Anthropological

Linguistics

An Introduction

William A. Foley
University of Sydney

CEtmollE INVESTIGACIONES
YESTUDIOS SUPE~IORES EN
ANTROPOLOG/A ¡OCW.
aIBL.HDTi¡CA

13 BLACI<WEU
Publishers

·n0670(ff'
 7
Color

The Neurophysiology of Color

U ndoubtedly the most influential and possibly the most robust elaims of
uni versal innate constraints 00 the semantic strucrure of certain cognitive
dornaios have beeo marle in the area of color terminology, starting with the
landmark study of Berlin and Kay (1969) and extended with subsequent
work by them and associates (Berlin and Berlin 1975; Kay 1975; Kay,
Berlin, and Merrifield 1991; Kay and McDaniel 1978; MacLaury 1987,
1991, 1992). The erosslinguistie study of color terminologies has beeome
something of the paradigm case for demonstrating the effeets of universal
innate biological cons traints 00 human categorizations of th e world. Color
has long beco a favored semantic dornaio in which to investigare issues of
the relationship between language and thought (Brown and Lenneberg
1954; Lantz and Stefflre 1964; Lenneberg 1953; Lenneberg and Roberts
1956; Stefflre, Castillo Vales, and Morley 1966). This is a tradition that
Berlin and Kay (1969) build on, but erueially they are attempting to show
universal constraints in this dornaio, rather than relativistic effects linked to
language differenees, as is the aim of the earlier work. The whole thrust
of work in color terminologies stemming from Berlin and Kay (1969) is to
demonstrate thar universal design [eatures of the human visual perceptual
system strongly eonstrain the systems of color terminologies found in the
world's languages to a very small and largely predietable subset of the very
large set of theoretieally possible, but aetuall)' unattested, types:

basic color categorics can be derived directly from the ncural response pat-
terns that underlie the perception of color . (Kay and McDaniel 1978:130)

predicting the composite color categories ofthe \Vorld's languages from prop-
erties of color vis ion that are independent of culture and of langua ge, biolog-
¡cal properties which are in fact independent ol' human experience per se being
widespread in genera other than Horno. (Kay, Herhn, and Merrifield 1991:18)
 Color \51

Thus, universal constraints in color categorization are directly based in

(primate) neurophysiology, and this is reflected in the color- naming systems
found in the world's languagcs. Cultural practices and human interests,
according to this view, play no role in the actual sensible experience covcred
by a given basic color rerm in a language, as this is informed strictly by
biological constraints. This, of course, is a claim already made familiar by
the work in ethnobiological classification and kinship airead y discussed, but
ir can be made more strongly here and tested more rigorously, given the
clearly restricted perceptual basis of th e domain and the greater knowledge
\Ve have of the human visual system, particularly the physiology of color
visiono Ethnobiology, by contrast, is a much mOfe open-ended dornaio, with
probably many mOfe perceptual fcatures relevant to a given c1assification
than just color, and not even the most extreme upholder of the pivotal role
of genealogy in the reckoning ofkinship relations would argue that this could
be reduced to simple perceptual universals. Of these three dornains, then,
color is quite unique and provides a particularly good arena lO study the ef-
fect ofuniversal innate constraints, biological ones ofhuman phY5iology, on
human categorization as revealed in linguistie systems of basic color terms.
Before eonsidering the actuallinguistic and categorization results of this
work, it is best to surnmarize what is presentIy known about color and the
physiology ofhuman color "ision (Davidoff 1991 ; Thompson, Palacios, and
Varela 1992). The physiology of human vision is constant across aH races
and populations of present-day members of the genus Hamo, provided, of
course, there is no individual pathology. AH of the colors that we see are a
eombination of six basie colors: red, yellow, grecn, blue, white, and black.
For example, turquoise is a combination of blue and green; orange, of yel-
low and red. Perceivablc color varies along three dimensions: hue, saturation,
and brighrness. Hue i5 the "coloredness" of acolar, ¡ts redness, ycllowness,
greenness, or blueness. These are the fundamental hues, defined as opposi-
tions of red to green and blue to yellow. Combinations are possible across
these oppositions, but not within them, as typically within a binary opposi-
tion, one poJe excludes the other. Thus, turquoise is a combination of grecn
and blue (across the oppositions), but there is no hue that i5 a combination
of yellow and blue. Not all colors have hue; white and black do not, nor do
their intermediate shades of grey. Colors with hue are known as chromatic
colors; those without hue, achromatic colors. Saturarion defines the strengrh
ofhuc within a given color. Saturated colors have vivid hues, while desatur-
ated colors are like pastels, closer to grey. Finally, brightness indicates the
light reflectance of a color, from dazzling to barely visible.
Wh y is color perceived along these dimensions? Simply because the
human visual system is structured in such a way as to revcal these dimen-
sions, aeeording to the now widely accepted opponent-proeess theory, pro-
posed in modern form by Hurvich and Jameson (\957), but traceable back
152 Universalism: Imzale COllstraints 01l Mind

