65 COLOR CATEGORIZATION IN NANUMEA, A POLYNESIAN ATOLL SOCIETY Keith S. Chambers University of Auckland Auckland, New Zealand This paper describes the color classification system and aspects of color semantics of Nanumea, a Polynesian atoll in Tuvalu, Western Polynesia. Since this is, to my knowledge, the first substantial description of a Polynesian atoll color classi- fication system, and data on Polynesian (or indeed, Oceanic) color categorization has been limited at best and widely scattered in published and unpublished sources, it is hoped that it will prove of some use as a contribution to the growing ethnographic litera- ture on world color systems.1 Utilizing the methods and approach outlined in Berlin and Kay (1969) and Berlin and Berlin (1975), the Nanumean color system can be characterized as Stage V: it includes basic color terms for BLACK, WHITE, RED, YELLOW, GREEN and BLUE. There are, in addition to these basic categories, several which appear to be "incipient" or emerging basic color terms or categor- ies. These additional categories are apparently being encoded in the order posited by Berlin and Kay (1969) for the evolution of basic color terms. Other aspects of Nanumean color semantics to be discussed here include secondary color terms, the ways in which color terms are qualified, and possible evidence in the Nanumean data for diachronic change in color classification. Also discussed briefly are recent claims that Polynesian cultures are "color indifferent," have poorly developed color lexicons, and generally display little interest in the domain of color. Methods Originally, fieldwork plans did not include a formal investi- gation of color classification, although both my wife Anne and I were familiar with the work of Brent Berlin and Paul Kay. After a number of months of residence on Nanumea, and with increasing fluency in the language, color terms were frequently encountered, and we decided to pursue the subject in more structured fashion. We took every opportunity to note down color terms and usage, and to question informants about color classification; in the course of the field period we gathered considerable data of both a systematic and an informal nature on the color lexicon, and the domain of color. Brent Berlin supplied us with the Berlin-Kay Munsell color array for use in the identification and mapping tasks. This board (cf. Berlin and Berlin 1975, Dougherty 1977 for other descriptions) has 320 Munsell color ships (each measuring approximately 1.5 x 2.0 cm.) mounted contiguously in spectral array. The chips vary in two 66 dimensions, hue along the long axis of the board, and brightness along the narrow axis, with all chips of maximum saturation. White and Black are included with the addition of a continuous row of 40 white chips (neutral hue, maximum brightness) bordering the bright (top) edge of the array, and a similar row of black chips (neutral hue, minimal brightness) bordering the lower edge. A border of neutral hue grey surrounds the array. A range of brightnesses in the neutral hues is provided in the addition, to the left side of the array, of a vertical row of ten chips ranging from black through grey to white. For the formal portion of the research informants were inter- viewed, and carried out naming and mapping tasks using a selection of colored objects and the color array. This phase of the research was carried out in the privacy of our house, away from onlookers, and in natural daylight. All phases of the research were conducted in the Nanumean language. Informants2 were asked to carry out three tasks.3 The first of these, elicitation, involved discussing the domain of color and asking the informant to name all the colors he/she could. Each informant was then presented with, in random order, each of 41 colored items, selected from among our household equipment, and asked to name its color. These items are listed in Table 1, and Figure 1 charts their locations on the Berlin-Kay Munsell color array. Finally, each informant was shown the color array and asked to map the maximum boundaries of each color he or she had named, as well as to indicate its purest or truest occur- rence (focus). A clear plastic film over the array allowed informants to draw in boundaries and foci with a black felt tip pen. GREY RED YR YELLOW GY GREEN BG BLUE PB PURPLE RP 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 _B 8 6 41 _ _ 7 _ 410 R~~~~~~~~~~2I - R 7 _ 7 __ 3 2 _ 18 34 _____I_24 _1E4 H I S _ _2 __ c1liil 14 TS 38 _ _ -~ _8 4C 40:____ -5t 1____ 11 ~~~~3C5 91? FIGURE 1. Distribution, on Berlin-Kay Munsell color array, of colored objects used in naming task. Numbers refer to objects listed in Table 1. 67 TABLE 1 Test Objects for Naming Task red notebook off-green book cover plastic casette tape box, beige-yellow plastic casette tape box, dark grey red Fujichrome slide box pink book cover orange book cover green Neopan film box beige folder yellow plastic Kodak film box yellow notebook metallic orange alarm clock blue Sanyo battery grey book cover gold metallic rim on alarm clock blue, blue-green book cover blue notebook chartreuse notebook reddish brown book cover greyish blue book cover black book cover metallic silver Sanyo battery green book cover flecked with black light blue metal can dark blue plastic pen orange plastic jar lid purple plastic jar lid light blue plastic cup brown pencil rose red pencil yellow-green plastic pen blue eraser brush red eraser brush chartreuse pencil black plastic pen purple pencil orange pencil pink pencil yellow pencil light blue pencil pink newspaper 7.5,10 R/5 10 Y/6,7 2.5 Y/6,7 GREY/3 7.5 R/3,4 10 RP/8, 2.5 R/8 7., YR/7,8 10 -GY/5,6 2.5 Y/7, 5 Y/8 2.5,5 Y/8 5 Y/8,9 5 RY/6,7 not coded; approx. 10 B/6 GREY/5, GREY/6 not coded 7.5 BG/4 10 B/4 5 GY/7 2.5 YR/4,5 7.5 B/3 GREY/1 not coded approx. 5 G/3 not coded; approx. 7.5 B/7,8 7.5 B/3 not coded; approx. 2.5 YR/6 not coded; approx. 10 PB/3 10 B/7 7.5,5 YR/3 5 R/4 10 GY/6, 2.5 G/6 2.5 PB/3 7.5 R/3 7.5 GY/7 GREY/1 2.5 P/2,3 2.5 YR/6 7.5 RP/6, 2.5 R/5 2.5 Y/8 7.5,10 B/5 5 R/8, 10 RP/8 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. Object Munsell Coordinates 68 Informants encountered no difficulty in carrying out our instructions in the naming or mapping tasks (though Hage and Hawkes 1975 report their informants had problems). It is likely that the procedure was simpler for Nanumeans than for Hage and Hawkes' in- formants, since most Nanumeans have been to school and are familiar with pens and pencils, and writing and drawing, and because there is in Nanumean an abstract noun meaning color (lanu). All inform- ants found it somewhat easier to map the ranges of their color categories than to specify foci. They were encouraged to mark multiple foci if they wvere so inclined, though most marked a single chip. One informants mapped boundaries using curved lines which cut across color chips, reporting that he felt the colors varied at that line. Another felt compelled to include the whole color array in her mapping, and so drew very wide boundaries. These unusual responses underscore recent criticisms of the color array (see Hage and Hawkes 1975 for instance) which point out that it may induce skewed responses. Berlin and Kay, now engaged in a widespread survey of color usage in societies previously not included in their original color term research, have modified their methodology and the color array in an attempt to eliminate these problems (personal communication from Brent Berlin). The Setting--Nanumea Atoll Northernmost of the nine atolls and small reef islands com- prising Tuvalu (formerly the Ellice Islands), Nanumea lies 5? south of the equator and just west of the International Date Line. With a population of nearly one thousand people living on 3.8 sq. km. of land, the island is densely settled. The vegetation and fauna are those typical of coral atolls (cf. Alkire 1978), limited but at the same time varied enough (since Nanumea receives abundant rainfall) to sustain the largely subsistence life style adequately. Fish, coconut, coconut sap toddy, Cyrtosperma "atoll taro," taro, and breadfruit are the staples of life, supplemented today by rice, flour and other commodities from the island's branch of the Tuvalu Cooperative store. With its single village occupying a peninsula of land facing both the sea and the calm lagoon, its thatched houses sheltered under palm and breadfruit, and a line of outrigger canoes beached in the shade of the lagoon shore, Nanumea is the essence of the Polynesian atoll society. Settled by traditional account (Chambers, Chambers and Munro 1978) some 26 generations ago, Nanumea has apparently remained an independent political unit