The University of Chicago Press and Wenner-Gren Foundation for Anthropological Research are collaborating
with JSTOR to digitize, preserve and extend access to Current Anthropology.
http://www.jstor.org
A Preliminary
Survey1
Ethnographic
Semantics:byB. N. Colby
DURING
THE LAST FEW YEARS semantics
marks
has figured detailsof meaningand meaningrelationships
prominently
in arguments
betweenanthropologists
who a new phase in descriptiveethnography.The
stressrigorousdescriptive
ethnography
are popularbecausetheyshowpromiseof
and anthropol- techniques
ogists-who
emphasize
selectivity;
thatis,
comparative
studies.Ethnograph- solvingtheproblemofethnographic
ers seek betterways to handle the semanticsof the theymaylead to psychologically
meaningful
elements
cultures
theyare describing
whileethnologists
look for of a culturewhichare analogousto psychologically
a cross-cultural
elements
of a language(e.g.,phonemes).
"language"in whichsemanticsis the meaningful
I shall speak mostlyof the broaderconceptionof unitscannotbe describedin termsof itsparts,but mustbe
structure-that
whichincludesbotha semanticsystem treatedas a whole.
andan organized
worldview.Sucha structure
is usually
Nida's lexical unit is semanticallyexocentric; the
seen to consistof organizedmemorytracesof those meaningof thewhole is not deduciblefromthemeanings
aspectsofpreviousexperience
whichenablemantocope of the parts. Exocentricformsare contrastedto endomoreefficiently
withfutureevents.Part of thesystem centricones,in which the constituentparts do summate
is builtfromwhat one learnsfromthe experience
of to provide the total meaning (Nida 1958: 286). The
others.Witheachnewobservation
and instruction,
the distinctionhas been made by others(e.g., Seguy's prisystem
is revisedand expandedintoa mentalimageof mary and secondary formations,1953), but the exotheworld (Humboldt1836) thatencompasses
all im- centric-endocentric
expression has gained the widest
portantaspectsoftheindividual's
surroundings,
animal, acceptance. Context is sometimescritical in such disvegetable,or mineral.This mentalsystemhas been tinctions,as Nida illustrateswith the example he is in
in different
conceptualized
ways and givendifferentthedoghouse,which can be endocentricin meaningif it
names,no 2 withthesamedefinition.
Amongthemare: applies to one's pet dog but exocentricif it applies to a
cognitivemap (Tolman 1948), cognitivestructure man who is in troublewithhis wife (1964).
(Bruner,
Goodnow,and Austin1956),image(Boulding Exocentric expressionsor lexical units have been
1956; Lynch 1960), Umwelt (Uexkull 1957), eidos labelled as idioms (Hockett 1958: 171) and lexemes
(Bateson 1958), model (Whorf 1956; Robertsand (Conklin 1962; Goodenough 1956; Swadesh 1946).4
Sutton-Smith
1962), mazeway (Wallace 1961c), and Exocentric expressionsof more than 1 morphemeare
infra-structure
(Levi-Strauss
1951). Someof thesecon- called idioms by Weinreich(1963: 145-46), who sugceptsreferto individualthought-structure,
othersto geststhatstatisticson the distributionamong languages
thought-structures
shared(inwholeorinpart)bysever- of the morpheme-to-lexemeratio (index of idiomaal individuals.
ticity)would be theoreticallyuseful.Determinationof
exocentricexpressionsdepends in part on one's conception of polysemy (p. 146) and "dead metaphor",
whichagain is problematicand usually treatedarbitrarThoughtand meaningare notisomorphic,
yettheyare ily (i.e., withoutculturallybased validation) by linguoftentreatedas such.Meaningandwordformarenotin ists.
1-to-1relationship,
yet theyalso are oftentreatedas
such.We expresssurprise
when2 meaningsof a word
impinge
uponourconsciousness
and wonderabouttheir
"continuous"or "discontinuous"
relationship
to each
other.Withoutvalid etymological
information
forunwritten
we nevertheless
languages,
preoccupyourselves
withproblemsof homonymy
and polysemyat theexpenseofmoreanthropologically
important
inquiry(unless we are developinga semantictheory,a task for
linguistics
ratherthanethnology).
A pause to reflect
uponwords,meanings,
andconcepts
maythusbeprofitable,beforediscussing
specificstudiesand approaches.
of inclusion)a linguistic
form,an unboundmorpheme,
CONCEPTUALIZED DESIGNATED REALITY
or a wordform.It has also beencalleda signor sign
vehicle(Morris1955), signifiant(Saussure1915), free The greatestproliferationof termsseems to exist for
form(Bloomfield
1933: 181),andname(Ullmann1957; what is usually treatedas "meaningproper," the con1962). ThoughLounsbury(1956: 190) rightlystates ceptualizeddesignatedreality.5The followinglistis far
thatlinguisticformsare essentially
irrelevant
for se- fromcomplete:semeor sememe(Bloomfield1933; Nida
manticanalysis,theyare usefulfor classifying
and 1951; 'Wonderly1952; and Goodenough 1956), ethnothedatapriorto analysis.
arranging
sememe,macrosememe(Greenberg1954), engram(Ull-
ANTHROPOILOGY
influencedby Russell (1940) and Skinner(1957), is with Wittgenstein,closer to a behavioristicoutlook. Ziff (1960:54) includes the
behavioristicelement among the points of his attack on Quine's
position.
The currentswing away frombehaviorismand back to an interest
in cognitionis undoubtedlyinfluencedby the place computershave
in our society; a good example is Miller, Galanter, and Pribram
(1960). The great interestin generativegrammarsand the associated terminologynow being used loosely in anthropological
writingsmay indicate the same influence among anthropologists.
7 To add to the confusion,philosophers sometimesdepart from
the ordinaryusage of connotationand make it synonymouswith
intensional meaning.
8 A considerationof sense again leads us from the meaning of
specified semantic units to the conceptualizationbehind various
combinationsof semanticunits. Somewherein between it is convenient to posit rules-cognitive procedures for interpretingthe
syntacticand semanticrelationsbetweenmessage elements.The use
of rules in ethnologyis discussed later.
ANTHROPOLOGY
Colby: ETHNOGRAPHIC
Vol. 7
No. 1
February1966
SEMANTICS
12).
The ;rrl
in
in nativetermin- This typeof'analysishad itsAmericanbeginnings
characteristic
may be least frequent
ologies(Frake1962),thoughthissometimes
dependson theworkofMorgan(1871) and Kroeber(1909),where
werediscussed
in termsof
systems
how highone goes in the system.At highercontrast kinshipterminology
or components.
levelsthetermsusuallybecomefewerin number,
but a limitednumberof discriminations
notalways.Sometimes
a higherlevelmaybe moreex- Takingtheterminologies
of 12 NorthAmericantribes,
haustiveand may involvemoretermsthanthe level Kroeberdistinguished
8 components
servingto define
generation;
immediately
below(Nida 1964).12
termsin someorall ofthe12 terminologies:
sexof relative;sexof
Much can be learnedfromthe analysisof various marriage;degreeof collaterality;
and vitalcondition
typesof synecdochic,
in speaker;relativeage in generation;
or part-whole,relationships
folk classifications.It is theoreticallyinterestingof connecting
relative.
analysisofEnglish
(whether
or space
oneis dealingwithobjects,relations,
Later,Sapirmadea componential
or timeunits)to distinguish
classinclusion
or "kindof" totalizers(1930). A "totalizer"is any termexpressing
judgment". . . whosefunctionit is to
from"partof" or "part-whole"relations a quantitative
relationships
thequantifi(Conklin1962: 129).
emphasize
thefactthatinthegivencontext
e.g.,
of as capableof increase,
Contrast-level
studiesshow that languagesdiffer able is notto be thought
moreat eachhigher
of totalizers,
levelofcontrast.
