

Altman, Cristina. 2003b. A Pesquisa Lingüística no Brasil (1968-1988). 2nd ed.

São Paulo: Humanitas. (1st ed. 1998.)
Altman, Cristina. 2004. “Tupi, Guarani and the Search for the Mother
Tongue”. Paper given at the II International Conference on Missionary-
Colonial Linguistics (1492-1850). São Paulo, 10-13 March, 2004.
Anchieta, Joseph. 1990[1595]. Arte de Grammatica da Lingoa mais Usada na
Costa do Brasil. 7th ed. São Paulo: Loyola. (1st ed., Coimbra: Antônio
Mariz, 1595; 2nd ed. by Julio Platzmann, Grammatik der brasilianischen
Sprache: Mit Zugrundelegung des Anchieta. Leipzig: Teubner, 1874; 3rd
ed., 1876.)
Bertoni, Moisés Santiago. 1916. “Influencia de La Lengua Guaraní en Sud-
América y Antillas”. Anales Cientificos Paraguayos 1.3-26.
Bessa Freire, José Ribamar. 2004. Rio Babel. A história das línguas na
Amazônia. Rio de Janeiro: EDUERJ-Atlântica.
Bessa Freire, José Ribamar & Maria Carlota Rosa, eds. 2003. Línguas Gerais.
Política lingüística e catequese na América do Sul no período colonial. Rio
de Janeiro: EDUERJ.
Cardim, Fernão. 1939[1625]. Tratado da Terra e Gente do Brasil. (Intr. and
notes by Baptista Caetano, Capistrano de Abreu, and Rodolfo Garcia). 2nd
ed. São Paulo: Nacional.
Couto de Magalhães, José Vieira. 1876. O Selvagem. Rio de Janeiro: Typ. da
Cruz, Aline da. 2005. O Resgate da Língua Geral. Modos de representação
das unidades lingüísticas da língua geral brasílica e do tupi austral na
obra de Martius (1794−1868). M.A. Dissertation. Faculdade de Filosofia,
Letras e Ciências Humanas, University of São Paulo.
D’Orbigny, Alcides. 1944[1839]. El Hombre Americano Considerado em sus
Aspectos Fisiológicos y Morales. Buenos Aires: Editorial Futuro.
Ehrenreich, Paul. 1892. “Divisão e Distribuição das Tribus do Brasil Segundo
o Estado Actual dos nossos Conhecimentos”. Revista da Sociedade de
Geographia do Rio de Janeiro 8.3-55.
Figueira, Luís. 1878[1687]. Grammatica da Lingua do Brasil. 6th ed. Facsimile
ed. by Julius Platzmann. Leipzig: B. G. Teubner. (1st ed., Lisbon: Manoel
da Silva, 1621; repr., Lisbon: Miguel Deslandes, 1687.)
Franchetto, Bruna. 2001. “Línguas Indígenas no Brasil: Pesquisa e formação
de pesquisadores”. Povos Indígenas e Tolerância ed. by Luiz Donisete
Benzi Grupioni, Lux Vidal and Roseli Fischmann, 133-156. São Paulo:
Gimenes, Luciana. 1999. As Formas de Saber sobre as Línguas do Brasil no
Século XVI. Uma contribuição para a historiografia lingüística brasileira
colonial. M.A. Dissertation. Faculdade de Filosofia, Letras e Ciências
Humanas, University of São Paulo.

Martius, Karl F. P. Von. 1845[1840]. “Como se deve escrever a história do

Brasil”. Revista do Instituto Histórico e Geográfico 5.381-401.
Martius, Karl F. P. von. 1863. Glossaria Linguarum Brasiliensium/ Glossários
de Diversas Línguas e Dialectos, que Fallao os Indios no Império do
Brazil/ Wörtersammlung Brasiliannischer sprachen. Erlangen: Druck Von
Junge & Sohn.
Martius, Karl F. P. von. 1895[1861]. “Carta do Dr. Carlos F. de Martius.
Contendo observações sobre botânica, vocábulos tupis e origem das tribus
americanas”. Revista do Instituto Histórico e Geográfico 58.59-68.
Martius, Karl F. P. von. 1969[1867]. Wörtersammlung Brasiliannischer
sprachen/ Glossaria Linguarum Brasiliensium/ Glossários de Diversas
Línguas e Dialectos, que Fallao os Indios no Império do Brazil.
Wiesbaden: Martin Sändig.
Martius, Karl F. P. von. 1905. “A Ethnografia da América. Especialmente do
Brazil. O passado e o futuro do homem americano”. Revista do Instituto
Histórico e Geográfico 9.534-562.
Mattoso Câmara, Joaquim. 1975[1962]. História da Lingüística. Petrópolis:
Vozes. (Transl. by Maria do Amparo Barbosa de Azevedo from the 1962
English Ms.)
Monteiro, John. 1994. Negros da Terra: Índios e bandeirantes nas origens de
São Paulo. São Paulo: Companhia das Letras.
Montoya, Antonio Ruíz de. 1876[1640]. Arte y Bocabulario de la Lengua
Guarani. Tomo Primero. Facsimile ed. by Julio Platzman in two separate
volumes: Arte de la Lengua Guarani and Bocabulario de la lengua
Guarani. Leipzig: B.G. Teubner. (1st ed., Madrid: Juan Sanchez, 1640.)
Noelli, Francisco Silva. 1996. “As Hipóteses sobre o Centro de Origem e Rotas
de Expansão dos Tupi”. Revista de Antropologia 39:2.7-53.
Prezia, Benedito A. 2000. Os Indígenas do Planalto Paulista — nas crônicas
quinhentistas e seiscentistas. São Paulo: Humanitas.
Rodrigues, Aryon. 1994. Línguas Brasileiras. Para o conhecimento das
línguas indígenas. São Paulo: Loyola.
Rodrigues, Aryon. 1996. “As Línguas Gerais Sul-Americanas”. Papia 4:2.6-
Rodrigues, Aryon. 2000. “Hipótese sobre as Migrações dos Três Subconjuntos
Meridionais da Família Tupi-Guaraní”. Atas do II Congresso Nacional da
ABRALIN. CD-ROM, 1596-1605. Florianópolis: ABRALIN.
Rosa, Maria Carlota. 2003. “A Língua Mais Geral do Brasil nos Séculos XVI e
XVII”. Bessa Freire & Rosa, eds. 2003.133-146.
Spix, John Batist & Karl F. P. von Martius. 1938[1823]. Viagem pelo Brasil.
[Transl. by Lucia Furquim Lahmeyer from Reise in Brasilien.] Rio de
Janeiro: Imprensa Nacional.
Urban, Greg. 1996. “On the Geographical Origins and Dispersion of Tupian
Languages”. Revista de Antropologia 39:2.61-104.

Varnhagen, Francisco Adolfo de (Viscount of Porto Seguro). 1841. “Sobre a

Necessidade do Estudo e Ensino das Línguas Indígenas do Brasil; lida na
sessão do 1o. de agosto de 1840”. Revista do Instituto Histórico e
Geográfico 3.53-63.
Varnhagen, Francisco Adolfo de. 1874[1849]. “Ethografia Indigena. Linguas,
Emigrações e Archeologia. Padrões de mármore dos primeiros
descobridores”. Revista do Instituto Histórico e Geográfico 12.366-376.
Varnhagen, Francisco Adolfo de. 1876. Introduction to the Vocabulario y
Tesoro de la Lengua Guarani, ó mas bien Tupi, por el P. Antonio Ruiz de
Montoya, natural de Lima, Misionario en la antigua reduccion de Loreto,
junto al rio Paranapanema del Brasil, Superior en otras y Rector del
Colegio de Asubpcion, etc. Nueva edicion: mas correcta y esmerada que la
primera, y con las voces indias en tipo diferente. Vienna: Faesy y Frick;
Paris: Maisonneuve Y Cia.
Zwartjes, Otto & Even Hovdhaugen, eds. 2004. Missionary Linguistics/
Lingüística Misionera. Amsterdam & Philadelphia: John Benjamins.
Zwartjes, Otto & Cristina Altman, eds. 2005. Missionary Linguistics II/
Lingüística Misionera II. Orthography and Phonology. Amsterdam &
Philadelphia: John Benjamins.

Linguistics Department
Institute of Language Studies, UNICAMP – BRAZIL


Since the 19th century, Brazilian linguistics has been involved in the
question of language relationships, both expressed and tacit.
In this work we will analyze how this question presents itself in historical
Brazilian linguistics by observing the concept of civilization and its
relationships with the concept of culture.
We analyze a period that goes from the end of the 19th century until the
Observing the relationships of the words in their authors’ texts, we analyze
the meaning of words such as civilization and culture.
We show how the concept of civilization, even though it may have
acquired a plural signification, maintains its sense of universal value and
functions as a justifying concept of the relationships among languages,
projecting itself, in the interior of scientific thinking, as a “legitimated”
substitute for colonization.

Since the 19th century, Brazilian linguistics has been explicitly or tacitly
involved in the question of language relationships. These relationships
appeared in Brazilian history, as well as in the rest of America, through the
colonization movement that has marked the American-European relationship
ever since the end of the 15th century.
In this work we analyze the manner in which this question presents itself in
Brazilian history, observing the concept of civilization and its relationships
with the concept of culture.
We study the specific period of the 1930s to the 1950s in which we
encounter the production of important Brazilian linguists, directly connected to
the first Brazilian courses of higher learning in Letters, created in the 1930s, in
São Paulo and Rio de Janeiro.

As we know, the word “civilization” appeared in the 18th century in the

French language and shortly thereafter in English. In the sense that interests us,
it appears in Mirabeau in 1756. With Starobinski (1989:16), we can say that:
“when calling civilization the fundamental process of history and, when
designating with the same word the final state resulting from this process, the
term contrasts in an antonymic manner with a state that was supposedly first
(nature, savagery, barbarism)”. Thus “civilization”, within scientific thinking,
projects itself as a “legitimated” substitute for colonization.
The fundamental division of the word, maintained unique by the antonymy
that is guaranteed in both cases, will assure that, in its history, civilization will
either appear in the singular, a civilization (a universal value), or in the plural,
the civilizations (the set of characteristics presented by the collective living of
a group or by a period of time, according to what Braudel (1987) tells us).
To know what civilization designates, I will analyze its enunciative
functioning in specific texts, observing the word based on the enunciation in
which it appears and then, based on this, arrive at the relationship of this word
with others in the text (Guimarães 2002, 2004a). These are the relationships
that constitute what I call the “semantic domain of determination” (SDD) of a
word (Guimarães 2004b) and, therefore, what it designates in the texts in
which it occurs.
I will specifically analyze the designation of civilization and culture
according to three Brazilian linguists, based on three of their central works:
Gramática Histórica (Historical Grammar) (1938) by Ismael da Lima
Coutinho, Estudos de Filologia Portuguesa (Studies of Portuguese Philology)
(1946) by Francisco da Silveira Bueno, and Introdução ao Estudo da Língua
Portuguesa no Brasil (Introduction to the Study of the Portuguese Language in
Brazil) (1950) by Serafim da Silva Neto. I will give particular attention to the
manner in which the concepts of civilization and culture present themselves to
these authors when they reflect on the implantation of Portuguese in Brazil.
This, in itself, already brings up the question of differences between peoples
and histories.

