Universality and Evolution of Basic Color Terms. Working

Home , Basic Color Terms, Brent Berlin, Color term

R E P O R T R ESUMES ED 019 655 AL 00.1 003 UNIVERSALITY AND EVOLUTION OF BASIC COLORTERMS. WORKING PAPER NUMBER 1. BY- BERLIN, BRENT KAY, PAUL CALIFORNIA UNIV., BERKELEY, LANG.-BEHAV.RES. LAB. EDRS PRICE MF-$0.50 HC-$3.24 79P.

DESCRIPTORS- *LANGUAGE UNIVERSALS, *VISUALDISCRIMINATION, *CULTURAL FACTORS; *COMMUNICATION (THOUGHTTRANSFER), *LANGUAGE RESEARCH, COGNITIVE MEASUREMENT,COGNITIVE TESTS, HYPOTHESIS TESTING, CODIFICATION, ANTHROPOLOGY,LINGUISTIC THEORY, BASIC COLOR TERMS,

THE RESEARCH REPORTED IN THIS WORKINGPAPER "STRONGLY INDICATES" THAT SEMANTIC UNIVERSALS HAVEBEEN DISCOVERED IN THE DOMAIN OF COLOR VOCABULARY.MOREOVER, THESE UNIVERSALS APPEAR TO BE RELATED TO THE HISTORICALDEVELOPMENT OF ALL LANGUAGES IN A WAY THAT CAN PROPERLY BETERMED EVOLUTIONARY. THE RESEARCH WAS CONDUCTED IN A GRADUATESEMINAR GIVEN IN THE DEPARTMENT OF ANTHROPOLOGY, UNIVERSITY OFCALIFORNIA, BERKELEY. STUDENTS AND THE AUTHORS SYSTEMATICALLYCOLLECTED DATA FROM SEVERAL INFORMANTS IN EACHOF SEVENTEEN LANGUAGES FROM A NUMBER OF UNRELATED LANGUAGEFAMILIES. AN ADDITIONAL THREE LANGUAGES WERE INVESTIGATEDIN DETAIL AFTER THE ORIGINAL RESEARCH SEMINAR WAS COMPLETED.THESE MATERIALS FROM TWENTY LANGUAGES WERE SUPPLEMENTED BYCOMPARATIVE DATA FROM THE LITERATURE, BRINGING THE SAMPLEOF LANGUAGES TO SO REPRESENTING A WIDE VARIETY OF MAJOR LINGUISTICSTOCKS. THE SEMINAR WAS DESIGNED AS AN EXPERIMENTALTEST OF THE FOLLOWING, LOOSELY STATED HYPOTHESIS-THEPREVAILING DOCTRINE OF AMERICAN LINGUISTS AND ANTHROPOLOGISTSHAS, IN THIS CENTURY, BEEN THAT OF EXTREME LINGUISTICRELATIVITY. PROPONENTS OF THIS VIEW FREQUENTLY OFFER ASA PARADIGM EXAMPLE THE ALLEGED TOTAL SEMANTIC ARBITRARINESSOF THE LEXICAL CODING OF COLOR. THE AUTHORS FEELTHAT THIS ALLEGATION OF "TOTAL ARBITRARINESS" INTHE WAY LANGUAGES SEGMENT THE COLOR SPACE IS A "GROSSOVERSTATEMENT." THEIR HYPOTHESIS WAS BASED ON INTUITIVE EXPERIENCEIN SEVERAL LANGUAGES OF THREE UNRELATED MAJOR STOCKS.THEIR FEELING WAS THAT COLOR WORDS TRANSLATED RATHERTOO EASILY AMONG VARIOUS PAIRS OF UNRELATED LANGUAGES FORTHE EXTREME LINGUISTIC RELATIVITY THESIS TO BE VALID. THEIR RESULTSSUPPORT THE ABOVE HYPOTHESIS AND CAST DOUBT ON THECOMMONLY HELD BELIEF THAT EACH LANGUAGE SEGMENTS THE THREEDIMENSIONAL COLOR CONTINUUM 'ARBITRARILY AND INDEPENDENTLY.THEY SUGGEST THAT ALTHOUGH DIFFERENT LANGUAGES ENCODE INTHEIR LEXICONS DIFFERENT NUMBERS OF BASIC COLOR CATEGORIES,THERE EXISTS UNIVERSALLY A TOTAL INVENTORY OF 11 BASIC COLORCATEGORIES FROM WHICH THE 11 OR FEWER BASIC COLORTERMS OF ANY GIVEN LANGUAGE ARE ALWAYS DRAWN--THESE CATEGORIESBEING WHITE, BLACK, RED, GREEN, YELLOW, BLUE, BROWN,PINK, ORANGE, AND GREY.(AMM)

1.1 1 U.S. DEPARTMENT OF HEALTH,EDUCATION & WELFARE OFFICE OF EDUCATION

THIS DOCUMENT HAS BEEN REPRODUCEDEXACTLY AS RECEIVED FROM THE POINTS OF VIEW OR OPINIONS PERSON OR ORGANIZATION ORiGINATINGIT. EDUCATION STATED DO NOT NECESSARILYREPRESENT OFFICIAL OFFICE OF POSITION OR POLICY. Universality and Evolution

of Basic Color Terms

"PERMISSION TO REPRODUCE THIS MM. MATERIAL HAS BEEN GRANTED

BY PA 0L.KAY

TO ERIC AND ORGANIZATIONS OPERATING UNDER AGREEMENTS WITH THE U.S. OFFICE OF EDUCATION. FURTHER REPRODUCTION OUTSIDE THE ERIC SYSTEM REQUIRES PERMISSION OF THE MIN OWNER." Brent Berlin and Paul Kay University of California, Berkeley

Working Paper Number 1, Laboratory for Language-BehaviorResearch University of California, Berkeley

AL 0 01003 Contents

0 Introduction 1. The hypothesis and generalfindings 1.1 Procedure 10 2 refining basic color terms 1.3 Mapping basic color terms 11.4 Universality of basic color terms 1.5 Inter-language vs. intra-languagevariability 1. 6 Category foci vs. categoryboundaries 2. Evolution of basic color terms 2.1 Basic color lexicon andtechnological/cultural complexity 2.2 The seven stages in theevolution of basic color terms 2.3 Supporting data and examples 2. 3.1Stage I systems 2. 3. 2 Stage II systems 2. 3. 3Stage III systems 2. 3. 4 Stage IV systems 2. 3. 5Stage V systems 2. 3. 6Stage VI systems 2. 3. 7Stage VII systems 2.4 Internal reconstruction of basiccolor terms 2. 5 Problematical cases 3. Summary and indications forfuture research Notes References cited -2-

0. Introduction

Ethnoscience studies, and studies of color vocabulary in particular, have firmly established the point that to understand the full range of meanings of a word in any language, each new language must be approached in its own terms, without a priori theories of semantic universals. H. C. Conklin (1955) has shown, for example, thatHanuri'oo "color'? words in fact encode a great deal of non-colorimetric information. The essentially methodological point made in such studies has been frequently misinter- preted by anthropologists and linguists as an argument against the existence of semantic universals. The research reported here strongly indicates that semantic universals have been discovered in the domain of color vocabulary. Moreover, these universals appear to be related to the historical development of all languages in a way that can properly be termed evolutionary.

1. The hypothesis and general findings

Research was conducted in a graduate seminar given in the Department of Anthropology, University of California, Berkeley. Students and the authors systematically collected data from several informants in each of seventeen languages from a number of unrelated language families. An additional three languages were investigated in detail after the original research seminar was completed.1 These materials from twenty languages were supplemented by comparative data from the literature bringing our present sample of languages to 80 representing a wide variety of major linguistic stocks. The seminar was designed as an experimental test of the following, loosely stated hypothesis: The prevailing doctrine of American linguists and anthropologists has, in this century, been that of extreme linguistic relativity. Proponents of this view frequently offer as a paradigm example the alleged total semantic arbitrariness of the lexical coding of color.2 We suspect that -3- this allegation of "total arbitrariness'' in the way languagessegment the color space is a gross overstatement. The hypothesis was based on the intuitive experience ofthe authors in several languages of three unrelated major stocks. Curfeeling was that color words translated rather too easily among various pairs ofunrelated languages for the extreme linguistic relativity thesis to be valid.Cur results support the above hypothesis and cast doubt on the commonly held belief thateach language segments the three dimensional color continuum arbitrarilyand independently of each other language. 3 It appears now that, althoughdifferent languages encode in their lexicons different numbers of basic colorcategories, there exists universally a total inventory of eleven basic colorcategories from which the eleven or fewer basic color terms of any given language arealways drawn. The eleven basic color categories are white, black, rezl, green,yellow, blue, brown, pink, orange, and grey. A second and totally unexpected finding is the following: if alanguage encodes some number n < 11 basic color categories, then there arestrict limitations on which n categories it may encode. The list ofU basic categories is partially ordered, producing seven equivalenceclasses4: I. white, black < 2, red < 3. green < 4. yellow < 5.blue < 6. brown < 7. purple, pink, orange, grey 1 If a language codes a category from the mth equivalenceclass, (m LI:1,

2f, ..*7), then it encodes all categories in each equivalenceclass r < m. Thus, excluding the empirically unlikely possibility of a one-termcolor vocabulary, there are just twenty-one possible basic color lexicons.5(Cee r. able 1; we are not of course referring to the phonological or syntacticfeatures of the colo.c lexemes). Moreover, if a color lexicon encodes six or fewer categories, we can predict exactly which onesthey will be.'1 his fact has clear implications for the evolution of color vocabulary in all languages.(In fact, the above ordering gives considerably over-simplified picture of thedetailed evolutionary sequence, which is presentedbelow). Table I: The Twenty-one PossibleBasic Color Lexicons Perceptual categories encoded in thebasic color terms: No. of basic a) 0 0 0 r..I bn color 0 0 a) ,m as 0 .) ro a) ,....,,t o rwl CO terms li) $4 a) 51 $.4 .4 o $.4 XI 04 P. 0 to Type $.4 ttO )4

1 2 + 2 3 + 3 4 + + + 4 5 + + 5 6 + + + OM. + WOO - WWI OMOP 6 7 + + + OMM 6.0 -- 7 8 + + + 8 + + + + + IMIM + 1000 -- NW. ~ + 4MW 9 8 + + el.. ..WO + 10 + + + + + + + IMMO 11 9 + + OM + -- 12 9 + + + + M.. 0.. + 13 9 + + + + Maw 14 9 + + + 4. + + Ma MOWS + 15 9 + + + + M1M + + 16 9 + + + + + + + WOO 17 10 + + + + + + MOO* + 18 10 + + + + + + + 19 10 4 + + + + + + + + 20 10 + + + + + 21 11 + + + + -4-

In sum, our two majorfindings are (i) the referentsfor the basic color terms of all languages appearto be drawn from a setof 11 universal perceptual categories,(ii) these categories becomeencoded the history of a givenlanguage in a(partially) fixed order. There doesaot appear to be anyevidence that differences incomplexity of basic color lexiconbetween languages reflect 6 perceptual differences betweenthe speakers of thoselanguages.