Wordsat thelower all,thewholeflock."Using16 categories
levelsusuallyadhereto the"perceptually
based on the following4
distinguish-Sapir made a classification
able objects,"whilehighercontrast
levelsreflect"con- componentdimensions:general(abstract)-specialized
ceptuallybased classifications"
non-evaluative(pure)(Nida 1964). Inves- (concrete);direct-calculated;
tigators
Sapir derivedthe
usuallyhaveworkedat thelowerlevelsandhave evaluated; and simple-modified.
experinot appreciatedthe ethnological
insightsthatcan be notionoftotalityfrom2 kindsofpsychological
of theway wordscon- ence:
gainedfroman understanding
trastat higherlevels,as suggested
by Hockett(1954: 1) thefeeling
to proceedaftera count,
ofrestor inability
113,118-19).
has beenmadeof a set or seriesor
formalor informal,
Of specialinterest
is theoccurrence
ofthesameterms aggregation
or unof inability
of objects;2) thefeeling
at different
hierarchical
levels.At 1 levelin Englishthe willingness
to breakup an objectintosmallerobjects.
termanimalsubstitutes
forsuchtermsas wolf,sheep, Thesefeelings,
to as
whichmaybe schematically
referred
respectively,
are correladog,cat,etc.,butat a higherlevelanimalcan substitute the'all' and the'whole'feeling
fromexperience
for man, fish,bird,and insect(Nida 1964). Frake, tiveto each other.Theyarisenaturally
studying
diseaseclassification
in theSubanunlanguage, withobjects(1930:7).
discovered
thatthetermnuka,meaningskineruption, This paperand Sapir's"Grading:a studyin semanexistedat morethan1 contrast-level
(1961). An erup- tics" (1951b) are importantlandmarksin semantic
tionof thistypemayhealwithoutcomplication,
butit theorythat deservemore attention.(See Weinreich
may also go on to developinto 1 of 23 moreserious 1963: 128-29 for a recentlinguisticdiscussionof
diseases.Thereforenuka is used forboth a terminal quantifiers,
and Quine(1959) fora logicaltreatment.)
diseasecategoryand a developmental
stageforother
Though linguists(Jakobson1936; Harris 1948;
diseases,as botha generaland a specificterm.What Wonderly1952) and others(Lounsbury
1956: 161-62)
differentiates
thistypeof stagefromotherstages,such analyzedpronounsand affixescomponentially,
the1st
as a stagein plantgrowth,
is thefutureindeterminacy.anthropologists
to publishrigorouscomponential
analThe progression
is stochastic,
withvariousalternative yses of meaning,followingKroeberand Sapir,were
possibilities
at each diseasestage.Such a classificationGoodenough
(1956).Their
(1951; 1956) andLounsbury
doesnotspecifysucceeding
stages;it merelylimitsand papers,whichappearedin thesameissueof Language,
structures.
It is botha description
and prediction.
arenowtheworksmostoftencitedin reference
to comof taxonomiesimportantin ponentialanalysis.
Anothercharacteristic
theanalysisof contrast-levels
was underlined
by Nida
thepurposeof
Beyondbettersemanticspecification,
(1964): termsat thesameleveloftenoverlapat many componential
units(Goodanalysisis tofindconceptual
points;thesamedesignated
realitycan be identified
by enough1956: 196, 198) or ". . . to revealthestructure
2 different
terms.Nida offerstheexampleof shipin of thelogicalcalculuswhichis employedin thegiven
English,whichcan occurwitheitherit or sheas a sub- taxonomyassociatedwiththe terms"(Wallace 1962:
stitute,dependingupon the context.Nida mentions 353). Anotherobjectivementioned
is to
by Lounsbury
othercharacteristics
revealedin chainanalysis.Colors discoverthestructure
behavior:"The
of non-linguistic
and numbers
can be brought
in linearrelation- systemof discriminatory
together
behavioris thenrelinguistic
shipsby thistechniqueto showoverlapping
areasand latedto thesystem
beof discriminatory
non-linguistic
and shifting
indistinct
boundaries(1964: 69). See also havior" (Lounsbury1956: 189). Goodenoughand
Conklin(1962).
Wallace emphasizepsychologicalcorrelatesin componentialanalysiswhile Lounsburyspeaks more of
COMPONENTIAL ANALYSIS
correlates.13
sociological
in componential
analysisis
The mappingof a domainin whichthe conceptual An additionalpossibility
embodiedin a
and hierarchical
segmentation
levelsare indicatedby to go beyondtheconceptualdistinctions
lexical unitsis a preliminary
step for another,more set of lexicalitemsto conceptsthatare not lexically
detailed,analysisin whichtherelevant(i.e., domainof each uniton a givenlevel is 13 Of course there is a trivial sense in which language relates to
related)signification
analyzed into componentsor distinctivefeatures.
See also Simpson's (1961:13-15) distinctionbetween a hierarchyand a key in which the idea of priorityis important.
12
ANTHROPOLOGY
as endocentric
the
to avoidtreating
versus"non-proximate"phor."It is difficult
brokendowninto"proximate"
exointo"known"versus"vague" meaningfora lexicalunitthatis psychologically
and in somecasesfurther
fromthesumof the
(lecture,Harvard University,1962). It is difficultto centric(wherethemeaningdiffers
of thewordparts),butmayonce have been
determinehow much of such classificationis native, meanings
thedead
Hockett(1954: 111) exemplifies
and how muchis imposed endocentric.
learnedby the ethnographer,
part of theanalyst'sthought-structure.metaphorproblemwith the Chinesetermfor train,
by a pre-existing
This is, of course,the most fundamentalethnographic hw6che.Hw6 "fire" and che "cart" were combined
steamlocomotivewas firstinproblem. Conklin spent enough time speaking the whenthe fire-spitting
language in the culture to have learned some of its troducedinto China. But hwocheno longerhas this
bythetermfor
as is demonstrated
meaning,
psychologically significantaspects. But whether the endocentric
in whichdyanltsignifies
representcrypto- electrictrain,dyanlt-hwoche,
aspectsof timeand space measurement
power."
types (Whorf 1956) of the native or of the ethno- "electric
The same processundoubtedly
occurredin Wissegrapher's language must be specified by the ethnographer.Approximationto implicitnative categoriesof mann'sstudy(1958; see Ullmann1962: 52) and is the
of
of Whorf's(1956) demonstration
timeand space,maynotbe necessaryif theethnographer chiefshortcoming
The same
literaltranslation.
relativity
through
can generatesentencesabout time and space which are linguistic
to demunderstandableto the natives,but then the problemis erroroccursin Ullmann's(1962: 122) attempt
onstratethatFrenchis more"abstract"thanEnglish
to determinewhen a native "understands."