1. Ismael da Lima Coutinho (1900 – 1965)

In the chapter “O Português do Brasil” (The Portuguese of Brazil) from
Coutinho’s Gramática Histórica (1938:322), the first appearance of the word
“civilization” occurs in a referential expression: “a notable civilization”. In this
first appearance, the word “civilization” is qualified as notable, which leads us
to the meaning of one civilization among others. The domain of what
civilization designates already appears predicated by a hierarchized difference,
between the notable and the not notable. The notability of the civilization
relates it to the notion of progress. Another aspect is that this expression is part
of a predicative expression in relation to the “Portuguese language” and

“Portuguese discoverers and colonists”. The “notable civilization” is then a

predicate of Portugal and, by extension, of Europe.
This referential expression (“a notable civilization”) is rewritten throughout
the text as “the civilized centers” (ibid. 323), as “the civilizing influx” (ibid.
326), and as “civilized man” (ibid. 326).
These three rewritings basically restate the name (civilization) of the
referential expression “a notable civilization”. At the same time, they receive
the predication of notable from this last expression. On the other hand, this
rewriting attributes exteriority to civilizations. If there are notable and not
notable civilizations, and we are therefore in the interior of the civilizations,
there is something that is outside of the “civilized center”. Thus, we are led to
another designation of the same text: “aborigine”, which also appears
articulated in a referential expression: “the brazilic aborigine” (ibid. 323). This
referential expression takes the entire set of Brazilian indigenous peoples as
one and the same. It appears articulated to the predicate “were concentrated in
the remote forests of Mato Grosso, Amazonas, Goiás...” (ibid. 323). By this
predication, aborigine is determined by the remote, the forests, which are the
opposite of the center of the civilization.
Therefore, it is particularly interesting that “a notable civilization” (ibid.
322) be rewritten as “the civilized man” (ibid. 326). In this way, Coutinho’s
formulation establishes an opposition between classes of the subject: on the
one hand, “civilized man” and, on the other, “brazilic aborigine”. Taking this
aspect a bit farther, in a very specific articulation we will encounter “the
civilized man combating ignorance” (ibid. 326). The opposition between
“civilized man” and “ignorance” projects ignorance to the “brazilic aborigine”,
thus predicating it fundamentally. By this approach, the author places
education in the domain of civilization’s designation.
I will not elaborate on this point but it is interesting to recall here that,
when speaking of what he calls the “caipira” (hillbilly or country) dialect,
Coutinho projects the predicate of “primitive” onto this dialect of Brazilian
Portuguese. According to the author, the speakers are therefore speakers of a
precarious Portuguese dialect that will disappear in virtue of its primitive
character and of the speakers’ ignorance.
These analyses allow us to make the following presentation of the semantic
domain of designation of civilization in Ismael da Lima Coutinho’s texts (the
line that separates the domain indicates the antonymic relationship between
what comes below the line and the SDD considered):


Civilization├─ education
forest├─ ignorance ─┤ caipira (hillbilly/country)

2. Silveira Bueno (1898 – 1989)

At the beginning of his Estudos de Filologia Portuguesa, Silveira Bueno,
when defining philology, brings us the word “civilization”: “A peoples’
knowledge of civilization at a given moment of its history...” (1946:5).
Throughout the text that introduces his work, we encounter “civilization”
rewritten as “cultural thinking” (ibid. 8), which appears to be almost a direct
paraphrase of civilization, and as “literary production” (ibid. 8), which presents
one of its aspects. A bit farther on, civilization appears rewritten as “culture”
and “civilizations” (ibid. 9) and again as “cultural data” (ibid. 11).
The main element of these rewritings is the relationship that is established
in a direct manner between civilization and culture, in virtue of the
constructions in which they appear. The first occurrence of civilization comes
in the referential expression “the civilization of a people” (ibid. 5). Observing
the way in which this referential expression is rewritten, we encounter: “the
cultural thinking of a people” (ibid. 8), “all of its literary production during a
determined period” (ibid. 8), “a civilization” (ibid. 9), “a culture” (ibid. 9), “the
civilizations” (ibid. 9), “the civilization’s data” (ibid. 11), “the sum of the
cultural data that these writings might bring” (ibid. 11), “state of civilization”
(ibid. 12).
We are now seeing the word “civilization” in the plural and, in this sense, it
clearly appears as a synonym of culture.
We now advance to the analysis of the articulations in which the word and
its rewritings occur. In the first case, we have the referential expression
determined by “of a people”, which makes the defined referential expression
signify the relationship “a civilization - a people”, just as for “cultural thinking
of a people”, going in the direction of the plural sense of civilization. There is
not one civilization; there are civilizations. Each people has its civilization as
well as its culture. Here it could be said that civilization is a synonym of
culture and, therefore, has no valuable element. Acceptance of this conclusion
would be a mistake.
Continuing the observation of articulations in which civilization appears, or
one of its rewritings, we encounter “civilization” as a determinant in “state of
civilization”, this expression being determined by “advanced” in “advanced
state of civilization” (ibid. 12). Therefore, if there are advanced states of
civilization, there are also nonadvanced states of civilization. It is of great
interest that we observe that an “advanced state of civilization” is the

determination of an “indicator”, forming the expression “state of civilization

indicator”. And this expression rewrites and specifies “the cultured language,
fixed by the school, by books” (ibid. 12). Therefore, “indicator of advanced
state of civilization” is a predicate of the “cultured language, fixed by the
school, by books”, which, at once, gives the relationship between written and
cultured. Thus, cultured and written appear as indicators of progress, as that
which predicates civilization positively.
The meaning of civilization, determined by progress, reappears as a value,
even though it does not directly install barbarism and the primitive as being
outside of civilization. They are inside civilization, but as that which has less
value and should be avoided, and, in this sense, determine culture as a
synonym of civilization.
Thus, we arrive at the following SDD (the ---- sign signifies a synonym

Education Progress
| |
┴ ┴
writing ─┤ Civilization ---- culture

3. Serafim da Silva Neto (1917 – 1960)

In Introdução ao Estudo da Língua Portuguesa no Brasil (Introduction to
the Study of the Portuguese Language in Brazil), Serafim da Silva Neto (1950)
introduces us to the word “civilization” at the very beginning of this work, at
the same time in which he characterizes it and presents its objective. He tells
This book does not intend to be more than a brief essay. Our objective was to
encounter support in Brazilian history, in the formation and growth of Brazilian
society, to put the language in its rightful place: the expression of society, inseparable
from the history of civilization. (1950:13)

In this introduction, civilization appears in a referential expression, in the

singular, dealing with the civilization. It will be rewritten in the sequence of the
text as “social-historic evolution” (ibid. 14, 15), as “civilized world” (ibid. 20),
as “cultured centers” (ibid. 28), as “cultural patterns” (ibid. 31), as “religion,
habits, language” (ibid. 95), etc.
In these rewritings, the word “civilization”, which reappears insistently,
does not receive distributive specifications in its articulation throughout the
text; in other words, rewritings that attribute it with the sense of the plural (the
This type of particularization appears with words that specify civilization
when rewriting it. For example: a) In “in the ethnography and social-historic

evolution of the Brazilian people” (ibid. 14), “social-historic evolution” is

specified by “of the Brazilian people” and in “the social-historic evolution of
Brazil” (ibid. 15), by “of Brazil”. b) In “the Cultured Centers” (ibid. 28), which
is already a plural expression, we have a referential expression whose
determinant is the (understood to be in a plural form); in addition to this,
“cultured centers” is part of an expression of place, thus signifying the
diversity of cultured places. c) In “its own cultural patterns” (ibid. 31), which
in itself opposes “its own patterns” to “other patterns”, “cultural patterns”, in
addition to being in the plural, is also in a referential expression.
Civilization is unique and characterized by particularizing specifications:
religion, habits and language; culture, cultural pattern, as well as rewritings
that predicate it: social-historic.
In this manner, the sense of civilization is formed as a value. This is clearly
manifested if we observe some specific articulations in which the word
appears. In a passage in which Silva Neto characterizes the relationship of
Portuguese with other languages existent in Brazil before the arrival of the
Portuguese, he says that it deals with “[...] Tracing the history of the
Portuguese language in these lands of Santa Cruz: its victory over exotic
languages and its progressive implantation [...]” (ibid. 15). In other words, the
Portuguese language conquers the exotic languages, which are languages of
speakers in a backward stage of the civilization process. Civilization then
continues to clearly operate its sense of universal value, which involves the
sense of progress. The exotic is backward and primitive, and it is natural that
civilization should prevail over it. And civilization has therefore an outside -
the exotic, the primitive, the backward - which, if included on the inside of
civilization, would decharacterize it, due to its backwardness.
The meaning of progress is particularly significant when Silva Neto
characterizes the “written language”, which, in his text, he rewrites as “a
superior product of intercommunication” and then rewrites again as “a
powerful reflex of civilization and maintainer of unity” (ibid. 19). This
rewriting of the “written language” is directly related to what civilization
designates and, at the same time, to the meaning of superiority that, when
attributed to the written language, is transitively attributed to civilization. The
same occurs with the “maintainer of unity” that, by rewriting the “written
language”, defines it and, in this sense, also becomes attributed to civilization,
such as the process (normative) that seeks the unity, the unity in progress,
which goes from the primitive to the superior.
On the other hand, at the beginning of the chapter “Diferenciação e
Unificação do Português do Brasil (Differentiation and Unification of Brazilian
Portuguese)”: “The decision to colonize Brazil brought thousands of settlers to
the great American possession and put two different cultures face to face: the
European culture and the culture practiced by the inhabitants of the land” (ibid.
30). Here diversity is marked by the plural of cultures. In this way, the

elements of particularization are the cultural elements (religion, habits,

language and ethnography). In other words, by its difference from other
languages, a language is an element of a culture different from others, but it is
an element of civilization if determined by its civilizatory value.
The above analysis leads us to the following SDD:


Culture ─┤people ⊥
┬ ├─ civilization ─┤ written language
| |
religion-habits-language ┴
exotic, primitive

An important characteristic here is the fact that we consider that the SDD
of a civilization determines the SDD of culture; in other words, it is not the
social and historic characteristics of a people that determine and predicate its
civilization, but the civilization as a value that predicates a culture.
We see here, in Silva Neto’s thinking, two distinct concepts: civilization
and culture. On the other hand, the concept of civilization is a principle of
valorization and, in this context, it separates the cultures and their elements
into civilized and noncivilized. Civilization is a norm of judgment.

4. Conclusions
The word civilization, although not designating exactly the same notion in
the three cases studied, always brings the sense of opposition to the
“primitive”, “exotic” and “ignorant”, either because it designates something
that has an antonymic relationship with these words, as in Coutinho or Silva
Neto, or because it designates something that these words have at their
negative pole. In other words, even if the plural of civilization is part of their
meaning, such as in Silveira Bueno, this plurality is hierarchized in the
direction of what progress, education and written language designate, for
In the thinking of the three authors, and of other earlier and later linguists,
including “civilization”, as they do, is part of the process that seeks an identity
for Brazil (its language), based on work using a model (the Portuguese
language). From these analyses, we see how the concept of civilization, even
having acquired the plural sense, maintains, in that moment of Brazilian
linguistic history, its sense of universal value and functions as a justifying
concept of the relationship mode among languages. In Brazilian conditions, the
meaning of “civilization” as a universal value, differently from its meaning in

the European conditions of the 18th and 19th centuries, separates itself from the
sense of colonization, because the value of civilization emerges to justify
Brazil as independent, and as having a language of its own (Portuguese,
spoken by the colonizer) and, at the same time, to sustain Brazil as a nation
integrated to the values of those from whom it has become independent. In the
cases that we have analyzed, to say that Brazil, its people and its national
language are “civilized” is a way of affirming independence by not affirming
the primitive. Therefore, Brazilian linguistic thinking takes on the concept of
civilization as a manner of formulating for Brazil that which the concept, in
principle, would deny it: a (sense of) belonging.


Bueno, F. da S. 1946. Estudos de Filologia Portuguesa. São Paulo: Saraiva.

Braudel, F. 1987. Gramática das Civilizações. São Paulo: Martins Fontes.
Coutinho, I. da L. 1938. Gramática Histórica. Rio de Janeiro: Acadêmica.
Guimarães, E. 2002. Semântica do Acontecimento. Campinas: Pontes.
Guimarães, E. 2004a. História da Semântica. Campinas: Pontes.
Guimarães, E. 2004b. “Civilização na Lingüística Brasileira no Século XX”.
Matraga 16.89-104.
Silva Neto, S. 1950. Introdução ao Estudo da Língua Portuguesa no Brasil.
Rio de Janeiro: Presença/MEC.
Starobinski, J. 1989. “A Palavra Civilização”. As Máscaras da Civilização, 11-
85. São Paulo: Companhia das Letras. 2001.