since settlement. To be sure, close cultural and linguistic ties bind the island with its neighboring Tuvalu islands. Tuvalu is, and Tuvaluans feel it to be, a single culture, despite minor inter-island variations. What little linguistic work has been done points to a Samoic parent language for Tuvalu, with the split between the parent and daughter language taking place about 1000 A.D. (Pawley 1966), although another esti- mate places the breakoff as far earlier, in the period 300-500 A.D. (Bayard 1976). The Tuvalu dialects vary significantly, and 69 Nanumeans claim there is a substantial Tongan component, both linguistically and genetically, in the island's past (A. Chambers 1975; Chambers, Chambers and Munro 1978). Although this assertion based on traditional history has not been formally investigated,4 the claim serves to underscore the fact that until archaeology is brought to bear on the question of Nanumean and Tuvalu origins the prehistory of this pivotal border region of Polynesia is likely to remain somewhat uncertain. The contact history of Nanumea, relevant for an understanding of possible influences in color nomenclature, has been one of limited communication with other settled areas, except for the nearest of the other Tuvalu islands. Traditional accounts record a series of raids by "Tongans" continuing for several generations after initial settlement, followed by a period of peace, and then a renewed time of unrest and attempted invasion by Micronesian speaking Gilbertese (K. Chambers n.d.). In the mid 19th century European whalers sometimes visited the island, followed later by occasional traders, and the landing in 1873 of the first of what was to be a series of resident Samoan Protestant pastors. The establishment of a British Protectorate over Tuvalu in 1892 marked the formal incorporation of the island group into the British sphere of Pacific affairs. Although the coming of the Samoan pastors and political association with Britain brought little immediate outward change in Nanumean life, it did result in gradual and far-reaching alterations in the structure of the Island culture. One major change involved the island's political system, which had evolved into a dualistic balance of power between hereditary chiefly lineages (aliki) and non-hereditary but achieved roles dominated by leading warriors (toa). The balance rapidly gave way when the toa of Nanumea converted to Christianity and accepted the nominal leadership of the resident Samoan pastor. The chiefs, their influence already undermined, were gradually relegated to figurehead status, and the traditional leadership role continued its decline throughout the 20th century, until virtually abolished in 1973. Nanumean society is structured today, as it was in the past, largely on the basis of kin relationship. Primary allegiance is to loosely defined extended family groups, which reside together form- ing a household unit. Under the leadership of the senior member, households control rights to land and produce, and allocate them to individual members. Island political authority is vested in an elected council, a system of government fostered by the British Colonial officials. Each Tuvalu island also sends one or more representatives to a national legislature which meets at the capital, Funafuti. Tuvalu is today an independent Pacific nation and member of the Commonwealth (see Macdonald 1975, Wilson 1978, and Connell 1980 for accounts of these recent political developments). External linguistic influences, which are relevant to an under- standing of possible inputs into the Nanumean color classification 70 system, have been varied but apparently slight over the centuries, with three contact languages being of most significance: Gilbert- ese (a Micronesian language), Samoan and English. The early clashes with Gilbertese raiders mentioned in traditional accounts probably left minor, but as yes uninvestigated influences. More recently, and of greater significance, opportunities for employment at "overseas" island phosphate quarries have become available to Nanumean men, and have brought them into close contact with the more numerous Gilbertese workers and the Gilbertese language. Today many Nanumean men understand some Gilbertese, and there are a number of obvious lexical borrowings. The Samoan language has been of recent importance since Samoan pastors organized an island school in the 1870s and staffed it until the 1960s. Teaching and, of course, preaching was largely in Samoan. The Samoan Bible and other religious literature was used, and Samoan was the first literate language (today replaced by the vernacular and English). The lexical influence of Samoan has been limited, however, and Nanumeans have kept the two languages separ- ate. All adult Nanumeans today know some Samoan, though few are fluent speakers. The official language of church and school has been Nanumean for more than a decade, and Tuvalu hymns and a Tuvalu language Bible are now used. Samoan lexical borrowings are usually readily apparent and are most prevalent as religious terms. Despite, then, an apparent multilingual milieu (since English, too, has been taught in the island primary school for some years), most Nanumeans today are essentially monolingual. There is no common second language, and children and parents use a single vernacular, which has maintained its integrity as a distinct dialect within Tuvalu. Everyone knows a smattering of Gilbertese, English or Samoan, but few individuals are truly fluent in any of these languages. Nanumean Color Categorization Avoiding, for the moment, considerations of possible internal variation in the Nanumean color classification system, Nanumeans appear to share a Stage V system, with basic categories consisting of WHITE, BLACK, RED, YELLOW, GREEN and BLUE. There is, in addi- tion, evidence that brown, purple, pink and orange have enough saliency for at least the first three of these to be considered "incipient" basic color categories. It seems clear that BROWN is very nearly a basic category, and PURPLE and PINK not far from universal acceptance. Nanumean Basic Color Categories The six basic color categories in Nanumean are tea WHITE Uli BLACK 70 kula RED sama, samasama YELLOW lanu launiu GREEN 5 puluu, lanu moana BLUE tea WHITE tea (derived from Proto-Polynesian *tea 'WHITE'6) was re- stricted in the mapping task to pure white by all but one informant (see Figure 2). The exceptional informant extended tea to include half of row nine, the brightest hues from green through red-purple. The fact that, inadvertently, few pale or whitish objects were included in the naming task may have resulted in undue restriction of the eventual range of tea, however. Kennedy (1931:102), for instance, reporting his experiences in the 1920s at Vaitupu in Tuvalu, claims that the WHITE term there referred to all light colors. This situation certainly does not hold in Nanumea today, where most informants are most comfortable restricting tea to pure white. It is conceivable, however, that a larger sample would result in some individuals extending their mapping of tea to in- clude other light (or "warm") colors. uli BLACK, "dark" Naming and mapping responses for uli are shown in Figure 3. As can be seen, only one item outside the range of neutral black or grey was called uli. This item, a greyish blue book cover of low saturation, was difficult for all informants to categorize, and one called it kaakii, BROWN. Mapping responses for uli were confined to pure black and dark GREY for most informants, but one extended it to GREY/4. In all cases focal black was confined to the bottom row of pure black chips. It is clear, both from the mappings and from experiences with Nanumeans in conversational settings that uli means both BLACK and "dark." Thus Nanumeans sometimes refer to themselves in contrast to Europeans as tino uli "dark people," their skin color being a light brown. It is possible that uli may have had the connotation of DARK-COOL at some point in the development of the language (Dougherty 1975, Kay 1975, Kay and McDaniel 1978 discuss the prob- able separation of all color categories from an initial DARK-COOL and LIGHT-WARM dichotomization). But Kennedy's statement (somewhat enigmatic since he then goes on to list 15 color terms) for Vaitupu in the 1920s definitely does not apply either there or in Nanumea today: it would seem that the only true colour- distinctions are between white or light-coloured and black or dark-coloured. A dark shade of almost any colour would be called uli (Kennedy 1931:102). 71 RED YR YELLOW GY GREEN BG BLUE PB PURPLE RP 5 10 5 10 5 1O 5 10 5 10 5 10 5 10 5 10 5 10 5 10 I! e . .