10 nineteenth-century work by Hering. In this theory, the human visual

system consists of three subsystems. The first subsystem signals differences
in brightness and is achromatic. The other two signal differences in hue; one
for the red--j¡reen opposition and one for yellow- blue. The relation between
these subsystems and actual neuronal connections is nor yet settled, bu!
someching like che foUowing is generally accepted. The retina contains cluslers
of cone ceUs which respond to contrasts in "hue," by differential responses
according lO lhe wavelengths of lhe received lighl. There are lhree cluslers
of cone ceUs: long wave, middle wave, and short wavc. The output of the
lhree cone types is reorganized al neurological levels higher than lhe retina
so lhal lheir signals can be addilively or subtractively compared. The differ-
ence between the signals from rhe long- and rnedium-wave receptors con-
stitutes the red-green subsyslem, while lhe difference belween the sum of
the signals from long- and medium-wave receptors and the signals from the
Shorl-wave constitutes the yellow- blue subsystem. The achromatic bright-
ness subsystem is the result of the summed aClivit)' of the long- and medium-
wave cones. As rhe tWD chromatic subsystems are made up of oppositions,
arre pole always excludes rhe other; an increase in blue is always at rhe
expense ofyellow. This system of oppositions explains in neurophysiological
reTms the difference between pure hues and secondary "mixed" colors. Pure
bluc results when the yellow- blue subsystem signals "bluc" (i.e. whatever
neuronal firing pattern that realizes "blue" is occurring) and the red-green
subsystem is neutral, signalling neither "red" nor "green." Turquoise, on
the other hand, results when the yellow- blue subsystem again signals "bluc,"
but the red-green subsystem signals "green." Purple, in like manner, is the
result of "blue" from the yel1ow-blue subsystem and "red" from the red-
green. Such secondary, "mixed" colors are thus cognitively "computed"
from inputs from the two subsystems (this process is modelled formally in
Kay and MeDaniel (1978)) and are not the responses 10 neuronal stimula-
tion in only one subsystem, as are the pure, primary opposing hues of the
two subsystems. Slight differences in hue of, say, turquoise reflect differ-
ences in the relative contributions of "blue" and "green" froro the two sub-
systems. Finally, white, black, and grey result when both che yellow- blue
and red- green subsystems are neutral, and the third subsystem of bright-
ness is opera ti ve; neurological-firing patterns for high brightness signals
"white" and its relative abscncc, "black," with "grey" in between.

Color Categorization

With this background in the neurophysiology of human color vislon in

mind, let us return to issues of anthropological linguistics, speeifieally the
constraints this physiology has upon the color categorization rcflccted in
 Color 153

languages. Al] of this work in universals of color categorization inaugurated

by Berlin and Kay (1969) uses the Munsell set of color chips, a set "of 320
color chips of forty equally spaced hues and eight degrees of brightness, all
at maximum saturation, and nine chips of neutral hue (white, black and
greys)" (Berlin and Kay 1969:5). These represent the controlled stimulus
for the required and evaluated responses of color naming. Speakers of
languages are asked ro provide the basic color term for each stimulus chip,
and to date well over 100 languages have been investigated using this meth-
odology, with from two (Oani) to eleven (English) basic color terms. A basic
color term is defined on the basis of a number of criteria such as: (1) it is
monolexemic, not composed of composite parts, excluding, for example,
bluish in English; (2) is not included, hyponymically, within another color
term, excluding, for example, English sear!et, which is a kind of red; (3) is
attributively not restricted, excluding English blond which is restricted 10
hair and wood; and, perhaps most problematically, (4) is salient psycholo-
gically, fOI example, listed first among rerms in the given dornain or most
widely known. Cornmonly, these critcria converge, but in individual cases
they may conflict, requiring additional criteria or creative decisions on the
part of the analyst.
Using these criteria to establish basic color terms and investigating their
usage as names for particular stimuli of colored chips, noticeable patterns
emerge. First, it is generally the case that regardless of the number of color
terms in a lan guage, the focal hue', the best exemplar of a named color, is
remarkably consistent across languages. Thus, a speaker of a language with
three basic color terms like Watam \ViII identify as Ihe best exemplar of
"red" about the same hue as will speakers of English of eleven basic color
terms, essentially "fire- engine red. n The boundaries, however, of what hues
count as "red" is much less stable, but surprisingly, within languages (i.e.
variation among speakers of the same language), as much as across them.
These prolOtypical effects are c1aimed by Kay and McOaniel (1978) to lie
in Ihe universal neurophysiology of color visiono Membership in the cat-
egory labelled by "red," for example, is established by Ihe proportion of
"red" response in the oeuroos of the red-green subsystem. Pure or focal
red occurs when the yellow- blue subsystem is neutraL Similar definitions
apply 10 the other focal hues: yellow, green, and blue. Less focal hues have
more partial responses in their own subsystem and greater or less contri bu-
tion from Ihe other subsystem. This neurophysiological description accounts
for our perception of primary colors through what Kay and McOanie! call
"fundamental neural responses," but is extendable to secondary, "mixed"
colors. For example, those hues labelled "purple" by English speakers are
the intersection of the neurophysiology underlying "blue" and "red," i.e. a
"blue" pattern of neuronal firing in the yellow- blue subsystem and a "red"
pattern in the red- greeo subsystem. In this way, the neuronal and cognitive
154 Universalism: bmate COl1straints on Mind

underpinnings to all the colors subsumed by the eleven basic color teTms of
English can be modelled. Seeond, universal panerns of eolor-naming sys-
tems emerge across languages. No language, for example, has a word for
"grecn," unless it also has a word for "red." T he maximum number ofbasic
color terms seems fixed at around a dozen, but a language may have much
less. If so, a given color term will be a composite category, covering a range
of stirnulus colors, for example, the native Trobriand color terminology has
four terms (Senft 1987); one of these digadegila eovers the speetrum range
of yellow-green-blue, although its prototypieal foeus seems to be "yellow"
(Senft 1987:329). The patterns of sueh eomposite eategories seems largely,
but not eompletely, predictable on neurophysiologieal grounds.