Universidad de Deusto


The work of early missionary linguists in Central and South America

received some criticism in the 19th and 20th centuries. This paper seeks to show
that such comments were unfair in that they lacked proper historical
contextualization. The first part focuses on the educational background of
early missionaries. The second part outlines the theoretical model under which
missionary grammarians operated. A perusal of their works reveals that they
analyzed exotic languages at two different levels: at a universal or notional
level; and at the level of usage or the ways universal features were manifested
in particular languages. The third part examines the linguistic terms particle
and case within the two-tiered model. The conflicting statements perceived by
some modern scholars regarding the application of such terms to linguistic data
vanish if one considers such statements within their theoretical framework.
Previous linguistic currents have to be judged within their own epistemological
framework; otherwise the risk of misrepresentations is great.

1. Introduction
The work of early missionary grammarians in Central and South America
received some unfair criticism in the 19th and 20th centuries. In Versuch einer
Analyse der Mexicanischen Sprache (1813; in Gesammelte Schriften 1905,
Vol. 4:237-238), Wilhelm von Humboldt (1767-1835) considered missionaries
as “somewhat unfitted for researching languages whose unusual structures
were totally new for them” and regretted that they forced those languages
“within the narrow rules of Antonio de Nebrija’s grammar or of any other
Spanish pedant”. Likewise, some modern historiographers encounter a number
of contradictions in early missionary grammars. Some of those judgments are
misleading in that the scholars who make them overlook the epistemological
framework within which missionaries operated and evaluate such grammars
from a completely different intellectual climate.

This paper seeks to reconstruct the theoretical and intellectual background

within which missionary grammarians developed their ideas. This procedure
will allow us to gain a better understanding of their grammatical efforts and of
the purpose behind them and will reveal that such criticism and apparent
contradictions in their works were inaccurate. The first part examines the
religious and educational background of early missionaries, in which Latin
and, to a lesser degree, Spanish played an important role. The second part
outlines their two-tiered theoretical model, which included a universal level
and a level of usage or actual speech. Finally, the third part centers on the
values missionaries assigned to metaterms such as particle, case, and
declension within the dichotomous model inherited from Europe.

2. Educational background, writing of grammars, and role of Latin and

At the beginning, the training of Franciscan, Dominican, Augustinian, and
Jesuit missionaries took place in Europe. They usually traveled to America
once they had received an adequate religious and humanistic education in the
schools of their orders and in the various universities. With the passage of
time, schools, language centers, and universities were inaugurated in America
and printing presses were established (1540 in Mexico and 1580 in Peru).
In New Spain, the linguistic situation was quite favorable since one
language, Nahuatl, was used over a vast territory. First, missionaries took notes
and wrote grammatical sketches, brief vocabularies, and basic religious texts,
which turned into more complex works as time went on. In the Prologue to his
Arte de la lengua mexicana (1547; see 1993:2), Andrés de Olmos (1491-1570)
states that he had examined what “sobre la materia…otros habían escrito [on
the matter…others had written]”, implying that he had benefited from previous
works; this was something quite common among missionary grammarians. 1
Nahuatl grammars circulated in manuscript form and were instrumental in the
learning of systematic rules and features with communicative relevance. Such
works were written with the purpose of helping to speed up the evangelizing
efforts of friars, and most certainly not to facilitate the understanding of those
exotic languages by European linguists interested in them for totally different
reasons, especially from the 18th century on. Thus, in the Prologue to his Arte,
Olmos (1993:10) notes “Finalmente oso afirmar que cualquiera que esta senda
siguiere o sabrá más de esta lengua mexicana o tetzcucana en un año que yo en
XX [Finally I venture to state that anyone that follows this path will learn more
about the Mexican or Tetzcucano language in one year than me in twenty]”.

Before Olmos finished his grammar, having resided in Mexico for nineteen years, Francisco
Ximénez (d. 1537) had composed an Arte of the Mexican language eight years after his arrival,
which was used by his fellow members in the Franciscan order. Alonso de Rengel (c. 1500-
1547) had also written another Arte of the Mexican language in the 1530s.

Also, in his dedication letter or Carta dedicatoria, Olmos (1993:8) mentions

that the readers or beneficiaries of his grammar would be beginners and
grammarians wanting to improve his work, “me bastará si algo sirviera a los
principiantes o si en verdad diere ocasión a otros más doctos que yo y de
mayor ingenio para que completen lo que aquí se halla [I will be happy if it
turned out to be of some use for beginners or if it really gave others, more
knowledgeable and more ingenious than me, the opportunity to complete what
they find here]”. In fact, Olmos was aware of his descriptive limitations
regarding sounds and prosody, two subfields that were treated more adequately
by Horacio Carochi (1579-1662) in the following century. 2
In general, grammatical works were composed by missionaries with the
best linguistic training and by those particularly gifted for such a task. The
best tools for their grammatical endeavors came from their knowledge of the
European linguistic tradition and their innate grammatical skills. Within the
Andean tradition, Domingo de Santo Tomás (1499-1570) in his Prologue to
King Philip of his Quechua Grammatica o Arte (1560; see 1994:1) notes that
“en quince años continuos que estuve en Perú…sería digno de reprehensión
como el mal siervo…que el talento que recibió del señor lo había tenido
escondido (principalmente el don de lenguas…), luego comencé a tratar de
reducir aquella lengua a arte [in the fifteen years I spent in Peru, it would have
been reprehensible like the bad servant if the talent he had received from God
had remained hidden (principally the gift for tongues), therefore I began to try
to reduce that language to grammar]”. Historian José de Acosta (1539-1600)
referred to Alonso de Barzana (1528-Cuzco 1596), one of the authors of the
Anómino (1584) containing a grammar and a vocabulary of Quechua and
Aimara. Acosta (1954:284) writes that Barzana “…les predicó, como una
hora, en la lengua aymara, con gran atención de los indios… […preached to
them for about one hour in Aimara, with great attention from the Indians] ”.
Besides mastering Quechua and Aimara, it is believed that Father Barzana also
had a practical knowledge of nine other Indian languages, for many of which
he had written grammars, vocabularies, and catechisms.
Clergymen quite often worked together as a team in their schools and their
own language centers (cf. Nágel Bielicka 1994). The Imperial College of
Santa Cruz in Tlatelolco (north of Mexico City) opened in 1533, where Andrés
de Olmos happened to be a teacher for some time. In addition, Tepotzolán
(also north of Mexico City) was the venue of a Jesuit school which had been
planned as a language center for members of their own order. Antonio del

The renewed interest in exotic language by European linguists in the 18th century is easily
observable from the large number of missionary grammars that circulated in Spanish America
in manuscript form until that time. These only appeared in print from the 18th century on, when
the study of indigenous languages received a new impetus among European scholars for other

Rincón (1556-1601) and his disciple Horacio Carochi resided for extensive
periods of time in the Jesuit College at Tepotzolán; both Rincón and Carochi
authored Mexican grammars. Within the Andean tradition, the Juli mission,
strategically located on the shores of Lake Titicaca, had had a Dominican
school at least since 1560 as well as a Jesuit center for linguistic activities (cf.
Calvo Pérez 1997:324).
Early Spanish missionaries learned Latin mainly with the help of Antonio
de Nebrija’s (1444-1532) Introductiones Latinae (1481). Nebrija’s Latin
grammar was a standard text in Spanish schools and universities; and the
various religious orders considered it a model to follow in the description of
indigenous languages (Aguirre Beltrán 1983:207). 3 In addition, the utilization
of Nebrija’s grammar had a utilitarian dimension since it was the model in
which missionaries had received their training: it contained metaterms with
which the evangelizers (i.e. the main readers of those grammars) were well
acquainted. In this respect, Domingo de Santo Tomás (1560, Prologue to the
Reader; see 1994:21) notes “porque este arte…se hace para eclesiásticos que
tienen noticia de la lengua Latina, va conforme al arte de ésta [as this grammar
is made for ecclesiastics who know the Latin language, it is laid out in
accordance with the grammar of that language]”. Nebrija’s Latin grammar
placed great emphasis on the study of formal traits due to the humanists’
concerns for the study of usage in Latin authors. However, conceptual
definitions also had a relevant role in his grammar. This notional nomenclature
was useful for missionaries because it was applicable to the study of languages
in general. It included terms such as noun, verb, neuter verb, active or
transitive verb, case, root, particle and composition; the latter term was
particularly helpful in the description of agglutinating languages.
Frequently, one is under the impression that while discussing an issue,
missionaries did not identify roots and other parts of words sufficiently so that
readers who were unacquainted with those languages (as was the case with
European linguists in later centuries) could understand their grammatical
explanations. At best, one would have to search for proper identification of
such forms in other parts of their grammars. In part, it had to be that way
because those works were written for clergymen living together with their
language instructors in their missions, and, more importantly, because those
preachers were in daily contact with the natives from whom they had already
been picking up linguistic information. Those grammars provided them with

In the early stages of the colonial period, in French Canada, missionaries used as their
grammatical model Johannes Despauterius’ (c. 1460-1520) Rudimenta (1514) and
Commentarii grammatici (1538), while in Brazil the most popular work in the last quarter of
the 16th century was De institutione grammatica (1572) written by Emmanuel Alvarez (1526-
1582) (cf. Zwartjes 2002:28).

linguistic accuracy and resolved grammatical questions, allowing them to

perform their pastoral activities more effectively.
Latin was essential in the composition of grammars in that up to Nebrija’s
time, grammar meant Latin grammar, since practically few grammars of
vernacular languages were on hand. In another of Nebrija’s treatises, his
Gramática castellana (1492), missionaries had Spanish terminology in a
vernacular language, which, as Nebrija states (see 1980:100-101, Prologue),
was loose and needed to be reduced to rules, “…acordé ante todas las otras
cosas reducir en artificio este nuestro lenguaje castellano…como vemos que se
ha hecho en la griega y la Latina, las cuales por haber estado debajo de
arte…todavía quedan en una uniformidad [above all I agreed to reduce to rules
our Castilian language…as we see it has been done for Greek and Latin,
which, since they have been under the rules of grammar, still remain
uniform]”. Domingo de Santo Tomás also expresses himself in similar terms
in his Prologue to the Reader of his Quechua Grammar (1560; see 1994).
Even though it was not reprinted for two centuries, Nebrija’s Gramática
castellana must have been useful for missionaries (cf. title of Book 5, “De las
introducciones de la lengua castellana para los que de estraña lengua querrán
deprender”), as it revealed how a language different from Latin could be
reduced to rules (consult Zwartjes 2000 for several studies on missionary
grammars in the Spanish tradition).

3. European theoretical background: Universal grammar and study of

From a perusal of both manuscript and published grammars, it is
discernible that, following the European climate of opinion in a conscious or
unconscious manner, missionaries analyzed exotic languages from a two-tiered
theoretical model: the universal or general level (the mental language) and the
level of usage or manifestation of the universal component in particular ways
in the various tongues. In Plato, we find a duality between the “Ideas” of
things, unchanging or true reality, and the changing or uncertain things of the
visible world. On the other hand, in his Categories (see Ross 1928, Vol. 1),
Aristotle considers the various ways things reveal their existence in nature and
concludes that all phenomena in nature can be reduced to two types of
categories or ways of existing: absolute or relative. As a mirror of nature,
language also possesses absolute and relative categories. Since Aristotle’s
time, grammarians have had a term to indicate the relative categories in
language. In his Rhetoric (Book 3, Chapters 5 and 12; in Ross 1946, Vol. 1),
Aristotle called them σύνδεσμοι, ‘links or relational particles’.

