Introduction
‘Universal Grammar’ (UG) is a technical term in what has become an increasingly successful
effort to construct a biologically based natural science of language and of its development in the
individual and the human species. Like any term in a science that has made progress since the
term’s introduction, its use and application have undergone several changes, some of them
major. And like any term in a formalized natural science that has made progress, its connection
(without the capitals) is uninteresting. My history of UG begins with its introduction within the
work of Chomsky. That history – a short one that begins about 50 years ago and ends for
Most readers of this volume expect something else. A few might expect a general discussion of
the uses of the term ‘universal grammar’ and its cognates and component terms and their
cognates in various languages at various times in the works of the thousands who made efforts to
make sense of language and universality. They will not get that here. They are unlikely to get
anything of that sort at all; it would go back to Plato and before, would fill several volumes, and
would fail to get published, or probably even read. The sensible reader looking for a history
might expect a discussion of figures, views, and texts prior to Chomsky’s work that detail
influences on his accounts of language and of UG. Remaining within the 20th century, several
have done this, including Chomsky. It is easy to find accounts of Chomsky’s instructors, of
various individuals and views he was sympathetic to and confronted, of the works he read and
was influenced by and reacted against as he developed his account of language and UG. This
can be good intellectual history, but with rapidly diminishing returns as one attempts to go
further back.
Abandoning intellectual history proper, Chomsky went further back on a different project –
looking for connections between his work on language and UG in the works of others from the
17th century to the 19th. While this project of looking for precedents and parallels continues, his
most expansive effort (and among the earliest) is found in his intriguing Cartesian Linguistics
(CL). A brief but complex and rich work, CL very selectively places various philosophers,
linguists, poets, and polymaths from the 17th to 19th centuries in a ‘rationalist’ or ‘Cartesian’
group with which he was and is sympathetic, or an ‘empiricist’ one that he took (and takes) to be
a methodological dead end. Chomsky explicitly denied that CL was a work in history; it is a
controversial, raising the ire of intellectual historians, linguists, and several others who read it. I
am in what may be the minority who find CL illuminating, plausible, and undeserving of most
(but not all) of the criticism it has received. I use the opportunity this chapter offers to try to say
why, and to do so in a way that will – I hope – be useful to those who are unfamiliar with CL
itself.
To anticipate objections of one sort, the history of the technical term “Universal Grammar” with
which I begin is integral to the discussion of CL and its different project too. As mentioned, CL
notes the efforts of some individuals from Descartes through Humboldt who explored different
aspects of an internalist and nativist strategy for the natural science study of mind and language.
That strategy culminates in the current biolinguistic program for the study of language and UG.
This is a program that in Chomsky’s and others’ hands is by the standards of natural science
research not just the most recent but the most successful research program for the study of
language and UG. It is most successful because it offers the best opportunity so far to explain
and describe the nature of ‘the human language’ and the specific contribution of biology to it.
Earlier efforts described in CL were on the right track in some ways, and current work indicates
how. Prior to Chomsky, however, the individuals CL places among the rationalists lacked the
specific formal tools, the creation of biology as a natural science and the branch of it that studies
organic development ( “evo-devo”), and the focused efforts of individuals who are consciously
engaged in making progress in advancing the natural science of language and UG. They also
often also held views and harbored unwarranted assumptions that prevented them from
Keep in mind that the labels “rationalist” and “empiricist” are themselves semi-technical terms.