Types of Basic Color Terrrlinologies

The simplest system of color namiog found in any language consists of

two teTms (considering the basis of the co lor dornain in oppositions, it is
impossible 10 eoneeive of a simpler system and still eall it "color"). This is
exemplified in Papuan languages of New Guinea like Oani (Heider 1972a)
and Australian languages, sueh as Burara (Jones and Meehan 1978). There
1S sorne debate as to whether the contrast in such systerns is truly one of
color (hu e), rather than brightness. Oani has been extensively studied by
Roseh (Heider 1971, 1972a, b). T his has two basie color terms - mili and
mola: mili contains black and darker browns and aH the cDoler colors, greens
and blues; mola covers white and the warrner colors, reds, yeHows, orange,
reddish-purple, pink and lighter browns, like European skin colors. The
foci of these color terrns \Vere highly variable across speakers: for mili rang-
ing from blaek 10 pure blue or pure green and for mola, burgundy, red ,
pink, brown, or pastel shades. Sorne speakers chose pure black and pure
white as the foei for mili and mola respeetively. On the basis of these
findings, Roseh argued that these terms denoted both hue and brightness,
glossing them as mola OARK /COOL and mili LIGHT /WARM, where
COOL and WARM deseribed the eool (green, blue) and warm (red, yellow,
orange) hueso Interestingly, in spite of Dani's ooly having two basic color
terms, Rosch has been able to demonstrate prototype focal effeets for a
range of hues lacking unique names in the language. Made-up names for
focal hues can be learned more rapidly and recalled more easily than those
[or more peripheral anes. Thus, made-up nam es for pure greeo and pUTe
blue are more accessible thao for turquoise (i.c. "mixed" blue--green), in
spite of the faet that Oani lacks unique basie color terms for all three of
these. This is evidenee for Kay and MeOaniel's (1978) claim for the uni-
versal innate basis for focal colors in the human color vision system.
Jones and Meehan's (1978) evidenee suggests that brightness is indeed
 Color 155

the basic dimension of contras! for the t\Vo BUfara basic color rerms, gUl1galtja
and gUlIgulldja. Gll1lgaltja refers to light, bright colors, like white, pastels,
and red. Its focal prototype is reflective aluminum foil, certainly suggesting
high brightness as its basic feature. AU other colors are covered by gltllgttlldja.
The BUfara system, then , is a more transparent brightness-based two-term
system [han the mixed brightness-hue system of Dani. In fact, MacLaury
(1992) suggests two paraUel paths of development of basic color terms, one
based on hue and ane on brightness, with crossover relations between them.
Burara is probably the most likely example of a basic color system grounded
in brightness contrasts, Dani exemplifying the more common crossovcr
between the t\VD dimensions.
When a language has thrce basic color rerms, the warm colors red-
orange--yeUow split from LIGHT /W ARM to form the foUowing three-
way contrast: LIGHT, WARM, and DARK/COOL. A true hue contrast
now definitely emerges. The focal hue for LIGHT is white, and for WARM,
red but DARK/COOL continues to have variable foci in black, pure blue,
or pure grecn. Watam is a good example of this type of language, with
wawar LIGHT: white, greyish, 1Ilbllk1llbllk DARK/COOL: black, dark
brown, green, blue, andyallp WARM: red, orange, yeUow, red-brown . This
early emergence of RED fits weU with Sahlin's (1976) claim of its psycho-
¡agical salience and emotional attractiveness.
With four basic color rerms, the possibilities now become quite complex,
with no less than five-attested sysCems (Kay, Berlin, and Merrifield 1991).
The two most common of these were uncovered as early as Berlin and Kay
(1969). In one, the WARM category, previously covering from red through
yeUow, but focused on red, splits into two distinct terms: RED and YEL-
LOW. This system is found in traditional, pre-Spanish Bisayan of the
Philippines (Berlin and Kay 1969:68): mabosas LIGHT / WHITE, maitu1Il
DARK/ BLACK, mapllla RED, and madurag YELLOW. In such systems
orange is divided between RED and YELLOW.
The other commonly attested four-term system separates the cool colors,
grecn and blue, [rom the DARK/COOL category, resulting in DARK/
BLACK and GRUE, a composite category made up of green and blue. In
such systems the focal hue for GRUE is either blue or green or bifocal in
both) never turquoise) i.c. the secondary "mixed" hue composed of these
two primaries. T his again is said to vindicate Kay and McDaniel's (1978)
universal neurophysiological account for basic color terminologies. Ibibio of
Nigeria is one such four-term system: afíá DARK/BLACK ébúbít LIGHT /
WHITE, Mdíddt WARM, awawa GRUE/focus in green (Berlin and Kay
1969:64).
The other three systems are more problema tic. In one, GRUE separates
from DARK/COOL and RED emerges as a distinct category from W ARM,
but "yellow" rejoins the LIGHT category. A neurophysiological account
156 Universalism: Innate COl1straints on Mind