A brief review of Thomas Aquinas’s (c.1225-1274) theory of knowledge

and of his doctrine of the word 4 helps to shed some light on early missionary
linguistics. In the Thomistic philosophical system, intellectual or intelligible
knowledge and sensible or sensory knowledge have their counterpart in human
discourse, where we find an inner word or mental language (the same for all
humanity) and an outer word comprising the diverse and changing tongues of
the various nations. 5 Medieval grammarians in the 13th and 14th centuries
operated within the Aristotelian and the Thomistic framework.
During the Renaissance, the medieval framework survives in the works of
a number of logical grammarians. Some features of this theoretical model are
not that obvious and are even difficult for the modern reader to perceive, due to
the passage of time and the subsequent loss of the intellectual climate hovering
around at that time. Grammarians such as the Italian Julius-Caesar Scaliger
(1484-1558) and the Spaniard Franciscus Sanctius (1523-1601) still worked
within the medieval philosophical model, a methodology which views human
rationality (i.e., ratio) as the source or the cause of the universal principles of
language. These general principles help to explain linguistic variation and the
particular usages in the various tongues in a better light.
Following the previous grammatical tradition, Scaliger (De causis linguae
Latinae 1540, Book 3, Chapter 72; see 1584) not only includes conjunctions,
adverbs, and prepositions as the linguistic manifestations of the relative
categories of nature in the Aristotelian sense, but also the accidents of nouns
and verbs such as case, gender, number, tense and so on (cf. Padley 1976:58-
77). In a similar vein, Sanctius (Minerva seu de causis linguae Latinae 1587,
Book 1, Chapter 2; see 1986) makes a tripartite division of the parts of speech
into noun, verb and particle. He argues that this tripartite division is universal
and as such is found in all languages. This conception of particle at the level of
universal grammar, of the causes that underlie the particular usages of the

For his theory of knowledge, consult Summa Theologiae, Book 1, quaestio 85, art. 1 (in
Opera omnia 1980, Vol. 2). For his philosophy of the word, see De veritate, quaestio 4 (De
verbo), art. 1, in Quaestio disputata I (in Opera omnia, 1980, Vol. 3); and De differentia verbi
divini et humani (in Mandonnet 1927).
This idea is an adaptation and development of Aristotle’s statement in De interpretatione,
Chapter. 1 (in Ross 1928, Vol. 1) and of the doctrine of the word as outlined by Augustine in
De Trinitate (Book 9, Chaps. 7 and 10 and Book 15, Chaps. 10 and 11; see 1968) from a Neo-
Platonic perspective. In Thomistic thinking, the knowledge of reality implies the formation of
passive impressions (species intelligibiles expressae) in the intellect (the simple and complex
concepts of the mind or inner word). Such universal concepts (the same for all humanity)
constitute the language of the mind in which we contemplate the ratio of reality, true being.
Now, the word which we bear in our mind (also called verbum cordis or the speech of the
heart) becomes known by bodily signs. At the level of the bodily senses, we find the outer
word (i.e., the sounds and thoughts of the senses). The word of the mind belongs to no tongue,
while the tongues of nations belong to the word of the bodily senses (cf. Arens 1980).

various tongues, will help us to better understand the meaning of the metaterm
particle in many missionary grammars.
Within this theoretical framework, Scaliger (1540, Book 3, Chapter
74:161; see 1584) clearly distinguishes between case and declension, the
former being used to portray the natural phenomenon as it is understood by our
intellect, while the latter represents this phenomenon as it is marked in
language. Declension signifies two things: (1) the act of inflecting as when
from ambulare one derives ambulatio, and (2) the inflection to which nouns
are reduced, as when we say first declension, second, third, and so on. Case is
not declension in that sense because cases are reduced to those marks. On this
matter, Sanctius (1587, Book 1, Chapter 6:17r; see 1986) claims that in every
noun, nature has established six parts or ways in which things relate among
themselves in the world and adds that since such division is natural, “it would
be necessary [a logical necessity] for the specific languages in the various
nations (omni idiomate) to possess the same number”. 6
Another pillar supporting this grammatical edifice touched on language and
its development in history. Relying on Genesis, Chapters 10 and 11, scholars
believed both in the existence of a first natural language as well as in linguistic
diversity resulting from the confusion of tongues in the episode of the Tower
of Babel and the subsequent dispersion of nations. Many authors felt that the
primeval language must have been perfect in that it reflected the rationality of
the language of the mind. In the Prologue to the Reader of his Vocabulario en
lengua castellana y mexicana (1555; see 2001:4), Alonso de Molina (c.1514-
1585) referred to this first language as follows, “Luego, después del diluvio en
toda la tierra no se hablaba más de una lengua, en la cual todos se trataban, se
comunicaban y se entendían… [Then, after the Flood, all over the Earth, only
one language was spoken, in which people treated, communicated with, and
understood each other]”. In the course of history, the original language
became figurative, diversified, and changed, even though the basic rational
principles of that first language, it was believed, still remained in the tongues
of the various nations. However, older languages (Hebrew) and classical
languages (Greek and Latin) were supposed to be more perfect and to have
changed less than others. In the 13th century, some authors still asserted that
only Hebrew, Greek, and Latin were rational languages; all the others were
considered ‘barbarian’. With the emergence of national languages in the

For this reason, Sanctius argues that Greek does not lack a sixth case or an ablative, even if
that language does not have a formal marker for it, in that Greek may express the function of
the ablative by means of the dative marker. It was not my intention to claim the existence of a
direct link between Scaliger and Sanctius, on the one hand, and missionary grammarians on the
other, but rather to outline the intellectual climate under which grammarians worked in 16th-
century Europe, and in Spain in particular where a strong revival of Thomistic thinking had
been taking place, mainly starting at the University of Salamanca and radiating from there to
othebr universities.

Renaissance, Italian, Spanish and other European languages acquired a certain

aura of respect.
Probably consciously or unconsciously embracing the previous view on
language and its development in history, Olmos (Arte de la lengua mexicana
1547; see 1993:61) writes “Primeramente se porna la conjugación, no como en
la gramática, sino como la lengua lo pide y demanda, porque algunas maneras
de decir que nosotros tenemos en nuestra lengua o la latina, esta no las tiene…
[First is given the conjugation, not as in the grammar (meaning Latin
grammar), but as this language asks and demands, because some ways of
speaking we have in our language and in Latin, this language does not have]”.
Olmos suggests that Latin and Spanish are more perfect and keep more rational
traits than Nahuatl. Hence “esta no las tiene” implies a lack of tenses in that
language and a value judgment resulting from his educational background. For
pedagogical reasons, Olmos breaks the order in which tenses and moods were
traditionally laid out, since he begins with those Latin and Spanish tenses and
moods that had distinctive markers in Nahuatl and continues with those that
had no markers, i.e. those Latin or Spanish tenses that had no tense equivalent
in Nahuatl and translated either through existing tenses or with the help of
other words (see Olmos 1993:62, 72).
On the other hand, Latin and Spanish were also considered individual,
changing tongues with their specific usages. Hernández Sacristán (1997:51)
appears to perceive this theoretical nuance between what is general and what is
particular to each language in missionary grammars, when he writes,
Ciertamente en muchas ocasiones la gramática latina se concibe como discurso
gramatical genérico[…] La distinción entre gramática y arte tendría justamente que
ver con la distinción entre discurso gramatical genérico y descripción de una lengua
con interés puramente instrumental.

[Certainly, on many occasions, Latin grammar is conceived as generic grammatical

discourse[…] This distinction between grammar and art (particular grammars) would
precisely have to do with the distinction between generic grammatical discourse and
the description of a language with a purely instrumental interest.] 7

Hernández is right when he notes that missionary treatises make a distinction between
grammar and art. In the European tradition, this distinction is clear if one reads theoretical
authors such as Scaliger and Sanctius, who study the causes of usage. However, in Nebrija’s
Institutiones Latinae, in his Gramática castellana, and in missionary grammars, the distinction
is implicit as a working assumption, but is not as obvious since both Nebrija and missionaries
are more concerned with the description of usage than with the general principles underlying
usage. In his Quechua Grammatica o arte (1560, Prologue to the Reader; see 1994:20),
Domingo de Santo Tomás defines art as giving rules “del modo de hablar de cualquier lengua”,
and adds “este negocio…requería más erudición en la lengua y conocimiento de la
significación y propiedad de los términos de ella (que es la materia del arte)”.

Yet, Hernández Sacristán seems to grasp only partially the epistemological

model missionary grammarians were working under when he further notes,
Pero el misionero no es absolutamente consecuente con esta distinción implícitamente
apuntada en su texto, porque el discurso gramatical genérico no deja de sentirse al
mismo tiempo como propio de una lengua concreta (la latina) que para determinados
efectos, puede ponerse en pie de igualdad con las lenguas descritas.

[But missionaries are not at all consistent with this distinction implicitly suggested in
their texts, because, at the same time, generic grammatical discourse is perceived as
part of a specific language (Latin), which, on certain issues, is placed on an equal
footing with the languages being described.]

Actually, for missionary grammarians, the fact that Latin was thought of as
particular and generic grammatical discourse at the same time is not an
inconsistency or lack of coherence but the result of their dichotomous
theoretical model. For practical or instructional reasons, missionaries wished
to contrast the specific features of Latin or Spanish usage with those of exotic
languages. But, on the other hand, the Latin language was believed to be more
perfect, more complete, and closer to the principles of universal rationality
than exotic languages and, consequently, the latter had to be explained on the
basis of the former.
In the light of the previous reconstruction of the missionaries’ theoretical
model, the modern reader can better understand the following passage taken
from Domingo de Santo Tomás’s Prologue to King Philip in his Quechua
Grammatica o arte (1560; see 1994:14):
Si viere por esta arte […] la gran abundancia de vocablos […] las diversas maneras de
hablar […] Y, brevemente, en muchas cosas y maneras de hablar tan conforme a la
latina y española; y en el arte y artificio de ella, que no paresce sino que fue un
pronóstico que españoles la avían de posseer. Lengua pues, S.M., tan polida y
abundante, regulada y encerrada debajo de reglas y preceptos de la latina como ésta
(como consta en esta arte) no bárbara, que quiere decir (según Quintiliano, y los
demás latinos) llena de barbarismos y defectos, sin modos, tiempos, ni casos, ni
orden, ni regla, ni concierto, sino pulida y delicada […], pues según el philosopho en
muchos lugares no ay cosa que más se conozca el ingenio del hombre que en la
palabra y en el lenguaje que usa, que es el parto del entendimiento del hombre.

[If you saw from this grammar […] the great abundance of words […] the diverse
ways of speaking […] And, briefly, in many things and manners of speaking,
conforming so much to Latin and Spanish; and in its particular grammar and rules,
that it seems that it was a prophecy that the Spaniards had to possess it. A tongue,
then, Your Majesty, so polished and abundant, principled and contained within the
rules and precepts of Latin as this one (as this grammar reveals), not barbarian, which
means (according to Quintilian and the other Latin authors) filled with barbarisms and
defects, with no moods or tenses, with no cases, no order, no rules, no agreements,
but polished and delicate […] for, according to the philosopher (i.e., Aristotle), in

many places there is no better thing to discover the genius of man than in the word
and in the language he uses, which is born from man’s intelligence.]

In it, the Dominican missionary, among other things, implicitly refers to the
two theoretical tiers under which they work: the level of the particular
characteristics of individual language and the level of the intellect. The
intellect is a faculty whose operations form universal concepts: the rational,
general and unchanging principles (the mental language) underlying particular