While they have historical significance, they are intended as labels for research strategies for the
study of mind. The rationalist takes seriously poverty of the stimulus observations with regard to
the acquisition of various mental capacities, including language – its combinatory mechanisms
and its primitive concepts and sounds – and assumes that it is reasonable in studying the mind to
suppose that at least some, and perhaps a great deal of what is ‘in’ the mind is actually the
1
‘Innate’ implies neither consciousness nor realization of a concept or a language at birth, or
ever. Generally (but perhaps not universally), an innate concept needs to develop to be available,
and it needs to be ‘triggered’ by some relevant ‘input’ to develop at all. The important
then it has some empirical warrant. In the case of the study of language specifically (and perhaps
other systems to an extent), the rationalist also takes seriously Descartes’s “creative aspect of
language use” observations and assumes that the proper object of the study of language is not its
use by humans (in ‘communities’ or not), but the (native) system(s) that provide the materials
needed for use – cognitive ‘tools’ offered by sententially expressed “perspectives” (Chomsky’s
term). The rationalist is also an ‘internalist’, drawn to assuming mental modularity, and – when
combined with the success of theories that proceed on the assumption that ‘content’ is innate – to
a kind of ‘constructivism’ that supposes that the mind’s internal systems (‘mental content’)
largely fix how the mind can conceive the world. The empiricist is the opposite: the empiricist
rejects nativism (especially in recent years with regard to language-specific notions), and is
externalist with regard both to ‘mental content’ and field of study. Modularity too is as a rule
denied. With regard to language, the empiricist assumes (in line with common sense) that
languages are public entities outside the head, perhaps institutions or practices of communities,
If we take the use of ‘Universal Grammar’ (UG) to be that found in at least one instantiation of
Chomsky’s Minimalist linguistic program, its immediate history could begin around the year
2000 with the introduction of different versions of the syntactic operation Merge, followed by
2005’s “Three Factors” paper and more recent work (e.g., 2008). Taken together, UG becomes
neonate infant’s genome-specified ‘initial state’ of a “language faculty.” At birth, the faculty
considerations are that the concept develop swiftly and internal systems both ‘anticipate’ the
nature of the concept and fix the nature of the relevant trigger(s).
(UG) is a biologically based developmental agenda in a human infant’s head that when given
input that the faculty specifies, automatically develops into a child’s I-language or I-languages.
The faculty so conceived is a nature-based ‘mechanism’ of sorts that yields a specific version (an
procedure that receives needed input and – relying also on non-biological (physical, etc) and
non-specific development constraints – yields a stable ‘final state’ in a child’s mind/brain by the
age of 3;5 or 4. The final state is thus the result of 1) genetic specification (biology), 2) ‘input’
effective computation.
This recent version of UG relies on insights from the early 1980s built into what was introduced
then and called the “principles and parameters” program. By introducing parameters thought of
sounds/signs, and perhaps concepts of human languages, the program offered a reasonable way
to address the issue of how a child acquires an I-language under the conditions described in the
“poverty of the stimulus” observations. It also offered the opportunity to extend the scope of
efforts to not just describe UG and the various I-languages that could develop from it, but to
explain how the uniquely human capacity to acquire and use languages came to be introduced to
the human species (evolution). In effect, it offered the opportunity to begin to include in the
explanatory task of the natural scientist of language the job of not just getting an answer to the
acquisition issue (popularly called “Plato’s Problem”) and what might be called – lexical item
‘choice’ aside – the “diversity of language” issue, but that of beginning to try in a serious way to
address questions concerning the biological basis of language. Due further to the (2005)
some parametric options no longer needed to be a part of the human genetic (biological)
endowment. Some, or possibly all, could be due to these ‘third factor’ constraints on the
development of what Chomsky sometimes calls “the language organ.” Maximizing third factor
constraints and the options they provide in an account of language growth in the child minimizes
the contribution that biology must make to the development of a specific I-language. In doing
so, it very considerably reduces the ‘load’ on an account of language’s evolution. By including
physical, chemical, computational, and other constraints among the contributions to the
Carroll (2005), among others), the evolution of language could now be seen as the result of a
single transmissible mutation in a single human being. The evolution of language would not,
then, need to build a complex developmental instruction set, nor require many millennia and
language, but there is the fact of what Jared Diamond called the “great leap forward” in human
cognitive capacities some 60 thousand years or so ago; the introduction of language at a single
step makes sense of that. Note that 3rd factor considerations add to the notion of the genetically
innate (UG) another form of innate contribution: the “non-genomically innate,” as Christopher
Cherniak (2005) puts it). Even though the current version of UG offers the opportunity to
The history of UG as currently understood could end here. According to it, UG (the biologically
innate) might consist solely of what Chomsky calls “Merge” in its mathematically and
empirically allowable forms, yielding recursion (Hauser, Chomsky, Fitch (2002) and follow-up
discussion) and hierarchy. Perhaps, however, UG also includes some parameters that turn out to