for this is no! apparent, nor is a diachronic mechanism which would yie1d
this four-term system from the previous three-term one ofDARK/COOL,
LIGHT, and WARM. Most problematic ofall are systems which have as a
basic color term a YELLOW /GREEN composite category, in which YEL-
LOW again splits from WARM, but this time merges with GRUE. How
can YELLOW be simultaneously LIGHT/WARM and DARK/COOU
Once previously thought to be rare, YELLOW /GREEN is now attested
on four continents: Asia (Hanunoo), Australia and Oceania (Gugu-Yalanji,
Trobriand), North America (Salish, Wakashan, Creek, Natchez, Karok),
and South America (Chacobo). There are two types of such systems. One
has a YELLOW- GRUE category and is exemplified by the traditional
Trobriand color terminology (Senft 1987): pupwakau LIGHT /WHITE,
bwabwau DARK/BLACK, bweyani RED, and digadegila YELLOW /GRUE.
Other languages with this system inc1ude Karok (Bright 1952) and Gugu-
Yalanji (Berlin and Kay 1969:70; Rivers 1901). This grouping of YEL-
LOW /GRUE poses formidable problems for Kay and McDaniel's grounding
of generalities of basic color terms in ¡nnate perceptual properties of rhe
human color vision system, specifically, the subsystems based on opposing
colors. Yellow and blue (member of the composite GRUE) are opposing
poles of the same subsystem and if these subsystems based on oppositions
are rhe universal grounding for human color categorizations, ir is .hard to see
how yellow and blue could be conflated in a single named category. This
serious problem is c1early seen by MacLaury (1992), who tries to solve it by
claiming this category is actually informed by a brightness dimension, flot
one of hue, these co10rs occupying a range of middle brightness. This is in
keeping with his postulation of the pivotal role of a brightness dimension in
the sequence of development of basic color terms, but it rather does con-
travene the strict Kay and McDaniel universalist interpretation in terms of
hue, and poses the question what favors the mediation of the brightness
dimension in addition to hue in certain color terminologies. Culture?
The other four-term system involving YELLOW /GREEN actually di-
vides GRUE into BLUE and GREEN, merging the former with BLACK
and the latter with YELLOW. Hanunoo of the Philippines (Conklin 1964)
is one such system: bi:ru DARK/BLACK/BLUE, lagtiq LIGHT /WHITE,
raraq RED, latuy YELLOW /GREEN. Other languages with this system
are Creek (Kay, Berlin and Merrifield 1991) and Shuswap (MacLaury 1987).
Given that these languages assign yellow and blue to different named cat-
egories, they do not pose the quite same problems for the universal neurophy-
siological basis of basic color terms in color oppositions like yellow-blue;
they do, however, pose problems of their own, such as how would they arise
from a previous three-term system ofDARK/COOL, LIGHT, and WARM?
Again the role of yellow is pivotal; it must migrate out of the WARM cat-
egory to join with the green focus of GRUE, a composite category already
 Color 157
1
separate from DARK/COOL. Thus, the ancestor of four-term systems
with YELLOW /GREEN may be the common type with the four terms
LIGHT/WHITE, DARK/BLACK, WARM and GRUE, as in Ibibio.
MacLaury (1987) proposes something like this for the history of the color
terminology of Shuswap, a Salish language of Canada in which the WARM
term extended to the green focus of GRUE (due to an operative brightness
dimension?) and simultaneously retracted fram red. Blue rhen merges wit~
DARK/BLACK. Halkomelem, another Salish language (MacLaury and
Galloway 1988), shows the opposite trajectory: GRUE is extended to the
yellow focus of WAR¡v¡, isolating red and simultaneously retracting from
blue, which again merges with DARK/BLACK. While not ruled out on a
neurophysiological account of opposing color hues, yellow- blue and red-
grecn, such an account cerra¡nly provides no straightforward explanarian
of such migrations of yellow or green. MacLaury's (1992) theory of a co-
emergent brightness dimension, with YELLOW /GREEN a category of
medium brightness, seems ro offer the only avenue of explanarian here.
Languages with five basic color rerms exhibir fewer possibilities rhan
those with four, there being three-attested types. The most common arise
from rhe [\VD most \Videl)' attested four-rerm systems, eirher by splitting
WARM into RED and YELLOW or by dividing DARK/COOL into
DARK/BLACK and GRUE, again with variable focus of GRUE in either
blue or green, or bifocal in both. AIl such languages ha ve DARK/BLACK,
LIGHT /WHITE, RED, YELLOW, and GRUE and are illustrated by
Tzeltal, a Mayan language ofMexico (Berlin and Kay 1969:82); 7ink' DARK/
BLACK, salk LIGHT /WHITE, cah RED, k'an YELLOW, andyafGRUE.
Uncommon, but attested, systems inc\ude those in which only GREEN is
separated from DARK/COOL, leaving blue to remain in the latter along
with black: DARK/BLACK/BLUE, LIGHT /WHITE, RED, YELLOW,
and GREEN. Chinook Jargon, a pidgin previously spoken in the Pacific
north-west of the United States and Cana da, is an example of this five-term
system (Berlin and Kay 1969:75): 1/7et DARK/BLACK/BLUE, Ik 71lp
LIGHT /WHITE, p,1 RED, ptcJh GREEN, and kawakawak YELLOW
(with pale greens). And there is at least one reported language (Kay, Berlin,
and Merrifield 1991) in which green migrates from GRUE to merge with
yellow to form a YELLOW /GREEN composite category, isolating BLUE
as a new uniquely named color category: DARK/BLACK, LIGHT /
WHITE, RED, YELLOW/GREEN, and BLUE. Kwak'wala (Saunders
1992), a Wakashan language of Canada, may also illustrate this last sys-
temo zulla DARK/BLACK, lIlela LIGHT /WHITE, tlakwa RED, ¡henza
YELLOW /GREEN, and zasa BLUE. .
The possibilities for systems of basic color terms from the minimum of
t\VD members up to five can be secn in 7.1.
Beyond tive basic color terms, there appears to be far less arder. Berlin
158 Universalism: lnnale ConstrainlS on Mind