-_1 I R G H N S S 1 a 7 6 5 4 3 2 l FIGURE 2. Range and focus for tea, WHITE. Contour lines represent range mappings for 80% agreement and maximum extension. Dots represent foci. The only object named as tea in the naming task was a metallic silver battery (not coded). RED YR YELLOW GY GREEN BG BLUE 5 10 5 10 5 10 5 10 5 10 5 10 5 10 PB PURPLE RP 5 10 5 10 5 10 FIGURE 3. Range, focus, and naming responses for uli, BLACK. Contour lines represent range mappings for 80% agreement and maximum extension. Dots represent foci. Integers represent naming responses and indicate the number of times an object matched to the chip in which the integer is written was named with the color term concerned. GREY GREY I R G H T N 10 9 8 7 6 5 4 3 2 1011- 1111 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I I I I 0 1 1 1 1 1 1 1 1 1 1 1 1 I I I I I I I I I I I I I I I I I I I I I I I I I I F I I I I I I I I I I - T -1-1- -fl I I I - I I I I I I I I I I I I I I I I - 11 - 1. 11 - - I I I 1 1 I I - 1 1 1 I I I I 1 2 1 1 --7T-- I I 5 72 If uli once had a wider application in Nanumean, today it is fairly well restricted to a few darker shades of GREY and to pure BLACK. This is in keeping with recent findings for the Stage IV and V individuals on the Polynesian outlier West Futuna as well. As individual color systems there approached Stage V, the former DARK/ LIGHT distinction became gradually more restricted and approached a pure BLACK/WHITE opposition (Dougherty 1975, 1977). kula RED Naming responses, range and foci for kula are shown in Figure 4. GREY RED YR YELLOW GY GREEN BG BLUE PB PURPLE RP 5 10 5 1O 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 _ 10 _ 1 R G 6 2 N -E 4 35 5r - 2 1 11 _H __ 1111111 _ FIGURE 4. Range, foci, and naming responses for kula, RED. Contour lines represent range mappings for 80% agreement and maximum extension. Dots indicate foci. Integers represent naming responses and indicate the number of times an object matched to the chip in which the integer is written was named with the color term concerned. In neither the naming nor mapping tasks was there any inclination to extend kula to include YELLOW. ORANGE was frequently included within kula in the mappings, while distinguished as separate by some informants in the naming task. It is clear from the mappings and remarks made by informants that ORANGE is not yet a separate category. It is discussed below under secondary terms. There is limited evidence that an original undifferentiated ROW (RED and YELLOW) category was once present in Nanumean. During the research period the term kula ("red") was occasionally extended in casual conversation to include some yellow. The most common usage of this sort is in reference to bananas or papayas which have ripened, changing in the process from green to yellow. Thus, koa kula te fui maika ("the cluster of bananas is yellow") was con- trasted to koa ppala te fui maika ("the cluster of bananas is 73 ripe") by one elderly individual to back his assertion that kula was indeed a color term (YELLOW) in this context, and not merely a reference to a stage of ripeness. Yet younger informants laughed at the idea that it might be possible to label YELLOW kula. For them, the phrase koa kula ("it is kula") refers unequivocally to the state of ripeness (of bananas, papaya, etc.), and not to the color of the fruit. The use of kula for YELLOVI by some informants (in addition to its universal meaning, RED) was demonstrated in another context, as I was discussing the use of a type of coral in lime making. Its color was described to me as being tai kulakula, which I inter- preted to mean "sort of reddish." When I asked the informant (a middle aged male) to indicate the exact shade of kula, I was sur- prised when he pointed to a yellow table. This extention of the term kula to include YELLOW, which may reflect an archaic usage, was not corroborated in the limited sample who participated in the formal naming and mapping tasks, since none merged RED or YELLOW. The sample included no elderly informants, however, and, as noted previously, it is likely that a more extensive survey would have revealed variation including some merging of RED and YELLOW to form a ROW category. While the separation of RED and YELLOW is a firmly established fact for all younger Nanumeans, and probably for most older ones as well (since instances of individuals using kula to mean YELLOW were very rare during nearly two years of fieldwork), it seems that for a few individuals a ROW category probably persists. sama, samasama YELLOW This term, found universally in the sample of individuals who participated in the naming and mapping tasks, and in very wide- spread use on Nanumea, is apparently a recent borrowing from Samoan7 (Samoan samasama "yellow," sama "tumeric powder," Milner 1966). An indigenous Nanumean term, felo or (alternately) fefelo (from PPN *felo "yellow, tawny") is often heard as a synonym of sama, although this was only used by one informant in the naming and mapping tasks. She indicated that, while the "true" Nanumean term was ffelo, for her sama was a more salient label for YELLOW. This view, apparently reflecting a relatively recent shift in the label for YELLOW (but not in the clearly established category itself) under influence from Samoan, is shared by many Nanumeans. Naming responses, range and foci for sama are shown in Figure 5. lanu launiu GREEN This term for GREEN, literally "coconut leaf color" (from lanu "color" + launiu "coconut leaf"), does not meet the first criterion set by Berlin and Kay (1969) for a basic color term since it is not monolexemic and can be analyzed to its constituent morphemes. Nonetheless, there is ample evidence to suggest that lanu launiu is indeed basic. It satisfies all other criteria set down for a basic 74 RED YR YELLOW GY GREEN SG BLUE PB PURPLE RP 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 B 8 5 -111 11 R 7 G 6 1m 13111111111]11]1 H T 5 N E 4 S _ _ I H F 1 1 I I I FIGURE 5. Range, foci, and naming responses for sama, YELLOW. Contour lines represent range mappings for 80% agreement and maximum extension. Dots indicate foci and predominant foci (heavy dots). Integers represent naming responses and indicate the number of times an object matched to the chip in which the integer is written was named with the color term concerned. IEY RED YR YELLOW GY GREEN BG BLUE PB PURPLE RP 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 I R G H T N E S i 10 9 a 7 6 5 4 3 2 1 FIGURE 6. Range, foci, and naming responses for lanu launiu, GREEN. Contour lines represent range mappings for 80% agreement and maximum extension. Dots indicate foci. Integers represent naming responses and indicate the number of times an object matched to the chip in which the integer is written was named with the color term concerned. GREY GR 41 151 - 1 4 [ 1 5-15 1 F] I IIH I. I -I - 121 1 1 1 1 I I F I I I F M I --III M I A 75 term: it is not included in some other color term, its use is not restricted to a particular class of objects, and it is highly salient in Nanumea. It is interesting that Kennedy (1931:102) appears to have dismissed lanu launiu as a true GREEN term for Vaitupu in southern Tuvalu, and to have classed it with other non- basic Vaitupu compounds such as lanu kefu "color of a dead leaf" (dark brown), lanu fulu lupe "color of pigeon feathers" (irides- cent), and lanu lau milo "color of a milo leaf" (light green). There are several such complex forms in use on Nanumea to indicate varying shades of greenish and yellowish, but in the mapping tasks it was apparent that they were all sub-types of lanu launiu, GREEN. These sub-terms are discussed under secondary terms below. The naming task and mapping and foci responses for lanu launiu (Figure 6) confirm what is readily apparent from the extremely fre- quent use of this term in everyday life to mean GREEN: the term is stable in use and extension. One informant (number 5) did extend her mapping for this category into 5, 7.5 and 10 YELLOW, but this appears to have been related to her desire to include the whole color array in the mapping task. The ranges she mapped were far more inclusive than those of any other informant. One other informant (number 4) extended her mapping of lanu launiu to 5 and 7.5 BLUE GREEN on the board. puluu, lanu moana BLUE As for GREEN, the native term for BLUE is lexemically complex; it means literally "the color of the sea" (lanu "color" + moana "sea"). But, as in the case of GREEN, there is good reason to conclude that BLUE is a basic color category in Nanumea. For one thing, there now exists a borrowed, monolexemic term which is considered to be synonymous with lanu moana by most speakers; this is puluu, derived from English "blue." There are two other BLUE terms in current use on Nanumea, ngau and unoko. Ngau (the name of a tiny bright blue fish which fre- quents coral outcrops in the lagoon shallows) was mapped by three informants as a part of a continuous BLUE category. For these informants the remainder of BLUE was labeled as lanu moana and/or puluu. Figure 6 shows the overlapping mappings for the various terms for BLUE. Unoko (possibly from unoko "bruise"?) was used by just one informant (number 1) and was used in addition to the term puluu. He considered them synonyms. Remarking on unoko he said, "this is a real Nanumean word, the ocean is unoko, the lagoon is unoko maamaa ('light' unoko). But puluu, probably an English word, is replacing unoko here." With the small sample size it is difficult to know whether unoko is generally widespread in use. It is glossed as "blue" in a word list of Nanumean compiled in 1967 (Ranby, in press). The Samoan term uno'o "be bruised or swollen" (Milner 1966) is cognate. 