be included in the evolved genetic specification of UG. That is a question for empirical study to
settle; it is reasonable now to believe it includes at least Merge. Or rather, the history of UG so
understood could end with this, if we were to assume that at the time(s) in the period between 50
and 100 thousand years ago at which the mutation(s) that led to the introduction of Merge plus
genetically specified parameters, humans had in place at least rudimentary operative concept-
genetic specification of ‘meaning’ (concept) and linguistic ‘sound’-acquisition systems, the study
One could also, however, include in the history of UG an earlier stage. That would include
discussion in the 1960s (Chomsky’s Language and Mind and Eric Lenneberg’s 1964 paper on
acquisition followed by his (1967), plus the early 1970s discussions (a conference on
as then understood to biology. The issue of how UG was conceived at that time was dominated
by the 1965 Aspects picture of UG and of how a child’s mind ‘chooses’ a grammar, given
relevant input. That view of acquisition had precedents in earlier work too (e.g., The Logical
Syntax of Linguistic Theory). It posed a serious barrier to making the study of language into a
natural science and accommodating the theory of language to biology. If UG is what biology
contributes to the growth of language, and if UG is rich and complex, its evolution cannot be
saltational, but must be gradual, requiring many millennia to gradually increase complexity in
the genetic instruction set. If so, accommodation to biology looks hopeless: there is no evidence
of language evolution of this sort (see in this regard Lewontin (1998)). One can, of course tell a
story of the sort told by Pinker and Bloom (1990), but that is not science. And given the current
view of UG, it is unnecessary: the much-simplified version of UG now available allows for a
saltational view of evolution. The Aspects account was inadequate on other grounds too, of
course. While it represented progress over anything that had appeared before, and progress
especially over so-called “empiricist” accounts, its solution to the acquisition and difference
problems was unwieldy and demanded too much ‘choice’. It also lacked theoretical simplicity,
its theoretical vocabulary lacked precision, and it placed a heavy load on computation.
One could also go back to even earlier works of Chomsky’s (Syntactic Structures, Logical
Structure of Linguistic Theory, Morphophonemics of Modern Hebrew, and articles that discuss
the mathematics of linguistic syntax) to bring out the fact that a major contribution to the effort
to describe UG, natural language, specific languages, I-languages, and the character of linguistic
processing. That had been completely missing in pre-Chomsky work, and its lack provides one
of several reasons for stopping the history of UG as a technical term at this point, going no
further. The reason is simple: an appropriate and explicit set of formal terms (one of several
demands made on a natural science) is needed if one is going to make any progress at all in
developing a science. Nothing of the sort was offered by the historical individuals Chomsky
considers in his Cartesian Linguistics, and the efforts of Zellig Harris, Bloomfield, and others in
the 20th century were flawed because of their assumptions about what a natural language is and
how to investigate one. Perhaps they and those who after Chomsky continued to attempt an
empiricist approach to language managed from time to time to come up with insights and
suggestions that when reconceived became useful; Harris’s “transformations” are an example.
But, lacking an understanding of what they had to deal with, they did not and could not do so in a
principled way, guided by the idea that language is a biologically (and not physically, etc) based
‘organ’ inside the human head that develops and operates in accord with its own internal
principles and procedures, so that its study is a study of a natural object, not a socially
constructed entity.
In the interests of completeness, the history of the technical term ‘Universal Grammar’ outlined
above could be supplemented with works from many others, including some of Chomsky’s
critics. It could include post-Chomsky 1950s contributions from Halle, Postal, Ross, McCawley,
Jackendoff, Pinker, and many others working both within and (for contrast, if nothing else)
outside the specific line of development found in Chomsky’s work while remaining within the
One might think CL is a work in intellectual history. First, there is the subtitle: “A Chapter in
the History of Rationalist Thought.” Second, in co-editors Halle and Chomsky’s Preface to the
first (1966) edition, there is the category of work in which Cartesian Linguistics is obviously
approaches to the study of language [in this case, Chomsky’s rationalist approach] in the
appropriate historical and intellectual setting.” Third, there is the remark in Chomsky’s
introduction (1966: 1; 2009: 57)2 that what follows in the text can be seen as a “discussion of the
history of linguistics in the modern period,” followed by the suggestion that the part of the
“modern period” that he aims to discuss has been ignored by “modern linguistics” (meaning that
However, just below (2; 57-8) that remark is an outline of what Chomsky actually attempts in the
volume. It is highly selective, with selection driven by the current (and continuing) “concerns
and problems” of someone like himself, engaged in constructing a science of language as he was
current work that seeks to clarify and develop these ideas. The reader acquainted
with current [1965-66] work in so-called “generative grammar” should have little
will, however determine the general form of this sketch, that is, I will make no
current work. My primary aim is simply to bring to the attention of those involved
in the study of generative grammar and its implications some of the little-known
work which has bearing on their concerns and problems and which often anticipates
2
From now on I omit dates in references to CL. The first number indicates pages in the first
(1966) edition, the number after the semicolon pages in the third (2009).