7.1 DARK/COOL: Grecn (G), Blue (Bu), Black (m) 2 rerms

L1 G IIT / W'\R.M" Red (R) Vello\\' V) White (W
I
:DARK/ COOL: G, Bu,
WARM: R, y
BU¡3 tcrm s
I L1GHT/ W

I 1-+ tcnns
DARK/ COOL: G, Hu , Bl DARK/ B1
R GRUE: G , Bu
y WAR.!\oI: R, y
L1GHT/ W LlGHT / W

1
DARK/ COOL: Bu, BI DARK/ BI DARK/ B!
R GRUE: G, Bu y IGRUE: G, Bu
Y/ G R R
UGHT/ W LlGI-IT/ W / Y L1GHT/ \V
I I
DARK/ BI
GRUE: G, Bu
R 5 tcrms
y
LlGHT/ W

DARK/COOL Bu, BI DARK/BI

G
R
y
""
Y/G
R
LIGHT/ W LIGHT/ W

and Kay (1969) in their early phase of the theory proposed the strong con-
straint that GRUE necessarily splits into GREEN and BLUE, followed by
the addition of BROWN, and then lhe emergence of PURPLE, GREY,
PINK and ORANGE in any order, but later research has shown this to be
false. Evidence above shows that GREEN and BLUE may already emerge as
distinct in a five-term system. It is also now weJl established that, contrary
to Berlin and Kay (1969), GREY, BROWN, and PURPLE appear as basic
color terros in no fixed order. Indeecl, GRUE may persist undissolved into
GREEN and BLUE well after GREY, BROWN, or PURPLE have been
labelled wilh their o\Vn basic color terms (Kay, Berlin, and Merrifield 1991).
It is important to note that MacLaury (199 1,1992) has found that prior
to their emergence as basic color terros both purple and brown show con-
flicting categorizations in different languages. Thus, in a typical five-term
system, purple may be assigned complete1y to the GRUE category or to the ~
boundary between GRUE and RED. Similarly, brown can be included in the
YELLOW or DARK/BLACK category. Variable categorizations like these
 Color 159

may be the souree of the largely unpredietable panern for their emergenee
as basie color terms. PI K and ORANGE do generally appear to be dis-
tinetly late, though see Saunders (1992:151) for a claim that ORANGE is a
basic color term in Kwak'wala, otherwise a language with a five-tcrm sys-
tem, as mentioned aboye. Finally, sorne languages may have mOfe (han one
basic color rerm corresponding to a single ane in English: for example,
Russian has two basie color terms for BLUE, gO/IIboj for LIGHT -BLUE
(i.e. "sky-blue") and sinij for OARK-BLUE, and Hungarian has two words
for RED, piros LIGHT-RED and voros OARK-REO (Wierzbieka 1990).