1.1. Procedure. Standarized color stimuli wereused in conducting theresearch. 7 hese 7 %;onsisted of a set of 329 colorchips provided by the MunsellColor Company. The set is comprised of(1) 320 color chips of 40 equallyspaced hues and eight degrees of brightness, all atmaximum saturation, and(ii) nine chips of neutral hue (white, black andgreys).'I he full array of chips wasmounted on stiff card- board and covered with acetate toform the array shown inFig.1.rfith the exception of our addition ofthe neutral hue series,these materials arethe same as thoseused by Lenneberg andPoberts (1953) in their classic cross- cultural study of English-Zunicolor terminology. Curmethod of obtaining the individual mappingsdiffers, however, from theirs, as maybe seenby comparing the following discussionwith their work. Data were gathered in twostages. First the basiccolor words of the language in question wereelicited according to ethnosciencetechniquesusing as little aspossible of any language otherthan the one under studyduring the interview. Secondly, each subject wasinstructed to map boththe focal point and the outer boundary ofeach of his basic color terms onthe array of standard color stimuli describedabove.

1. 2. refining basiccolor terms.

7 here is in everylanguage an indefinitely largenumber of expressions that can denote color.Note, for example, thefollowing vnglish expressions: l'igure HUE The Stimulus Materialsfor Obtaining Used Mappings of Basic Color Terms* 0ru a,I-141 m p4 ari-74 .6.0 0 1. D .1.1) U 1 kJA 1 ti 1,_ t,, Pi g o Pi

1 i 1.1

7 T1 65 4 I I 1 43 11 I 9 , I 1 * Numerals refer to the Munsell system of cotter notation. -5-

(a) crimson, (b) scarlet,(c) blond,(d) (e) bluish,(f) lemon-colored, (g) salmon-colored,(h) the color of the rust on Txaunt's old Chevrolet. Cn the other hand, psychologists,linguists, and anthropologistshave long operated with a concept"basic color term" or "basiccolor word", which ex- cludes forms such as (a)-(h) andincludes forms like black,white, red,rem, and so on. "Basic colorterm" does not have a unique,accepted operationalde- finition. Cur operational procedurefor the determinationof basic color terms is as follows. Ideally, eachbasic color term shouldexhibit the following four characteristics: (i) It is mono-lexemic, i. e.,its meaning is notpredictable from the meaning of its parts.9This criterion eliminatesexamples (e)-(h) andperhaps also (d). (ii) Its signification is not includedin that of any othercolor term.'i his criterion eliminates examples(a) and (b) which are bothkinds of red for speakers of English. (iii) Its applicability must not be restrictedto a narrow classof objects. This criterion eliminates example(c) which may be predicatedonly of hair, furniture and perhaps a few otherthings,. (iv) It must be salient for informants,tending to occur at thebeginning of elicited lists of color terms,stable in its reference,occurring in the ideo- lects of all informants, etc. Thiscriterion eliminates all theexamples (a)-(h), most particularly (h). Criteria (i)-(iv) suffice in the vastmajority of cases todetermine the basic color terms in any language.7he few doubtful cases thatarise are hand- led by the following subsidiarycriteria: (vi) 7 he doubtful form should havethe same distributionalpotential as the previously established basicterms. For example, inEnglish, allowing the suffix -ish,e. g. ,redish, whitish, greenish, etc. ,but not *scp.rletiish, *blue- greenishetc. (vii) Names of objects characteristicallyhaving the color in question are suspect, e. g., gold, silver,ash. This subsidiary criterionwould exclude orange, in English, ifit were a doubtful case onthe basic criteria(i)-(iv). (viii) Recent foreign wan words aresuspect, other thingsbeing equal. (ix) In cases where lexemic status isdifficult to assess,morphological complexity is given some weight as asecondary criterion. The7,ngoish term blue -green might be eliminated bythis criterion.

1. 3. Mapping basic color terms.

No mapping of color terms wasattempted until eachinvestigator had elicited verbally the basic color termsin each language. Mapping wasaccom- plished by use of acetate strips cutto the size of thestimulus board. Eachin- formant was given a black greasepencil (china marker) andasked, for each basic color term, x: (i) Please indicate all those chipswhich you would under anyconditions call x. (ii) Please indicate the best, mosttypical examples of x. Cur querries were designed to getat the total area of abasic category and to determine, as well,its focus or most typicalmember(s).ry he mapping procedure was carried out at leastthree times, at one weekintervals, for each informant. Cften, we had access toonly one informant perlanguage. }owever, in the case of 'zeltal. (Mayan), it was possible toconsult 40 informants. 'The languages we selected wererelatively diverse genetically.lhe%choice of each, however, waslimited by informant availability.All informants were native speakers of theirrespective languages and, withthe exception of thezel- tal individuals, resident inthe San Francisco Bay Area.Cur primary data include basic color categoriesfor the following languages:

1. (Lebanese) Arabic 2. Bulgarian -7

3.Catalan 4.Cantonese (Chinese) 5. Mandarin (Chinese) 6.English 7.Hebrew E. Hungarian 9.Ibibio 10.Indonesian 11.Japanese 12.Korean 13.Porno 14. (Mexican) Spanish 15.Swahili 16.Tagalog 17.Thai 18.Tzeltal 19.Urdu 20.Vietnamese

1.4. Universality of basiccolor terms.

After each language wasmapped, we made acomposite for all languages of the foci of all basiccolor terms. The compositeis given in Fig. 2,where letters indicate the twentylanguages for which we havesystematig.-.,ally collected data. When severalneighboring ch.ps are markedby the same letter,it indicates that severalchips were judged to beequally good representativesof the focus of a category.Fig. 2, although a brute summaryof the cial:a and consequently rather hard toread, nevertheless showsthe considerable extentto which the foci of colorcategories are similar amongtotally unrelated languages.Both (i) the large blank areas(over 707 of the surface of thechart) and (ii) the close Notes to Figure 2

Dotted-lines on chip YELLOW RED 10, brightness 7, indicate overlap of the categories orange and yellow. It is affiliated with yellow for Tzeltal and Cantonese, with orange for Swahili. b) Arrow emanating from chip BLUE 2.5, brightness 3 indicates its affiliation with the category 'green' for Vietnamese. Likewise, chip RED PURPLE 10, bright- nese 3 is affiliated with 'purple'for Korean. c) * indicate chips chosen as category foci for each of the 20 languages. d) Where a letter occurs more than once, several chips were judged to be equally good representatives of the focus of the category. Composite of Foci of Basic in 20 Languages Figure 2 Color Terms r.T4 HUE E-4z 0 z111 z ,c4PI 10 A A 5 10 10 5 10 10 5 10 10 10 9 2.5 57.5 10 Hi /41 fillin,S p 8 x11 KA A iCt x "&USPvE ECT 7f, J a 5 Ct VT If7 E 7 IIs n X X X X x. K T1) 5 6 H Vli tiV nr,e8 )17cE X aX K it It .ktA V liU T S A a I X H lbA K 11 Tz SP1 E 6 Cat 4 aEuCUSA 5T S8 iSA ESr V vIT$2 /V 3 3 11 :11, la Bc HiPI 5 s;Ma A3 I/Pn li rsrHAa' i A H 1...... 4 ...... LE 3r Ag E sr li 21 Key j AECt--CantoneseC--CatalanB--Bulgarian -- - Arabic- English H-JIIbHb-Hebrew --Indonesian- --JapanesHungarian -Ibibio K--KoreanSp-S--SwahiliPM--Mandarin - -Porno- Spanish UTz--TzeltalTh-TV--Vietnamese --Tagalog- -Urdu -Thai -8- clustering into discrete, contiguous areasof the foci of the variouslanguages, attest to the failure of the strictlinguistic relativity hypothesis. This effect can be seen moreclearly in Fig. 3. Asshown in Fig. 2, informants frequently designate morethan one chip as the focusof a color term. Fig. 3 is based on a calculationfor each language of thecenter of gravity of the focus area for each basic color termin each language; eachof the eleven areas, with an associated gloss (e. g., 'white','red') includes foci for anumber of languages equal to thecorresponding numeral (e. g. ,the area 'red' includesthe foci for all 20 languages, the area'orange' includes foci for11 languages, the remaining 9 languages in the samplelacking a term for thiscategory, and soon). The results shown in Fig. 3 supportquite strongly our initialhypothesis: color categorization is not random and thecenters of basic colorterms are very similar in all languages.

1.5 Inter-language vs inter-4.nformantvariability

Further evidence for the cross-languageuniversality of color foci isthe fact that variability in the location ofcolor foci appears to begreater among the speakers of a givenlanguage than between languages.Whether or not the observed result - that inter-informantvariability in a given languageexceeds inter-language variability - wouldattain statistical significanceunder various sampling assumptions, from thefact that intra-languagevariability absolutely exceeds inter-languagevariability, we can confidentlyreject the hypothesis of greater intra-languagevariability in favor of the nullhypothesis of no difference. The question of inter- vsintra-language variability wasassessed as follows. The only language onwhich we have reliable datafor a substantial number of informants is 'I zeltal.Data were collected forforty 'I zeltal informants. Cf these, thirty-onelocated the center ofya:s in the green area and nine in the blue area, a fact discussed indetail below (section 2. 4. 3.).'3 zeltal has five basic color terms: 'white','black', 'red','green' (i. e.,Eli), and 'yellow'. HUE Normalized Foci of Basic Color Terms in 20 Languages Figure 3 01-1 2.5 5 7.5 10 5 10 10 iu iv i 4...1 i. ./ 98 pink 76 Atil Milli11111ppli 5 A IP.1 III r r.nr 19 ' illigEillh., 'rNI e wan 'gree .,, . 1 tpurpLe1 11111 Mr Mill Note: Numerals refer to the Munsell system of color notation 1 gi 1,,, -.9-

From the thirty-one nti-normal informants a random sample of ten was selected. The restriction to zai-normal informants biases the ' Tzeltal sample toward homogeneity and thus in the direction opposite from the point being made here. There are no other languages in the basic sample of twenty containing just five terms. However, three languages Japanese, Korean, and Cantonese can be reconstructed to five-term systems on the basis ofinternal evidence (see section 2.4 for detailed discussion of internal reconstruction). The foci for white, black, red, green, and yellow in these three languages and in rI zeltal were used for the inter-language comparisons. The bias,if any, introduced by using languages with varying numbers of color terms should be toward greater between-language variation, again in the direction opposite from the observed result. The foci for Cantonese, Korean, and Japanese were each obtained from a single informant. In order to introduce a finer co-ordinatesystem for computing inter- focus distances each unit of hue and brightness was sub-divided into four units, yielding 160 units of hue and 36 of brightness. That is, each box in Figure 1 is considered a square of 4-unit side rather than 1-unit side. For each of the forty-five pairs of Tzeltal informants, the Euclidean distance between their foci for each of the five color categories was calculated, resulting in 225 distances (45 informant-pairs x 5 color categories). The over- all mean of these 225 inter-informant (intra-language) differences is 4. 47 units on the fine scale, or roughly one and one half chip widths. The distances between foci were calculated for each of the five categories, for each pair of languages. Two slightly different sets of foci were used for T zeltal in these comparison: (1)the five foci of a single informant selected at random from the sample of ten, and (2) the five mean foci for the ten informants - that is, the points given by the arithmetic means of the hue and brightness co-ordinates of the ten informants for each category.'The different ways of treating Tzeltal did not materially affect the result. The inter-language distances for each of the five foci were then averaged foreach pair of languages to give a single mean distance measure for every languagepair. The results of -10- these computations are summarized in TableII, which shows that every inter- r language distance exceeds the mean inter-informantdistance for the sample. of Tzeltal informants. The above results, we think, are interestinggiven the traditional anthro- pological attitude concerning the nature of human colorcategorization. While it can be argued thatbilingualism in English affects the results to someextent (cf. Ervin, 1961), we find it hard to conceive that Englishcould influence the place- ment of the foci of categories in these diverse languagesin such a total fashion. Moreover, the work completed with forty Tzeltal informantsvarying from pure Tzeltal monolinguals to perfect Tzeltal-Spanish bilingualsindicates that our results are not skewed due to bilingualism. Finally, thefact that inter-individual differences in a given language appear to be as great or greaterthan inter-language differences weakens considerably thepossible objection on the basisof bilingualism of the informants consulted.