Further, in some remarks in his 1970s televised and then published discussion with Foucault, he
says of his approach to the texts of Descartes and other historical figures who figure in CL:
philosophy, but from the rather different point of view of someone who has a
stage people may have been groping towards these notions, possibly without even
realizing what they were groping towards. …. One might say that I’m looking at
accurate account of what the thinking of the 17th century was – I don’t mean to
demean that activity, it’s just not mine – but rather from the point of view of … an
art lover who wants to look at the 17th century to find in it things that are of
particular value and that obtain part of their value … because of the perspective
It is clear: for Chomsky, Cartesian Linguistics is not a work in intellectual history. Rather – as
suggested above – it details ways in which various individuals contributed to (and in the case of
strategy for the scientific study of mind and language initiated (Chomsky’s title suggests) at
around the time of Descartes, and – to an extent that Chomsky considers significant – by
Descartes. Its focus is fixed by work current in 1966 and – given now the history in part I – its
more successful successors. And it illuminates both historical figures and works and current
work.
On reflection, no one should expect to extend the history of the technical term “UG” in a science
invented in the 20th century back to the 17th, or hope to find a science invented in the 17th (or
according to some of Chomsky’s critics, well before) that extends to 1966. Physics and
chemistry plausibly have such histories, but not the science of language.
What is Cartesian about Cartesian linguistics? Descartes’s connection to linguistics then and
now might appear remote. He had nothing of interest to say about language itself. He did,
though, have interesting and important things to say about language use and the implications of
these observations for natural science. He also was with Galileo and a few others one of the
originators of the methods of modern formal naturalistic theory construction. And he made some
interesting and relevant observations concerning innateness, including some that appear to
demand what we would now think of as biological (or at least biological, physical, chemical,
computational…) explanation. Perhaps it is these three contributions taken together that led
Chomsky to call his study “Cartesian Linguistics.” All contributions are novel with Descartes in
some identifiable way. Each deserves attention. Each has important implications for the science
of language and mind, not always recognized by Descartes or others at his time – and in one
In his Discourse and in slightly different forms elsewhere Descartes offered what is plausibly the
most articulate and relevant of the very few efforts available at his time to describe natural
science methodology. Some might hold that that honor belongs to Bacon. But Bacon’s dicta
bear primarily on gathering and organizing data or evidence, not on the more central matters of
constructing theories – indicating what they should aim for and accomplish.3 Others might point
out that Descartes left things out, such as Galileo’s emphasis on simplicity – but it appears in
Descartes’s methods in different form. To see that he acknowledged its importance and
naturalistic scientific research, I begin by listing those desiderata. There are seven.
First and second, a successful natural science must be both descriptively and explanatorily
adequate. Third, it should employ appropriate formal symbols, and fourth, it should aim towards
simplicity. The latter is hard to define but readily recognized by practitioners of a science. Fifth,
natural science theories should allow for and seek accommodation with other sciences. In the
case of the science of language, the obvious candidate is biology.4 Sixth, natural science must
aim for objectivity, where this amounts to ensuring as well as possible that law statements track
the structures and events of natural objects and processes as best we can understand them
(through successful theories). And seventh, natural science theories should improve in one or
more the above six desiderata. To do so indicates that the theories are correct. Descartes did not
bother with progress. He should have. For Descartes, natural science amounted to his contact
mechanics, which he believed applied to all ‘extended substance’. He was much too confident
about his theory’s longevity, as Newton’s law of gravitation fifty years later showed. Newton’s
law postulated what was for common sense an arcane force far removed from the
commonsensical notion of action through contact. Those familiar with the science of language
3
For more discussion of this and related matters, see my (forthcoming-b).