Universal Constraints on Basic Color Terrrlinologics

Mueh of the early enthusiasm in the researeh tradition inspired by Berlin

and Kay (1969) lay in a belief that this work uneovered strong universal
constraints in ane linguistic and cultural dornain, contrary ro [he reccived
wisdom of relativism current in (he mid- 1960s. However, as sketched aboye,
(he curren! findings in this field represen! a significant rerrear from rhe
early strong universals proposed in Berlin and Kay (1969), with multiple
possibilities now available to languages in their basic color terminologies
espeeially in four- and five-term systems. Still, it is equally apparent that
constraints are operative, for nor all possibilities are attested: for exarnple,
there is no four-term system with ORANGE, LIGHT /WHITE, YEL-
LOW, and GRUE. And for three-term systems only one possibility exists:
OARK/COOL, WARM, and LIGHT / WHITE. Why not, COOLlGRUE,
OARK/ BLACK, and LIGHT /WARM? In other words, why does WARM
split from LIGHT befare COOL splits from OARK? Phrased in a different
\Vay, why does the presence of a distinet COOL/GRUE term always imply
the prior ernergence of a WARM terrn, but not vice versa? Such questions
dernand answers and do argue for universal constraints on the structur-
ing of this linguistic dornain. The SQurce of these constraints is quite pos-
sibly neurophysiological, to be found in the strueture and funetioning of
the human color vision system, but Kay and McOaniel's (1978) aeeount
in terms ofHurvieh and Jameson's (1957) opponent-proeess theory of color
vision, with systerns of opposing hues, yellow- blue and red- green, is clearly
not eompletely adequate, failing especially in the case of languages with
the eomposite eategory YELLOW /GREEN/BLUE, like Trobriand (Senft
1987). The neurophysiologieal aecount requires opposing colors in the same
subsystem, like yellow and blue, to be in separate labelled categories, yet
c1early they are noto The future suecess of the neurophysiologieal explana-
tion of findings of this researeh tradition may founder on sueh cases. As
the opposing hue hypothesis is what is eontradieted here, MacLaury (1992)
tries to accouot for such systerns in terrns of brightness, arguing that this
160 Universalism: bmate Conslraints 011 Mind

represents a second universal, neurophysiologically based, dimension in color

terminologies. This move appears necessitated by the data, but if it should
turn to be unmotivated, then the findings of this research tradition will no
longer be supported by a neuroph ysiological base. And if this is the case,
what is the SOUTce of the discovered constraints on basic color terminologies?
Culture? This, contrary to the whole intent of this tradition, right back to
Berlin and Kay (1969), opens the door to relativism, a point not lost on
MacLaury (1992: 137):

Inserting a category of ye/low- green- blue ioro the sequence cas[s daubt upon
[he presumed conncerion betwcen panhuman visual physiology and the widely
observed regularities of color categorization; ir calls into question the vcry
notioo of color- category univcrsals.
Since the yellow-greell- bluc category represents the extreme, dropping it
into the universal sequence concedes the debate to the relativists .

Relativist Responses to Proposed Universal s of

Color Terminologies

So what is the relativist response to this imposing body of work and how
do they account for the universal constraints on basic color terminologies
already mentioned? It is perhaps best articulated in Sahlins (1976), but
see also Lucy (1996), Saunders (1992), and Tornay (1978). The basic point,
of course, is that cultural practices are a crucial mediating force in color
naming and the systems of basic color terms. They argue that culture must
be the crucial autonomous intermediary between any innate and hence
universal neurological perception of co lor stimuli and the cognitive under-
standing of these. This point is echoed linguistically by Wierzbicka (1990)
who notes that the meaning of a color term in a language cannot possibly be
a neural response to a color chip, but rather the cognitive understanding the
Native speaker of the language has of that term: "language reflects what
happens in the mind, not what happens in the brain" (Wierzbicka 1990: 163).
The basic thrust of the relativist critique of the Berlin and Kay (1969)
inspired tradition is to invert its determinism: "it is not, then, that color
terms have their meaning imposed by the constraints of human and physica1
natures; it is that they take on such constraints in so far as they are mean-
ingful" (Sahlins 1976:3), i.e. meaningful in a culrurally constructed sym-
bolic system; symbols for practical public action, as in Geertz (1973). Herein
lies the rub; Berlin and Kay (1969) and associates regard a particular color
as a label given in response to a controlled stimulus, a Munsell color chip,
an act of naming an objective sensible difference. The language of color,
thus, is reduced to a nomenclature of objective pure color referents in a
controlled sensible world . The basic color terms in a language are separated
 Color 161

semantically froro other words denoting color on essentially this principIe:

they denote color and nothing eIsc, whereas secondary color rerms have
additional denotations and eonnotations. But to relativists like Sahlins (1976)
and Luey (1996), it is exaetly this striet separation whieh is at issue. And
where is the meaning of the color terms in all this? The meaning of a color
term is its cognitive understanding, the culturally defined relations ir en-
gages in and activates, not its mere recognition and labelling. Color terms in
a given culture do not mcan Munsel1 chips. And from this paint of view,
there is no basis foc a semalltic separation of basic and secondary color terms.
The relativist critique thus rejeets the strietly referential/behaviorist theory
of meaning, that words are simply labels foc perceived stimuli, implicit in
the methodology of Berlin and Kay (1 969), and subsequent work. This
assumes an objective pre-given reality which language simply labels. In the
tradition of basic color terminology, this pre-given reality is the separate
domain of color, embodied in the contrastive dimensions of the Munsell
color chips. Relativism challenges this assurnption and clairns that any such
dornain is not a pre-given, easily isolatable piece of reality) but is cultur-
ally constructed, seamlessly linked tú other parts of that culture's symbolic
arder and meaningful practices. By isolating color as the only dimension in
a pre-given dornain of reality, Berlin, Kay, and associates create a completely
artificial situation, parallel aeross cultures and languages, whieh, by defil/i-
tion, should result in their universal findings. Combined with their rejec-
tion ofsecondary color terrns and the connotative, non- "color" meanings of
basie color terms in languages, Berlin and Kay's (1969) findings are vi rtually
assured (Luey 1996). Luey (1996) diseusses how different the color termino-
logy of Hanunoo looks from its presentation in Berlin and Kay (1969:64),
if the full detail of Conklin's (1964) ethnographie description is systemat-
ized. Berlin and Kay (1 969) eharaeterize Hanunoo simply as a four-term
system, now known to be bi:ru DARK/ BLACK/BLUE, lagtiq LIGHT/
WHITE, raraq RED and la/l/y YELLOW /GREEN. But Conklin (1964)
provides mueh fuller information. Besides this four-fold contrast in hue/
brightness, "this classification appears to have certain correlates beyond what
is usually considered the range of chrornatic differentiation, and which are
associated with nonlinguistic phenomena in the external world" (Conklin
1964: 191 ). Thus, the four terms eontrast aeeording tú semantie dimensions
other than color ascribed to their typical referents. There are three of these
(Conklin 1964:191):