If objectswere named exclusivelyaccordingto phys- or Germanin thatsuchwordsas ashtrayand Aschenwordforms
teapotandTeekanne,arecomposite
ical properties,theircomponentialanalysis would not becher,
and the'iere
arenot.
be such a problem.But if objectsare recognizedalso by whileFrenchcendrier
function,analysisbecomesinfinitelymorecomplexand
more culturallyrevealing.If we are tryingto find the
particular attributesimportantin the perception of
coins,we may concentrateon type of metal, condition
of edges, and size. This is not likely to be culturally
revealing.But if we add a fewfunctionalor experimental attributes-thesound of a coin when droppedor the
desiredresponsefroma slot machine-we move into a
area of investigation.
less scrutablebut moreinteresting
Silver dollars signifygood luck to childrenand may
recall Las Vegas for adults. This is still trivial but
approaches what mightbe classed as a culturallyimportantbit of Americana.
of definitionby attributeis much
The indeterminacy
increasedfor entitiesmore abstractthan coins. Studies
of child behaviorsuggestthat size, shape, and color are
not properties by which objects are 1st identified
(Church 1961: 5). Perceptual attributesof color and
formstabilize in the child's identificationalapparatus
at later developmentalstages-at least as a basis for
concept formation(Church 1961: 11). The most imof criterialattributesis undoubtedportantdeterminant
ly the particularpurpose behind the symbolizationsof
a situation.The learningprocessmay involve thebuilding of a seriesof theoriesconcerningtheworld of reference and the words which symbolizeits many aspects.
We categorizeeventsand objects and pick ready-made
words fromthelanguageforthem.The learningcriteria
for such categoriesare probably differentfor each individual. A baby learnsthata rattlemakesa noisewhen
shaken, and that a cat scratcheswhen squeezed. He
learns otherthingsabout rattlesand cats; but does he
use noise and scratchingas criterialattributesfor the
restof his life?Meaningmay be learnedin thesameway
large buildingsare constructed.A scaffoldis necessary
at the beginning,but on completionof the buildingthe
scaffoldis removed and forgotten.How then can we
analyze meaningentirelyon the basis of attributescaffolds? Clearly, learningabout objects for the 1st time
and identifyingthem subsequentlyinvolve different
attributes.Word meaningsdevelop; they are not immutable (Vygotsky1962: 121).
In determiningthe componentsof a conceptualized
designatedreality,anotherproblemthat is particularly
serious for the non-nativespeaker is the "dead meta10
SEMANTIC RULES
1963b)and genTransformation
models(Levi-Strauss
erativegrammars(Chomsky1957; 1961) offernew
as well as in
semantics,
in ethnographic
possibilities
generalsemantictheory.Basic to thesenew developmentsis theidea of rules,in manyrespectssimilarto
treatment
(1953).
Wittgenstein's
is
The mostrecentand complexworkon semantics
basedon semantic
rules:Katz and Fodor(1963) present
constituents,
objecdescribing
a semanticmetatheory
of a workablesemantictheory.
tives,and constraints
Accordingto theauthors,a semantictheoryaccounts,
ornonto thecontext
(eitherlinguistic
withoutrecourse
a senforthespeaker'sabilityto interpret
linguistic),
tenceofhislanguage.A theory
whichaccountsforconof an utterance
on theinterpretation
textualinfluences
of
all thespeaker'sknowledge
wouldhaveto represent
untheworld,a requirement
whichtheauthorsconsider
realistic.
has2 basiccomponents:
Theirtheory
(1) a dictionary
whichsupplieseverypossiblemeaningof a lexicalitem
in anysentence
and (2) projection
ruleswhichselectthe
meaningofeachlexicalitemin eachgramappropriate
of thatsentence
maticalstructure
by takingaccountof
thesemantic
relations
between
morphemes
andtheinteractionbetweenmeaningand syntactic
structure.
Lexicaldefinitions
in thedictionary
have2 parts:(1)
2 typesof componential
(e.g.,
markers:grammatical
malenoun,verb) and semantic(e.g., human-animal,
whichare specificdefifemale)and (2) distinguishers
nitions(e.g.,a distinguisher
for1 meaningof bachelor
is "whohas nevermarried";foranother,
"whohas the
1stor lowestacademicdegree").Thus thevariousmeaningsof each lexical itemare subjectedto a limitedcornponential analysis plus more specific definitions.The
CURRENT
ANTHROPO
LOG
ELICITING PROCEDURES
AND PROGRAMMED SPECIFICATION
VALIDITY
validity is best attained throughsuch pro-
of Descriptive
one can learnsomething
Workingwithinformants,
ceduresas
specification.But once the data
or related have beenprogrammed
theboundaries
of synonyms
and dimensions
recorded and arranged for analysis, other
tech- kinds of validity assume importance.Much of the litframe-and-substitution
wordsby distributional
niques(Nida 1964; Stefflre1963). These techniques, eraturethesedays is concernedwiththe"emic problem"
in use and shouldbe or psychologicalor cultural reality.'6Are the compothoughnot new, are increasing
the "rapportbetweenwords" nents in a componentialanalysis really indigenousto
helpfulfordetermining
conceptsthatare thepeople usingtheword forms,or are theysimplycon(Whorf1956: 67) or fordiscovering
being venientconstructsof the analyst?Are we searchingfor
relatedina waypeculiartotheculture
cognitively
analyzed.
somethingthat really exists,for "God's truth,"or are
is to use deliberate
error we simplyrearrangingthingsat will forpurelypractical
Anotherelicitingtechnique
re- ends?". . . theGod's truthman doesn'tbelievehe'll ever
to stimulus
in reference
objectsto evokecorrective
plies, presumablyon the same contrastlevel as the find God's truth,but he does believe it exists,and that
used lexeme(Frake 1962: 81-82).Landar by tryingand workinghe can gradually approach it
erroneously
foreliciting.
See Kelly asymptotically"(Householder 1952: 261). During the
(1960) has useda matrixsystem
of semanticmap- last decade, emphasis has been on the "God's truth"
discussion
(1955: 59-60) forfurther
of promptingposition,thoughthenew interestin generativegrammars
pingand Quine(1960: 30) fordiscussion
translation(i.e., of a language
and elicitingin "cradical"~
with littleor no contactwith any language already described).
Vol. 7
No. 1
February1966
as a modelforanthropological
analysismay bringa appointing.Frake (1961) brieflyhintsat social contexts
shifttowardthe"hocuspocus"or pragmaticdirection. that relate to skin disease and its terminologybut does
The new emphasisis on informant
responsesto pro- not specifythemin therigorousdetailneededto validate
ductionsor behaviorsthatare generated
by a seriesof the arrangementof his lexical units. More relevantis
rulesorcalculiwhichthemselves
arenottested.