1.6. Category foci vs category boundaries

An immediate result of the mapping procedure was thatjudgements of category foci were highly reliable on repeatedtrials with the same informant and also across informants. It was very rare that a categoryfocus was displaced by more than two adjacent chips onrepeated trials. On the other hand category boundaries were not reliable, even for repeated trials withthe same informant. This phenomenon was also reflected in the relative easewith which informants accomplished the focus-designating task in contrast with theboundary-mapping task. Many subjects engaged in long hesitations withregard to the latter, deman- ded clarification of the instructions, etc. In fact, in marked contrast tothe foci, category boundaries proved to be so unreliable, evenfor a given informant, that they have been accorded a very secondary placein the analysis. Conse- quently, whenever we speak of color categories,either above or below, we refeE to the foci of cate gories, rather than to theirboundaries or total area (volume), except when specifically stating otherwise. Two alternative interpretations of thisresult suggest themselves. First, it may well be that the primary storageprocedure in the brain for thephysical reference of color categories (1. e. ,their meaning) is concernedwith points (or very small volumes) of thecolor solid rather than extendedvolumes. Some sort of secondary processes, of considerably lowersalience and intersubjectiveho- mogeneity, would then have to account for theextensions of reference to points in the color solid not equivalent to (orincluded in) foci. Current formaltheo- ries of lexical definition do not appear able todeal with such phenomena na- turally. If empirical results of this kindaccumulate, simple Boolean function theories of lexical definition may have to berevised in favor of more powerful formalisms.10We do not have space here to pursue thematter further, especially since there appears some reason to suspectthat color (and perhaps a few other semantic domains such as smell andnoise) may have unusual lexical properties. The alternative explanation for the superiorreliability of category foci to category boundaries is that this is anartifact of our experimentalprocedure. In retrospect, we find nothing in ourprocedure which might plausibly be argued to produce such a bias, although othersperhaps can. Moreover, the fact that the evolutionary scheme, including thedata from the 60 additionallanguages ordered by it, works so well in terms of foci seemsto argue against interpret- ing the apparent reality of foci as an artifact.

2. Evolution of basic color terms

Cur second major result is that there appears tobe a fixed sequence of evolutionary stages through which a language must pass asits basic color vocabulary becomes enriched over time. This conclusion is based in part on the findings ofuniversality for the eleven basic category foci, in part on thenon-randomness of their distribution across contemporary languages(and certain logical consequences of the Table II Comparison of Meandistances in the Location of Five ColorFoci Among Four Languages and Among TenSpeakers of OneLanguage

4.47 I Mean inter-focusdifference for ten Tzeltal(th-norma0 informants

II Mean inter-focusdifference for all pairsof four Languages 1. Japanese - Cantonese 4.43 2. Japanese - Korean 4.30 4.16 3. Cantonese - Korean 4. Japanese - Tzeltal 3.84 (0 one Tzeltal informant [(ii) mean focus for tenTzeltal informants 3.00] 5, Cantonese - Tzeltal 2.74 (0 one Tzeltal informant [(i0 mean focus for tenTzeltal informants 3.72] 6. Korean - Tzeltal 2.30 (0 one T zeltal informant [(i0 mean focus for tenTzeltal informants 3.16] -12- particular distribution found), and inpart on additional dataand arguments to be introduced below. An important methodological consequenceof the universalityfinding is that it has allowed us to expand ourdata base from the twentylanguages treated experimentally to a larger number,reported with varyingdegrees of precision, in the general literature. Cnce thebasic universal categoryfoci are established, meaningful comparison can often be madeof literary accounts ofvarious color nomenclatures.11It may well be that someof the motivation for thetraditional relativistic position has derived from aconfusion of noncomparabilityof des- criptions of systems with randomvariation of structure amongsystems. As shown in 1, the basic colorcategories are partiallyordered, in seven equivalenceclasses, such that if a languageencodes a categoryfrom a given class m, it must encode categoriesfrom each prior class r(r < m). This empirical generalization holds, notonly for the original twentylanguages investigated, but for all 80 languages in oursample (with the minorexceptions discussed in section 2.5.). There seems nogood reason to supposethat this generalization, which applies so clearlyin the present, shouldnot apply also in the past. At least we know of no resultfrom historical linguistics--or anyother discipline -- which would impel such anotherwise unmotivatedcomplication of assumptions. Accepting then, that 1holds equally for existinglanguages at prior times in their individualhistories, it follows that for alanguage to gain or lose color terms it must do so in justthe order specified by 1.Although it is logical- ly as possible for languages tolose basic color terms asto gain them overtime, this appears never, or at most veryrarely, to happen empirically.In our con.. sideration of 80 languages with afair amount of assesxnent ofcomparative and internal historical evidence, wehave so far found no indicationof loss of a basic color term.12Hence, the seven equivalence classes 1 may be interpreted as seven evolutAanary. .Eltlx2sof con.jk. 1,,...,e)sity f basiccolor lexicon which have the properties(1) that a given language L, at a givenpoint of time, can be assigned to one and only onestage and (ii) that i.f L is, at agiven moment, in stage m (m < 11), then L musthave previouslypassed through stages1,2, m - 1 in thatorder.13

2.1 Basic color lexicon andtechnological /culturalcomplexity

In addition to the factthat the stages ofcomplexity of colorvocabulary have a temporal ordering,there appears to be apositive correlationbetween general cultural complexity(and/or level of technologicaldevelopment) and Euro- complexity of color vocabulary.All the languages ofhighly industrialized pean and Asianpeoples are Stage VII,while all representativesof early Stages limited tech- (I,II, and III) are spoken bypeoples of smallpopulation units and nology, located in isolated areas.However, this kind ofcorrelation cannot be established with any greatprecision until conceptssuch as "level oftechnolo- gical development" and"degree of culturalcomplexity" are betterunderstood theoretically and more preciselymeasured than they areat present. Suchin- formation as we have on this score,vague as it maybe, suggests thatthe se- accompanying, and quence ofelaboration of color lexicon is anevolutionary one perhaps a reflex of,increasing technological andcultural advancement. The total vocabularies oflanguages spoken bypeoples possessingrela- tively simple technologies tendto be smaller thanthose of highlycomplex civi- lizations. Moreover, it seemslikely that the earliestlanguage(s) spoken by man hadextremely smallvocabularies, perhaps not manytimes greater than the repertories ofdiscreet verbal symbolsavailable to living apesand mon- 14 keys. Increase in the number ofbasic color terms maythus be seen as part richer of a general increasein vocabulary, a responseto an informationally cultural environmentabout which speakers mustcommunicate effectively. The above argument is notoffered as conclusive but as aplausible speculation regarding thecultural evolutionarymechanisms accounting forthe growth in size of basiccolor lexicon. In any case,the argument isaddressed only to the problem of increasein size of colorvocabulary and makes noattempt to -14-

explain the particular order in which colorfoci universally become lexically encoded. We return very briefly to the lattertopic after presenting the seven stages in the evolution of basic color vocabulary.

2.2. The seven stages in the evolution ofbasic color terms

Stage I in the evolution of lexical colorcategories is represented byjust two terms, (1) black plus most darkhues and (ii) white plus mostlight hues. For convenience we will call these categoriesBLACK and WHITE. Stage I is represented in Fig.4.15 At Stage IL a third category emergeswhich we call RED. RED includes all reds, oranges, most yellows, browns,pinks and purples (including violet). WHITE and BLACK continue to segment themiddle -range hues. Stage II is represented in Fig. 5. At Stage III, the reduction in area of WHITEand BLACK continues.A new category, GREEN,becomes encoded at this time whichincludes roughly English green-yellows, greens, blue-greens,blues and purple-blues.WHITE and BLACK become more restricted at thisstage to hues of high and lowbright- ness, I. e., toward the topand bottom of the color chart.

There appears to be some variability atStage III between languages in which the boundary of the categoryGREEN includes blues to the exclusionof yellows, as represented in Fig. 6a, andlanguages in which GREENincludes most yellows, leaving most blues andblue-purples included in BLACK. The minor variant is represented in Fig.6b. Note that the variation at StageIII concerns only theboundaries of categories. The order of emergenceof foci is not affected. Stage IV sees the emergence of YELLOW,which includes roughly En- glish yellows and oranges, Normallythe YELLOW focus occurs in theformerly RED area (Stage III, majorvariant) but occasionally in the'GREEN' area

L11111110.11111100001111111111______sr 11 .. 0,,.; ... - ...... t; l :.:. ' °. :0:...... :::: 4: ' " :t :4**' :::'::: '' . .::*::"". .:: ::":.:.-..::: ..... :*.':. ;4.; :4.4.....%....: ..;;;;4:.; 44 . :.*::-.4:4. :4.1.4:' : 4: *4.4 :. 4.

: ::.*.t'ee .4%1' *4 : .44:4 :::44.4*.-*: *..%::*4 y::*:*...;.4,8.4 4:.' ::.' .:.:.:.:::. ::: .; 4' 4.:: ; *: . 4. .___*.*: :'::.:4:"" 4: .4 BLACK

Figure 4 Typical Stage I Basic Color Lexicon Figure 5 Typical Stage II Basic Color Lexicon Typical Stage III Basic Color Lexicon, .

Figure 6b Typical Stage III Basic Color Lexicon Minor Variant -15-

(minor variant). RED continues to encompassthe areas of English red, some yellow reds, purple and purplereds. Presumably, BLACKand WHITE continue to be deprivedof hue reference at this time,becoming more and more restricted to the neutral values.Stage IV is seen in Fig. 7. At Stage V the focus of blue emerges,normally from the GREEN area but perhaps occasionally alsofrom BLACK .In either case, GREEN nowbe- comes green. At thisstage, BLACK and WHITE arefully reduced to blackand white; i, e., to neutral values. TheRED area is probablyalso reduced with respect to purples and violets.Stage V is depicted in Fig.8.

Stage VI is the last at which asingle focus appears,brown. At Stage VI both RED and YELLOW become even morerestricted in area althoughit is not until Stage VII that they become redand yellow. Stage VI is seenin Fig. 9. When the color lexicon expands beyond seventerms, i. e. ,beyond Stage VI, there is apparently a rapidexpansion in filling out thefull roster of eleven basic color categories. Thisconclusion is suggested by thefact that, for the 80 languages investigated,only 4 color lexicons belongto types other VII, than 1,2,3,4,5, 6 and 21. (See Tables Ill andIV). Apparently, at Stage the remaining basic categoriespink, orange, grey and purple areadded to the lexicon very rapidly ancl, as far as wehave been able toascertain at the mo- in no particular/ ment, /order. Our data now suggestthat purple and pinkprobably arise from RED while orange becomesisolated from YELLOW . Thereis, however, some evidence toindicate that orange may have, in some cases,arisen from hues RED.Grey represents simply the encodingof mid-brightness neutral between black and white. Stage VII systems include all eight-,nine-, ten-, and eleven-term systems and thus include types7-21. As shown in Table III,of the 20 Stage VII systems so far encountered,13 are of type 21,i. e., contain all elevenbasic categories while 11 of the fifteenpossible types of Stage VII arenot represented. (cf.note 5, also section 2« 3.). An eleven termStage VII system is seenin Fig. 10. Figure 7 Typical Stage IV Basic ColorLexicon white

...