76 Figures 7 and 8 show the naming responses for the various BLUE terms, as well as the range and foci of this category. Despite some variation in a label for this category, mapping and naming responses show that BLUE is a clearly established and basic cate- gory in Nanumea. It seems likely that the original term for this category was lanu moana "sea color" (Snow [1971] accepts lanu moana as the basic BLUE term for Samoan). What has apparently happened in Nanumea is that secondary sub-terms, perhaps of limited distri- bution in the population, refer to various shades of BLUE: ngau to a bright blue similar to hhat of the ngau fish, unoko possibly to a darker bruise-like shade. But, as Table 2 makes clear, these various BLUE terms are nearly eclipsed in frequency of use today by the borrowed term puluu. It may be, as Dougherty (1975:218-9) has speculated for several color terms in West Futuna, that the amnbigu- ity of reference of several competing local terms has led to the ready acceptance of a new, monolexemic, and context-free basic term from another language, in this case English. A further indication of the instability of the label for BLUE is the response of inform- ant number 2, who easily mapped BLUE and labeled it lanu moana. For her the label puluu referred to PURPLE. Yet, in her more spontaneous naming responses, both BLUE and PURPLE were labeled puluu. PURPLE, an emerging basic category, is discussed below. GREY RED YR YELLOW GY GREEN BG BLUE PB PURPLE RP 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 R 7 _ _|?{X4W]- _2 5 1-_______ G 6 H T 5 4 N S 4_____XT 1 3 le I I I 2 . li I _7 _ I r g~~~~~~~ I I _, I _ I _11~17f I I 1111t FIGURE 7. Range, foci, and naming responses for BLUE (several terms). Contour lines represent range mappings for 80% agreement and maximum extension. Dots indicate foci and predominant foci (heavy dots). Integers represent naming responses and indicate the number of times an object matched to the chip in which the integer is written was named with one of the four color terms indicating the BLUE range. 77 RED YR YELLOW GY GREEN BG BLUE PB PURPLE RP 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 I IIII H II III I _1 II 1 1 X i _ _ _~~F 3. a. labelled both lanu moana and he ngau. b. labelled puluu. Informant no. RED YR YELLOW GY GREEN BG BLUE PB PURPLE RP 5 10 5 10 5 10 5 1O 5 10 5 10 5 10 5 10 5 10 5 10 III MI IC _ 11 1 I _ _ _ _ llmIIlililil I 1I IL _ 111 _ _ _ _ Ll~~~~~~~ibilTlll 4. a. labelled ngau. b. labelled puluu. Informant no. RED YR YELLOW GY GREEN BG BLUE PB PURPLE RP 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 10 9 e 7 6 5 4 3 2 Informant no. 5. a. labelled lanu moana. a'. labelled ngau. b. labelled puluu (subsumes a and a'). FIGURE 8. Division of the BLUE category by three informants. See text for glosses of individual terms. GREY 10 9 a 7 6 5 4 3 2 R G N T I GREY 10 9 8 7 6 5 4 3 2 B R G H T N E S S I GREY B R G H T N E S 1 78 TABLE 2 Elicitation Frequency of BLUE terms in Naming Task * glosses for these terms English "blue". appear in the text. All approximate Secondary Terms This completes the inventory of terms which I consider to be the basic color terms in Nanumea. There are a number of other terms which I consider secondary, because they fail to meet several of the defining requirements for basic terms outlined by Berlin and Kay (1969:5-7). Among these secondary terms are two which appear to be very close to becoming basic terms. These "incipient" basic terms are kaakii BROWN, and violeta PURPLE. kaakii BROWN This term (from English "khaki") was used by all informants, and the range for the category (Figure 9) is consistently and obvi- ously BROWN. The focus is somewhat uncertain. Kaakii seems to be generally less cognitively salient and common as a color term than the preceding basic terms. But most important in concluding that Informant Color Terms Totals puluu lanu moana ngau unoko 1 4 0 0 4 8 2 9 1 0 0 10 3 8 1 1 0 10 4 6 1 4 0 11 5 6 2 1 0 9 Total 33 5 6 4 48 Total as % of BLUE terms 68.8% 10.4% 12.5% 8.3% 100% used in naming task 79 GREY RED YR YELLOW GY GREEN BG BLUE PB PURPLE RP 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 1 _I 98 2 G 6 4 2 T 5 ____ 3___ '_Ll _________II_111 N E 4 ____3 _____: _ 2 _ L _ _ l7 1_ I I IIT111 IX i H -- - I I I10 LX0 E Ed IIIIIT FIGURE 9. Range, foci, and naming responses for kaakii, BROWN. Contour lines represent range mappings for 80% agreement and maximum extension. Dots indicate foci. Integers represent naming responses and indicate the number of times an object matched to the chip in which the integer is written was named with the color term concerned. it is still less than a basic term are the naming responses. These included item 14, a grey book cover (GREY/6, GREY/5), item 3, a yellow-beige plastic box (2.5Y/6,7), and item 20, a grey-blue (7.5 B/3) book cover (see Table 1). These responses are at odds with the range as mapped, and indicate that the term kaakii is still sufficiently unfamiliar to be extended to a range of rela- tively desaturated hues for which no ready term exists. It is interesting to speculate on the extension of kaakii to include some greys; this is possibly an indication that a GREY category, as yet unlabeled, is developing in Nanumea. violeta PURPLE PURPLE also seems to be an "incipient" basic category, with the borrowed term violeta (from English "violet") most often used to denote it. Four informants used the term violeta or a recogniz- able variation of it. The fifth used u (from English "blue") to indicate PURPLE. Mapping responses for violeta were fairly consistent; foci were generally clustered between 2.5 and 10 PURPLE on the color array (Figure 10). Again, however, naming responses were inconsistent and included a grey and a pink chip. Neverthe- less, PURPLE appears to be generally recognized and is distin- guished both from BLUE and from RED. piniki PINK Another borrowed term which labels a category which is also probably a potential basic category is piniki (from English "pink"). 80 RED YR YELLOW GY GREEN SG BLUE PB PURPLE RP 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 5 1o _ __ _- _ _ _ __ __ _ ___ _ ___ 141E W _ I R G H T N E S S 4 10 9 8 7 6 5 4 3 2 FIGURE 10. Range, foci, and naming responses for violeta, PURPLE. Contour lines represent range mappings for 80% agreement and maximum extension. Dots indicate foci. Integers represent naming responses and indicate the number of times an object matched to the chip in which the integer is written was named with the color term concerned. YR YELLOW GY GREEN 5 10 5 to 5 10 5 10 BG BLUE PB PURPLE RP 5 10 5 10 5 10 5 10 5 10 FIGURE 11. Range, foci, and naming responses for piniki, PINK. Contour lines represent range mappings for 80% agreement and maximum extension. Dots indicate foci and predominant foci (heavy dots). Integers represent naming responses and indicate the number of times an object matched to the chip in which the integer is written was named with the color term concerned. GREY GREY RED 5 10 p I G H T N E S I 10 91 B| 7 6 5 4 3 2 I 4 4 I U 81 The naming and response and range and foci mappings are shown in Figure 11. Four informants used the term piniki, while one (informant 1) appeared not to know it, but correctly mapped PINK, calling it tai teatea, tai kulakula "a little whitish, a little reddish." Later in the interview he seemed spontaneously to recall the word piniki, but mapped it as the first four chips of 2.5 RED/2-5, which overlapped almost entirely with his earlier mapping of kula RED. It seems likely that a wider sample would reveal variability in the use of PINK ranging from complete integration of this category with RED, to complete separation. But