4
Current empiricists (and many students) prefer neurology. A look at empiricist efforts indicates
that they assume without defense a picture of neural nets congenial to their apriori assumption of
learning plasticity (with no consideration of evo-devo and the like).
should think in terms of Deep Structures and the more arcane-looking successor denizens of the
Exercising critical scrutiny and some charity, one can find a large majority of these desiderata in
the Discourse’s four rules. Descartes’s much-disputed first rule is not to accept as true anything
unless one is certain of it, where certainty amounts to having the claim “present itself to my mind
so clearly and distinctly that I had not occasion to doubt it” (CSM I: 120). Certainty – especially
Descartes’s psychological-looking version of it – does not belong in natural science. Clarity and
distinctness, however, resonate with some of the seven desiderata and with current practice in the
natural sciences. They resonate with the demand to break things down into simples (what in
Descartes’s time would be called “corpuscles”), to employ precise formalization (for Descartes,
mathematics and geometry), and to seek simplicity in the theories one constructs. A
contemporary version of some of these desiderata appears in Colin McGinn’s (1994) CALM
(“combinatory atomism and lawlike mappings). Notice that insisting on “clarity and
distinctness” distorts matters when applied to objects and events as understood in the
commonsense domain, where interests of humans play central roles in ‘defining’ things. Into
what does one analyze a shirt – sleeves, shoulders, main body, collar, cuffs? What about buttons
and zippers? Individual threads? The strands that make up the threads? The molecules of which
Lycra is composed? The point: do not start; you lose the shirt. Shirts are ‘tools’ of a sort
manufactured for human use, and their parts – whatever they might in any specific case be –
subserve the relevant purpose. Analysis into ‘simples’ misses the point. With the natural
science of language, on the other hand, the lexical item “shirt” might be analyzed into ‘sound’
– depending on one’s view of syntax and morphology – ‘formal’ features too. At each level of
‘grain’, one needs arguments that that degree of simplification is adequate for the purposes of the
theory and its lawlike principles: with semantic features, for example, it is useful to break a
word’s meaning down into more primitive features because this allows for a plausible account of
swift meaning/concept acquisition, and provides a way to account for differences between
different I-languages, even though from the point of view of syntax, a word’s semantic features
might be treated as a unitary whole. As these and many more examples show, natural science
hypotheses introduce concepts alien to and out of reach of commonsense understanding with its
innate concepts.5
The second rule adds little to the first: “divide each of the difficulties I examined into as many
parts as possible as may be required in order to resolve them better.” Again, this applies usefully
in science (with ‘possible’ depending on theoretical purposes), not to commonsense objects and
commonsense actions and events. A walk to the store is not understood as n numbers of steps, or
m number of left knee joint flexions, etc. “John walked slowly on Saturday to the store,” on the
5
After the invention of natural science in the 17th century, no one should be surprised – for
reasons like those above and others – that the entities and concepts of the natural sciences (or at
least, those of the more advanced forms) are remote from common sense and its form of
understanding the world, or what Descartes called “bon sens.” Descartes himself did not fully
realize this implication of his method. He argued that good sense or common sense (sometimes
also identified by him with ‘reason’) must be supplemented with his method (which yields
science) to reach truth; see, for example, “The Search for Truth” in CSM II and its portrayal of
Polyander’s development. That the results of following the method surprise the person of
unaided common sense indicates that they have as a result come to understand in a way that they
had not before.
Descartes seems to have recognized the distinction in another form, however. He
distinguished the concepts (‘ideas’) that appear in science from the innate ‘common notions’ of
common sense. The distinction appears in several places, including his reply to objections to the
sixth meditation, and is implicit in his reply to Regius in Comments on a Certain Broadsheet. In
a letter to Mersenne in CSM III (182-3), he explains with the example of the sun. There is for
common sense the ‘common notion’ we have of the sun (rising and setting, etc…) and for
science the “invented” or “made up” concept the scientist creates in creating a natural science
theory. That example would be significant to his audience.
other hand, might from the point of view of Pietroski’s semantic theory (2005) be understood in
a ‘neo-Davidsonian’ way: event e is a walk, and e is slow, and agent is John, and….
The third is to “direct … thoughts in an orderly manner, by beginning with the simplest and most
easily known and by supposing some order even among objects that nave no natural order of
precedence.” Descartes no doubt associated this rule with his foundationalism; we must ignore
that aspect. The rule is also, however, a way of anticipating the notion of the unity of the
sciences – something Descartes insisted on in the Rules too. If it is read as a demand for some
kind of reductionism in the sciences, it is unwarranted and too strong, given what is actually
between appropriate sciences, however, we find in this rule an early version of the desideratum
of evolution.
And the fourth and last: “throughout to make enumerations so complete, and reviews so
comprehensive, that I could be sure of leaving nothing out (CSM 120). Here there is little need
for charity: this looks like the demand that a theory of descriptively and explanatorily adequate
to its domain. Of course, as mentioned, Descartes thought that natural science had a single
domain – the ‘extended’ – but we can ignore the several problems with that. And we can and
should recognize that one can have natural sciences of the mind that are descriptively and
explanatorily adequate to their specific domains – vision, perhaps, or language – although not
language use.