4. Use of linguistic metaterms within the missionaries’ intellectual

In the linguistic descriptions of the period, missionaries subsume under the
term ‘particle’ consignifying or relational elements such as prepositions,
adverbs, conjunctions and affixes. What all of these have in common is that
they are dependent on absolute categories as observed earlier with regard to the
linguistic model outlined by some theoretical grammarians in Europe. In his
Arte de la lengua mexicana (1547), Olmos understands by particle the
relational elements in Mexican or Nahuatl. Indeed, he groups under this
category forms which in modern linguistics would be considered derivational
affixes (such as diminutives), inflectional affixes (for instance, case, number
and verb tenses), and indeclinable parts of speech (such as adverbs,
prepositions, and conjunctions), regardless of whether the latter occurred in
composition or as separate forms. This classification has been viewed as
opaque and ambiguous by some modern critics. However, on this point this
Franciscan monk was following an old idea from the grammatical tradition.
From a contemporary standpoint, Nahuatl is a language without case
markers, in which syntactic relations between arguments (subject or object)
and the verb are not marked in the noun phrase but in the verb. Such markers
would be named agreement classifiers today. Olmos (1993, Part Two, Chapter
7, “Los verbos activos y algunas partículas que se juntan a ellos”) discusses a
number of elements incorporated into the verb, which mark the animate or
inanimate nature of the argument. Thus, he states (p. 110) “que esta partícula
tla denota que la acción del verbo a quien se junta puede generalmente
convenir o puede pasar a cosas inanimadas nitlatlaçotla [that added to the verb,
this particle tla denotes that the verbal action can generally take, or can pass on
to, inanimate things, nitlatlaçola]”, (where ni is ‘I’, tla indicates the ‘indefinite
inanimate nature of the object’ and tlaçola is the ‘present indicative of love’),
which he translates as ‘yo amo algo’. Olmos shows that both tla and te are
particles revealing the active (transitive) nature of a verb with either an
indefinite inanimate object in the case of tla or an indefinite animate object in
the case of te, expressed or understood “sub intelecto [sic]”. By referring to
particles rather than to pronouns, Olmos is here stressing the relational

character of those elements. It is such analyses indirectly showing that exotic

languages had their own operating mechanisms that gradually contributed, as
the years went by, to changes in the epistemological framework guiding
grammatical analysis.
Although ‘particle’ is used in the Aristotelian sense of ‘relational element’,
this metaterm is not univocal. In Arte en lengua michuacana (1574), a
Mesoamerican language with numerous agglutinating elements, Juan Bautista
Lagunas (d. 1604) virtually equates ‘particle’ (from Latin particula, ‘small
part’ or ‘partecilla’ as Nebrija translates this word into Spanish) with what we
understand today as ‘morpheme’. Indeed, in the verbal form irhah/tsi/ta/ni
(irhah ‘round shape of thing’, tsi ‘on top of something’, ta ‘cause something to
be placed or to put’, and ni ‘infinitive marker’, i.e., ‘to put something round on
top of something’), he labels as particles the shape classifier ihra, the locative
classifier tsi, the causative marker ta, and the infinitive marker ni (cf. Lagunas
2002:231-237, 240). Accordingly, he calls all the smallest meaningful units in
the verb ‘particle’. 8 Hence, the notion of morpheme appears in a larval state
in the early missionary tradition; it shows up disguised under expressions such
as particle and diction. 9 Missionaries needed some sort of broad working
concept to describe non-inflectional languages and they found it in these
expressions. However, generally, they were not interested in proposing or
finding new precise descriptive terminology as a modern linguistic theoretician
would. Their main goal was using expressions that would allow their
prospective readers to understand their descriptions without paying much
attention to how well-defined those expressions were.
Some modern scholars also perceive contradictions and inconsistencies in
the utilization of the word case and declension in missionary grammars.
Rodrigues (1997:380-381) notes that Alonso de Molina (Arte de la lengua
mexicana y castellana, 1571), after writing that “en esa lengua ningún nombre
varía ni declina por casos, así como la latina [contrary to Latin, in that
language no noun varies or is declined by case]” (see 1945:6v), provides the
singular and the plural of god, declaring “declínanse de esta manera:
Nominativo, teutl; genitivo, teutl; dativo, teutl; acusativo, teutl; ablativo, teutl;
plural, nominativo, teteu; genitivo, teteu; dativo, teteu; acusativo, teteu;
ablativo, teteu” (1945:7). Rodrigues observes a contradiction here because he
does not completely understand the way case and declension are used by
missionaries. In fact, what Molina means is that, on the one hand, the notional
values of the various cases have no declensional endings in Nahuatl, and, on
On this same issue, cf. Monzón (1997:137) with regard to Arte de lengua de Michoacan
(1558) by Maturino Gilberti (1498-1585).
In Peru, Domingo de Santo Tomás, in his Quechua Grammatica o arte (1560), also seems to
utilize the word diction in the broad sense of ‘morpheme’. For this Dominican monk, that
word is interchangeable with other terms such as particle, article (meaning noun affixes; for
example, case) and pronoun (see 1994, Chapters 2 and 3).

the other, that number distinctions are marked by means of declensional

markers (nouns are declined for number), i.e., the singular teutl becomes teteu
in the plural. In fact, following the grammatical tradition, even if nouns are not
declined for case, he provides a noun paradigm for pedagogical reasons which
shows that nouns have no case differences and that number has declensional
distinctions or endings in that language; this noun paradigm is important for
missionaries learning that language.
Monzón (1997:141-143) notices a similar incongruity in the description of
case and declension carried out by Gilberti and Lagunas in their respective
Artes (1558 and 1574) of the Michoacan language. 10 Monzón shows surprise
because, after acknowledging the absence of several cases in Michoacan,
Gilberti and Lagunas “ambos presentan el mismo paradigma de la declinación,
paradigma que no tiene en cuenta [both present the same declensional
paradigm, a paradigm which does not take into consideration]” their earlier
statement since Lagunas declines Angel in seven cases and Gilberti cuiripu
‘person’ in six (see Gilberti 1558, Part One:12r, in 1987; and Lagunas 1574:58,
in 2002).
When Gilberti (1558, Part Two:57r-60r; see 1987) writes that in
Michoacan there are only three cases: nominative, accusative and vocative, he
uses case in the sense of case-ending, not in a conceptual or notional sense.
That is (if we understand declension according to tradition), in that language
we only encounter three case-endings or declensional endings, which are the
absence of a marker for the nominative, -ni for the accusative (the –ni ending
in the dative and ablative would derive from the accusative), and –e for the
vocative. In addition, seemingly following Nebrija’s analysis of case,
declension and preposition, in both his Latin and his Spanish Grammars,
Lagunas (2002, Part One, Chapter 3:57, 58) admits a fourth case, i.e., the
genitive, in that the notional value of case may also be marked by an
independent particle such as eueri for the genitive, as in Angel eueri ‘del [of
the] Angel’. Finally, Lagunas suggests that some independent particles (for
example, himbo; cf. Latin ab, ex and pro as well as Spanish contra, entre and
cerca) do not show case values but are there to add slight differences of
meaning to basic case values. 11

Cf. also Hernández de León-Portilla (2003:15-16, 19-20) regarding certain inconsistencies in
the analysis of case and declension in Gilberti and Antonio del Rincón.
Lagunas’s analysis of case and declension is more complex than we have outlined here. If
we took it into consideration, his detailed presentation would allow Gospel preachers to
express themselves with “reglas de elegancia para hablar congruamente” (2002, Part One,
Chapter 17:125). He utilizes the word elegancy in the way the Italian humanist Laurentius
Valla (1407-1457) made it popular throughout Europe during the Renaissance, i.e. with the
special meaning of speaking a language with propriety and congruence in the manner of good
authors and good speakers. The same approach had been followed by Gilberti, who presented

Within their theoretical model, it appears that missionaries employed the

word ‘case’ with a double meaning: (1) Case in the sense of a formal marker
that differed in the various individual languages; here the most frequently used
term was ‘declension’; and (2) case in the sense of notional value applicable to
all languages. Accordingly, Gilberti and Lagunas refer to case in sense (1),
when they claim that in Michoacan there are only three or four cases,
respectively. On the other hand, when they present the case paradigm, they
seek to exemplify how case roles or their universal values are formally
manifested in that language by means of declensions or postpositions, and how
for some cases those values may, in turn, be modified by other postpositions to
add variety of meaning to such case values.

5. Closing remarks
Missionaries were simple practitioners of an epistemological model in
which the notional domain of language, the intelligible world, had precedence
over observable forms (either thought or voiced) in the various languages. In
addition, they were bound by their own practical and instructional needs,
which made Latin and European languages an effective contrastive means for
the speedy learning of exotic languages by their own fellow clergymen. It
would be wishful thinking to want them to work with other theoretical models
stemming from changes experienced by the sciences in the 16th, 17th, and 18th
centuries. Those changes brought about a shift towards a greater emphasis on
the study of sensory forms to the detriment of the study of the mental or
notional domain of language. In his Essay Concerning Human Understanding
(1690), John Locke (1632-1704) transmitted and refined ideas that had already
been floating around in European intellectual circles for some time. In his
time, the practice of observation and experimentation had already brought
about good results in the fields of astronomy and natural sciences such as
botany, zoology, physics, and geology.
This new intellectual climate contributed to the loss of supporters of
linguistic innatism (see Formigari 1970) and of practitioners of sacred
historicism involving attempts to reconcile Genesis with the history of the
languages of the various nations. However, the grammars and dictionaries
written by the missionaries also contributed to this loss. Such works
increasingly revealed that certain formal features and their parallel semantic
structures in languages from different geographical areas were open to
systematization and generalization. The descriptions carried out by
missionaries, a comparison of their works, and a gradual and better knowledge
of the sounds, the morphological structures, as well as of the syntactic and
semantic features of scores of tongues allowed researchers to perceive that

the basics or rudimenta of the noun declension in Part One, p. 12, and offered a more detailed
analysis of case and declension in Part Two, p. 57r-60v.

these had their own independent mechanisms, formal systems and subsystems
which, at times, were shared by other tongues. Instead of seeking to continue
with a model of universal grammar in the medieval and Renaissance guise,
justifiable when fewer languages were known, a novel and somewhat
revolutionary framework was developing, in which scholars searched for
general features of languages on the basis of observable and autonomous
characteristics. Modern linguistic typology and general linguistics founded on
empirical observations, as practiced by later linguists, including Humboldt
himself, would have been unimaginable or would have been delayed without
the work of early missionaries who were but a cog in the chain of such
historical developments (cf. Breva-Claramonte 1999, 2001 and 2004:61-62).
In sum, linguistic currents and grammatical traditions must be evaluated
according to their own merits and achievements and within their own
epistemological context rather than from a later perspective. Otherwise, we
run the risk of misinterpreting them and mistakenly finding errors and
inaccuracies that prevent us from determining their true place in the history of
our discipline.


Acosta, José de. 1954. Obras del Padre José de Acosta. Estudio preliminar y
edición del Padre Francisco Mateos. (=Biblioteca de Autores Españoles,
73). Madrid: Ediciones Atlas.
Aguirre Beltrán, Gonzalo. 1983. Lenguas vernáculas. Su uso y desuso en la
enseñanza: la experiencia de México. México D.F.: Centro de
Investigaciones y Estudios Superiores en Antropología Social.
Arens, Hans. 1980. “‘Verbum cordis’: Zur Sprachphilosophie des
Mittelalters”. Historiographia Linguistica 7:1/2.13-27.
Augustine of Hippo, Saint. 1968. Sancti Aurelii Augustini opera. De
Trinitate libri XV. Pars XVI, 1 and 2 (Corpus Christianorum, Series
Latina L-La). Cura et studio W.J. Mountain, auxiliante Fr. Glorie.
Turnholti: Typographi Brepols Editores Pontifici.
Breva-Claramonte, Manuel. 1999. “Morphosyntactic Analysis of Exotic
Languages in Lorenzo Hervás (1735-1809)”. History of Linguistics 1996:
Papers from the Seventh International Conference of the History of
Linguistics (Oxford, 12-17 September 1996), ed. by David Cram et al., Vol.
I, 81-89. Amsterdam & Philadelphia: John Benjamins.
Breva-Claramonte, Manuel. 2001. “Data Collection and Data Analysis in
Lorenzo Hervás: Laying the Ground for Modern Linguistic Typology”.
History of Linguistics in Spain ed. by E.F.K. Koerner & H.-J. Niederehe,
vol. II, 265-280. Amsterdam & Philadelphia: John Benjamins.