for many Nanumeans PINK is well on its way to becoming an established and labeled basic category. tongalaa ORANGE The term tongalaa (from too + -nga "setting" + laa "sun") was used and its range and foci mapped by all informants (Figure 12). One informant used the term olenisi (from English "orange") as a synonym for tongalaa. While it is generally used in reference to ORANGE, this category is not firmly established. For one informant tongalaa was named as a sub-type of RED (kula), and she called it kula tongalaa ("sunset red"). For several other informants tongalaa as mapped also fell within the range of their RED map- pings, though they did not overtly equate the two categories. But it was the naming task which clearly demonstrated that ORANGE is not a fully separate category. Item 12 (Table 1), an orange enamel alarm clock (5 YR/6,7) was variously named as kaakii ("brown"), tai kulakula ("sort of reddish"), kula ("red"), and lanu lau pulaka GREY RED YR YELLOW GY GREEN BG BLUE PB PURPLE RP 5 10 5 10 5 10 5 10 5 1O 5 10 5 10 5 10 5 10 5 10 I '8 R H N E I I I H_______1__ I I I I IT 1l FIGURE 12. Range, foci, and naming responses for tongalaa, ORANGE. Contour lines represent range mappings for 80% agreement and maximum extension. Dots indicate foci and predominant foci (heavy dots). Integers represent naming responses and indicate the number of times an object matched to the chip in which the integer is written was named with the color term concerned. 82 ("color of pulaka leaf" [a yellow-brown when dead]). A bright orange plastic jar lid, item 26 (ca. 2.5 YR/6) was named as tai kulakula ("reddish"), and as lanu lau pulaka. Other Secondary Terms A number of other secondary terms in use in Nanumea are sub- types of categories already discussed. Lanu lau pulaka (lanu "color" + lau "leaf" + pulaka "Cyrtosperma chamissonis"), i.e., "[dead] pulaka leaf color" was used by two informants, and refers to a mid-brightness yellow-brown; it falls into both YELLOrW and BROWN mappings. Lanu lau maika ("banana leaf color") and lanu lau olesi ("papaya leaf color") were used as synonyms by one informant for a yellow-green shade, which falls entirely within the extension of the composite YELLOW mapped by all informants. One anomalous borrowed term used by a single informant (num- ber 3) seemed to be idiosyncratic: kulini (from English "green") referred for her to several chips of brownish-red hue (5 YR/3-5, 2.5 YR/3) which were otherwise almost entirely included in the composite range of BROWN. Finally, two sub-types of RED were named by four informants. Kasimia (English "cashmere") was restricted to the deep reds and red purples of low brightness (2.5-7 R/2; 7.5,10 P/2; 2.5-10 RP/2), while akanta (probably from Gilbertese akanta "shrub of the genus Acanthus [whose] flowers are white, or pink to purple," Luomala 1953:50) was used by two informants to refer to bright reds (2.5 R/4,5; 7.5,10 RP/5). This latter term is also the local Nanumean name for the bougainvillea, and uncommon and imported shrub, which exists on Nanumea only in its bright purple form. Restricted Secondary Color Terms kelo blond, of hair, people; also used for pale colored fish. piho kelo! ("blond head!") is a derogatory epithet. leu to ripen, turn yellow or reddish. Used only of Pandanus (cf. W. Fu- tuna hleu, Dougherty 1975:44). aafu ripening stage, half green, half yellow, of bananas, papayas. laea to be black, dark; used only of uto variety coconuts. kafaaui a variety of black or dark coconut (from kafa, pertaining to sennit 83 pulepule taaselesele vali kele, kelekelea aulo siliva malifalifa hinaa cord, or husks used for making it + -ui "dark"?). The suffix -ui is probably cognate with PPN *usi "green," Tbngan uiui "black," and West Futuna wiwi GRUE. Samoan -ui "dark, of plants" is cognate: Biggs 1979; Branstetter 1977; Dougherty 1975; Milner 1966. spotted, patterned (cloth, leaves, dogs, pigs, etc.). striped (of cloth, fish, etc.). dye; a general term for any dye (e.g., vali kula, "red dye"). earth, soil; dirty (from PPN *kele "earth, dirt"). gold (precious metal); gold colored (cf. Samoan 'auro, from Tahitian auro, from Latin aurum: Milner 1966). silver (precious metal); silver colored (from English "silver"). shiny (of metallic objects). Used by some informants in response to gold or silver colored objects. One informant described a metallic silver color as malifalifa tea "white shining-ness." grey; white: of hair only (PPN *sina). The following secondary terms, included in Ranby's vocabulary list of Nanumean (Ranby, in press), were not encountered or elic- ited during our fieldwork on the island. hengahenga kefukefu mata lelefu lenga yellow (PPN *sengasenga or sengisengi, "daybreak"; Proto- Samoic Outlier *sengasenga "yellowish": Branstetter 1977). light of color, faded (PPN *kefu "reddish, of hair"). yellow, of eye or skin (cf. mata "eye"; "to have the appearance of"). yellow medicinal grease from Samoa (prob. an introduced term; cf. PPN *rengarenga "yellow"). 84 Qualification or Modification of Nanumean Color Terms There appear to be three ways in which color terms are quali- fied or modofied in Nanumea: by the addition of a modifying term; by reduplication; or by compounding or conjoining two color terms. These forms of modification parallel those observed by Dougherty (1975) in the Polynesian outlier West Futuna. By far the most common method of qualification in Nanumea is to add an adjective after the color term itself. Interestingly, the only qualifiers noted which are color specific are applicable only to WHITE, BLACK and RED, the three categories which Berlin and Kay found to be encoded first in all languages: WHITE tea tea tangihi, "very white"; tea can also take either hinahina or mahavinavina in same postposed modifier position. All are intensifiers. BLACK uli uli tamumu "jet black" (cf. Bellonese tamumu "reddish", Kuschel and Monberg 1974:224). RED kula kula mii "very red." Three additional qualifying adjectives referring to hue, brightness or saturation are not restricted in use to any particu- lar colors: maina bright; e.g., puluu maina "bright blue." maamaa light, of value, weight; e.g., puluu maamaa "light, pale blue." mau deep, fixed, permanent; puluu mau "deep, dark blue." Finally, there are general qualifiers which can be used in reference to colors, but which do not refer specifically to color quality. These qualifiers are used more commonly in non-color situations. kii very, e.g., kula kii "very red." tai sort of, e.g., tai kula "sort of red." hee loko not very, e.g., hee loko kula "not very red." tinaa pure, true, e.g., tinaa kula "pure or true red." 85 tonu true, correct, e.g., tinaa n-gau tonu "pure, true blue." maalosi strong, -er, e.g., tai uli maalosi "a bit blacker." maalie slightly, e.g., tai kula maalie "slightly reddish." Reduplication is common in Nanumean generally. With color terms reduplication normally signifies a less than "pure" color, removed from the standard focal hue: thus uli BLACK, but uliuli and tai uliuli "blackish," and similarly tea WHITE, but teatea, or tai teatea "whitish." Interestingly, however, reduplication can also signify intensification of the color: thus, for one informant an ORANGE object was named as kulakula ("off red"), while in another instance a pure RED was described as kulakula mii "very very red." Finally, in the case of the borrowed term sama or sama- sama ("yellow"), the single and the reduplicated form appear to be equivalent in meaning. It may be that the inherent semantic ambiguity of reduplica- tion for Nanumean color terms inclines users to add the unambiguous specifier tai "sort of, almost" to reduplicated forms when the intent is to indicate a non-focal color. Thus tai kulakula could only indicate "sort of reddish," while kulakula might refer to a focal red. There is at least one unusual case of the shortening of a formerly reduplicated form: felofelo "yellow," heard in old songs, is not used in this form today. Instead one hears ffelo or fefelo, which are apparently identical in meaning. This shortening of a basic color term may be an instance of Zipf's "Law of Abbrevia- tion," which posits that frequently used lexemes become progressive- ly phonemically shortened (Durbin 1972). Like reduplication, compounding of color terms occurs when the color is non-focal, novel or unknown. Examples of compound re- sponses we received are: launiu uli "dark, blackish green" for naming task object 23, a dark green flecked with black. kula pa uli "red very black" for object 30, a rose red. kaakii pelaa kula "brown that is reddish" for object 19, a reddish brown. kaakii tea "white brown" for object 14, a grey. 86 kaakii kula, kaakii kelo "red brown," "blond