::...... : : :;:::::- :;: est's.

RED ..:0

black

Figure 8 Typical Stage V BasicColor Lexicon white

. '.if::::::,ir...f.. .:: :::. : a.. ...1:4 .1 .. .'1::::14 4. .... :; ...,:.:ift% :: green 40.;;:::: ::;:%:.:::::. :;::.i::%-..::::t. °: ...:;.. ..' .%.4 ::; ...,,,: : 0 4....tje,t;;%1::.ii..:.i:.....:t.e"." ,,':: : il . :: : : : /d %. el I ..... : 4 * 41 brown 00 2 ::15;::**. RED II% NI . .. :"...2 6%"*.v: :j..:Iiii:.4 . t`:,.*.i.l.e:t Tr black

Figure 9 Typical Stage VI BasicColor Lexicon

OMNI 111111111111111111i kis . Figure 10 Typical Stage VII BasicColor Lexicon, Eleven Term System*

* The eleventh category, grey,cannot be depicted on the above diagram given theconventions discussed in note15. Table III Distribution of 80 Basic Color Lexicons Among the Twenty -one Theoretically Possible Types, with Indication of Evolutionary Stage.

Type No. of Basic Stage No. of Examples Color Terms

1 2 I 5 2 3 II 21 3 4 III 10 4 5 IV 13 5 6 V 7 6 7 VI 4

7 8 VII PRO 8 8 VII 2

9 8 VII 11111 10 8 VII 1 11 9 VII 12 9 VII Oft 13 9 VII MO 14 9 VII 15 9 VII 1 16 9 VII 17 10 VII a 18 10 VII 19 10 VII 20 10 VII IMP 21 11 VII 13 Total Stage VII20* Grand Total 80 *Catalan, Cantonese, and Vietnamese are Stage VII systemsbut are not typed. See Sectica 2.5.

eilkor -16-

2.3. Supporting data and examples

Our search of the literature forreports on colorterminologies is not complete. To date we have gatheredreasonably reliable information on60 languages in addition tothe twenty languages forwhich we have experimentaldata. The results from all, reliablyreported languages areconsidered here and con- form almost totally to ourproposed evolutionary sequence.First, we give a few examples of each stage, withemphasis on the earlier, moreinteresting stages.

2.3. 1. Stage I systems

Originally, we had no hope ofdiscovering an extantexample of Stage I. We were thus pleasantly surprizedto receive from K. F.Koch (1966) the following report on a New GuineaHighland group called theJa1;, whose language has tentativelybeen affiliated with the Dani(Non-Austronesian) group. In a report made in ourseminar, Koch, who was totally unawareof the theory, stoutly resisted our suggestionsthat Jal; might have morethan two true color terms.Jale is Stage I, having basic color terms onlyfor 'BLACK' and'WHITE' (see Fig. 11). There are otherterms which, in highlyrestricted contexts, refer to certain hues. Theseterms, however, are restrictedalmost exclusively to particular substances orobjects, e. g. , mut 'red soil',plan; 'name of plant whose leaves are used torub yarn, dying yarn a greencolor', etc. Koch reports that when he requested aJai; native to do something with a 'green'object, by using the termpiano, he was consistently misunderstandood.He sub- sequently learned to use the termsig 'BLACK' or11818 'WHITE' depending on the degree of brightness thatthe particular 'green'represented. This was made even moreobvious when he reported thatthe appearance of blood is2t 'BLACK; exactly what 'blood (red)'should be at Stage I due to itslow brightness. That we should find aStage I system in Highland NewGuinea is consonant with the association of simple color lexiconwith simple technological andel:L.:tura' development. Figure 11 Inferred ColorCategories forJale, Representing Stage I -17- Paliyans, a A very interestingvariant of Stage I isrepresented by the technologically marginal groupof Southern India.Our data comes from a re- returned cent, unpublishedmanuscript by P. M. Gardner(1966a) who recently from field-work amongthis group. ThePaliyans speak a dialectof Tamil, a Tamil has a major Dravidian languagewith about 30,000,000speakers. Plains clear Stage V colorterminology, given byGardner as venal'white', karuppu 'black', sivappu 'red', paccai'green', manjal 'yellow'and nilam 'blue'. However, in Paliyan Tamil,cognates of five ofthese six terms are re- tained, but with radicallyaltered meanings.Discrimination isencoded uniquely 16 their glosses and on thedimension of brightness. Paliyan color terms with standard Plains Tamil sources aregiven in Table IV. Gardner indicates thatthere is considerableoverlap in usagebetween each term in the seriesand its neighbor(s).The effect on the usageof the terms surface of the extent to whichbrightness comes from light sourceas against properties is not entirelyclear from Gardner'spreliminary manuscript.The for very most extreme terms,velle and karuppu are notreserved exclusively is velle extreme degrees ofbrightness. Gardner says"The usual leaf on a tree on its uppersurface and ka, j:12L.1 u on itslower surface."(1966a). He also notes, however, that the sameleaf may well benilam or sihappu (presumablyon both sides) if seen in differentlight. In any case, whether ornot Paliyan canbe construed as aperfect exam- plar of Stage I colorlexicon it is certainly avariant of this basictype. For example, "sihappu waselicited for dark shadesof red, yellow, green,purple and black" (1966a). Of particular interestfor the generalevolutionary hypothesisis Gardner's conviction that these peoplehave a minimum ofshared culture. Hespeaks of "imprecision and lack of elaborationin the most basic aspectsof Paliyan sub- sistence relatedclassifications... highyly idiosyncratictaxonomy... de-emphasis on bothverbal communication andformality of expression[ski" (1966a, see 17 also Gardner1966b:397-399). Another example of a StageI system is takenfrom W. H. R. Rivers'treatment of the Westerntribes of the TorresStraits of New Guinea(including the islands of Mabuiag,Muralag, Badu, Moa andSaibai) (1901). Of the sixterms which Rivers elicited as"names in general use", four areclearly descriptive Table IV

Paliyan and Plains Tamil BasicColor Lexicons

Paliyan Plains Tamil vele 'illuminated (sometimes 'bright')'< venal 'white' manja 'bright' < manjal 'yellow' nilarn 'of medium brightness' < nilam 'blue' sihappu 'dark' < sivappu 'red' karuppu'dark or in shadow' < karuppu 'black' -18- expressions formed by the name of somenatural object plus aproductive suffix -dgamulnja most appropriately glossed'it looks like'. Thus, we seekul- kadgamulnga 'red and purple' < kulka 'blood', mum am 'yellowand orange' < mur 'yellowochre',ildem....mulna, 'green andblue'< it 'gall-bladders bile', maludiamulnga 'blue and gr.Deni

2. 3. 2. Stage II systems[introduction of RED]

Stage II is exemplifiedby the Africanlanguages Tiv, Tonga and adialect 18 blues, and some of Shona. P. Bohannan notes that"In Tiv... all green, some grays are ii.But very light bluesand light grays are pupu.Nyian, which covers brown, also covers all warmcolors through red toyellow. The distinctionbe-. tween ii and tamactually is not in termsof color, but in termof what we would white are all call shade--darknessand lightness. Verylight blue, gray, or there is a pupu. 1i meansdark and covers all darkcolors and black -- unless warm colorpresent; brown, red andyellow are all nylan.Tiv can distinguish colors and do color-blindtests, but their culturedoes not require orallow- - that they make someof the color distinctionsthat Westerners make.Westerners are the mostcolor-conscious of peoples"(1963:35-36). A schematicrepresentation of Tiv is seenin Fig. 11. Shona, a Bantu languageof Rhodesia, mayrepresent Stage II,if we interpret the reportin H. Gleason correctly.Apparently, there exists aunitary Figure 12 Inferred Categories for Tiv, Representing Stage II -20-

term including reds and purples. The tworemaining terms not only encompass the mid-spectrum hues, but black and white aswell. Thus, Gleason notes"in- terestingly enough, citema also includes blackand cicena white"(1961:4) The basic color terms of the Tonga groupstudied by Colson consistof three apparently cognate forms with the Shonamaterials reported above:eisia 'black and all other dark colors',Cituba 'white and all other light colors',and Cisubila 'red, including oranges and dark yellows'. Theapparent cognate pairs citemaratuba and cicenaaisia suggest that the glosses forthese forms may be interchanged in Gleason's account. On very speculative evidence, we mayinterpret Kirchoff's report(1883: 546) on certain Australian languages as State II systems,having definite color names for only white,black and red. This fact, however,is tentative in that it is impossible to determine if any ofthe terms are descriptives.Some of Rivers' materials indicate that the term forred,oti,elicited by Kirchoff

1. from several Queensland natives, is reallythe word for blood whichwould make these groups Stage I. Other State II groups for which we haveearly historical data are the Tshi of West Africa: fufu 'white', tuntum'black' and koko 'red', theTodas of India's Nilgiri Hills, the Bantu reportedin Buchner (1883) and aNew Cale- donian group.19 The Nasioi of Bougainville are a clearcontemporary example of Stage11 if the term for red is indeed a basic termand not siLiply a descriptive.Other- wise, they are Stage I. E. Ogan reports:"I worked a total of twenty-six months among Nasioi speakers. Theonly words I heard in regular usewhich might be described as"color terms" were kakara 'white', 'black, dark' and ereren 'red'. The etymology ofthe last word is clear: ererl 'blood'.I know of no such etymology for the other twowords" (1967). Nasioi is a non-Malayo- Polyneslan language. -21-

2. 3. 3. Stage III systems [Introductionof GREEN]

. 20 Apparent Stage III terminologies areHanun 8o, Bassa andIbibio. We use Ibibio as theexamplary case of Stage III,since E. Kaufman gatheredthe data using our experimental method. Ibibio basic color terms glossed'WHITE', 'BLACK','RED' and 'GREEN; and are seen displayed in Fig.13. Hanun6o, also Stage III,is a minor variant of thisstage. The term for 'BLACK' inHanun8o, (ma)biru ranges over black, violet,indigo, blue, dark green, dark gray anddeep shades of other colorsand mixtures; 'WHITE' i9 ranges over white and light tintsof other colors andmixtures; 'RED' (ma)rara' includes maroon, red, orange,yellow, and mixtures inwhich these qualities are seen to predominate;'GREEN' (ma)latuy includes greenand mixtures of green, yellow and lightbrown. Bassa, a member of the Kwa branchof the Niger-Congofamily found in Liberia, is also probably StageIII. According to Gleason(1961:4) Bassa has one term including purples, greens,and blues while anotherterm encom- passes reds, yellows and oranges.As Gleason's chartrefers only to non-neutral hues, we assume that Bassa alsohas terms for neutral'BLACK' and 'WHITE'. The appearance of purple aswell as blue in 'GREEN'show Bassa to be an extreme form of the majorvariant of Stage III.