The four rules do not touch on progress; Descartes foolishly believed that his method quickly
yields truth. (Note the confidence he displays in Discourse V concerning his contact mechanics.)
Nor do they bear explicitly on objectivity understood as ‘built into nature’ – or at least, nature as
we can understand it. That can be assumed, however; it is presupposed in what he has to say
about his contact mechanics and its scope – apparently, everything in the domain of ‘body’ or
extension. Progress and perhaps objectivity aside, then, his rules anticipate our current view of
Descartes’s second unique contribution to the science of language lies in his observations
concerning the creativity of language use and the implications he drew for the natural science of
mind. Appearing in part V of the 1637 Discourse (with no precedents of which I am aware) and
Language use appears to be free of internal or external stimulus control: I can think (a dominant
example of language use) about anything, anywhere, under any description, without regard to
external or internal causal influence. Stimuli of either kind might “incite and incline” me to
think or speak one way or another, without in any way causing it. In effect, an audience or I
might find in some prior thought or external condition a reason for me to think or say such-and-
such. But neither prior nor simultaneous condition constitutes a cause. Those committed to a
causal program (e.g., Fodor’s ‘computational theory of mind’) will claim a cause of some sort,
but that is all it is – a claim, which appears to be inconsistent with these observations and
number of forms. The unbounded forms can be – and in a generative theory are – the
computational products of a finite set of operations on a finite set of words or lexical items.
6
For discussion, see the relevant appendix in Chomsky and McGilvray (forthcoming).
There is no upper limit, even with regard to a specific ‘inciting or inclining’ circumstance.
Someone stumbling over another person’s foot could be described appropriately in many ways,
with no upper bound on the ways that are appropriate. 3) While not being caused and while
taking any number of forms, given a specific circumstance or question posed (or some other
‘reason to speak or think’), what a person says or thinks remains appropriate, or ‘reasonable’.
The creative aspect observations are to be taken as a group. One can program a computer to
modeling a form of stimulus freedom and unboundedness. But there has been no success in
programming a computer that can meet all three conditions at the same time. Notice further that
it is important that these observations bear not on the nature of whatever computational system
generates sentential expressions, but on the use to which such expressions are put by human
beings. Humans can (and do) have computational systems in their heads that could yield endless
this competence. But the use of the expressions of the expressions by humans appears to be
beyond the reach of the sciences of language and mind, at least as we humans understand the
nature of natural science so far. In effect, we can and do have sciences of language and its
development in an infant, but no science of language use. Only an internalist (‘in the head’)
approach to language – without regard to use by people – allows for an adequate (by the
standards above) natural science of human I-languages and the ways that they develop. No
adequate explanatory natural science of language has, or apparently can, deal with the uses to
which a specific person puts what his or her computational system offers. These facts about use
and the scope of a science of language conflict with deeply held views of philosophers,
psychologists, and other cognitive scientists, including many in the 16th and 17th centuries and
still now.
Descartes made another important contribution to the science of language and mind. He was not
the first to offer some form of poverty of the stimulus observations; perhaps Plato was first,
although Plato did not actually state any in detail. Descartes made some progress in that regard
in his several discussions of the concept of a triangle and in the ways he expressed the conditions
for development or acquisition for what he called “adventitious” (‘triggered’ or ‘occasioned’) but
still innate concepts. Further, he was arguably among the first to think about what kinds of
constraints these observations imposed on the study of mind – a job done in much greater detail
soon afterwards by Ralph Cudworth and pursued in more detail with regard to language by
Humboldt. Descartes noted that being innate does not require actualization/activation, but is
compatible with needing some kind of ‘triggering’ event or ‘input’ in order to develop; these are
what he called “adventitious” ideas. Further, given the analogy to diseases that might arise in
some families (CSM I: 303-304), a concept’s development might not only require triggering, but
might require a course of development or growth, one that we (not he) would seek to describe
and explain by use of biology and the other sciences involved in “evo-devo.” In addition, he
held on reasonable grounds that the nature of a concept that develops depends not on the event(s)
that trigger or begin its development, but depend instead on the internal system(s) that fix its
nature. And finally, he noted that not all concepts are innate; some are ‘invented’, or ‘made up’.
The latter include the important concepts invented by the scientist when s/he constructs a theory.