Fenton's description (1940) of the classification of
Burling(1964) hasattackedcomponential
analysisby medicinalplants into a hierarchyof formsimilarto the
showingthevirtually
infinite
numberofwaysa lexical sequentialpatternin prayersto the spiritforces.Both
set can be componentially
divided,if thereis no way are based on the principleof relative stature.Though
of checking
whichcomponents
represent
"God's truth" Fenton omits detail, he indicatesthat such a principle
and whichare constructs
of the analyst.Answering also seems to underlie the Iroquois conceptionof the
Burling'sattack,Hymes (1964) stresses3 meansof treeof life."9
validatingcomponents
in theprocessof approximating
"God's truth":elicitingprocedures,
use of context, CONTEXT
and prediction.
It is perhapssignificant
that,withthe
exceptionof elicitingproceduresin Conklin (1962) In Hjelnmslev'sview, meaningdoes not existapart from
and Frake(1962),noneofthesehasreceivedsubstantial context (1961: 45).2? Many linguists,anthropologists,
treatment
in eithertheclassic1956papersofLounsbury and humanistsshare this emphasis on context rather
thanwords in isolationor in paradigms.The contextual
and Goodenough
or laterworks.
valid- viewpoint emphasizes (1) the influence of adjacent
The primary
meansofestablishing
descriptive
to words, sentences,and paragraphs on the meaning of
ityis simplyinformant
response.
Hymes(1964)refers
the "questiondependent"aspectof semanticanalysis. specificlexical units-textual or linguisticcontextand
in thefieldprovide1 means (2) theactual situationin whichthe speecheventoccurs
Propereliciting
techniques
of alternative
ofeliminating
a largenumber
componen- -non-linguistic or behavioralcontext.2"
contextin whicha heareris situatto arriveat whatis assumed The non-linguistic
tialanalysesin theattempt
ofthenatives.When ed already limitshis expectancy.He becomesmobilized
thecognitive
torepresent
processes
he meansmainlyan af- (Church 1961: 28) or "set" to make certain interpreHymesspeaksof prediction,
informant
firmative
responseto thecorrectnamingof tations rather than others in a continual process of
experience." In a particular sphere of
objectsin theenvironment,
showingthatthemeaning Ccthematizing
activity
he
expects
to encounterconceptsdirectlyrelathasbeenattainedbytheinvestigator.7
is thatthevarious ed to that activitymore than he expectsto encounter
The drawbackto sucha criterion
semantic
principles
and components
appliedby thein- conceptsalien to it. Non-linguisticcontextthusprovides
he
decides
whether
objectis desig- a kind of semanticredundancy(pp. 57-58, 97).
when
an
vestigator
Bateson (1960) outlinesa sequence of contextsin an
natedby a specificlexicalunitmaynotalwaysbe conopen
and possiblyinfiniteseries.One contextmay comsciousto him.Chomsky'sremarks(1962: 528-29) on
theunconscious
contributions
a readermakesto a tra- pletely reverse the meaning of a message normally
whilelearninga foreign
ditionalgrammar
languagecan given in another context. To indicate the messagebe appliedequallyto theinvestigator
whenhe predicts clarifyingfunctionof contextBateson uses the term,
and testswordusagein an alienlanguage.Even if the meta-message.22 Here we are dealing with what Vyinvestigator
is fullyconscious
ofall-thesemantic
criteria gotskyand Paulhan call the sense.23
he usesin testing
wordusage,an affirmative
informant "A word in a contextmeans both moreand less than
doesnotnecessarily
meanthecriteria
response
arethose
19 Principles of classificationinvolving religious beliefs are cerusedby nativespeakers."8
among the more interestingand revealing relationshipsin
Predictionof correctword usageis different
from atainly
culture.An example worthfurtherstudyis the possible relationof futureeventsor data characteristics
in- ship between Hindu dietaryconcernand classificationof birds by
prediction
accessibleto theinvestigator
at thetimeof prediction. the way theyhandle food.
correlates
Lounsbury's
sociological
validatecomponents 20 See also Chao's "principle of total accountability"(1953:379).
discover the extentto which relatively"autonomous" words
ofthislattertype.It is probablythe areToindependent
through
prediction
of contextfor theirmeaningrequiresan ambitious
promiseof thistypeof validationthathas attracted programof experimentalsemantics(see Stefflre1963).
to componential
many anthropologists
analysis.Un- 21 The division of contextinto linguisticand non-linguisticaspects
itis precisely
fortunately,
inthisareathatcomponential follows Morris' division into pragmaticsand semantics-syntactics
(1955). Wittgenstein(1953:5) includes both contextsas essential:
analysisand contrast-level
analysishavebeenmostdis- "I
shall also call the whole, consistingof language and the actions
See Hoijer (1958) and, again, Sapir (1951a).
Predictive validity concerns representativenessor replicability.
If data have been elicited from a single informantand analyzed,
what are the chances that anotherinformantwill produce the same
results? Obviously the answer depends upon the nature of the
data and the inferencesmade. Most linguisticwork is done with
a handful of informants.Other types of work require a much
larger sample before the data "stabilize." We have very little
systematicinformationon the degree of semantic sharing (see
Quine 1960:8, 13, 272).
18 Hymes (1964:117)
refers to this as the sorting problem.
Replication of semantic componentsis the goal of many anthropologists but usually not the concernof those who use a generative
grammaras a model. Katz and Fodor, e.g., are interestedonly in
economical and correct interpretation,with the emphasis that
semanticmarkers(components)or projectionrules be judged not
alone but in combinationin specificapplications.
17
12
ANTHROPO
LOGY
No.
February 1966
13
and traditionof a
The attitudesand values,theexperience
people,inevitably
becomeinvolvedin thefreight
of meaning carriedby a language.In effect,
one does not translate
languages, one translatescultures. Ethnography
may, in
fact,be thoughtof as a formof translation
(Casagrande
1954:335-40).
26
ANTHROPOLOGY
Vol. 7
No. 1
Febrmary1966
AFFECTIVE RELATIONSHIPS
Osgood and his associates(1957) have experimented
of objectsalonga seriesof bipolar
withthedescription
In spite of misguidedextrapolationssometimesfound in studies
of cognitivedevelopmentin children,much can be of interestto
anthropologists.For a short review of writingson child development,see Solley and Murphy (1960: chap. 7); for a recentextensive treatmentsee-Luria and Yudovich (1959).
29
15
dimensions
suchas hard-soft,
are beyondthescopeof thissurvey.Yet
good-bad,fast-slow,
etc., inadequately,
whichis calledthesemantic
which
Factoranalyses theyarepartofthevastcultural"encyclopedia"
differential.
of theresults
havesuggested
thatamongtheunderlying lies behindword usage and whichis presumably
the
factorsare an evaluative,a potency,and an activity basisforanthropological
concernwithsemantics,
word
factor.The evaluativefactor(by far the strongest) formsand theirsignificata
beingvehiclesforthe desuggeststhat the semanticdifferential
is a measure scription
of culturalstructures.
primarily
of theaffective,
or evaluative,meaningthat
objectsand conceptshaveforindividuals.
CONCLUSION
Affectivemeaninghas rarelybeen treatedin the
anthropological
literatureexceptin termsof general In thissurveyI have hardlytouchedupon theintercognitivestructuring.
Barnett(1953) is amongthefew relations
betweenultimate
goalsand method.Whatwe
anthropologists
who have consideredat lengththe will learnfromethnographic
semantics
is treatedonly
degreeof affectin partsof thecognitive
and brieflyand generally
structure
in mostof thestudiescitedwith
howaffectcan sometimes
reinforce
to thepoint the notableexceptionof Nida (1964). The answeris
rigidity
ofan austistic
disregard
ofimportant
neweventsin the important,
though,becauseit has a directbearingon
environment.