Breva-Claramonte, Manuel. 2004. “Contexto epistemológico y modelo

doctrinal de las ideas lingüísticas de Lorenzo Hervás (1735-1809)”.
Revista Española de Lingüística 34.39-68.
Calvo Pérez, Julio. 1997. “La gramática aimara de Bertonio (1603) y la escuela
de Juli”. Zimmermann 1997.321-338.
Formigari, Lia. 1970. Linguistica ed impirismo en el seicento inglese. (=Bibl.
di Cultura Moderna, 695). Bari: Laterza.
Gilberti, Maturino. 1987[1558]. Arte de la lengua de Michuacan. México:
Hecho en casa de Iuan Pablos impresor. Edición facsimilar con una
introducción histórica por Benedict J. Warren. Morelia, Michoacán,
México: Fimax Publicistas Editores.
Hernández de León-Portilla, Ascensión. 2003. “Las primeras gramáticas
meso-americanas: Algunos rasgos lingüísticos”. Historiographia
Linguistica 30:1/2.1-44.
Hernández Sacristán, Carlos. 1997. “Categoría formal, categoría funcional y
teoría de la traslación en las primeras gramáticas del náhuatl”.
Zimmermann 1997.43-58.
Humboldt, Wilhelm von. 1903-1936. Gesammelte Schriften. Ed. by Albert
Leitzmann, Bruno Gebhardt & Wilhelm Richter. 17 vols. Berlin: B. Behr.
Lagunas, Iuan Baptista de. 2002[1574]. Arte en la lengua michuacana.
Transcripción: Agustín Jacinto Zavalas. Zamora, Michoacán: El Colegio de
Michoacán. (This work also contains his Dictionario [sic].)
Mandonnet, R.P. Petrus. 1927. S. Thomae Aquinitatis opuscula omnia. Tomus
quintus. Parisiis: Sumptibus P. Lethielleux, Biblipolae Editores.
Molina, Alonso de. 1945[1571]. Arte de la lengua mexicana y castellana.
Facsimile reprint. Madrid: Ediciones de Cultura Hispánica.
Molina, Alonso de. 2001[1555]. Aquí comienza un vocabulario en lengua
castellana y mexicana. México: Por Juan Pablos. Edición facsímil con
estudio preliminar de Manuel Galeote. Málaga: Universidad de Málaga.
Monzón, Cristina. 1997. “Terminología y análisis de la estructura morfológica
en el ‘Arte en Lengua Michoacana’ de fray Juan Baptista Lagunas (siglo
XVI)”. Zimmermann 1997.133-145.
Nágel Bielicka, Federico Beals. 1994. “Aprendizaje del náhuatl entre
franciscanos y jesuitas en la Nueva España”. Estudios de Cultura Náhuatl
Nebrija, Antonio de. 1980[1492]. Gramática de la lengua castellana. Estudio
y edición de Antonio Quilis. Madrid: Editorial Nacional.
Nebrija, Antonio de. 1981[1481]. Introducciones Latinae. Salmanticae.
Facsimile reprint. Salamanca: Universidad de Salamanca.
Olmos, Fray Andrés de. 1993[1547]. Arte de la lengua mexicana. Edición y
estudio introductorio, transliteración y notas de Ascensión y Miguel León
Portillo. Madrid: Ediciones de Cultura Hispánica, Instituto de Cooperación

Padley, G. Arthur. 1976. Grammatical Theory in Western Europe 1500-1700:

The Latin Tradition. Cambridge: Cambridge University Press.
Rodrigues, Aryon D. 1997. “Descripción del tupinambá en el periodo colonial:
el Arte de José de Anchieta”. Zimmermann 1997.371-400.
Ross, William David, ed. 1908-1952. The Works of Aristotle. Translated into
English. 12 vols. Oxford: Clarendon Press.
Sanctius, Franciscus. 1986[1587]. Minerva seu de causis linguae Latinae.
Salmanticae: Apud Joannem & Andraeam Renaut, fratres. (Facsimile
reprint, with an introduction by M. Breva-Claramonte. Stuttgart-Bad
Cannstatt: Frommann-Holzboog).
Santo Tomás, Domingo de. 1994[1560]. Grammatica o Arte de la lengua
general de los Indios de los Reynos del Peru. Valladolid. Estudio y
transliteración por Rodolfo Cerrón-Palomino. Madrid: Ediciones de
Cultura Hispánica, Instituto de Cooperación Hispanoamericana.
Scaliger, Julius-Ceasar. 1584[1540]. De causis linguae Latinae, libri tredecim.
[Heidelberg:] Apud Petrum Santandreanum.
Thomas Aquinas, Saint. 1980. S. Thomae Aquinitatis opera omnia, curante
Roberto Busa S.I. 7 vols. Stuttgart-Bad Cannstat: Frommann-Holzboog.
Zimmermann, Klaus, ed. 1997. La descripción de las lenguas amerindias en
la época colonial. (=Bibliotheca Ibero-Americana, 63). Madrid:
Iberoamericana; Frankfurt am Main: Vervuert.
Zwartjes, Otto, ed. 2000. Las gramáticas misioneras en la tradición hispana
(siglos XVI y XVII). Amsterdam: Rodopi.
Zwartjes, Otto. 2002. “The Description of the Indigenous Languages of
Portuguese America by the Jesuits During the Colonial Period: The Impact
of the Latin Grammar of Emmanuel Alvarez”. Historiographia Linguistica

University of North Alabama


Chinese character writing, based only partially on phonetic considerations,

presents a challenge to the seemingly straightforward relationship between the
written record and etymology, where graphemic recordings of phonetic forms
(words) in a number of languages are compared to determine whether they
share a common ancestor. The very concept of ‘word’, vis-à-vis ‘character’ is
a complicated category in Chinese. Steinthal investigates Chinese character
writing and ‘character etymology’, taking note of its etymologically baseless
homographs and graphemic ‘synonyms’, and the problems these create for
reconstructing earlier stages of the language. A writing system which in effect
inverts the writing-etymology relationship by offering character etymologies in
place of word etymologies clearly demands that the very premises of
etymology be interrogated to achieve a more accurate assessment of its
intrinsic limitations.

In his 1854 essay “Zur vergleichenden Forschung der chinesischen

Sprache”, for which he shared the prestigious Prix Volney, Heymann Steinthal
(1823-1899) set out to apply the methodology and rigor of the comparative
method to the study of Chinese. It is important to note that, at this time, the
genetic relationship of Chinese to other members of what has, only relatively
recently, come to be called the Sino-Tibetan language family 2 had not been
established, nor were available dictionaries and other sources about the
language consistent or reliable, all of which makes Steinthal’s inaugural effort
all the more impressive. The obstacles Steinthal faced in applying the

This research was supported by the University of North Alabama, and by Friends of the
Department of Foreign Languages. Special thanks are due to Ms. Margaret Chien for her
patience in working with me in my studies of Chinese, as well as for her many valuable
insights into the spirit of the Chinese language and its unique system of writing.
Cf., for example, Matisoff 2003, whose ‘Sino-Tibetan Etymological Dictionary and
Thesaurus Project’ is aimed at reconstructing this important language family.

comparative method to the classification of Chinese were formidable. In terms

of available reference works, the extant dialects of Chinese had, for the most
part, been poorly recorded by missionaries and linguists using a welter of
notational systems, and the native sources, while helpful, were riddled with
inconsistencies, many born of the rampant and pernicious homophony
inevitable in a language with a very limited inventory of syllables. Most
importantly, the principles underlying the many layers of the Chinese writing
system themselves blur the boundaries between phonetic and symbolic
representation to such an extent that attempts to establish etymologies are truly
confounded at every turn. In examining Steinthal’s analysis of the status of
etymological issues in the Chinese language, I hope, simultaneously, to call
attention to certain weak spots in the underpinnings of conventional views of
etymology in general.
Steinthal’s manuscript essay was first published in the Prix Volney project,
under the editorship of Joan Leopold, one hundred years after his death in 1899
–and nearly a century and a half after he wrote it. As Jerold Edmondson, who
edited Steinthal’s essay within this important collection, points out in his
accompanying essay entitled “Steinthal and the History of Chinese
Steinthal was the first to apply systematically the principles of historical and
comparative linguistics to the Chinese language. His use of the comparatist’s tools on
large quantities of data from three contemporary varieties of Chinese—Mandarin,
Min and Cantonese—and his relating these to the rhyme tables of the Song Dynasty
was truly a stunning accomplishment for someone who never left Europe, especially
in light of the imperfect sources he had to work with. Materials about spoken Chinese
were, in the middle of the nineteenth century, limited in scope and often riddled with
extravagant exaggerations of fact and errors of transcription. (1999:381)

Moreover, Edmondson adds, Steinthal’s analysis of historical Chinese

phonology was so ahead of his time that “…some aspects of [his] views about
the syllable in Chinese could have been taken from the pages of a modern
journal” (1999:403), specifically his innovative insights into the evolution of
tones, tonogenesis. For our purposes here, it is important to take note of this
achievement since Steinthal’s analysis of a proposed “…phonological
dependency between tones and initials” (1999:390), in which he showed a
“…great sophistication about the physiology of sound production” (1999:406),
at the same time makes it reasonable to believe that his etymological efforts
have risen above the limitations of the data available to him, and the challenges
which the nature of the Chinese language itself presents to the etymologist.
In a section dealing with “The general characteristics of Chinese in regard
to its suitability for comparative study,” Steinthal sets the stage for his more
detailed assessment of “The Chinese writing system as an aid to etymological

investigation”. 3 Here he observes the simple syllable structure of Chinese, its

lack of grammatical inflections and morphological alternations otherwise so
helpful in internal reconstruction, and, in particular, the problems associated
with homophony, which he deems “…the most vexatious of all enemies of
etymology” (1854:426), all the more so in Chinese, where, in light of its
minimal syllable inventory, it is especially challenging to determine from
which homophone a given meaning derives. Could the unique writing system
of Chinese perhaps compensate for the shortcomings of the phonetic evidence
in detecting etymons? Yes and no. As Steinthal is quick to point out:

It must however be noted that a symbolic system of writing like the Chinese, which
developed out of pictures with original and derivative meaning, and which yet bears
this character of pictographic writing, is thoroughly etymological in nature, but
certainly not to the advantage of the etymologist, but rather to lead him astray.

In other words we must differentiate, in Chinese, between etymology in the

conventional sense—in which word relationships are determined on the basis
of either shared sounds or regularly derived reflexes of once common
sounds—and the ‘etymology’ of the written signs or ‘characters’, an attempt at
graphemic reconstruction teeming with crisscrossed semantic and phonetic
associations. Clearly, this is quite a different kind of ‘etymology’, and yet, in
the face of meager phonetic evidence for conventional reconstruction of words
and word elements (as opposed to reconstruction of written signs which
originated as pictograms), every possible clue is too precious to disregard.
Steinthal points out that, even in the old, classical literature, the written
characters were largely non-transparent to the first Chinese writers, who, he
notes, “…already no longer knew the etymology of their words” (1854:438).
When Steinthal declares this system of writing to be “thoroughly etymological
in nature”, he is referring to the manifest compositional nature of the written
characters. And yet he cautions us not to take the symbolic meaning of a
written sign as necessarily reflecting linguistic etymology. As often as not, the
composition of the written characters reflects simply what early writers held to
be the meaning of the constituent elements. In other words, the Chinese
writing system presents us with as much folk etymology as etymology. From
the point of view of the Humboldtian concept of ‘inner language form’, so
prominent in Steinthal’s language theory, 4 this is not necessarily a completely
bad state of affairs: after all, if the writing system symbolically registers
concepts, it stands to reason that it might offer a unique glimpse into the way
those concepts came to be imagined and composed, both initially and
subsequently. In passing, and as a testament to the bewildering barriers
Section titles are Edmondson’s translations; cited passages are my own translations.
For more on the concept of inner language form see Christy 2002 and Christy 1985.