First, there is the opposition between light and dark, obvious in the con-
trasted ranges of meaning of lagtiq and b'::ru. Second (here is an opposition
between dryness or desiccation and wetness or freshness (succulence) in visible
components of the natural environment which are reflected in the terms raraq
and latuy respectively. This distinction is of particular significance in terms
162 Ul1iversalism: hmate C01lStraints 011 Mind

of plant life. Almost al! living plant types possess sorne fresh , succulent, and
ofren " grecnish" parts. Tu cat any kind of raw, uncooked [Dod, particularly
fresh fruits Of vegetables, is known as pag-laty-ulI «latuy). A shiny, \Ver,
brown-colored section of newly-cur bamboo is maiatuy (llot mararaq). Dried-
out 0 1' matured plant material such as certain kinds of ycllo\Vcd bamboo or
hardened kernels of marure Of parched com are mararaq. Tu hecome desic-
cated, ro lose a1l moisrure, is koown as mamoraq «paraq " desiccation" ... )
... A rhird opposition, dividing (he [\VD already suggested , is rhar of deep,
unfading, indelible, and hence ofrcn more desired material as against pale,
weak, faded, bleached or "colorless" substance, a distinction contrasting mabi:m
and mararaq with malagtiq and malatuy. This opposition holds for manufac-
tured items and trade goods as well as for sorne natural products (e.g. red and
white trade beads, red being more valuable by Hanunóo standards; indig(}-
d yed corton sarongs, the most prized being those d yed most oftcn and hence
of the deepest indigo color - som etimes obscuring completely the designs
fonned originally by white warp yarns; etc.).

Conklin condudes by alternatively glossing the [our terms as DARKNESS,

LIGHTNESS, DRYNESS and WETNESS, respeetiveJy, and notes that
(Conklin 1964: 192):

what appears ro be color "confusion" at first may result from an inadequate

knowledge of the internal structure of a color (sic) system and from a failure
to distinguish sharply between sensory reception on the one hand and per-
ceprual categorization on the orher.

ReJativists like Luey (1996) and Saundcrs (1992) eharge that Berlin and
Kay, and subsequent workers, only get the results they do by bracketing out
such rich cultural information about the mcanings of the basic color terms
and by foeusing solely on the ehromatie information they denote. Luey notes
that adcquate Native knowledge (the goal of any truly cognitive anthropo-
logy) of ¡he Hanunoo system eould never be aehieved by eodifying labelling
responses in the Hanunoo language to color stímuli provided by Munsell
chips. The thrust of the relativist critique ís to daim that basie color terms
stand in meaningful relationships to each other not only according to chro-
matic contrasts, but many other culturally defined dimensions, and, further,
they individually and as a group may enter into relations with many olher
tenns and semantic domains, as part and parcd of the wider meaningful
practices which make up the culture. Colors as perceptual structures are
merely raw materials of cultural production, the handmaiden of meaningful
practices: virginal white versus promiscuous red. Colors, too, are good to
think with. Saunders (1992:219) summarizes lhis view nieely:

1 propose te rdocate colour semantics out of the psychophysical domain

of nomologically necessary "pure" perception into the domain of value, as
 Color 163

[ha! which most fuIly characterises rhe mental (or psychological). Instead
of cvolutionarily ernerging neuro-based epistemological primitives (funda-
mental ncural responses / basic color rerms) 1 propose to Jocate a quality like
Kwak'wala Ihmxa [YELLOW /GREEN] ... in "intenrional spacc" and char-
actcrise it in terros wcdded to ao intentional view of [he wodd, ¡.c. in purposivc,
indexical, and interactional rerms. Then lIun.m ... have rneaning only in [he
normativc setting of [he customs, regular beliefs and desires, and appropriate
behaviour of a Kwakiutl ... community [i.c. embodied practices], as COffi-
ponents of vistas on ly discernible from thcir intenrional points of vicw. Cat-
cgorisation in terms of"colour" or whatevcr, would then be inextricably bound
up wÍth having states of rnind which cannot be understood exccpt by those
who enjoy the appropriate kind of intentional liaison with the world [i.e.
structural coupling), that ¡s, by people who have the skills to use terrns like
Ihenxa . .. mauve, or Pmssian h/ue in the appropriate contcxts - an intentional
liaison only secured when one is part of(or becornes part of) a cornrnunity for
which il is utterly natura.! to idcntify fcatures of the world with these notions.