brown" for object 19, reddish brown. kaakii uli "black brown" for object 20, a greyish blue. The Salience of Color: Polynesian Studies Studies of color classification and semantics in Polynesia have been limited. Rivers' pioneering work in Oceania (discussed in Berlin and Kay 1969) dealt not with Austronesian speakers, but with Torres Strait peoples culturally similar to native Australians. Apart from some meager information on color terminology available in ethnographies of Polynesian societies published in the 1930s (e.g., Kennedy 1931 for Vaitupu, Beaglehole and Beaglehole 1938 for Pukapuka, and Beaglehole 1939 for Tonga), there has been little interest in color research in Polynesia until the past decade. Berlin and Kay's original (1969) sample of ninety-eight languages included just three from Polynesia. More recent field research (e.g., Snow 1971 for Samoa, Monberg 1971 for Tikopia, Kuschel and Monberg 1974 for Bellona, and Dougherty 1975, 1977 for West Futuna) has broadened knowledge of Polynesian color systems, as has Bran- stetter's (1977) valuable comparative study, which draws on dictionaries and other lexical sources to reconstruct Proto- Polynesian color terminology. Interestingly, recent field studies of Polynesian color systems have generally stressed the low level of salience of the domain of color for Polynesians. Thus, Dougherty (1975:27) notes that in West Futuna there is no single term for "color," that color terms are seldom used in natural conversation, and that colors seem to be seen as specific properties of objects in contrast to abstract qualities which can recur in various contexts. Commenting on Vaitupu, in Tuvalu, Kennedy (1931:102) felt that "for all prac- tical purposes, except in the making of pearl shell bonito lures, colour played a very small part in the mental activities of the race." And Kock (1965), remarking on a general "color indiffer- ence" among Polynesians, pointed out that fragrance often seems to eclipse color as a salient characteristic of experience. To these remarks must be added those of one of my informants (number 1), at the beginning of one interview: We don't use many color names. We compare things to various leaves at various stages of growth. We also mix terms and say things like 'sort of reddish and sort of darkish too' (tai kulakula, tai uliuli hoki). Yet the abundant data from Nanumea would seem to at least partially contradict these statements, since the domain of color is at least of nominal significance in daily life. Color is certainly an 87 abstract quality, and not particularly "contextualized" and limited to specific natural objects as Kuschel and Monberg (1974) assert for the Polynesian outlier Bellona. Nanumea has an abstract term for color (lanu), which can be and is used both with basic color terms and with secondary and context bound terms. And color terms are used in enough natural descriptive contexts, as the following examples show, to indicate at least the relative salience of the domain. For instance, in chants used by fishermen to goad fish into biting, taunts are used. One of these is palu kelo! ("pale/blond palu fish," Ruvettus sp.). To be blond, kelo, is not desirable in Nanumea. Takuo o tai, tou tala felofelo! ("tuna of the ocean, your yellow dorsal spines!") is another such taunt. The same chant for calling tuna includes the archaic and partly unintelligible phrase ko tapiki uli, ko tapiki tea . . . ("your black , your white . . \) Similarly, a popular faatele (dance and song) extolls a pearl- shell lure, calling it taku paa samasama ("my yellow lure"); in another song a lure is fondly described as "my black-tailed lure" (taku paa maiuli). One type of decorative design for a woven mat is said to be patterned after the brilliant colors of the uloulo fish, a type of parrot fish with red stripes on green. The mat design is called matauloulo ("face of the uloulo"). Color is also frequently used descriptively where one speaker is ignorant of the object being described. Thus, the example already cited where a type of coral I had not seen was reported to be kula in color. A type of Pandanus fruit which is distinctively red, and good for making flower wreaths, was described to me as kulakula mii "very very red." Folktale narrators utilize color terms, as this example, describing a small spider-like creature, shows: "ko te manu kamone telaa, e pelaa e fefelo . . ." "it's that small animal, it's sort of yellow. These examples--and many more could be adduced--serve to demonstrate that color terms, and the comain of color generally, are drawn on frequently in daily life in Nanumea. Comparative information now available on color systems in Polynesia also allows the remark cited above on the relative sali- ence of the domain of color to be placed in broader perspective. Branstetter's (1977) compilation of data on twenty-four Polynesian languages shows that the majority have Stage III systems (in Berlin and Kay's 1969 schema), while some are at Stage IV or higher. In all of the languages she surveyed there are, in addition to basic terms, a large number of secondary color terms and modifiers. In the original world sample of 98 cultures examined by Berlin and Kay (1969), only a third (33) were found to possess a Stage V or higher color classification system. It would thus appear that, while the 88 domain of color in some Polynesian societies may be less cognitive- ly or psychologically salient than it is in the Western societies most researchers have come from, the claim that Polynesian societies are "color indifferent" remains undemonstrated. Until further data are available, it seems best to accept that the domain of color in Polynesian societies is apparently of comparable sali- ence to the same domain in other small scale world societies. If, as has been suggested (e.g., Berlin and Kay 1969, Naroll 1970, Hays et al. 1972, Bolton 1978), the growth of color categorization and terminology indeed shows a positive correlation with degree of cultural and technological complexity, then efforts should be directed to understanding the causes leading to such development. Conclusion This survey of color classification in Nanumea has provided data which support the hypotheses of Berlin and Kay (1969) and revisions suggested by Hage and Hawkes (1975), Berlin and Berlin (1975), Kay (1975) and Dougherty (1975, 1977). Specifically: (1) Nanumean basic color categories conform to the universal color foci originally posited by Berlin and Kay. According to their classification, the Nanumean system is Stage V, with six basic terms: BLACK, WMITE, RED, YELLOW, GREEN and BLUE. (2) The foci of Nanumean basic and "incipient" basic colors are in close accord with the predictions of Berlin and Kay, and conform closely to the composite mapping of color foci from 24 societies presented in Dougherty (1975:206-208). These foci, and those from Nanumea, are presented in Figures 13 and 14. (3) While evidence for internal synchronic variation in Nanumea is limited due to the small sample size, the order of acquisition of color categories (here following Kay and McDaniel 1978 in considering the process to be one of successive differenti- ation of categories, rather than merely encoding of foci) in Nanumea appears to be following the evolutionary sequence posited by Berlin and Kay (1969). No basic categories have been acquired out of the order predicted. Four "incipient" basic categories are at present in various stages of progress toward apparent universal acceptance in Nanumea: BROWN, PURPLE, PINK and ORANGE, listed here in order of apparent cognitive salience. For the Berlin and Kay schema to hold, BROWN must appear before PURPLE, PINK or ORANGE; the Nanumean data for BROWN indicate that this color is indeed closest of the incipient categories to complete acceptance. (4) There is internal linguistic evidence of another type which also supports the Berlin and Kay hypothesis of the sequence of encoding color terms. Durbin (1972) has suggested applying the ideas of George K. Zipf to this problem. According to Zipf, with increasing frequency of use words become abbreviated in length. Dubrin speculated that, if Berlin and Kay's evolutionary sequence 89 RED YR YELLOW GY GREEN BG BLUE PB PURPLE RP 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 1v11111E YELSI I I I I I I I I I I I I I I _ _1 I I I PIK I 1 1 T -HGT I I I T 1TR 1TTT TT 1 I TItB _ _ RETIITT11TE I llIlL_ _ __ FIGURE 13. Universal foci from 24 1975:207). Hatching indicates basic non-basic categories. languages (after Dougherty categories, dots represent I R G H T N E 10 9 8 7 6 5 4 3 2 RED YR YELLOW GY GREEN BG BLUE PB PURPLE RP 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 