Degree of commitment
to the existing theeffort
and complexity
ofoursemantic
analysis:
cognitivestructure
is 1 of Barnett's3 mainfactorsin
without
anyfurther
explanation:
"What
culturalchangeand process(theother2 beingthefre- If I tellsomeone
menowis composite,"
he willhavetheright
quencyand claritywithwhichrepeatedeventsaliento I seebefore
For there
thecognitivestructure
are perceived,and theimport- to ask: "Whatdo you meanby 'composite'?
are
The
all
sorts
of
things
that
that
can
mean!"
question
anceorutility
oftheevents).The degreeofcommitment"Is whatyou see composite?"
makesgoodsenseif it is
dependson the degreeto whichsubstructures
are em- alreadyestablished
what kind of complexity-that
is,
beddedin an overallstructure
and on the degreeof whichparticularuse of the word-is in question...
emo,tional
investment
in substructures.
Quine (1960) (Wittgenstein
1953:22).
and Festinger
in
(1957) have treatedthischaracteristic
Complexity keynotesthe situation in ethnography
is deeplyembeddedin the
detail.If the substructure
overallcognitivestructure,
thena changein the sub- today. We are in a positionsimilarto the linguisticone
structure
is morelikelyto have extensive
repercussionsa numberof decades ago, beforephonemicsbut afterthe
throughout
the entiresystemthan changesin sub- greatproliferationof phoneticsymbols.Linguistsbegan
to realize that the numberof sound distinctionswhich
structures
lyingmoreat theperiphery.
The studyof affectas revealedthroughsemantics could be recordedfora languagewas virtuallyinfinite.
has been primarilyin the hands of literarycritics. The distinctionsselectedin a notationwere thus quite
Hymes(1961a: 337) has calledtheattention
of anthro- arbitrary.But with the advent of phonemics,the numpo,logists
to the writingsof Burke(1957), and Arm- ber of symbolsnecessaryforrecordinga givenlanguage
strong(1959) has used Burke'sschemein analysisof was reducedto the phonemesin that language,ranging
folktalediscourse.Richards(1936; 1953) is an out- anywhere from 13 to 45 (Hockett 1958: 93), and
standingliterarycriticfromwhosewritingsanthro- arbitraryselectionwas no longer a major problem.
pologists
canprofitconsiderably.
Ofparticular
semantic Ethnographystill lacks anythingcomparable to the
is Empson'sbook(1952),w-hich
interest
describes
many phonemicprinciple. As in the early linguisticperiod,
kindsof denotativeand affective
meaningin various new methodsforrecordingthe overwhelmingdetailsin
combinationsand permutations.
He indicateshow the streamof humanbehaviorboth by tape and filmor
singlewords or sentencescan simultaneously
carry by complexnotations(Pittenger,Hockett ,and Danehy
severaldifferent
denotativemeaningsin 4 kinds of 1960) have concentratedattentionon,the question of
"equations."The equationsare combinations
of the which ethnographicfactsare importantand which can
followingmeaningtypes: head, or main meaning; be safelyignored.Unless the ethnographercan gathera
centralmeaning;rootmeaning;primarymeaning;and prodigiousamountof data, usinga teamof ethnological
topical meaning.Idiosyncraticfeelingsare distin- specialists,he must be quite selectivein recordinghis
guishedfrom shared feelings,and "emotions"are data.
Determiningwhat factsshouldbe notedand analyzed
definedas whatis leftwhensenses,implications,
and
is less bothersometo the hypothesis-minded
investigamoodsare eliminatedfroma word.
Someof thesetypesof meanings,
are called "hyper- tor,who entersthe fieldwitha limitednumberof variables already in mind and recordsonly what is useful
semanticized"
by Weinreich
(1963: 118):
in his hypothesis-testing.
It is theethnographer
entering
in the"standard"
Whereas
useoflanguage
thereceiver
of an unknownor inadequatelystudiedarea who feelsthe
a message
mustonlydecodeit,notdecipher
it (crackthe problemof ethnographicselectivitymostacutely.Fearcode),in "hypersemanticized"
thecommon
language
code
thatsuchcataloguingsas theHuman RelationsArea
is modified
ad hoc,and thefavorably
inclinedreceiver ing
of themessage
mustguessthecodemodification
before
he Files are no less arbitrarythan what he would do himself,he falls back on the 1 index specificto the culture
canproperly
decodethemessage.
he is studying,the 1 least likely to omit important
Weinreich
thinksit pointlessto concentrate
on such elementsof theculture-the lexicon.
specialeffects
without1staccounting
forthesemantic While the procedure of the hypothesis-mindedinof languagein morestandarduse. Neverthe- vestigatoradvances anthropologicaltheoryat a faster
workings
ina morepatentformmayalert pace, the materialgatheredis likely to be of less use to
less,suchspecialeffects
us to characteristics
thatare also part of standarduse.
more sophisticatedfuture anthropology.Conversely,
The conceptual and affectiveaspects of cognitive the ethnographerconcernedwith detailed cataloguing
structurewhich relate to semantics,here touchedupon of the culturethroughits lexiconwill contributeless to
16
CURRENT
ANTHROPO
LOGY
Abstract
Vol. 7
No. 1
February1966
17
Comments
By E. PENDLETON BANKS*
ticularly disappointing.
By WALLACE L. CHAFE*
Berkeley,
Calif.,U.S.A. 29 iii 65
ANTHROPO
LOGY
No. 1
February 1966
20
The followingcomments
are intended
not as criticism
of Colby's paper but
as supplement.As Colby is aware,
advances
therehave been significant
in the fieldof ethnographic
semantics
since his paper was drafted(see his
to review
addendum).It is appropriate
someof the resultsof thisworkhere,
howeverbriefly,becausetheylead to
certainreinterpretations
and clarifications of the literaturecovered in
Colby'sexcellentreview.
On thetheoryside,thefieldof ethnographic semantics now contains a
ANTHROPO
LOG
Colby: ETHNOGRAPHIC
was made by Conklin (1962:135) but
SEMANTICS
of membership,
grid(e.g.,minimalness
Conklin 1962, inclusionof speaker,
Berlin1963). The followingcomments
apply regardlessof whetheran etic
grid is in any way involvedin the
settingup of thesemanticdimensions.
of semantic
(Isolationand justification
in the case wherethereis
dimensions
no readily available etic grid is an
extremelyimportantmethodological
in this
problem.Major contributions
area have beenmade by Metzgerand
Williams[1962; 1963a,b, c].