His example (CSM III: 182-3; see also n.5 above) contrasts the innate although adventitious
‘common notion’ of the sun with the concept the scientist invents. The invented concepts can be
taken to be products of his method, which can also be called a procedure for what Chomsky calls
“science formation,” a procedure only underwritten guided by the ‘light of nature’ we all have
available, but only occasionally follow. The concepts that are innate (‘anticipated’ by the mind,
as in Cudworth’s (1996) “proleptic”) limit the domain of common sense. Scientific concepts are
a different matter. Descartes was likely alone in bringing together for the first time these
I am not suggesting that Descartes himself or any of the others portrayed as rationalists in CL
recognized all of the implications of these contributions of Descartes, or that any of them prior to
Chomsky and the introduction of the relevant formalism, the development the science of biology,
and the focused efforts of many working within a particular field actually managed to develop a
natural science of language – or even realize that that was what work within the rationalist
strategy as applied to language in the head could lead to. Notoriously, and yet given what was
available to him, Descartes denied that there could be a natural science of mind or language at
all. And Descartes was certainly no scientist of language; nor he fully recognize the implications
of his contributions. Rather, I am suggesting that these crucial contributions to the natural
science of language as we now understand it, are the reason CL is ‘Cartesian’. They explain why
Chomsky chose to give Descartes the honor of initiating a research strategy and a natural science
methodology that leads to a remarkable degree of success in the study of language and mind.
Space limitations prevent discussing any of the other individuals mentioned in CL and their
contributions towards refining the rationalist strategy for the study of mind and its current forms.
Perhaps the discussion of Descartes’s contributions and of what has become of Chomsky’s
science of language are enough to explain briefly what went wrong with some of Chomsky’s
critics’ efforts.
II.3.
I will be very selective in the individuals and critiques discussed. I do not discuss Hans
Aarsleff’s two attempts to denounce CL or George Lakoff’s early review. I also ignore many
later efforts to criticize CL (they continue to appear) that are derivative or simply misunderstand.
Some reviewers did read CL carefully and attempted to address it – or at least parts of it. The
two I focus on are Robin Lakoff (1969) with her review of a then-new edition of Lancelot and
Arnauld’s Grammaire générale et raisonnée, and Vivian Salmon (1969) reviewing CL itself.
Karl Zimmer (1968) deserves discussion too, but space limitations prevent anything but mention.
I emphasize that I have had the advantage – and as the discussion above should indicate, it is an
advantage – of seeing what has become of the study of language after almost five decades of
exercising a rationalist research strategy for the study of mind that adopts natural science
methodology. I can see what even Chomsky in 1966 could not have seen – how important
certain aspects of Descartes’s contributions proved to be. If Chomsky could not fully appreciate
them in the 60s, Robin Lakoff and Salmon would appreciate them less. And they had an
additional excuse. Chomsky was sometimes careless in what he wrote. In addition to factors
mentioned before such as a subtitle that gives the impression that CL is history, CL includes
remarks like this: “the Port-Royal Grammar, in which a Cartesian approach to language is
developed, for the first time” (33;79). This looks like a precedence and influence claim and is
easy to read historically. Caveats to the effect that is not what Chomsky is doing tend to be
ignored. In any case, they do have an excuse for treating CL historically, and for focusing
almost entirely on the claim to originality and the ‘Cartesian’ nature of the Grammaire. I do not
think that this completely exonerates them, given Chomsky’s claims that he is not trying to trace
distinctly Cartesian. Lakoff argues that Lancelot developed some of the Grammaire’s
‘Cartesian’ insights (specifically concerning ellipsis and deletion) in an earlier work, and
emphasizes that in a late edition of that work, Lancelot attributed his insights in turn to the 16th
century’s Sanctius. If she was correct, the Grammaire was neither original nor Cartesian.
Salmon in her historical tour de force traces the Grammaire’s notions of generality/universality
sometimes remote. Her erudition is impressive. Generally, if they correctly identified what
Chomsky believes makes the Grammaire ‘Cartesian’, the idea that the Grammaire first applied
Both, however, misidentify precisely what Chomsky must have seen in the Grammaire and other
Port-Royal work. He sees in the ‘Cartesian’ (actually, Aspects-like) contributions the initiation
of a natural science. No one doubts that the Grammaire had precedents; Chomsky emphasizes as
much (97n.67;134n.67). But plausibly Lancelot and Arnauld – and perhaps especially Arnauld –
launched a novel project that puts a different cast on – for example – the Grammaire’s view of
the role of something like Aspects’ Deep Structures. In the Grammaire and before, something
like them are introduced as explanations. But what sort of explanations? Arnauld, who honored
Descartes’s method, gave these explanations a then-novel cast. They became natural science
explanations.