Chinese presents to the etymologist, it is worth noting that, to date, there is no

equivalent for Chinese word etymology of, say, Skeat’s Etymological
Dictionary of the English Language or Kluge’s Etymologisches Wörterbuch.
Indeed, the term ‘Chinese etymology’, as referenced in current sources, seems
inevitably and exclusively to be paired with the idea of character etymology,
and, in my own experiences querying native speakers, more often than not I am
asked “What do you mean, etymology of the word?”, suggesting that the idea
of the etymology of the spoken, as opposed to written, word has no
psychological reality.
The concept ‘word’ is, to be sure, problematic in Chinese, where the
syllable is the most salient linguistic unit. While we are accustomed to saying
a given word has a given meaning, things are not always so clear-cut in
Chinese. Steinthal (1854:425-6) cites as an example the Chinese word for
house, fáng zi [房子], and notes that, by itself, fáng has no meaning in speech.
Yet in the phrase yì jiān fáng [一 間房] ‘one [classifier] house’ fáng does
indeed mean house, even without the following zi. Steinthal concludes that
“The representation of composite Chinese words through the simple elements
of which they are formed proves that these monosyllabic elements are true
monosyllabic polysemic words, and not just meaningless syllables”
(1854:426). Yet precisely because second-tone fáng corresponds to a
multiplicity of meanings (I count five second-tone fángs), it is unintelligible in
speech when isolated from a context. From this we may gather that the concept
of ‘characters’ is in fact more relevant than that of ‘word’, and that the issue of
meaning has one sense, in relation to speech, and another in relation to writing.
In Chinese, a language especially rich in synonyms, many words consist of
two semantically related morphemes, a pattern that, on the one hand, may have
evolved to counterbalance the effects of homophony, and, on the other, mirrors
the Chinese predilection for coupling elements as evidenced also by the way
so-called ‘radicals’ (determinatives) are paired with ‘phonetics’ to give a clue,
in the writing, as to the meaning and the sound, respectively. Thus while, as in
the example above, a given syllable such as fáng can be analyzed as having
meaning, in actual practice combinations of the sort chī fàn [吃飯] ‘eat’, shuō
huà [談話] ‘talk’, péng yǒu [朋友] ‘friend’ are the rule in the vernacular.
However, in the laconic classical literature—where, as Steinthal claimed, logic
replaced grammar—isolated characters bereft of relational terms predominate,
addressing the eye rather than the ear. 5
The traditional distinction drawn between morphemes and words is not so
clear cut in Chinese, largely because of confusion about the linguistic status of
the third term, character, which is a unit of pronunciation as well as the
fundamental unit of the writing system. Just as it is often difficult in Chinese

Cf. Steinthal 1854:434.

to determine what belongs to speech, and what to writing, it is similarly not

always easy to avoid confusing the terms ‘character’, ‘word’, and ‘morpheme’.
A few examples will highlight the dimensions of this confusion. 6 While some
characters have meaning, for example 好 hǎo ‘good’ and 人 rén ‘man’, others,
such as 葡 pú and 萄 táo, have no meaning individually, though in
combination, 葡萄 pútáo, the meaning is ‘grape’. Owing to homophony, such
determinations as these can only be made by inspecting the characters since
there are a half dozen or more second-tone pú’s and táo’s. I might add that, in
dictionaries, 7 both pú and táo are glossed as meaning ‘grape’, though in actual
practice, this is not accurate. The same can be said of péng yǒu ‘friend’, cited
above, and countless other examples. When we also consider that this same pú
has the meanings ‘crawl’ and ‘vine’, meanings that would be evident in other
combinations and contexts in both speech and writing, it becomes ever more
difficult to say that pú has no meaning, and is therefore non-morphemic, in the
combination pútáo. The concept of ‘character’, we might say, simply does not
mesh well with the concept of ‘morpheme’.
As for what we call ‘words’, some correspond to an individual character
while others are comprised of two or more. Since most morphemes in Chinese
are monosyllabic, it is easy to see why they are often identified with characters,
even though they are not identical. Morphemic analysis, as we have just seen,
forces us to label some characters, in some combinations, as being
meaningless. And while from a purely synchronic perspective we might say
the –ty of ‘loyalty’ is meaningless, this is really not comparable to saying that a
character, which, unlike –ty, has visually real graphemic individuality and
identity, is meaningless.
As if this were not already sufficiently confusing, a single character can,
moreover, represent completely different, albeit semantically related,
morphemes. The character 樂 , for example, corresponds to both lè, as in 快樂
kuài lè ‘happy’ and yuè, as in 音樂 yīn yuè ‘music’, the basis for the shared
sign apparently being the idea that music produces the effect of pleasure and
joy. 8 As for the etymology of this now highly conventionalized character, it
presumably depicts “…a wooden…support on which…a drum and…bells are
hung,” the drum in the middle with the bells on either side (Wieger 1965:224). 9

These examples are taken from Zhenhua 2000:1-8.
Cf. Fenn 1942, Wieger 1965, and Yeh 2001.
Cf. Wieger 1965:224.
Cf. also Wilder & Ingram 1974:155-156, where it is pointed out that the “drums are on the
sides and the bell is in the middle”. The addition of the grass radical to this character, it is
noted, yields the character for yao ‘medicine’, the idea being that “…vegetable substance[s]
…will restore the proper functioning of the body; restore harmony [emphasis added]”.
Vegetable substances were first used as medicines. Cf. also Hongyuan 1997:178.

Like yuè, the character 快, cited above in the compound kuài lè ‘happy’,
has several readings. This “character 10 may be a picture of a man drawing his
bow or fitting the thimble before he draws. It occurs as a part of characters to
give sometimes the sound jue, sometimes the sound guai (or kuai)”
(McNaughton & Ying 1999:124). This character also figures in Steinthal’s
explanation of why phonetic information alone can never be sufficient in
establishing etymons in Chinese. Taking as point of departure the form 決 jué,
which, with the ‘water’ radical, 11 signifies ‘the flowing of water through a
ruptured dike’, then ‘decide’, ‘definite’, ‘steadfast’, ‘certain’, and, he claims,
‘fast’—though this character is actually, or now, written with the ‘heart’
radical—Steinthal wonders how certain we can be that we are in fact dealing
with related concepts from a single semantic nucleus. How can we be sure that
this does not, instead, represent the convergence of several semantic nuclei?
After all, the idea of ‘rushing water’ is the virtual opposite of the sense
‘definite’, ‘steadfast’, and indeed there exists a variant form written with the
‘ice’, rather than the ‘water’, radical, suggesting ‘firmness’, ‘steadfastness’.
For that matter, he adds, “…what is actually carrying this meaning? -a word, or
a written sign? The entire semantic development could belong to the sign
inasmuch as the word itself originally had neither the purely material nor the
purely abstract meaning, but rather something in between, for example ‘fast’.
How was this word to be represented ideographically in writing? The writing
system could obviously achieve this only in a symbolic fashion and seized on
the image of rapidly flowing water. In this way the word received a material
sense, which, in the beginning, it never had. More precisely, the word actually
did not receive this meaning, but rather the image sign was used symbolically,
and the symbolic meaning of the image signs grew ever more absolute”
Noting that another character, 訣, with the same pronunciation jué, has the
meanings ‘separate’, ‘go away’, and even ‘sorcery’, Steinthal cannot help but
wonder whether we really still have here the same word, or rather several
In the final analysis all these meanings can be united; the like meanings of both signs
are perhaps separated by fine, synonymic differences, and yet we have here not two
synonymous words, but rather two synonymous signs; the small difference in
meaning belongs not to the word, but is instead brought out through the form of the
sign and its original meaning. (1854:435)

Thus jué, written with the ‘water’ radical, contrasts with jué, written with the
‘horse’ radical to give the synonymous meanings ‘fast like water’ and ‘fast like

I.e., the component to the right of the ‘heart’ radical.
The component on the left is the ‘water’ radical.

a horse’ but these semantic distinctions belong not to language, but to writing.
For this reason, Steinthal concludes:
This sign 12 is not translatable into any language, precisely because it does not speak,
but rather paints; the painter may translate it into a painting, no word is so pregnant as
to express an entire image. In this potentiality lies the achievement of the old Chinese
literature; but the etymologist! he seeks language in vain. (1854:436)

The issue of identity and difference is also central to Steinthal’s

explanation of just how sound came to be separated from meaning in rebus
writing. It is simply not sufficient to say that homophony opened the door to
rebus substitutions since this leaves unexplained how the ideographic sign,
contrary to its very nature, could acquire phonetic value. 13 He finds the
solution in signs in which a constituent element functions both ideographically
and phonetically. Using the ideogram for ‘stick’, 丈 zhàng, he notes that,
coupled with the ‘hand’ radical, the idea of ‘hitting’, also pronounced zhàng,
could be written, as could the idea of ‘supporting’ or ‘leaning’, again zhàng, by
adding the ‘man’ radical. In this group of signs, each was “…doubly bound up
with the other, and the difference of each from the other was part of its essence
and content, and none of these signs could be written without simultaneously
calling the others to memory” (1854:431). To this forward-looking statement
of the differential nature of the sign, 14 Steinthal adds that the common
pronunciation of these signs, all pronounced zhàng, coupled with the image-
sign for ‘stick’, the nucleus of their semantic kinship, guaranteed that the
sound zhàng would be evoked in whatever combination the ‘stick’ image
appeared, such that, to write the idea ‘sickliness’, also pronounced zhàng
though semantically unrelated to ‘stick’, one needed only to add the ‘sickness’
radical. In this way the figurative meaning of the image receded, and the
sound image was born. Once this metamorphosis has taken place, the
divergent meanings maintain their separate identities despite sharing a
common pronunciation. Grammaticalization, 15 the process whereby lexical
items come to serve as markers of grammatical function, presents us with
allied phenomena: in a sentence such as ‘I can’t wait to go’, the primary
meaning of ‘wait’, ‘bide time’, simply does not come to mind. Here the phrase
‘can’t wait to’ functions as an auxiliary to signal impatience with respect to the
action of the main verb.

I.e., 決 (minus the component on the left, the three-line ‘water’ radical, here included since
the component on the right no longer exists as an independent character).
For a discussion of the role of homophony in Steinthal’s theory of writing see Steinthal 1851
and Christy 1995.
Cf. Christy 1999.
See Christy 2003 and 2000 for further discussion of grammaticalization and the lapse of the
etymon from consciousness.

In seeking to reconstruct Chinese word etymologies by correlating dialect

variants, Steinthal sought to overcome the limitations posed by forms which,
having undergone drastic phonetic loss, particularly loss of final consonants,
had become homophonous with semantically unrelated forms. His approach
was entirely in keeping with the methodology of nineteenth-century
comparative philology, as described, for example, by Whitney in his Language
and the Study of Language of 1867:

In the view of the present science, while each existing dialect is the descendant of an
older tongue, so other existing dialects are equally descendants of the same tongue.
All have kept a part, and lost a part, of the material of their common inheritance; all
have preserved portions of it in a comparatively unchanged form, while they have
altered other portions perhaps past recognition. […] Each has saved something which
others have lost, or kept in pristine purity what they have obscured or overlaid: or
else, from their variously modified forms can be deduced with confidence the original
whence these severally diverged. […] Thus the deficiencies of the evidence which
each member of a connected group of dialects contains respecting its own genesis and
growth are made up, in greater or less degree, by the rest, and historical results are
reached having a greatly increased fullness and certainty. (1867:240)

While Steinthal was keenly aware of the many difficulties involved in trying to
reconstruct Chinese—“but the etymologist! he seeks language in vain”
(1854:436)—and the ways in which this state of affairs also called attention to
some of the basic assumptions of etymological practice, the views he advanced
in his Prix Volney essay, first published 145 years later, seem to have had no
noticeable effect on the contemporaneous practice of comparative philology,
not an entirely surprising result given that this yet young science was
overwhelmingly focused on western languages recorded in alphabetic writing
systems. Steinthal consulted both dialect variants and the Chinese writing
system itself for clues to help in recuperating word etymologies. Yet precisely
because this writing system is an amalgam of now phonetic, now semantic
encodings, which overlap and crisscross in a dizzying array of ways, its value
to the word-, as opposed to character-etymologist, is negligible. Etymology,
at least in the conventional sense, is based on written records of sounds.
Chinese presents us, on the one hand, with pictographs, which, properly
speaking, have no pronunciation but rather only a meaning, which itself is then
bound to sound. On the other hand there are composite signs in which one half
points to the meaning and the other to the sound. The only problem is the
meaning clues are for the most part extremely vague and the clues to the sound
as often as not involve sound loans which themselves reflect expired
pronunciations. The closest we come to phonetic writing is with the use of
homophonic elements to encode individual syllables, with no further reference
to the signs’ semantic origins. As Steinthal concludes, none of this offers the
etymologist much help. To chart the history of words, rather than characters,

in Chinese requires scrutiny of extant dialects and such early analyses of the
language as found in the rhyme books and studies of word initials and finals. 16
But that is another story, and one that is still unfolding. I for one look forward
to the day when I can thumb the pages of an etymological dictionary of the
Chinese language. How very much we take for granted!