Exactly paraUel comments are found in relativist critiques (Schneider

1984) of proposed universals of basic kinship term categories discussed in
the previous chapter. Motlzer is nat just callateral, first-ascending generatían
female, contrasting to azmt, daughler and father; she is the bcarer of semantic
dimensions that account for expressions like necessity is the mother of úwell-
!ion; slte mas like a mOlher to me; maybe even tlzat stonll mas a real mother.
Similar questions apply to extensions of the Kannada word for " mother,"
a1l1ma, ro cover "woman," Hgoddcss," "pox," and "help!" (Bean 1981). Only
a culturally sensitive reading, in which particular embodied practical en-
gagements with the world are appropriate, can truly account, a relativist
would claim, for the semantics of basic kinship terms.
If \Ve concede aU this, then, what explanation is available for the uni-
versal canstraints on basic color terminologies unearthed by Berlin and Kay
(1969) and others? Why does WARM always precede COOL/GRUE, for
example? If such constraints reaUy are valid (see Lucy (1996) for sorne
serious doubts), then an explanation is required, and simple neurophysio-
logy, for reasons discussed above with regard to the YELLOW /GREEN/
BLUE compasite category, may not provide a complete answer. Sahlins
(1976) does, howevcr, lacalize the strongest constraints in universals, albeit
culturally mediated, of human experience, underlain by biology. WARM
precedes COOL/GRUE because "red is to the human eye the most salient
of color experiences. At normallight levels, red stands out in relation to all
other hues by virtue of a reciprocal, heightening effect between saturation
and brightness ... Red, simply, has the most color; hence its focal position
in the contrast ofhue to achromaticity (lightness/ darkness)" (Sahlins 1976:4-
5). Sahlins cautions that the fact that the early emergence of WARM is
rooted in biolagically based human experience does not imply, however,
164 Universalism: Innate Constraints on Mind

that WAR.M: means this experience. The meaning of WARM is in the

oppositions it shares with DARK/COOL and LIGHT /WHITE, and these
are culturally constructed. Thus, in a language with a three-term system,
WARM is a difference that makes a difference. Like aU differences, such as
between day and night, W ARM versus DARK/COOL must be perceptu-
aUy discriminable, and WARM emerges early beca use it scores so highly
here on several dimensions. Sahlins (1976: 12) summarizes:

No less than any other code, a system of color meanings must be grounded
in a corresponding set of distinctive perccptual properties. Hence, the natural
corTeJa tes of color words: they comprise the rninimal distinctions on the
ohjcct plane ~ of lightness/darkness, hue/ neutrality, uniqueness/admixture,
and the like - by which differences in meaning are signalled.

The members of a culture ¡nfuse these distinctions with meaning and

employ them in the course of embodied practices in the service of their own
symbolic ends, as they do with phonetic distincrions, such as voiced- unvoiced.
As mediums of symbolic ends, color categories are cultural norions, enacted
in the sense of Varela, Thompson, and Rosch (1991:171): "experiential,
consensual and embodied: they depend upon our biological and cultural
history of structural coupling."

Sununary

The systems of color terminologies among the languages of the world present
a promising case for the establishment of universals in human categoriza-
tion, due to the pan human neurophysiology of human visiono The thrust
of work in the tradition stemming from Berlin and Kay (1969) has becn to
prove exactly this claim, locating established universals in the systems of
basic color terminologies in the mechanisms of human color vision and
arguing further that cultural interests and practices play no role. They have
determined a typology of basic color terminologies varying from a min-
¡mum of two terms up to a maximum of eleven, with the actually attested
systems restricted to a very smaU and mainly predictable subset of the very
large set of theoreticaUy possible, but actuaUy unattested, types, a finding
strongly in support of universal constraints in this dornain. Further, the
focal hue of a basic color term like "red," for instance, remains the sarne
across languages, regardless of whether it comes from a three-term system,
where the term covers the range from red through yellow, or an eleven-
term one, more evidence that universal perceptual biologically based con-
straints are operative. Relativists respond by claiming that this research
tradition gets the results it does almost by definition, by bracketing out a1l
 Color 165

the cultural information in the meaning of color terms, so that only the non-
cultural perceptual components can reveal themselves. They argue that Ihe
meaning of a color term is not a labelling response 10 a color stimulus, but
the full culturally defined relations it engagcs in and activates, its role in the
ongoing social coupling of a people.

Further Readillg

Color has also been much wrinen about. Davidoff (1991), Hardin (1988), Lamb and
Bourriau (1995), and Ottoson and Zeki (1985) are good introductions to the field,
but see Thompson (1995) and Thompson, Palacios, and Varela (1992) for an altern-
arive. There is a large amount of literature in the Berlín and Kay tradition, bu! the
basic saurces are Berlín and Kay (1969), Kay and McDaniel (1978), Kay, Berlin and
Merrificld (1991), and MacLaury (1992). Important relativist responses are Lucy
(1996) and Sahlins (1976).