5 10 p I N K I~~~~~~~~~~~P N 111111x11111111111111111RI11E11N _R El E____:Za PI T UPIL I I FIGURE 14. Foci of the basic Nanumean color categories (hatching), and the non-basic categories (dots). Predominant foci of the non-basic categories are indicated by heavy dots. GREY WHITE GREY 10 9 8 7 6 5 4 3 2 1 / I R I G H T N E S s BLACK GREY WHITE L ACK 90 held, then terms for BLACK and for WHITE should be phonemically the shortest color terms in any given language, followed in order of increasing length by terms for RED, GREEN, YELLOW, BLUE, BROWN and then PURPLE, PINK, ORANGE and GREY. Considering the Nanumean lexi- con with this hypothesis in mind, the terms for BLACK and for WHITE, uli and tea, are indeed the shortest in the Nanumean inventory. RED and YELLOW follow, with four phonemes each (kula, ffelo/sama). GREEN and BLUE (lanu launiu and lanu moana) are both lexemically complex terms, and apparently more recently derived. All the other terms are recent borrowings from contact languages. Again, the Nanumean data lend support to the Berlin and Kay thesis of the temporal ordering of the acquisition of color terms. More recent studies of non-Western color classification systems have led to some modification of Berlin and Kay's original position. Thus Kay (1975:260), drawing on revisions suggested in numerous recent studies, suggests that the encoding sequence be revised to the following: WHITE [GRUE] [yellow] green purple [RED] and brown pink Li [yellow] ~~~~~blue Iorange BLACK [Yellow] [GRUE] grey Stages: I II IIIa/IIIb IV V VI VII The revision now postulates that, instead of GREEN being encoded before BLUE, an undifferentiated GRUE category (GREEN and BLUE) will emerge, with its focus in either green or blue. A further revision of the sequence has been suggested by Dougherty (1975:213), after intensive work with West Futanese color classification. Examining additional evidence from Stage I systems in New Guinea, she suggests that the original black/white opposition should be one of opposition between LIGHT-WARM (high brightness and long wave length hues) and DARK-COOL. She also posits that, instead of RED at Stage II, a combined RED and YELLOW category is encoded (actually, separated from the other light and warm hues), which she calls ROW. The next step is either the separation of ROW into RED and YELLOW, or the separation of GRUE from the DARK-COOL category. The full sequence is diagrammed on the following page. Another suggested revision is that GREY can be encoded at any point from Stage III to VII. The Nanumean data lend only slim support to these revisions in the encoding sequence. There is now no evidence for an original LIGHT-WARM/DARK-COOL opposition, although of course there may have once been such a situation in Nanumean. Kennedy's remarks for Vaitupu, already quoted above ("it would seem that the only true colour-distinctions are between white or light-coloured and black or dark-coloured") tend to lend support to the idea of an original 91 LIGHT-WARM [GRUE] RED +1 PINK [ROw] L:LJIGREEN1 [BROWN] PURPLE [RED + 1 BLUEJ _DARK-COOLJ YELLOWFI [GRUE IORANGEJ I GREY Stages: I II III IV V VI VII a&b light/dark distinction, at least for southern Tuvalu. But the contemporary situation on Nanumea is one of virtual total restric- tion of uli to BLACK or darker shades of GREY. There is some support for the idea that RED and YELLOW were once a unitary category for Nanumeans in the occasional extension today of the RED term to include YELLOW. But for the majority of speakers today the two categories are entirely separate. It is with the GRUE category that the Nanumea data are most at odds with the revised encoding sequence. On the basis of internal variation in the application of a label for BLUE, and the lexemic complexity of both the BLUE and GREEN terms, these categories appear to have been encoded relatively recently, and after RED and YELLOW. And yet there appears to be no evidence for the prior existence of an undifferentiated GRUE category. It would seem reasonable to expect that if there had been a GRUE which has only recently separated into GREEN and BLUE, some lingering indication might be found, i.e., some informants would fuse the two categories, or that a term for GRUE might still be known. Yet neither lanu launiu nor lanu moana ("color of coconut leaf" and "sea color"), as clear of reference as they are, could have served as a label for an undif- ferentiated GRUE. It is possible then that either the Nanumean data provide evidence of an exception to the encoding sequence, and that GREEN and BLUE were encoded separately, not from a composite GRUE cate- gory, or that the composite GRUE once existed, but both the original term and its referent have been lost to Nanumeans after the encoding of a separate GREEN and BLUE. The derivation of Nanumean GREEN and BLUE categories remains somewhat of a puzzle, and any conclusions regarding a former GRUE category and term in Nanumean must await further research utilizing a larger sample of informants. Branstetter's (1977) findings regarding the process of naming new categories are relevant here: she has shown that for some Polynesian languages when new categories are labeled the new names are drawn from former secondary terms, or even from old basic terms formerly used. It is possible that a similar process may have occurred in Nanumea. 92 Finally, it should be stressed that the data presented here are to be considered a first step toward a complete description of color categorization in Nanumea. The sample size is small, a fact which precludes meaningful conclusions on the interesting question of internal variation within what I have characterized here as the Nanumean "system." It is possible that further research will reveal that nothing quite so unitary as a system is present, and that in fact Nanumeans are at various stages of color lexicon development. A useful extension of the present work would also include more emphasis on color nomenclature and concepts for objects occurring naturally in the Nanumean environment. Thus the colors and patternings of living fish, the varieties of hue and pattern in plants, the variation in color of the coconut husk, and so on, ought to be given greater attention. In this sense Kuschel and Monberg's (1974) call for attention to "contextualized colors" is justified. Despite these caveats, present data show that the domain of color in Nanumean is one that appears to be basically analogous to the domain as conceived in Western society. Color terminology refers to color in its measurable, abstract sense. Nanumeans do not, apparently, extend their terms to connote non-colorimetric information as has been reported for the Hanun6o (Conklin 1955), or the Polynesian speakers of Bellona (Kuschel and Monberg 1974). Colors are quite easily conceived and discussed apart from the objects they characterize, and the Nanumean basic color categories belong to the group of eleven universal categories described in the work of Berlin and Kay (1969) and subsequent researchers. Notes Field research was carried out on Nanumea over a 17-month period from June 1973 to January 1975. I was accompanied by my wife Anne, who was also carrying out doctoral research in anthropology. We collaborated in most of our research areas, including the present topic, and have shared our notes freely, so that this paper is based on data gathered jointly. I am also indebted to Anne for her critical comments and ideas since the fieldwork period. I also wish to thank Ralph Bolton and Janet Wynne Dixon Dougherty for their comments on an earlier version of this paper, and Brent Berlin for airmailing a color array board to the central Pacific. In Nanumea itself many people assisted us in investigating the domain of color, but I must thank particularly Salailoto Make, Memelisa Valo, Laina Teuea, Talofa Haleti and Laumata Tealei. Support during fieldwork was provided by Training Grant GM-1224, National Institutes of Health (National Institute of General Medical Services). 2Informants who completed the listing, naming and mapping tasks are listed below. Although the small size of the sample precludes any real discussion of intracultural variation based on the sample 93 alone (but see below for considerations of variation based on other evidence), it is felt that the consistency shown in results of the various tasks is an index of the reliability of those results. Still, in light of the extensive intra-cultural variation reported for another Polynesian society (Dougherty 1975, 1977), it seems likely that a larger sample, particularly one including a more heterogenous mix of older persons, would reveal some variation in color classification among Nanumeans, and that this variation would be likely to be related to age and experience. Investigation of such variation must await future research. Informants: 1. Male, 24 years old. Bilingual in Tuvalu and English. Edu- cated to mid-secondary level, which included seven years in a church-run boarding school on the southern Tuvalu island of Vaitupu. This informant used some Vaitupu forms in his speech, and in many cases was unable to distinguish these from Nanumea variants even when they were pointed out to him. No specifically Vaitupu color terms were known or used by this informant. 