2. 3. 4. Stage IV systems[Introduction of YELLOW]

Stage IV is represented by manylanguages of the world.For Africa we find this stage attested in adialect of Shona distinct fromthat reported above with terms for 'BLACK' nema,'WHITE' Cena, 'RED' cuku,'GREEN' pfumbu 21 and 'YELLOW'ilara. Ibo and Urhobo, Nigerian languages,probably represented pure examples of Stage IVuntil recent times. In Ibo wefind basic terms for 'BLACK'21i, tWIFTE1 nzu, 'RED' uhui, and 'YELLOW'odo. The aborigi- Figure 13 Ibibio, Representing Stage III -22- nal term for 'GREEN' has apparently beenreplaced and we find merely a descriptive phrase agwokwondu meaning roughly 'it has thecolor of leaves'. In Urhobo we find basic terms for 'BLACK' .)12a1, 'WHITE',fuafu, 'RED' ,-)babare, and 'YELLOW' odo. The term for 'GREEN'has apparently been lost under pressure from English and replaced by theEnglish loans grini 'green' and blu 'blue'. In North America we find Stage IV representedby Eskimo, with terms for 'BLACK' girnitak, alcuctak, 'RED' aupaluktak,'GREEN' tumuli- tuk, and 'YELLOW' quksutak.(Graburn, 1967). In Central America we find Stage IV systemsin many of the aboriginal languages of the area. To our knowledge, all of the Mayanlanguages of Mexico and Guatemala exhibit Stage IV color terminology. Wehave chosen Tzeltal, a Mayan language of Chiapas, Mexi co, as theexemplary case of Stage IV as we have collected rather extensive data from this languageutilizing the experimental methods discussed earlier.22 Tzeltal has five basic color terms which are9i1110 'BLACK', sak 'WHITE', cah 'RED', vi 'GREEN' and lean'YELLOW'. The distribution of these terms may be seen in Fig. 13. The treatment of the category nts 'GREEN' inTzeltal is of particular interest in that some of the data suggest that thislanguage may now be transitional from Stage IV to Stage V. Of the 40 Tzeltal informants for which we havegathered experimental data, 31 indicate that the focal point of ya falls preciselyin the area of the spectrum which corresponds to English green. Ingeneral usage, the maximum extension of m includes greens, blue-greens, bluesand some blue purples. However, when greater specification ofzai is requested, many informants restrict the term almost exclusively to greens and someblue greens. 'Blues' and 'purple blues' are recognized as a distinct area onthe speart,m and are designated by a descriptive phrase, 'ihk' 'ihktikiiiya7:Jal 'blackish green' or simply 9ihki9ihk'tik 'blackish'. In at least one instance, aninformant referred to sak

e . : ::/%;*.i::,*i.:441...:;..41* *!****6 '. * %I" *0:

. 717: ,Ii:.

::.;,/.*:; 1,**:*......

3%1.: 9ihk

Figure 14 Tzeltal, Representing StageIV -23- this area by the Spanish termasul 'blue'. The remaining 9 informantsin our sample of 40have essentially the same maximal extensionof zaLi as the previous31 individuals (i.a., over greens and blues) but thefoci of the category issquarely in the blue area. When greater specificity wasrequested for the greensand blue-greens, descriptive phrases were often utilized, e. g.,saksaktikiyaiial 'whitish green'. How might these data concerningthe foci of 'GREEN' inTzeltal best be interpreted? Cur mostplausible explanation is onewhich suggests thatTzel- tal may be moving from Stage IVto V and the ambiguityof the foci forzsi re- flects this transitional period. Itis apparent to all Tzeltalspeakers thatni; includes two major perceptual centers, greenand blue. In contactwith speakers of Spanish overthe last 400 years, thisfact has most likely been accen- tuated many times. Speakers ofTzeltal respond by reducingthe extension of ai in instances where specificity is requiredeither to greens(for most informants) or to blues (for theminority) and treat the remaining areawith descriptive phrases. Tzeltal maycontinue for many generationto rely on such descriptives to designate what isclearly an incipient colorcategory best glossed as 'blue'. It is ourprediction, however, that asTzeltal speakers become more exposed toSpanish in the schools will eventually berestricted entirely to greens and that asul or someother Spanish term willbe encoded for the perceptual category'Wale', making Tzeltal alegitimate V system.

2.3.5. Stage V systems [Introductionof blue]

Thus far, Stage V colorlexicons have been foundonly in Africa and Southern India, although Mandarin mayalso, on further research, proveto be Stage V as well. We havealready discussed Stage VPlains Tamil (seeabove) which exhibits terms for 'white'vellai,'black' karuppu, 'RED' sivam,'green' 23 zaccai, 1YELLOIATI man al and'blue' nilam. The inferred distributionof these terms is seen in Fig. 15. :v.:...:....%:.: : .. . .. '0:::::!::....:;...... "..I :::.::::;...: ... : .... :.: ...... `.::::=::::.... ::...... u...... %*. 4°".:"%:::::.*::0:...i:.., . . ::.": %:...::.:::;.::". . .. . ' ::!..* :. ...*:.".,:;.f::::: : . *". %:::".::: ::::.%::::* ...... 1.....: .4?:...; ...... ::...::.;:.:.10 c..e.::::v:. ../Sty....9fIi."...... :.. c.::;.:**1:1:1..;..';i: ,:::......

*:e..:::::.:*ifi:.:::'...*:::::::*i:::'.....:;?:3.f.4:.f.:::..:1:*.j;:.:::::i.:'.:.*::....:.el.:.: ...... :...... ;,.. ...:; ...... -.....t. ,.::::.:...::::.:!!?::;??*.*:**::i*:::.. :...... :...... :.:...... :-....:,...:: karuppu 11.10.m.l.vonolaimom

Figure 15 Inferred Categoriesfor PlainsTamil, Representing StageV -24-

In Africa, Stage Vvocabularies are foundfor the Nupe, theMassai, 4 terms of these Bedauye and at least somedialects of Hausa. The basic color African languages arepresented in Table V.

2.3.6. Stage VI systems[Introduction of brown]

sample but Stage VI systems arerather sparselyrepresented in our North America. thus far we have foundexamples in SouthernIndia, Africa, and the state of Wash- The color terminologyof Nez Perce, anIndian language of ington, is depicted inFig.15.25 Stage VI is also seenfor some dialects ofMalayalam of SouthernIndia, mannS. e. g. ,velll 'white, iadupL 'black',Cuw:dpe3 'RED',paCel 'green', 'YELLOW'nil& 'blue', tavita 'brown'(Goodman 1963:9-10). Bari and the We have at least twoexamples of this stag()in Africa, the 'green', Siwi. Bari terms are-kwe 'white', -ruo'black', -tor 'RED',-...9122 aztuf :tom 'YELLOW', -nxi..../Eys'blue' and .-jere'brown'. Siwi terms are 'white', amilal 'black',ozgahh 'RED', owrarr'green', lasfarr'YELLOW', 26 asmawee'blue' and lasmarr'brown'.

2. 3. 7. Stage VIIsystems

Stage VII is representedby 20 of the 80languages in oursample, and languages. It may varying Fimes of thisstage are foundwidely in the world's of terms as des- eventually prove possibleto establish somepartial ordering however, offer a cribed for the preceedingstages. We cannot,at this time, Many more sequence forthe appearance ofbasic color termsafter stage VI. historically, be- languages must beexamined in detail, bothsynchronically and fore we can present a moredefinite report in thisregard. Noting only languagesin our experimentalsample, we maysuggest red, tentatively that Urdu is"early" Stage VII having termsfor black, white, xcimux ..,,f7;; ...... , ...... :::.....*:.:!..:"." ::::...... " ** *: *: :6: :.i:::.:;.::.....-::...... *:.;.:...... *::.:..,:::..:-...::',:*::.":...... :.....:;::.:;::::0...... -*** .EJ4 *.: :::"..*:::.*: ":.*::: "..' :.:::::::.::::.:::::*:...... ! .1 . i ::. ...%:...... :.:*.:* : :* V:: : . , . ...-- -::: 7. ...Kmaqsmaqs :.'.....:....:.:...... ::...... _ .: .... :...7.-: ....: :.".%.":::::..:: : :, :' .

.6 : LIt' IL:6 .1..64 ....!f:::.::.....1.....7.Li.:ex.:::....t . . .. 1 :. I.:4:i.e.: ---: . , ...... *:.i.:.:-...?::: i' % ku.skus .- . -::;:.:.V.::....:* , 9ilp'ilp . ',.'....!%.*:suk ' uys uk ' 7 ..:,;:..c....i...'...f. . ..:*:...... -s-vs-...:..s ** 4 le ..."-tr.,.------;;;;;...... 9 ilp ' ilp , ...... :.;:::...... :://...:.:::.:.::::.:::::::...... ::...... et"...... c : I **:*. :** :.:* :.:::°:%*:.:::::::*....:::::::::.:.:.!::;::::.::.;.... :i -" :*--.:*.-":*.7..:::::.::::.:t' .....:::.:::.:,:;:;.....,:.:...:.:...... ;:":'.""'II' xapcayx mawNwwwmllir

Figure 16 Inferred Categories for NezPerce, Representing Stage VI Table V Basic Color Terms in FourStage V African Languages

Gloss Massai Bedauye Hausa Nupe white eborr era bald biSkim black erok hadk fari zlikb RED enyki adaro ja dzilrii green mbusth stay algashi align 4.- YELLOWngirro asfa rawaya w91.13111, blue ainyori delft shudi dOfa green, yellow, blue, brown,and purple, but lacking orange,pink and grey. Likewise, Cantonese has yet to addbrown, purple or orange toits basic inventory. (cf. section 2.5. ). Tagaloglacks a term for orange asdoes Vietna- mese. Finally, Catalanlacks pink and orange terms. Hungarian presents a special case. Ithas bE sic terms forthe ten basic categories exclusive of red and twobasic terms for red.Should this finding be born out on further research, it maybe possible to suggestadditional developmental stages other than those alreadymentioned.Similarly, Russian, as well as severalother Slavic languages is reportedto have two basicterms for blue. A total summary of the availabledata relevant to theevolutionary hypothesis is given in Table VI. .Allinterpretable reports found aresummarized in this Table which gives for eachlanguage listed, its stage, type,and the source of the data. (The actual categoriesfor each color lexicon type aregiven in Table 1). All languages examinedconfirm the evolutionary hypothesis ineach detail except as noted in section 2.5. English, atypical Stage VII eleven term system has been depicted in Fig.10, 2.4. Internal reconstruction of basic colorterms

The principles of internal linguisticreconstruction outlined byEdward Sapir (1916), and employed byRomney (1967) in his treatmentof Yuman kinship suggest that several of thelanguages in our sample haveonly recently ac- quired basic color terms characteristicof Stage VII. Korean is aninteresting example which illustrates the effectof foreign influence on theformation of new color terminology. Koreanhas basic color terms (boundforms accompanied by a suffix meaningroughly 'color') for 'BLACK','WHITE', 'RED', 'GREEN',and 'YELLCW'. These expressions areclearly indigenous Koreanforms. Terms for pink, orange, (chestnut)brown, brown, green, blue,purple and grey, however, are of obvious Chinesederivation as can be observedin the following forms: Old Korean terms --kamata 'BLACK', haia...ta'WHITE', palgata 'RED' Table VI

Classification of 80 Languages in tP,rmsof Evolutionary Stage of Basic Color Lexicon

Stage I (BLACK, WHITE)

Language Area Source Jale New Guinea K. - F. Koch,(1966) Murray Island New Guinea Rivers (1901) Ngombe Africa Stapleton (1903) Paliyan South Irella Gardner (1966a) Torres Straits New Guinea Rivers (1901)

Stage a (BLACK, WHITE, RED)