However, the discussionhere is
limitedto the formalrelationsof the
and doesnot
lexemesto thedimensions
problemof
cover the veryimportant
operationsfor obthe ethnographic
[see p. 11].)
tainingtheserelationships
of all
When the featuredefinitions
thelexemesare known,thebasicprobthe cognitive
lem for representing
of the domainis to decide
structure
whether,or to what extent,informantsapply the semanticdimensions
simultaneously as against sequentially.
amongthe comFormalrelationships
can providesugponentialdefinitions
gestive,but not conclusive,evidence
on this point.
andmostelegantformal
The simplest
consonantwitha psychologstructure
application
ical theoryof simultaneous
of dimensions
is theparadigm.In order
to definethe notionof paradigm,we
1st considera set of thingswe may
events."If
call "minimalclassification
we selectany 1 featurefromeach of
and take the
the semanticdimensions
intersection
(conjunction)of all the
we have selected,theresultis
-features
a minimal classificationevent. For
A
example,with2 binarydimensions
and B, thesetof minimalclassification
events is [albi, alb2, a2bi, a2b2] (where
concatenationsymbolizesintersection).
The semanticstructureof a domain
is characterizedby a perfectparadigm
if and only if each componentialdefinitioncorrespondsto a unique minimal
classificationevent,and conversely.An
interestingresult of this definitionis
that in a perfect paradigm, for any
pair of features (ai, a2) on a given
dimension A, there exists a pair of
lexemes in the domain whose com-
Descendant
Nonlineal
Child
Consanguineals
Vol. 7
Son
Grandchild
Ancestor
Parent
In-Law
Grandparent
Daugh- Grand- GrandGrand- Grad- Father- Mother- Son- Daugh- Sister- Brotherter
rdauh- Father Mother father mother in-law in-law in-law
ter- in-law in-law
son
21
FIG. 2.
A perfectparadigm.
contains8 lexemes)
A
B
=
=
[d]
[a,, a2]
[b,, b2]
[c1,
C2]
Componential
definitions:
L1 > d a, bi cl
L2 > d a1 bi c2
L3 > d al b2 cl
L4 > d a1 b2 C2
L5 > d a2 bi cl
L6 > d a2 bi c2
L7 > d a2 b2 cl
L8 > d a2 b2 C2
Box Diagram:
a1
a2
cl
L1
L5
C2
L2
L6
cl
L3
bi
C2
L4
L7
I
L8
~b2
22
key
this structure.
3. A box diagram is impossible for this
structure.
4. Only thebottomrow of nodes is labeled;
hence no taxonomy.
CURRENT
ANTHROPO
LOG
FIG. 3c.
L15]
Lg
a2
a,
Ll
Llo
bi
b2
bi
b2
Li2
L13
L14
L15
Ci
C2
C1
C2
C1
C2
C1
C2
Li
L2
L3
L4
L5
L6
L7
L8
L15]
Dimensions: as in Figure 3b
Componentialdefinitions:
(1) L1, L2, ..., L8 as in Figure 3b
Lg > d a, b, d or d al b2 e or
d a, c2 f or d a2 C2 g *
Llo > d a, b, d or d a, b2 e
Ll > d a2 cl f or d a2 C2 g
L12 > d a, b1 d
L13 > d a, b2 e
L14 > d a2 cl f
L15 > d a2 C2 g
Key:
d
-- head term for
Lg
a,
a2
Ll
Llo
b2
bi
L12
d,
L1
L13
d2
L2
whole
domain
el
L3
e2
L4
c2
cl
L14
fl
L5
L15
f2
g1
g2
L6
L7
L8
Vol. 7
means logical
Colby:
ETHNOGRAPHIC
SEMANTICS
worksin
to the list of distinguished
ethnographiclexicography,viz., the
EnciclopediaBororo,vol. 1, by the
Salesian FathersCesar Albisettiand
AngeloJaymeVenturelli(1962). This
splendid ethnographicdictionaryof
1047 pages is the 1st in a projected
seriesof 4 volumeson theBororo,the
contentsof which are announcedas
follows:vol. 1, Vocabularyand Ethnography;vol. 2, Language,Legends,
and Proper Names; vol. 3, Chants;
vol. 4, Acculturation.
A list of significant
earlyexamples
such as
of ethnographic
lexicography,
given by Colby in his footnote10,
should not omit the Vocabulariode
la lengua Aymara of P. Ludovico
Bertonio(1612) or the Dictionnaire
[sic] and Dictionnaire
caraibe-franqois
ofP. RaymondBreton.
franqois-caraibe
By
LOUISE
E. SWEET*
ANTHROPO
LOG
Colby:
ETHNOGRAPHIC
SEMANTICS
of anthrobackground
not new-it is merelya long-neglected basic linguistic
continuationof the old, tried and true pological data-seekingall of us, inmethodsof Boas and all the founding cludingthose who sufferunder that
fathersof our field. True, they were still pejorativeappellation"linguist,"
followed by a generation of much are goingto haveto do. A linguistically
lesser stature, with a kind of inbred sound theoryof semologyis still not
ofitsteachings
fear of language and its uses, and the attained;theapplication
of a theoryof cul"great thinkers" of that genteration to theformulation
went in for theory, or what they tureis not evenbegunD
thoughtwas theory,withoutbothering
variations.
Rather,it assumestheculturaltradi- with troublesomedata. There are now By FRANCIS LEE UTLEY*
tion and turnsto matterswhichare once again some anthropologistswho
Ohio, U.S.A. 22 iii 65
Columbus,
if ethno- insiston recordingdata, includinglanperipherallyethnographic,
and guage and remarks about other data, Colby's compact and useful report
graphicat all-the intrasomatic
psychologicalrelationsof symbolates, and who don't care if the theorydoesn't takes account of foreign studies:
if you will, to their carriers.This- fitthe data. In thissensethereis a new French,German,Italian, Portuguese,
seemsto be the area of investigation,interest,but it is renewed rather than Polish,Russian.Unlikesomebranches
of linguistics,
semanticscannot long
new.
but it is not veryclearto me yet.
My
There are few anthropologistswith survivein an insularatmosphere.
I had moreoverthoughtthat the
arenotcriticaloftheauthor,
validityof proposedanalogiesbetween sufficientlinguistic sophistication to comments
had make the analogies that Colby says but merelyhealthyresponsesto his
and culturalphenomena
linguistic
and anthropologists "press on toward." freshviewof a richand excitingfield.
challenged,
beenfairlysuccessfully
1. Apropos of The Lexical Unit
certainlythe questionablescientific Lounsburyin the 1962 paper cited has
has been? not fallen "short of the mark"; he has (p. 4), Katz and Fodor's view that
statusof theemicstandpoint
recentlyarguedby a culturalanthro- limited himself to taking a 1st and semanticsis what remainswhen you
subtractthe grammarimpliesthat all
pologistwho still proposesto start necessarystep.
or strucgenerative
whether
It is here that I begin to feel grammars,
withindividual
cultural"observations"
Must
bewildermentat Colby's review. I ad- tural,existonlywithinsentences.
behaviors(Harris1964:133-50).
to publish mire his diligence in reading so much syntax eternallyrule out pronounIn the displayof efforts
and publishand publish,I beginto see of what I would myselffind great dif- antecedentrelationshipstranscending
angels on all us pinheads. Survey ficultyin going throughwithout utter the sentenceboundaries?Are "later,"
articlestendto revealthisclearly.That boredom. But there is no citation of "hence,"and "indeed"too cosmicto
In any event,
is 1 of theirvirtues,at least,and the actual data, of conclusions that have belongto thegrammar?
of privateterminologies,been arrived at. We are told that so- semanticsdoes transcendthe sentence
proliferation
which Colby laments,for what are and-so (and his name is legion) talks withits "context."We may,however,
that the
sometimescommonplacesmay be a about such-and-such in this or that not have fullydemonstrated
rather journal or book. What little indication paragraphis not a genuine"linguistic
symptomof neo-scholasticism
or summarythereis of what is actually unit"ratherthana merewritingconthan new discovery.
said is so sketchyand out of context vention.I have heardHopi tapeswith
contoursas a markof
finalintonation
as to make no sense to me.