2. Female, 17 years old. Monolingual, with local island primary school education. Minimal experience away from Nanumea, al- though had spent one year on neighboring Nanumanga. 3. Female, 41 years old. Monolingual, primary school education. Entire life, with exception of two years on Banaba, spent on Nanumea. 4. Female, 34 years old. Bilingual, Tuvalu and Gilbertese, as result of having lived for six years in the Gilbert Islands. Primary school education (Nanumea). 5. Female, 21 years old. Monolingual, primary school education. No experience living away from Nanumea. 3As noted below, questions regarding color usage and instructions for the various tasks were relatively straightforward. Since Nanumean possesses an abstract term for color (lanu), it was possible to phrase all instructions will a minimum of circumlocu- tion. These instructions and questions were: Listing task: "List for me the names of all the colors you know" (Fakaholo mai ingoa o lanu katoa e ke iloa). Naming task: (with household items) "What color is this thing?" "What is this color?" (E lanu fea te mea nei? He aa te lanu nei?) Mapping task: "Try to enclose all colors which belong to the color name . Make a line around the color 94 name . If it is , it lies inside the line, if it is not it lies outside." (Taumafai o faka puipui a lanu katoa e kau i te lanu _ Kae fai e , e moe iloto i te laina, kae fai e hee , e moe itafa.) Foci identification: "Where is the pure , the very purest ? Make an 'x' on the correct . (Tehea te tinaa , tinaa ailoa? Fai he 'ekis' i te tonu.) 4Interestingly, there is a degree of color term evidence in favor of the Nanumean claim for T'ngan origins. Nanumean color terms (presented below) are closely cognate to Tongan terms (Branstetter 1977, Beaglehold 1939), while relatively dissimilar to Samoan terms (Branstetter 1977, Snow 1971). 5The Nanumean language features both phonemic double consonants and vowels, indicated here by doubling the letter. Double consonants are occasionally in free variation with reduplicated forms (as in the synonyms ffelo and fefelo), apparently indicating an ongoing process of change involving loss of medial vowels in reduplicated forms. To avoid confusion the double consonant ngng is indicated by a bar, as ng 6Proto-Polynesian (hereafter PPN) derivations are from Biggs 1979, and Branstetter 1977). The asterisk * preceding Proto-Polynesian and other forms indicates that they are reconstructions. 7Since the Nanumean reflex of PPN *s is /h/, lexical forms incor- porating /I/ appear to be recent borrowings. 8While it is not possible here to draw conclusions regarding the possible division of BLUE into a light and a dark range, this development has been posited as currently in process for some Stage VII color lexicons (see for instance, Berlin & Kay 1969, Harkness 1973, Bolton 1978, Kay and McDaniel 1978). 9It should be noted that, while not considered here, there have been other recent proposals for further refining of the encoding sequence, either through simplification (e.g., Witkowski and Brown 1977), or by specification of alternative choices at each stage (e.g., Kay and McDaniel 1978). The neurological and psychobiologi- cal foundations of color naming and salience are the subject of much interest as well (e.g., Bornstein 1975, Ratliffe 1976, Ember 1978, Von Wattenwyl and Zollinger 1979). 95 References Cited Alkire, William H. 1978 Coral Islanders. Arlington Heights, Ill.: AHM Publish- ing Company. Bayard, Donn 1976 The Cultural Relationships of the Polynesian Outliers. Otago University Studies in Prehistoric Anthropology, Volume 9. Beaglehole, Ernest 1939 Tongan Colour-Vision. Man 39:170-172. Beaglehole, Ernest and Pearl Beaglehole 1938 Ethnology of Pukapuka. Honolulu, B. P. Bishop Museum (Bulletin no. 150). Berlin, Brent and Elois Ann Berlin 1975 Aguaruna Color Categories. American Ethnologist 2:61-87. Berlin, Brent and Paul Kay 1969 Basic Color Terms: Their Universality and Evolution. Berkeley, Calif.: University of California Press. Biggs, Bruce 1979 Proto-Polynesian Word List II. Working Papers in Anthro- pology, Archaeology, Linguistics, Maori Studies, University of Auckland, no. 53. Bolton, Ralph 1978 Black, White, and Red All Over: The Riddle of Color Term Calience. Ethnology 17:287-311. Bornstein, Marc H. 1975 The Influence of Visual Perception on Culture. American Anthropologist 77:774-798. Branstetter, Katherine B. 1977 A Reconstruction of Proto-Polynesian Color Terminology. Anthropological Linguistics 19:1-25. Chambers, Anne F. 1975 Nanumea Report: A Socio-Economic Study of Nanumea Atoll, Tuvalu. Wellintgon: Victoria University Department of Geography. Chambers, Keith S. n.d. Manuscript Collection of Nanumean Traditional Narratives. Chambers, Keith S., Anne Chambers, and Doug Munro 1978 Sapolu, S. Percy Smith and a Tale from Nanumea. Journal of the Polynesian Society 87:29-40. 96 Conklin, Harold C. 1955 Hanun6o Color Categories. Southwestern Journal of Anthropology 11:339-344. Connell, John 1980 Tuvalu: Independence or Dependence? Current Affairs Bulletin 56(9) :27-31. [Department of Adult Education, University of Sydney] Dougherty, Janet Wynne Dixon 1975 A Universalist Analysis of Variation and Change in Color Semantics. Ph.D. dissertation (Anthropology), University of California, Berkeley. 1977 Color Categorization in West Futunese: Variability and Change. In Ben G. Blount and Mary Sanches, eds. Socio- cultural Dimensions of Language Change. New York: Academic Press, pp. 103-118. Durbin, Marshall 1972 Basic Terms--Off Color? Semiotica 6:257-278. Ember, Melvin 1978 Size of Color Lexicon: Interaction of Cultural and Biological Factors. American Anthropologist 80:364-367. Hage, Per and Kristen Hawkes 1975 Binumarien Color Categories. Ethnology 14:287-300. Harkness, S. 1973 Universal Aspects of Learning Color Codes: A Study in Two Cultures. Ethos 1:175-200. Hays, David G., Enid Margolis, Raoul Naroll and Dale Revere Perkins 1972 Color Term Salience. American Anthropologist 74:1107-21. Kay, Paul 1975 Synchronic Variability and Diachronic Change in Basic Color Terms. Language in Society 4:257-270. Kay, Paul and Chad K. McDaniel 1978 The Linguistic Significance of the Meaning of Basic Color Terms. Language 54:610-646. Kennedy, Donald Gilbert 1931 Field Notes on the Culture of Vaitupu, Ellice Islands. New Plymouth, N.Z.: Polynesian Society, Memoirs 9. Kock, Gerd 1965 'Farbenindifferenz' bei Pazifischen Volkern. In Carl A. Schmitz and Robert Wildhaber, eds. Essays Presented to Alfred BUhler. Basler Beitrage zur Geographie und Ethnol- ogie, Ethnologische Reihe, Band 2:235-242. 97 Kuschel, Rolf and Torben Monberg 1974 'We Don't Talk Much about Color Here': A Study of Color Semantics on Bellona Island. Man (n.s.) 9:213-242. Luomala, Katharine 1953 Ethnobotany of the Gilbert Islands. Honolulu, B. P. Bishop Museum (Bulletin no. 213). Macdonald, Barrie 1975 Succession in Defence of Identity: The Making of Tuvalu. Pacific Viewpoint 16:24-44. Milner, G. B. 1966 Samoan Dictionary: Samoan-English, English-Samoan. London: Oxford University Press. Monberg, Torben 1971 Tikopia Color Classification. Ethnology 10:349-358. Naroll, Raoul 1970 W7hat Have We Learned from Cross-Cultural Surveys? American Anthropologist 72:1227-1288. Pawley, Andrew 1966 Polynesian Languages: A Subgrouping Based on Shared Innovations in Morphology. Journal of the Polynesian Society 75:39-64. Ranby, Peter In Press A Nanumea Vocabulary. Pacific Linguistics. Ratli f f, Floyd 1976 On the Psychophysiological Bases of Universal Color Terms. Proceedings of the American Philosophical Society 120(5) :311-330. Snow, David L. 1971 Samoan Color Terminology: A Note on the Universality and Evolutionary Ordering of Color Terms. Anthropological Linguistics 13: 385-390. Von Wattenwyl, Andr4 and Heinrich Zollinger 1979 Color-Term Salience and Neurophysiology of Color Vision. American Anthropologist 81:279-288. [Wilson, John F.] 1978 Tuvalu Achieves Independence. Commonwealth Law Bulletin 4:1003-1009. Witkowski, Stanley R. and Cecil H. Brown 1977 An Explanation of Color Nomenclature Universals. American Anthropologist 79:50-57.