Arawak South America van Wijk(1959) Baganda Africa van Wijk(1959) Bantu Africa Rivers (1901) Bulu Africa von Hagen(1914) Ila Africa Smith (1907) Kongo Africa Stapleton (1903) Lingala Africa Anderson (1966) Nasioi Bouganville Ogan (1967) Ndembu Africa Turner (1966) [New Caledonia group] South Pacific Rivers (1901) Ngbandi Africa Lekers (1908) Porno California Corson (1966) Poto Africa Stapleton (1903) [Queensland group] Australia Rivers (1901) Table VI (Page2)

Language Area Source Sango Africa Anderson, (1966) Shona Africa Gleason (1961) Tiv Africa Bohannan (1963) Todas India Rivers (1901) Tonga Africa Colson (1966) T shi Africa Rivers (1901) Yibir Africa Kirk (1905)

Stage III (BLACK, WHITE,RED, GREEN)

Bagirmi Africa Gaden (1909) Bangui Africa Stapleton (1903) Eassa Africa Gleason (1961) Batak Malaya van Wijk(1959) Pullom Africa Nylander (1814) HanunOo Philippines Conklin (1955) Ibibio Africa E. Kaufman(1966) Pout Africa Faidherbe (1882) Somali Africa Kirk (1905) /Xam Africa Bleek (1956)

Stage IV (BLACK, VJHITE,RED, GREEN, YELLOW)

Chinook Jargon Canada T. Kaufman(1967) Eskimo Canada Graburn. (1967) Ewe Africa Migeod (1908) Daza .A Erica Le Coeur (1956) Duhomeen Africa Delafosse (1894) Ibo Africa Goldberg (1966)

-.-...... _- _ Table VI (Page 3)

Language Area Source Papago Southwest U.S. O'Neale and Dolores (1943) Shona Africa Goldberg (1966) Songhai Africa Prost (1956) T zeltal Mexico Berlin (1967) Tzotzil Mexico Collier (1963) Urhobo Africa Goldberg (1966) Wolof Africa Rambaud (1903)

Stage V (black,, white, RED, green, YELLOW,blue)

Bedauye Africa Reinesch (1895) Hausa Africa Robinson (1925) Masai Africa Hinde (1901) Mandarin China McClure (i966) Nupe Africa Banfield (1915) Samal Philippines Geoghegan (1967) Plains Tamil South India Gardner (1986a)

Stage VI (black, white, RED, green, YELLOW,blue, brown)

:Bari Africa Owen (1908) Nez Perce Northwest U. S. Aoki (1967) Malayalam South India Goodman (1963) Siwi Africa Walker (1921)

Stage VII (all eight-,. nine-, ten- and eleven- termlexicons)

A rablz Lebanon Kay (1967) Bulgarian Europe Forman (1967) Table VI (Page 4)

Lanctuake rea Source Catalan Europe Corson (1967) Cantonese China Stross (1967) Dinka Africa Nebel (1948) English U. S. Berlin and Kay (1967) Hebrew Israel McClaren (1967) Hungarian Europe Madarasz (1967) Indonesian S.E.Asia Madarasz(1967) Japanese Japan McClure (1967) Steger (1967) Korean Korea Madarasz(1967) Nandi Africa Hollis (1909) Russian Soviet Union Slobin (1967) Spanish Mexico Stross (1967) Swahili Africa Madarasz(1967) Tagalog Philippines Frake (1967) Thai Thailand Forman (1967) Urdu India McClaren (1967) Vietnamese Vietnam Madarasz (1966) Zuni Southwest U.S. Lenneberg & Roberts (1953) -26- norata 'YELLOW', and parata 'GREEN'. Chineseloans -- punhon-sek 'pink' tyn-sek 'orange', kal-sek 'brown', pam-sek'(chestnut) brown', nok-sek 'green', tiion-sek 'blue', tiia-sek 'purple', and ke-sek'grey'. Cantonese Chinese is also only recently Stage VI,judging by internal reconstruction of its color vocabulary. In dictionariesof 100 years ago, the term for 'pink' sui does not occur. The presentmeaning can be shown to be best translated as 'water colored', The terms lok 'jadecolored' and la:m 'ar- tificial blue' are also recent category labels which now segmentGREEN. We mays therefore, reconstruct aStage IV Cantonese with the followingterms: bak 'WHITE', hek 'BLACK', log 'RED', sel 'GREEN' and mu)'YELLOW'. That we find fui 'grey' is somewhat anomalous. However,there is some evidence that it refers to'ashes', and, if so, can be eliminated. The internal reconstruction for the African CreoleSwahili is relatively interesting in that it may be Stage II, having ancient termsonly for 'BLACK' rieusi, 'WHITE' rmze and 'RED' iiekundu. The term for green1c.ic5mi may be new in that it might be, glossed as'leaf green'. The remaining terms aredes- criptives or loan words, i. e.,kiLtvu 'grey' < 'ashes', ki*Cunwa 'orange' < 'orange fruit', hudhuruni 'brown' < Arabic 'brown',klmandjano 'yellow' < 'turmeric', bulu 'blue' < English 'blue',urudtv. 'purple' < Persian 'purple'. The appearance of new terms in some of the Europeanlanguages is also indicative that on internal evidence these languagesreconstruct to earlier stages. Bulgtorian, for has borrowed terms oranzh

2.5. Problematic cases

While the vast majority of all eighty languagesexamined to date con- -27- form to our notions on the universality of color term foci as well as the evolutionary sequence of basic color terms, thereare four problematic examples for which we require more data. The first problem is presented by Catalan. Catalan is clearly Stage VII but there appears to be some doubt, at least for the informant consulted, that the term for black isa basic, rather than secondary, color term. Corson(1966) reports that his Catalan informant realized that English 'black' was not a 'kind of grey' but consistently maintained that Catalan negre was a kind of giris 'grey'. This is the only example in our data where the status of black as a basic color term is questionable, and more data is clearly needed from additional Catalan speakers. Our data from Cantonese and Mandarin Chinese present several problems that effect to some extent the evolutionary hypothesis. We have, as in- dicated above, treated Mandarin as an example of Stage V, with terms for black, white, RED, green, YELLOW and blue. There is also a term for grey which is the same as the word for 'ashes'. For one informant (McClure 1967) this term is reported as basic; for several other informants (Madaraz 1967) grey is given as a tertiary form. Given these discrepancies, and the pattern pressure from the remaining data, we tx eat Mandarin as tentatively Stage V and plan to obtain more data on this language in future research. The problem with OantorEse is analogous. There exist terms for white, black, RED, green, YELLOW and blue (which would make it Stage V) but as well terms for pink and grey. Thee is no term for brown, this category being included in yellow (Stross 1957). There is reason to require more data for this language, however, in that several of the forms (pink, blue, grey) appear to be very recent and their status as basic terms is as yet unclear. Japanese also presents a problem with respect, not to its current state; but to its internal reconstruction. On internal evidence, the term for Japanese 'blue' ao(i) is of apparent greater antiquity than 'green' midOri(iro). Moreover, there is some evidence that ao(i) had an extension at one point in time over greens and blues. Should this be the case, wewould have a situationwhereby the unitary term GREEN (at Stage IV) has afocal point in 'blue' andeventually reduces to blues exclusively with the later appearanceof the term for 'green' (at Stage V). If these conjectures are born out byfurther work, we wouldhave no alternative but to treat Japanese as acounter-example to the evolutionary sequence of the categories blue and green.However, alternative andequally plausible interpretations can be made whichconform to the theory here presented. Final decision of the matter must awaitfurther research. Finally, Vietnamese must be mentioned inthat it appears to lack a term for 'blue' but has basic terms forblack, white, red, green,yellow, pink, purple, brown, and grey. In sum, of the 80 languages considered(i) there is no counterexample to the finding of universality of the elevenbasic color category fociand (ii) there are just two serious candidates forcounter-examples to the evolutionary ordering- -(a) the absence of brown in Cantonese,and (b) the absence ofblue in Vietnamese.

3. Summary and indications for future research

Our results to date show that (i) thereexists universally a totalinven- tory of eleven basic color categoriesfrom which the eleven or fewerbasic color terms encoded in any given language arealways drawn. (ii) There is a fixed partial ordering of the basic categoryfoci according to whichthey become lexically encoded in a given language over time.This ordering is black, white < red

Figure 17. Stage III System in Terms of a Developmental Paradigm -30-

Indo-European color terms,(iii) studies in a cluding a full-scale study of 26 variety of languages on the orderin which childrenacquire color terms, (iv) careful attempts to correlatestage of colorlexicon developmentwith various indices ofcultural/technological complexity,(v) experimental studies to determine whether the"salience ordering" interpretationof 1 can be sup- ported by non-verbal responsedata,(vi) a more comprehensivecheck of the existing literature for alluseable reports of colorterminologies. -31-

NOTES

1. We thank the members of the research seminar for their active collabora- tion in this study and comments on the present report. They include Chris- topher Corson (Catalan and Pomo), Sylvia Forman (Bulgarian and Thai), Keith Kernan (Samoan, but not included in the report), Paul Madarasz (Hungarian, Indonesian, Mandarin, Korean, Swahili and Vietnamese), F. atherine McClaren (Hebrew and Urdu), Erica McClure (Mandarin and Japanese), Peter Steager (Japanese) and Brian Stross (Cantonese aril Spanish). We also thank Charles C. Frake who is responsible for the Tagalog materials and Elain Kaufman for collecting the Ibibio data. The authors are responsible for data on English, Arabic and 'I zeltal. Vie are grateful to our colleages in the Department of An- thropology, University of California, Berkeley, for their critical remarks on an earlier version of the manuscript. In addition, we have benefitedfrom the useful comments of Wallace L. Chafe, Harold C. Conklin, George L. Cowgill, Christopher Day, A. Richard Diebold, Jr., Pricilla Diebold, Marshall E. Durbin, John L. Fischer, Joseph Greenberg, Dell H. Hymes, Terrence Kauf- man, Richard Lee, Floyd Lounsbury, Nicholas A. Hopkins, DuaneMetzger, Barbara MacRoberts, Eugene Cgan, Patricia Porth, John M. Roberts, David M. Schneider, Thomas A. Sebeok, Dan I. Slobin and the late Morris Swadesh.

2. For example, Verne Ray claims "...there is no such thing as a natural division of the spectrum. Each culture has taken the spectral continuum and has divided it upon a basis which is quite arbitrary" (1952:252). Inperhaps the most influential of standard linguistics texts in the United States, H. A. Gleason notes, "There is a continuous gradation of color from one end of the spectrum to the other. Yet an American describing it will list the hues as red, orange, yellow, green, blue, purple, or something of the kind. There is nothing inherent either in the spectrum or the human perceptioi of it which would compel its division in this way" (1961:4). -32-

3. Throughout this discussion, when we speak of color categories in a given lexicon, we refer to the meanings of native lexemes in terms of the three psychophysical dimensions hue, saturation, and brightness. It has been de- monstrated, e. ,by Conklin (1955) that color lexemes may well include, along with information concerning these particular psychophysicaldrmensions, other sorts of information, e.g., succulence versus dessication. Similarlyin Tzeltal, secondary color lexemes indicate not only features of surface texture but refer as well to features of shape and consistency. Moreover, it has been argued, to our minds quite convincingly, that to appreciate the full cultural significance of "color words" it is necessary to appreciate the full range of meanings, both referential and connotative, andnot restrict oneself arbitrarily to hue, saturation and brightness. We thusmake no claim, in fact we specifically deny, that our treatment ofthe various color terminologies presented here is an ethnographically revealing one. Thisstudy is admittedly etic rather than emic: the data are abstracted and removed from their proper cultural context. However, we will not accept the strictureof- fered by some ethnographers that such an abstraction always and necessarily renders data meaningless. The high degree of pattern found in the datais suf- ficient justification of the process dictating its selection. We thus interpret the pattern found in our results as representing legitimate linguistic and cultural universals. Given the well known variability in the structure which various lexicons impose on their "field properties",it appears that our choice of semantic dimensions for cross-linguistic investigation was a fortunate one.