By G. L. TRAGER*
The headings that Colby uses are in paragraphs,and there are certainly
Buffalo,N.Y., U.S.A. 25 iii 65
some instancesusefulguidesto thecited paragraph categoriesin English of
Colby writesas if "semantics"(a bad material, but most of them are either which the topic-sentencestructure,
signiwrittenand certainly
termfortechnicaluse) weresomething non-scientificordinary words or are commonly
mayalso be found
new in anthropologicaldiscussions. pseudo-scientifictechnical terms that ficantforsemantics,
Fromtheoccasional"anthropological" are undefined and often undefinable. in oral discourse.
2. Surely much can be done on
remarksof Herodotusdown to the Since I forone don't exactlyknow what
discussion "reality" is, I find it hard to figureout Polysemy and Homonymy (p. 5)
latestmathematico-structural
methods.Generative
hasbeeninvolved. what "perceived reality" may be and with synchronic
semantics
ofkinship,
have been muchconcerned
are made by ob- how it differs from "conceptualized grammars
Our investigations
ofhumanbehavior,including reality."My Taos friends,forinstance, withbothsemanticand syntacticamservation
and can providedepthlevel
and by re- "perceive" time in the Taos language biguities,
speech,by askingquestions,
cordinganswersand other language differentlythan I do in English; but analysesfor their resolution.Ogden
data. In all such activitysemantics- when they speak English,I thinkthey and Richards(1923) provideda techwhich
polysemy
the meaningof what is done as "ex- "perceive" the same thingsbecause they niqueforabstractional
plained"bylanguage,and themeaning "conceptualize" them in the same lan- ought to be more currenttoday; it
of language-is basic. I supposethere guage. And when I "conceptualize" in maynotworktoo well whenopposing
is some use in such special termsas Taos-let's say when I think of or ideologuesare stubbornabout special
but it can convincetheman
semantics"and "ethno- utter the term t'oyna- then I "per- meanings,
"ethnographic
but the uses should ceive" what the Taos does, in this case of goodwill. We shouldnot,however,
logic semantics,"
ofethdimensions
be thoroughly
explainedand theterms a "person," but a person who is an eschewthehistorical
carefullydefined.
Indian (maybe excluding Navahos!),
and philology.The "rigor"
nohistory
I considerColby's "morespecific" and very definitelynot a person like attainedby insistingon an arbitrary
to myself.This argumentationof mine is, "now" may provide a semanticdessemantics
definition
of ethnographic
be tautologicaland, in fact,meaning- I suppose, really directedat the people criptionwhichis quite uninteresting.
less. All aspectsof meaningin a lan- Colby cites and fromwhom he getshis This does not make meaning unguage are "culturallyrevealing."
interesting.
termsand headings.
Semantic Structures
Good ethnographers have always
3. Levi-Strauss's
Colby's review has perhaps 1 really
paid attentionto "minute details" of useful aspect, for me. It shows how (p. 6) ominouslyresemblethe Platonic
meaning. The "new phase in descrip- far from any real resultswe still are, archetypes, whereas Saussure's photive ethnography"that Colby notes is and how much hard work in the very nemes seem based on measurable and
Vol. 7.
25
of men as
features."The exclusiveness
mustnot createmodels time,and the generations
perceivable"distinctive
latterstatementis especiallytrue of whichsoon break down becausethey well as of languagesare transformed,
Jakobson'sbinaries,yet I have seen have no humaneinterest.The deter- especiallyif eithertheoryor data has
studentsflounderwith a set of such minationof meaningdomain"by the been at any time too long in the
epochshould
binariesas thoughthey were as ar- common morphemeshared by the ascendant.No scientific
bitrarilymanipulableas Plato's or variousmeanings"may lead to sterile lack surgesforthin either,and neither
has absolute primacy,though there
or the mechanisticdescription.
Pythagoras'world of numbers,
whichwould may be strategicor relativeprimacy
7. ElicitingProcedures
English suprasegmentals,or symsystematicallydesirableat any moment.Generative
in a skewed- force"theethnographer
metrical
phonemepatterns
to the anthroso physically to learn correctword usage in a grammaris as thrilling
up world. If something
fromthe specifieddomain of the language" pologistas to the linguistbecauseit
testableas phonemessuffers
archetypaltemptation,what is to sound dangerouslyprescriptive.We promisesto ordermassesof data hard
happento conceptsas studiedby Frake delightin Panini'sand Sahagun'sdes- to digestin thelump;justso was, for
criptionsbecause we cannot explore a while,structural
grammar.Generaand Conklin?
of theircultureor its tive grammarhas movedus fromthe
4. The Domain (p. 7) of color is, theconnotations
cloisterto themarketplace
and madea
Yet manyfreevariations.
as Colby says,quite ethnocentric.
colorsaresurelytestablediachronically 8. It is heartening
to see (note 22) firmalliancewithsemanticsand psybutby itspreoccupation
and synchronically
by physicalcorre- the awarenessamong anthropologistscholinguistics;
seemedto
lation betweenphysicalspectraand of their literarycompeers;Kenneth withtheoryit has sometimes
theirimmediately
visibleequivalents: Burke,above all, writesbooksso com- distrustdata and even to prescribe
viewing an old
the whitecliffsof Dover, the blue of pact that a page a day is a seminal rules. Ethnography,
ratios and long-studiedculture, requires
the sky at highnoon on a clear day, experience,
and his scene-agent
the red of iron-bearingrocks, the provide critiquesfor the behavioral theory;viewinga new one,it requires
asksformoretheory
orangeof an orange(thoughwe may sciences and perhaps the physical data. The restudy
have to face evolutionarychange sciencesas well. Though Ortega y thanthe pilot project,yet we always
for
in livingobjects).Thisdomain,apt for Gasset'stheory
of "silences"maybe too need more data for time-depth,
genuine rigorousanalysis,has been mysticalformostof us, the technique gap-fillingand discovery of gaps
scarcelytapped.
of thesilencehas beenof highvalue in obscuredby thetheoryof themoment,
and for the testingof prediction.Ethnoneed not con- thestudyof Chaucer,Shakespeare,
5. The ethnographer
thanethaspects T. S. Eliot,and it maybe of value also graphyis moredata-centered
finehimselfto non-referential
nology,which,becauseit is cross-cul(note23).
knowthat in ethnography
of reality;manysemanticists
patternis denotedthrough 9. Conceptualizationcertainlyde- tural,demandsmanytentative
eventheperceivable
Withina culture,
withor ingsforcomparison.
fromcarbon servesstudyby ethnographers,
multi-levelabstractions,
even
atom throughwood up to chair and withoutbenefitof Whorf.A striking in short,data maybe interesting
furniture.