Christy, Craig. 1985. “Humboldt’s ‘Inner Language Form’ and Steinthal’s

Theory of Signification”. Semiotics 1984 ed. by John Deely, 251-259.
Lanham, Maryland: University Press of America.
Christy, Craig. 1995. “From Concept to Word: The Homophonic Principle
and Steinthal’s Theory of Writing”. History of Linguistics 1993: Papers
from the Sixth International Conference on the History of the Language
Sciences (ICHoLS VI), Washington, D.C., 9-14 August 1993 (= Studies in
the History of the Language Sciences, 78) ed. by Kurt R. Jankowsky, 199-
207. Amsterdam & Philadelphia: John Benjamins.
Christy, Craig. 1999. “Between Intellect and Intuition: Saussure’s
‘Anagrams’ and the Calculus of the Auri-(Ora-)cular”. Interdigitations:
Essays for Irmengard Rauch ed. by Gerald F. Carr, Wayne Harbert, &
Lihua Zhang, 557-568. New York & Berlin: Peter Lang.
Christy, Craig. 2000. “Bréal and the Semantics of Etymological
Development: On the Need to Forget in Order to Re-member”. The
History of Linguistic and Grammatical Praxis: Proceedings of the XIth
International Colloquium of the Studienkreis ‘Geschichte der
Sprachwissenschaft’, Leuven, 2-4 July 1998 (= Orbis Supplementa, 14) ed.
by Piet Desmet, Lieve Jooken, Peter Schmitter & Pierre Swiggers, 519-
527. Leuven, Paris, & Sterling, Virginia: Peeters.
Christy, Craig. 2002. “From Articulation to Comprehension: Steinthal and the
Dynamics of Linguistic Intangibles”. Chajim H. Steinthal:
Sprachwissenschaftler und Philosoph im 19. Jahrhundert (= Studies in
European Judaism, IV) ed. by Hartwig Wiedebach & Annette
Winkelmann, 3-16. Leiden, Boston & Köln: Brill.
Christy, Craig. 2003. “Tooke’s ‘Abbreviation’ and Bréal’s ‘Latent Ideas’: A
New Perspective on Grammaticalization”. History of Linguistics 1999:
Selected Papers from the Eighth International Conference on the History
of the Language Sciences (ICHoLS VIII), Fontenay-St. Cloud, 14-19
September 1999 ed. by Sylvain Auroux (= Studies in the History of the

Bernhard Karlgren (1889-1978) pioneered this type of analysis with his Old Chinese
reconstructions. Cf. Karlgren 1992 and 1949. Norman 1988 provides a useful overview of
Karlgren’s method.

Language Sciences, 99), 237-246. Amsterdam & Philadelphia: John

Edmondson, Jerold A. 1999. “Steinthal and the History of Chinese
Linguistics”. The Prix Volney: Contributions to Comparative Indo-
European, African and Chinese Linguistics: Max Müller and Steinthal (=
The Prix Volney Essay Series, III) ed. by Joan Leopold, 381-413.
Dordrecht, Boston & London: Kluwer Academic Publishers.
Fenn, C. H. 1942 (1st ed. 1926). The Five Thousand Dictionary. Cambridge,
Mass.: Harvard University Press.
Hongyuan, Wang. 1997. Vom Ursprung der chinesischen Schrift. Beijing:
Karlgren, Bernhard. 1949. The Chinese Language: An Essay on its Nature
and History. New York: Ronald Press.
Karlgren, Bernhard. 1992 (1st ed. 1954). Compendium of Phonetics in Ancient
and Archaic Chinese. Taipei: SMC Publishing.
Matisoff, James A. 2003. Handbook of Proto-Tibeto-Burman: System and
Philosophy of Sino-Tibetan Reconstruction. Berkeley: University of
California Press.
McNaughton, William, and Li Ying. 1999. Reading and Writing Chinese: A
Guide to the Chinese Writing System. Boston: Tuttle Publishing.
Norman, Jerry. 1988. Chinese. Cambridge: Cambridge University Press.
Steinthal, Heymann. 1851. Die Entwicklung der Schrift. Berlin: Dümmler.
Steinthal, Heymann. 1854. “Zur vergleichenden Erforschung der chinesischen
Sprache”. The Prix Volney: Contributions to Comparative Indo-European,
African and Chinese Linguistics: Max Müller and Steinthal (= The Prix
Volney Essay Series, III) ed. by Joan Leopold, 1999: 414-498. Dordrecht,
Boston & London: Kluwer Academic Publishers.
Whitney, William Dwight. 1867. Language and the Study of Language. New
York: Scribner.
Wieger, L. 1965 (1st ed.1915). Chinese Characters: Their Origin, Etymology,
History, Classification and Signification. A Thorough Study from Chinese
Documents. New York: Dover.
Wilder, G. D. & J. H. Ingram. 1974 (1st ed. 1922). Analysis of Chinese
Characters. New York: Dover.
Yeh, The-Ming, ed. 2001. Far East Pinyin Chinese-English Dictionary.
Taipei: Far East Book Company.
Zhenhua, Guo. 2000. A Concise Chinese Grammar. Beijing: Sinolingua.

University of Ottawa


Recourse has been taken to Thomas Kuhn’s model of scientific change in

order to address the Neogrammarians’ “revolution” in linguistics. But the
Kuhnian approach seems inadequate in view of the absence of any “crisis” at
the empirical level prior to the Brugmann/Osthoff manifesto, and of any
ensuing relation of “incommensurability” between paradigms, granted that a
“paradigm shift” did occur. We aim at showing that Imre Lakatos’
“Methodology of Scientific Research Programs” is better suited as analytic
framework in the epistemological assessment of the Neogrammarian
movement. In particular, Lakatos’ notions of “hard core” and “negative
heuristic” of a “scientific research program” provide a better understanding of
the impact of the two main tenets of the movement. This view suggests that the
Neogrammarians, while aiming with their first principle (the exceptionlessness
of sound laws) at strengthening the notion of explanation in linguistic theory,
integrated the principle of analogy (which formerly functioned as “negative
heuristic”) within the “hard core” of their scientific program.

1. Introduction
A series of works from Hermann Osthoff (1847-1909), Karl Brugmann
(1849-1919), Hermann Paul (1846-1921) and Berthold Delbrück (1842-1922)
appeared in the years 1878-1880 that gave rise to the so-called Lautgesetz
Controversy. A group of young linguists, soon to be known as the
Neogrammarians, announced a decisive break with the scientific tradition in
which they were trained. A manifesto written by Brugmann and co-signed by
Osthoff proclaimed two basic principles that were henceforth to underlie the
scientific status of the study of language. They also put forward general views
on methodology and language theory, and denounced as metaphysical the
attempts at reconstructing the Indo-European source language.
I wish to briefly consider the methodological and epistemological aspects
of the Neogrammarian movement with the view of showing that a) Thomas

Kuhn’s (1922-1996) views and categories concerning theory evolution meet

with important difficulties here, and b) that Imre Lakatos’ (1922-1974)
“Methodology of Scientific Research Programs” seems better suited as analytic
framework. In particular, Lakatos’ notions of “hard core” and “negative
heuristic” allow for a cogent assessment of the epistemological function of the
Neogrammarians’ basic principles.

2. The background
In 19th-century Germany, the study of language evolved into an
autonomous university discipline in the form of historical comparative
linguistics. How this came about is well known. In his influential paper “On
the Language and Wisdom of the Indians” (1808), Friedrich Schlegel (1772-
1829) had called for a thoroughgoing study of the “internal form” or “inner
structure” of languages belonging to the Indo-European family; “Comparative
grammar”, he wrote, “will give us totally new insights into the genealogy of
Indo-European languages, much in the way comparative anatomy has shed
light on higher natural history”. 1 Some ten years later, three seminal works
from Franz Bopp (1791-1867), Ramus Rask (1787-1832) and Jacob Grimm
(1785-1863) set the ground for the transformation of this agenda into a genuine
scientific research program. 2 The second edition (1822) of Grimm’s Deutsche
Grammatik contained his ‘paradigmatic’ sound law on consonant mutation.
This prepared the ground for further development of Indo-European studies as
a scientific discipline based on explanatory phonetic laws and involving
recourse to hypotheses pertaining to the original Indo-European language. In
terms of textbooks, this process of maturation can be dated from Bopp’s
Comparative Grammar (1833-1852) to August Schleicher’s Compendium
(1861-1862). The Neogrammarians took the 4th edition (1876) of this latter
work as basis for their research. During this evolution, linguistic theory was
permeated by general Romanticist views construing language as a living
organism; indeed, linguistic theory construed itself as a cousin of anatomy and
biology. 3

3. The controversy
The story took place in Leipzig. Schleicher’s colleague in Leipzig, Georg
Curtius (1820-1885), had two young students, Karl Brugmann and Hermann
Osthoff. In 1876, Curtius was away on a trip and Brugmann took over

Schlegel 1808[1977]:28.
Bopp (1816), Rask (1818), and Grimm (1819).
Ever since Schlegel’s original construal of comparative grammar as vergleichende Anatomie,
natural science metaphors became gang und gäbe in the field. According to Christmann
(1994:203), and Koerner (1995:58), Schlegel most probably borrowed the expression from
Georges Cuvier’s (1769-1832) Leçons d’anatomie comparée (1800-05).

responsibility as co-editor for the 9th issue of Curtius’ Studien zur griechischen
und lateinischen Grammatik [Studies in Greek and Latin Grammar].
Brugmann inserted in the issue an article of his own dealing with one of the
pending questions on nasal consonants in common Indo-European and in
which he defended views opposite those of Schleicher. Back in Leipzig,
Curtius disavowed his young collaborator and disclaimed any responsibility for
the claims of Brugmann’s article. Brugmann ceased his co-editorship and,
together with Osthoff, started a new journal: Morphologische Untersuchungen
auf dem Gebiete der Indogermanischen Sprachen [Morphological Studies in
Indo-European Languages]. The first issue appeared in 1878, opening with a
Preface that had all the looks of a manifesto, proclaiming among other things a
new, truly scientific program for the study of language. The propounded views
diverged fundamentally from those of Curtius and, above all, Schleicher, who
had dominated the field of historical comparative linguistics in the 1860s.
The young scholars attacked the project of reconstructing the Indo-
European source language, which had so far been the main aim and center of
the entire field of comparative philology. Indeed, theoretical research had been
mainly directed at reconstructing this language, whereas the younger living,
contemporary languages were, “with a certain disdain, left out of consideration
and branded as degenerate, sunken, aging phases”:
It is not on the basis of hypothetical original linguistic structures, nor on the basis of
the oldest documented forms of Indic, Persian, Greek, etc., whose earlier forms can
only be inferred by hypothesis, that we should, in general, form a picture of the
evolution of linguistic forms, but, in accordance with the principle that one should
proceed from the known to the unknown, on the basis of those linguistic
developments whose earlier phases can be read, over relatively long periods, from
available sources so that the starting point is certain and established. 4

Osthoff and Brugmann argued that greater weight should be lent to

contemporary, living languages and dialects, and less to inevitably futile
attempts at reconstructing the presumed original Indo-European language:
Only those comparative linguists who manage to leave the hypothesis-laden
atmosphere of the workshop where the Indo-European root forms are wrought and to
enter the clear air of tangible actual reality so as to gain insight into things that grey
theory will never show […] – only those can achieve an adequate picture of how
linguistic forms live and change. 5

They announced that the works of Wilhelm Scherer (1841-1886) and

August Leskien (1840-1916) had given the initial impulse to a new approach
which they themselves applied, adding that they were followed in this

Osthoff and Brugmann (1878), quoted by Seuren 1998:90.
Seuren 1998:91.

approach by other young linguists who presented themselves under the banner
of the Neogrammarian Movement (die junggrammatische Richtung). 6
The Preface proclaimed two basic principles that were to underlie the
scientific status of the study of language. Following Leskien, they proclaimed
first of all the absolute exceptionlessness of sound laws:
First, all phonetic change, in so far as it occurs mechanically, proceeds according to
exceptionless sound laws. That is, the direction of the sound shift is identical for all
members of a speech community, except when dialect split occurs, and all the words
in which the sound subjected to the change appears in the same relationship are
affected by the change without exception. 7

The second principle acknowledged an important role for analogy in

linguistic change, namely in all stages of the development of languages:
Second, since it is clear that form association, that is, the creation of new linguistic
forms by analogy, plays a very important role in the life of the more recent languages,
this type of linguistic innovation is to be recognized without hesitation for older
periods too, and even for the oldest. 8

Wherever the premise of the inviolability of sound laws failed, analogy

(the precise nature of which need not concern us here) was to be applied as an
explanation. Thus, exceptions were to be understood as a (regular) adaptation
to a related form. These two principles ensured Gesetzmäßigkeit, or
nomological regularity, of linguistic phenomena.
The manifesto also formulated other important tenets of the new movement
consisting in general views on methodology (insisting that sound c