4. The relation designated '<' is defined as follows. For any distinct a, b .E white, black,...,grey;', a < b just if for any language L, if b is lexically encoded in L, then a is lexically encoded in L. For example, green < brown indicates that any language which has a basic color term for brown will have a basic color term for green. Thus '<'is irreflexive, assymmetric and tran- sitive (these are also the properties of 'less than' among thereal numbers),It -33- will be argued below that '<' canalso be given the additionalinterpretation 'becomes lexically encodedearlier in the history of everylanguage (than)'.

5. That is, there are twenty-onepossible basic color lexiconsgiven (1) the or- deing 1 and (ii) the plausible andempirically valid assumptionthat a one-term color lexicons make no sense.If a color lexicon could encode anynumber of the eleven basic categories(including' zero), there would be a211=2,048posed bilities. If we exclude lexicons of zeroand one term, but do notimpose t h e constrains of ordering1,there are still211-12=2,036 possibilities. There are thus 2015 more lexiconsallowed if the constraints ofordering 1 are not introduced; noneof these 2015 has been foundto occur in our larder sample of 80 languages. (SeeEvolution of basic color termsespecially Table .111 and surrounding discussion.)

6. A concern with theevolution of color terminology maybe traced to the early work of W. E.Gladstone (1858) and Lazarus Geiger(1867). Our attention was called tothese materials after the presenttheory had been developed through a reading of W. H. R.Rivers' Torres Straitsresearches. Rather than discuss indetail the contributions of thesewriters, we quote fromRi- vers, who offers aconcise statement of the ideascurrent at that time.(An ex- cellent summary is also found inSegall, Campbell andHerskovits, 1967). "The colour vision of primitive races hasexcited interest mainlyin its philological aspect and has beenconsidered especially in relationto the hypothesis that there has beenconsiderable modification of the colour sensein man within historical times. Thisquestion was first raised byGladstone (1858:(3) 457J, who from a close study of theepithets for colour used byHomer came to the conclusionthat the people of that age couldhave distinguished little more than differences of brightnessand darkness. Geiger[1667 :16] later advanced the view that there had been adefinite evolution of the colour sensein man; that at one period of hisexistence he had distinguished nothing morethan differen- -34- ces of brightness; that red had been the coitt,:r firstdistinguished and that the discrimination of other colours had deviecce la the same order as that of the arrangement of the colours in the spectrums the power of seeirig blue andviolet having been the last to cc;:velop. These views of Geiger were based entirely on philologual evicieia*:e derived from a wide study of ancient literature. He was gsv,ppu.. ed by Magnus [1E77] also on philological grounds, butit was generally held that these writers were not justified in their conclusions and that theclose relation between language and sense which these authors supposed to exist was far from being a fact. It was also found by Virchow... [1870] and othersthat savages might have exactly the same peculiaritiesof colour nomenclature which are found in ancient literature and might yet have awell-developed colour sense... "(1901:48). As noted above, we also have found no evidence relating differencesin color nomenclature to differences in perceptual ability. The most recent and intriguing cross- c'iltural comparison ofcolor terminologies is that of van Wijk (1959). Noting the discrepancies in thestructure of color lexicons in the languages of the world, van Vrijk attempts to showthat the differences are based on the relative importance of the brightness vshue dimensions in color perception. He attempts to show that societies nearthe equator have color lexicons which focus primarily on thebrightness dimension while societies nearer the poles focus predominantly on hue. In vanWijk's words, "...it is striking that the brightness terms occur in theregions close to the equator... it is in these regions that the average intensityof light is greatest. The average intensity decreases as one reaches thehigher latitudes. Peoples living in the higher latitudes generally use a color [i.e., hue] nomenclature, peoples living in the tropics roughly speaking have brightnes nomenclatures, as far as we can judge from the available data. These circumstan- ces lead us to the hypothesis thatoptological characteristics of perceptions of light are originally conditioned by the properties of thelight, in such a man- ner than anoptotnical system with specific brightness terms is formed where -35- the intensity of light is greatest, and an optologicalsystem with specific colour terms is formed where the intensity of light issignificantly less and the wavelength of the light e, hue] is ofcorrespondingly greater importance" (1959:131). Interesting as van Wijk's thesis may be,it leaves much to be desired as an explanation of the patterndescribed in this report. Cne of the most serious problems with his treatment is the failure to reportterms for neutral hues in languages which are supposed to be examplesof terminologies based on hue considerations. Certainly, these language must possessterms which reflect at least the contrasts in brightness marked by thecategories BLACK and WHITE. A more refined version of van Wijk's thesismight characterize tropical systems as brightness dominated and temperate systems asbrightness plus hue dominated. Such a reformulation would bring thetheory into conformity with the facts as a rough correlational statement.Still the correlation would be far from perfect and the explanation in terms ofgeographical differences in the mean intensity of sunlightstill insufficient. The major point missedby van Wijk is that brightness is a major dimension of contrastin all color systems. As a color system introduces hue contrasts, theimportance of brightness does not diminish - the systems simply becomes morecomplex. For example, the relatively late foci crown, pink, and grey, which areabsent from just those tropical terminologies van Wijk wishes to characterize asbrightness dominated, are based almost exclusively onbrightness contrasts. At the time of introduction of these terms, foci of virtually itentical hue arealready present in the lexicon: respectively yellow, red,black/white.

7. The complete set of color chips may be obtainedfrom the Munsell Color Company, 2441 North Calvert Street, Baltimore, Maryland,21218, U. S. A. at approximately $ 100.00 per set. Munsellcode specifications for the chips utilized in our stimulus materials may be obtainedfrom the authors on request. -36-

8. See, for example, Conklin (1962, 1934), Frake (1962, 1964), Metzger and Williams (1963), Black and Metzger (1965).

9. cf. Conklin (1962).

10. Lounsbury's (1964a, 1964b) work on generation-skewed kinship systems may have a related significance. Althoughhe does not present his findings within a fully explicit theory of lexical definition, he shows thatrecursive rules are involved in specifying the signification of many kinship terms.

11. However, verbal reports of color terminologies, especially thosefrom the earlier literature, must still be interpreted with care. Cne notinfrequent- ly encounters statements such as, "The natives have no words for greenand blue. " This statement is ambiguous between (i) a lexicon in whichneither the green nor blue centers have become encoded and(ii) a lexicon in which the green center has been encoded while the blue has ncA,with the boundary of the 'green' category very likely including most blues. Neecriess to say,statements of this sort tan also be given a wide variety ofinterpretations quite at odds with our scheme. Consequently, reports containing suchambiguities haw been excluded from consideration. There remain a 'fairly large proportion of the reports in theexisting literature, which can, now that we know the universal basic categories, be - unequivocally interpreted. These reports usually convey the meaningof color words from exotic languages by means of English, or other Europeanlanguage4, glosses. Without knowledge of the eleven universal categories,such reports are difficult or impossible tointerpret and compare. Given this knowledge, the vast majority can be assigned a unique interpretation in termsof objective color measurement«

12. By "loss of a basic color term" we mean here"proceeding from the state

11111*-, -37- of having some term for a given category tohaving no term for thatcategory". Vie do not mean that languages JrAevPr"lose" color terms in the sensethat, say, a native form encoding a given cc .1y is replacedby a foreign form. InBi- sayan, for example, beldi'green'( Spanish verde) no doubtis the result of replacement of an earlier Eisayan form ratherthan the encoding,under Spa- nish influence, of a previously non-encodedcategory.

13. In 'Language L' we include previousstages of L that wouldcommonly be called by different names. For example,if L is Modern English(and the moment is the present) L also includesMiddle English, Cld English,Anglo Saxon, Proto-Germanic, Proto-Indo-European, and so on.

14. That is, about a score at the outside,hence, perhaps no morethan a few hundred forms. Cf. Wallace Chafe(1967).

15. In this and subsequent diagrams,the following conventions areused.(1) The upper and lower bands ofthe diagram represent a rowof forty chips of neutral hue and brightness nine and onerespectively, (that is purewhite and pure black). The reasonfor this is that the rectangularcolor chart used as stimulus may be thought of as atwo dimensional projectionof the surface of the color solid (which latter hasroughly the shape of two fatice-cream cones joined at the ice-cream). Thestimulus chart bears approximatelythe same relation to the color solid as aMercator px ojection does to ahollow globe. In particular, Just as the Northand South poles are"stretched" in a Mer- cator projection along the entirelength of the upper and lower edgesof the map, so in ourtwo dimensional color chart thevertices of the two cones(1. e. the pure white and pure blackpoints) are represented by the superiorand infe- rior edges of the chart, or, atlest, should theoretically be sorepresented In fact, as mentionedabove (Figure 1), neutral hues wereactually presented. to subjects as a separate array.However, in discussing the results,it is -38- easier to treat white and black as the upper and lnweredges of the hue-brightness chart. (ii) Unstippled areas of all figures containing the nameof a category denote the focus of that category.(iii) Stippled areas indicate presumedmaximal extensions of a category. (iv) Unman ^ked areas are thosewhose category affil- iations, if any, are in doubt. The reader's attention is drawn to the fact thatWHITE and BLACK are identical to white and black in regard to foci. They differonly in terms of boundaries.The same holds for RED/red, GREEN/green, andYELLOW/yellow introduced below. Given our convention of referring always tofoci in general discussion, whatever holds for white holds for WHITE, etc.,unless the context clearly indicates that boundaries are at issue.

16. Gardner's terminology is a bit unorthodox in that he makes ananalytical distinction between brightness due to the intensity of the light source(which he calls "illumination") and brightness due to the properties of thereflecting surface (which he calls "brightnee s"). He notes that the Paliyansclassify these together en the dimension niram.

17. The Paliyan data invite another speculation, plausible butlacking in direct support* Several lines of evidence, including the distinctphysical differences between Paliyans and Plains Tamil speakers, indicate thatthe Paliyans may at one time have spoken a language unrelated to Tamiland perhaps to anything else. Parallels can be seen in the loss of their native languageof various Pig- mold groups, e. g., the acceptance of Bantu by the Pigmiesof the Central Con- go. We may speculate that theoriginal language of the Paliyans contained terms for BLACK and WHITE, i.e., for two degrees of niram'brightness'. In lear- ning Tamil, the Paliyans encountered additional termsfor the quality of reflected light and simply accepted the Plains Tamil meaningsof these in so far as the dimension of brightness was concerned. Note that(1) the Paliyan terms for -39- the extremes of brightness are cognatewith the Plains Tamil termsfor white and black,(ii) the Paliyan terms for secondarydegrees of brightness and darkness are respectively the PlainsTamil terms for yellowand red (note in Figure 3 that yellow and red are thebrightest and darkest of thebasic catego- rias other than white and black); and(iii) whereas Plains Tamillacks a term for grey, blue is taken in Paliyan as themiddle-brightness category.(Note in Figure 3 that blue is par excelencethe middle-brightness categoryother than grey).

18. For the Tiv, see P. Bohannan(1963:35-36), for Tonga we havetaken the data of Elizabeth Colson (1966); theShona data are from Gleason(1961:4).