We need merelysay that exampleis in thenamingof places in when it is what Bloch calls "bullaillustration;
witheasily America (and presumablyother co- macow"or impressionistic
thereis value in beginning
testableequationsbetween)word and lonialareas),wherethe"model"known in the comparisonof cultures,bullaof data defeats
objectin one'sown languageand even as mapsproducedthemanybranches, macow and scattering
greatervalue in usingsuch equations in Long Island,and perhapsevenOregon all interest,since contrastswithout
disciplinebecomefalse and even eththe language of another:tree-arbre, fromWisconsin(see Utley1963).
Vietnamese
10. Colby shows how computers nocentric.
parasol-du.
radish-Rettich,
themorphemeThe "operation"mustnot-beconfused can helpin determining
withfinaltruth,
"God'struth" lexemeratioor indexof idiomaticity,By ROGER W. WESCOTT
whether
more modest.We must in machinetranslation
or something
and polysemy,
Chambersburg,
Pa., U.S.A. 22 iii 65
not abdicatefrommappingcultureat and in theswingback frombehaviora higherContrastLevel (pp. 7-8), ism, now that our faultyperceptors Colby's erudite survey is at once
and disturbing-stimulating
as Hockett has shown us. Yet such are aided by rapidcalculators.Yet we stimulating
ethnolinthat we have becausenothingso interests
levels as Conklindeals with do not mustalways remember
bethetelescopeor, guistsas meaning,and disturbing
always convince;as Quine says (see not seenGod through
eludes
Translation
p. 13-14),thetroublewith despitea well-known
story,watcheda cause nothingso consistently
Whorf'sand Levy-Bruhl'sviews of computorinventGod or produce a them.
savagelogicis thatwe musttrusttheir "Cogito, ergo sum." The present I share with Colby "the exciting
translations.The linguistmay have "mentalism"of computerscholarsis hope thata breakthrough...is near"
But I cannot
techniquesfor getting behind the probablya meretemporarystrategy, in semanticinvestigation.
translations
whichthelogiciandoesnot a loose use of intuitionism.
It may sharehis implicitfaiththat the best
so
have; if so, he has to workhard and comefromthegenerative
grammarians,of analytictools is a bibliography
of meanwho have broughtfreshair into the exhaustivethat"thefortress
ingeniously.
by overturn- ing" will inevitablybe overwhelmed
6. Katz and Fodor with their wholefieldof linguistics
fetishes: by sheermass of namesand titles.I
SemanticRules(p. 10-11)have broken ing a numberof structuralist
Saussureanrigidities, ratherlook to a creativecombination
valuable new ground,thoughsome anti-mentalism,
and
syntax defeats them; my colleague thedisregardof thedata of Otto jes- of imagination,thoughtfulness,
me that the persen,theprimacyof phonology.
But persistencesuch as that shown in
CharlesFillmoreinforms
is a case in the intuitionalapproach does cause Lamb'srecentwork(1964) on semantic
construction
"centailment"
point. Ruling out "contextualin- sometroublewhenwe approachnon- structure.
of an writtenlanguages,and some anthro- However,I am glad to join Colby's
fluenceson the interpretation
utterance" seems a methodological pologistsare therefore
justlywary of quest for "meaningful
elementsof...
to the usual it.
necessity,corresponding
culture... analogousto... phonemes"
phosamplingausterity
of linguistics:
11. The manynewmodelsdescribed -a quest 1st enunciated,I believe,
nemics;paradigms;scantyinformant by Colby remindus thatwe are still by Clyde Kluckhohn(1949). These
26
ANTHR
O PO LOG
Reply
By B. N.
Colby: ETHNOGRAPHIC
SEMANTICS
a difference,
Keesingrefers
to thelimitedcommu- pology.Thereis certainly
nicationwithanthropologists
in other forexample,betweenthedata-lessspesubfields.This is true of some folk culations of Wallace and the less
highlydata-oriented
folk
science cnthusiastswho look down theoretical,
upon otherworkin ethnology
(except sciencestudiesof Conklinand Frake.
that
kinship)as "soft."Thislimitedoutlook Yet theyall sharetheassumption
is anothermanifestation
of the con- cultureis a set of mentalcodes and
servatismin the field. It sometimes rules.
includesa blanketrejectionof crossChafe takes the narrow view of
culturalresearch(morefrequent
in in- semanticswhen he distinguishes
betformaldiscussions
thanin print,how- ween linguistic and extralinguistic
of experience,
ever)whichis ratherunusualconsider- structuring
and he takes
ing that many of us defineanthro- thebroad view whenhe speaksof all
aspectsof languagesas beingculturally
pologyas a comparative
science.
Fischerbringsout the need to em- revealing.
Thisinconsistency
illustrates
phasize the cross-cultural
aspect; and 1 of the mainpointsI triedto make
in theprocessmorestudyof theinter- concerning
thebroadviewattackedby
relationshipsbetween ethnographic. Chomsky,whichis held by mostanethnologic,
and ontogenetic
semantics thropologists,
and the narrowview,
wouldindeedbe important
theattempt
and useful. whichunderlies
to develop
Keesing cautions against exclusive a semantictheorywithoutconsidering
context.My apparent
attentionto lexicallylabelled items. non-linguistic
paradox
Sweet, Wescott,Banks, and Chafe failureto explainthisseeming
forBanks' comments
as
make the samepoint.Few people are is responsible
likely to disagreehere; the question well.
turnsratheron degreeof emphasis. The otherdistinction
whichI probAn overemphasison ethnographic ably did not make clear enoughwas
semanticscan eclipseotheraspectsof thatbetweenmeansand ends in ethethnography.
However,unnamedcate- nographicsemantics.My definition
goriesof thoughtwere consideredin was in termsof ends.I certainlydid
can be
both Lounsbury'sand Goodenough's notmeanthatculturalpatterns
1956papersand theworkincodability discoveredwithoutsemantictheory.
is concernedpreciselywith the inter- Semantictheoryis highlyimportant,
relationbetweenthelexicallylabelled but it is more the goal of linguists
The anthropologist
and philosophers.
and thelexicallyunlabelled.
also is theviewof cul- should be concernedwith using this
Conservative
culture.This
tureas a mentalcode or set of rules. theoryforunderstanding
to me,I links,of course,to the criticismof
Thoughthisviewis attractive
am notsureit is themostfruitful
one. triviality.Unless a classificationof
that
I think1 reasonit has caughton is plantsleads to an infra-structure
otherdomainsof
thatit is mentally
It reduces helpsus understand
assuaging.
to a setof logicalrelation- the culture,it is likelyto be anthroeverything
shipsthatare clear and unambiguous. pologicallytrivial.On theotherhand,
witha greatconcernfor it would be considered less trivial
This,together
in thesearchforstepby step by someone interested primarily in
technique
COLBY
No. 1
February 1966
27
Cited
References
28
CURRENT
ANTHROPOLOGY
Vol. 7 .No.
1 .February 1966
Colby:
ETHNOGRAPHIC
SEMANTICS
HARRIS,
HJELMSLEv,
L.
1961.
English
edition,
ANTHROPOLOGY
Colby:
No.
1 .February
1966
ETHNOGRAPHIC
SEMANTICS
31
[WG*]
Rubricsof
32
competitive
behavior.
The
C U RR ENT
AN TH RO P OL OG Y