Lakoff in her effort to attribute ‘Cartesian’ insights to Sanctius and before focuses on Lancelot
and ignores Arnauld and his contribution. Chomsky in CL, however, notes (35f.;) that some of
the Grammaire’s more important claims concerning how ‘hidden’ structures explain a person’s
understanding of a sentence (where heard forms do not) are expressed in the same way or more
fully in the Logique that Arnauld published with Nicole two years later. Plausibly, it was
Arnauld, not Lancelot, who contributed to the Grammaire the clear explanation of differences in
‘idea’ with its intension and extension (97-8n.70;134-5n.70). Given the fact that Arnauld
corresponded with and appeared friendly with Descartes plus the acknowledged degree to which
Descartes’s early Rules played a role in the Logique, and given further that the Discourse’s four
rules compress the earlier Rules and also yield much of the method of natural science research, it
is plausible that Arnauld’s contribution to the Grammaire served to initiate a natural science of
language. He could have been unaware that he had launched the study of language on a path
did.
What about Salmon? After Descartes, there was what amounted to a test for a natural science of
mind: a candidate must show evidence of adopting a rationalist research strategy and provide
some indication that it can progress towards satisfying the desiderata of natural science research.
Someone who declares that they have a science of language on offer – such as the speculative
grammarian Campanella that Salmon mentions (1969: 172-3) as among those who anticipated
the Grammaire’s insights on universality – might be able to meet this test. Being influenced by
Galileo (if he was) would not be enough; his efforts would have to indicate that he was in some
7
Descartes might have anticipated something of this sort. In the Optics and elsewhere he noted
that ocular convergence yielded visual depth, and pointed out that the blind can approximate
such calculations by using converging sticks. Descartes had available the rudiments of a
computational science of vision. For discussion, see part III of my introduction to CL’s third
edition.
way committed to internalist and nativist assumptions. If his efforts cannot meet the test, the
It is of course obvious that the Grammaire, even if supplemented with the contributions of
Beauzée and Du Marsais, was not by the standards of natural science a great success. Chomsky
points out the failings – among them, being insufficiently explanatory in what we can see as a
natural science way (57-8;96-7). We can see them even more clearly now, because we can also
see the failings of Aspects and its Deep Structures too – what Chomsky had achieved by 1966,
and his object of comparison. But the failings that Chomsky lists in CL – and other failings
mentioned below – do not challenge my primary point. The novelty lay in beginning to conceive
of the science of language as a rationalist form of natural science of mind. Linguistics could not
be Cartesian until after Descartes and some evidence that his methods and observations (and
Another problem with the Grammaire was that while it anticipated a solution, it did not deal
adequately with the creative aspect of language use observations – another of Descartes’s
contributions to the science of language, even if he did not fully realize their implications. But
we know that only in hindsight. We now know that the only plausible way to deal with the
creativity observations is to ensure that language proper (as opposed to its use) is captured by a
theory of the operations of a modular internal system that is not creative as such itself, but that
can yield an indefinitely large number of understandable (though not necessarily usable)
use, only of an isolated generative system that makes creative use by people possible.
I should mention that taking creativity into account has important consequences for the
discussions of language and mind in the 17th century, and before and since. Since it is people
who think and reason (typically using language to do so), there is no natural science of thinking
and reasoning. The right way to look at the matter is to see language as making linguistically
expressed thought possible. The reader can easily draw the implications for the unexamined
assumption – exhibited in the work of the majority from Aristotle to Descartes and beyond – that
the right way to look for universal principles on which to base a theory of language or mind is to
look to reason, thought, and logic. The creativity observations demand a view of language and
Nor did the Grammaire adequately deal with Descartes’s poverty of the stimulus observations
(and related cluster of observations). They too are dealt with properly only by recent work on
evolution and development (evo-devo), in ways only now beginning to be explored. But its
III. Conclusion
It is only recently that linguistics has come to deal fully with Descartes’s contributions to the
study of mind and language. Nevertheless, the Grammaire apparently did start something new
and Cartesian. In depicting the beginnings of a rationalist and naturalistic science of language,