19, For the Todas data, see Magnus(1880); for Tshi, see Riis(1853). The Bantu material is reported in Buchner(1883), and the New Caledoniareport is found in Moncelon (1886). All aresummarized in Rivers(1901),

20. HanunLo is from Conklin(1955), Bassa from Gleason (1961) andIbibio from E. Kaufman (1966).

21. The data from this dialect ofShona, Ibo and Urhobo arefrom G. A. Gold- berg (1966).

22.T1m Tzeltal dtta were taken in the field from40 informants rangingfrom total Tzeltal monolinguals to perfectTzeltal-Spanish bilinguals. While ahigh intensity lamp was utilized as theconstant light source for thelanguages examined in the seminar, sunlight wasutilized in collecting the Tzeltalmaterials for practical considerations. Colorterminology in Tzotzil, Tzeltal'sclosest relative, has been studied byGeorge Collier (1963). WhileCollier's methods were not the same asthose employed in the researchreported here, it was clear from his report thatTzotzil represents Stage IV aswell. Administra- -40-

corroborates tion of ourexperimental procedureto two Tzotzilinformants Tzotzil beingalmost identi- this finding, thefoci of categoriesin Tzeltal and cal.

23. See above,section 2.3.1.

Massai data arereported in 24. Referencesfor the Nupe areBanfield (1915), (1895) and Hausaterms are Hinde (1901),Bedauye materials arein Reinesch from a biblio- found in Robinson(1925). Much of ourAfrican materials come graphy generouslycompiled by PhilipAnderson (1966).

25. Data are fromAoki (1967).

26. For Bari, seeOwen (1908); forSiwi see Walker(1921). regard and lends con- 27. The Tzeltalexperimental work isenlightening in this siderable validityto our earlierfindings.

Russian children. 28. See, forexample, the workof Z. M.Istomina (1963) on -41-

EIBLICGRAPHY

Anderson, Philip African languages.Unpu- 1966 Color categorysystems in some blished ms.Berkeley.

Aoki, Harumo 1967 Personalcommunication.

Bonfield, A. W.and J. L.Macintyre Society forPromot- 1915 A grammar ofthe Nupelanguage, London, ing ChristianKnowledge.

Berlin, Brent Unpublished ms.Berkeley 1967 rIzeltal colorterminology .

Berlin, Brentand Paul Kay 1967 English colorterms. Unpublishedms.Berkeley.

Black, Maryand DuaneMetzger study of law.In the Ethno- 1935 Ethnographicdescription and the graphy of Law.L. Nader(ed. ) AmericanAnthropologist. Special Publication.

Bleek, DortheaF. Haven: AmericanOriental So- 1956 A Bushmandictionary . New ciety .

Bohannon, P. Holt, Rinehartand Winston. 1963 Social Anthropology.New York. -42-

Buchner, 1E83 Das Aus land

Chafe, Wallace 1966 Explorations in the Theoryof Language. Unpublished ins.

Collier, George 1963 Color Categories inZinacantan. Harvardhonors thesis. Unpublished ms.

Colson, Elizabeth 1936 Personal communication.

Conklin, Harold C. 1955 HanunOo color categories. SouthwesternJournal of Anthro- pology 11:339-344. 1932 Lexicographical Treatment ofFolk Taxonomies.UAL 28: 119-141 1964 Ethnogeneological method. InW. Goodenough(ed.) Explo- rations in CulturalAnthropology.

Corson, Christopher 1967 Seminar paper on colorcategories. Dept. ofAnthropology University of California,Eerkeley. Unpublished ms.

Delafosse, M. 1894ManuelIDab.omeen. Paris, Ernest Leroux. -43-

Ervin, Susan M. 1961 Semantic shift in bilingualism.American Journal of Psy- chology, 74:233-241.

Faidherbe, L. L. C. 1682 Grammaire et vocabulaire de lalangue Paul. Paris,Mason- neve et Libraires.

Forman, Sylvia 1967 Seminar paper on color categories.Dept. of Anthropology University of California. Berkeley..Unpublished ms.

Frake, Charles C. 1937 Tagalog color terms. Stanford.Unpublished ms.

Gaden, H. 1909 Essai de la grammaire de lalangue Baguirmienne. Paris Ernest Leroux.

Gardner, P. M. 1966a Personal communication. 1966bSymmetric Respect and memorateKnowledge: The Structure and Ecology of Individualistic Culture..Southwestern Journal of Anthropology 22:389415. Geiger, Lazarus 1880 Contributions to the History of theDevelopment of the Human Race. London: Trubner andCompany.

Geoghegan, William 1967 Personal communication. -44-

Gladstone, W. E. Cxford Jni- 185C Studies on Homer and theHomeric Age. London. versity Press.

Gleason, H. A. 1961 An Introduction toDescriptive Linguistics.New York: Holt, Rinehart and Winston.

Goldberg, Berke- 1966 Ibo, Shona and Urhobocolor terms. Unpublished ms. ley.

Goodman, John Stuart 1963 Malayalam color categories.AL 5(5):1-12

Graburn, N. H. H. 1967 Eskimo color terms.Unpublished Ins.Berkeley.

Greenberg, 3. H. 1963 The Languages of Africa.Pub. 25, Indian Univ.Research Center in A nthro.,Folklore and Linguistics.

Hinde, (Mrs. )Hildegarde 1901 The Masai Language.(London, CambridgeUniversity Press).

Hollis, A. C. 1909 The Nandi; their languageand folklore.(Oxford, Clarendon Press). -45-

Istomina, Z. M. color in earlychildhood.Soviet 1963 Perception andnaming of Psychology andPsychiatry1:37-45.

Jacobson, Romanand MorrisHalle Mouton, TheHague. 1962 Fundamentals ofLanguage.

Kaufman, Elaine Unpublished ms.Berkeley. 1966 Ibibio colorcategories.

Kaufman, Terrence Unpublished ms.Berkeley. 1966 Chinook Jargoncolor terms.

K ay, Pahl Unpublished ms.Berkeley. 1937 Lebanese Arabiccolor terms.

Kirchhoff, Vega expeditionEd, 1 5.42. 1833 Lie Vissenschaft.Ergebnisse d.

Kirk, J. Ir.C. language.CambridgeUniversity 1905 A grammarof the Somali Press.

Koch,I?" -F. 1966 Personalcommunication.

Le Coeur, Institute Franais 1956 Grammaire et tentedTeda-Data. Dakar. DV\ friqueNoir-IFAN. Lee, Richard 1967 Personalcommunication -46-

Lekers, P. Benjamin 1952 Dictionaire Ngbandi. Tervurer. Commisionde Linguistique Africaine.

Lenneberg, Eric H. and Roberts, John M. 1956 The language of experience: A studyin methodology. Memoir 13 of IJAL.

Lounsbury, Floyd 1964a A formal account of the Crow andCmaha-type kinship terminologies. In E lorations Cultural Anthropolo: Essa s Presented to George Peter Murdock, W.H. Goodenough, ed. New York, McGraw Hill. pp 351-393. 1964b The structural analysis of kinshipsemantics. In Proceedings of the Ninth International Congress ofLinguistics. (Cam- bridge, Mass.). The Hague. Mouton pp1073-1093.

Madarasz, Paul 1937 Seminarpaper on color categories.Dept. of Anthro. Uni- versity of California. Berkeley. Unpublished ms.

Magnus, 1377 Die geschichtliche Entwickelung desFarbensinnes. Leipzig

McClaren, Kathrine 1967 Seminar paper on color categories. Dept.of Anthro. Um versity of California. Berkeley. Unpublished ms.

McClure, Erica 1937 Seminar paper on color categories. Dept.of Anthro. Uni- versity of California, Berkeley.Unpublished ms. -47-

Metzger, Duane and Gerald D.Williams weddings. 1963 A formalethnographic analysisof Tenejapa Ladino American Antlustd5:1073-1101.

Migeod, Frederick W. H. Trub- 1908 The Mende Language.London, Kegen,Paul, Trench, ner & Co.Ltd.

Moncelon, Paris, p. 708. 1886 Bulletin de laSocietd'Anthropologie. IX.

Nebel, 1948 Dinka grammar, Verona,Missioni Africane.

Nylander, Gustavus 2,, London. 1814 Grammar and vocabulary ofthe Bul lom language. Ellerton and Henderson. Ogan, E. 1967 Personal communication O'Neale, Lila M. andJuan Dolores 1043 Notes on Papago colordesignations. AA45(0:3E7-97.

Owen, Roger C. R. 1908 Bari grammar andvocabulary. London.3. & E. Bumpus Ltd.

Prost, 1956 La langue Sonay. Dakar.IFAN,

Rambaud, J. B. 1903 La langue Via loisParis. ImprimerieNationale. -48-

Ray, Verne F. of human color per- 1952 Techniques andproblems in the study ception. SJA8:251-259. Trans. 1953 Human color perceptionand behavioral response. New York Acad.of Sciences, ser.2, 16(2):98-104.

Reinesch, Alfred Holden. 1895 Worterbuch der3edauye-sprache. Wein.

Riis, Cdschi-sprache. Basel. 1853 Elemente detlwapim-dialects der

Rivers, W. H. R. (ed. ),Reports on 1901 Introduction and vision.In A. C. Haddon the CambridgeAnthropological Expeditionto the Torres Straits, Volume IIPhysiology andPsychology, part I.

Robinson, 1925 Hausa dictionary.London. CambridgeUniversity Press.

Romney, A. Kimball terminology. Studies 1987 Internal reconstructionof Yuman kinship in SouthwesternEthnolinguistics ed. DellH. Hymes with William E. 7.ittle.Mouton & Co. TheHague. Paris. pp. 379-388.

Sapir, Edward culture: A Study 1916 Time perspective inaboriginal American in method. InDavid Mandelbaum(ed. ) SelectedWritings of Edward Sapir inLanguage, Culture andPersonality. Berke- ..49N.

Press, 1949. ley and LosAngeles: Universityof California

Segall, M. H., D.T. Campbelland M. J.Herskovits perception.Indianapolis. 1966 The influence ofculture on visual Eobbs-Merrill.

Smith, Edwin W. London. CxfordUniversity 1907 A handbook ofthe Ila language. Press.

Slobin, Dan I. 1967 Personalcommunication.

Stapleton, languages.Bolobo. 1903 Comparative grammarof the Congo B. M. S.Press.

Steager, Peter Dept. ofAnthropology. 1967 Seminar paper oncolor categories. University ofCalifornia. Berkeley.Unpublished ms.

Stross, Brian Dept. ofAnthropology. 1967 Seminar paper oncolor categories. University ofCalifornia. Berkeley.Unpublished ms.

Turner, Victor ritual. InAnthropological 1956 Colour classificationin Ndembu approaches to thestudy of religion.Banton, ed.Associa- tion of SocialtnthropologyMonography. -50-

Virchow, 1878 2eitsh. f. Ethnol.Ed. X. 1879 Zeitsh. f. Ethnol.Ed. XL von Hagen, Eerlin. Druckand Verlag von 1914 Lehrbuch derEulu-sprache. Gebr. RodetzkiHofbuchhandlung.

Talker, 17. Seymour Paul, Trench,Truber 1921 The Siwi language.London. Vegan, & Co. Ltd.

ijk, H. A. C.W. van and brightnessnomencla- 1959 A cross-culturaltheory of colour 15.2: ture. Bijdragentot de Taal-,Landzen Volkenkunde 113-137. The Hague.