Beruflich Dokumente
Kultur Dokumente
Black, White, and Red All Over: The Riddle of Color Term Salience
Author(s): Ralph Bolton
Source: Ethnology, Vol. 17, No. 3 (Jul., 1978), pp. 287-311
Published by: University of Pittsburgh- Of the Commonwealth System of Higher Education
Stable URL: http://www.jstor.org/stable/3773198
Accessed: 22/02/2009 08:29
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=upitt.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.
University of Pittsburgh- Of the Commonwealth System of Higher Education is collaborating with JSTOR to
digitize, preserve and extend access to Ethnology.
http://www.jstor.org
II
BLACK
TERMS
WHITEJ
III
IV
GREEN-YELLOW
REDN
-BLUE'YELLOW--GREEN"
VI
VII
PINK
BROWN-- PURPLE
ORANGE
GRAY
288
ETHNOLOGY
argument. Naroll (I970) and Hays et al. (I972), for example, reported results of
malan Spanish and Mam), but also that the continuing evolution of basic color
terms in Mam follows the sequence proposed by Berlin and Kay. Studying
intracultural variability in color lexicons, Dougherty (I974: I3) concluded that
289
Kay 1969: 6) must be more salient than nonbasic color terms, and presumably
the terms found earlier in the evolutionarysequenceshould be more salient than
those found later in the sequence. Berlin and Kay (I969: 6) mentioned three
indicators of "psychological salience." A color term high in salience should
appearat the beginning of color-termlists given by informants,it should exhibit
"stability of reference" with respect to both informants and occasions of use,
and it should exist in the ideolects of all informants.3According to Hays et al.
(I972), salience refers to "prominence" and "conspicuousness."Does relative
salience among basic color terms in fact correspondto the Berlin-Kay evolutionary sequence? In some cases, salience data have supported the evolutionary
order, while in others they have not.
Probably the strongest evidence showing an association between salience
measures and the evolutionary sequence was published by Hays et al. (1972).
Examining frequencycounts of color words in written texts of seven languages
(English, French, German, Hebrew, Romanian, Russian, and Spanish), these
investigatorsfound that the basic color terms used most frequently are those of
supposedgreater antiquityin the Berlin and Kay scheme; the correlationswere
impressivelyhigh (rs = .76 and above). Additionally, Hays et al. showed that
the frequencywith which ethnographersdiscuss specificcolors in their writings
agrees well with the proposedevolutionaryorder.
A series of experiments by Rosch (Heider I97I, I972a, I972b; Rosch I973,
1974, I975) demonstratedthat focal colors are more salient than nonfocal colors,
but, for the most part, her studies failed to support the hypothesis that the
relative salience of focal colors would conform to the predicted evolutionary
sequence. Nonetheless, she did conclude that primary colors tend to be more
salient than nonprimary colors (indicators of salience in her experiments included length of color names, latency of naming, short- and long-term memory,
and ease of learning color names). Rosch's studies, like those by Naroll and
Durbin mentioned above, would supporta two-stagedevelopmentaltheory. The
failure of her data to confirm the Berlin-Kay evolutionarysequence should not
be interpreted,in Rosch's view, as evidence againstBerlin and Kay. She (Rosch
I973:341) states that "there need be no simple relation between order of
individual acquisitionof terms and linguistic evolution. The present study only
representsa case in which a possible evolutionaryorder could have been, but
was not, reflected on the level of individual learning."
Pollnac (I975) asked I02 Baganda informants to list all of the color names
they could think of. The twelve terms at the head of the resultantcomposite list
with respect to frequency of mention and mean rank included all of the basic
color terms designated by Berlin and Kay. But Pollnac (I975: 93) pointed out
that the relative salience of these terms "does not follow Berlin and Kay's
evolutionary sequence."While the Baganda data indeed do not exactly parallel
the Berlin-Kay sequence, it should be noted that there does exist a statistically
significantcorrelationbetween the Berlin-Kay sequence and Pollnac's materials
(absolute frequency of mention, Spearmanrho = .54, p < .05; mean ranks,
Spearman rho = .57, p < .o5). To account for Pollnac's data on chromatic
color terms, again, a two-stage theory might be sufficient since YELLOW,
GREEN, RED, and BLUE (the primary colors) are more salient in the
Baganda list than the other basic color terms.
290
ETHNOLOGY
TABLE 1
Relative'Salience
(Frequencies)
Terms
Color
Terms
SAN JUAN
CHAPA
Berlin-Kay
Color
of Basic
Evolutionary
Ran
Frequency
Frequency
Frequency
Rank
BLACK (Negro)
1.5
37
41
WHITE (Blanco)
1.5
27
26
RED (Rojo)
54
50
YELLOW (Amarillo)
4.5
49
2.5
53
GREEN (Verde)
4.5
49
2.5
52
BLUE (Azul)
38
49
BROWN (Cafe)
11
25
PINK (Rosado)
9.5
19
19
PURPLE (Morado)
9.5
10
14
ORANGE (Anaranjado)
9.5
16
27
GRAY (Gris)
9.5
11
Freq
TABLE 2
Relative
Color
Color
Terms
Terms
Salience
Berlin-Kay
Evolutionary
Evolutionary
Rank
of
(Mean Ranks)
Basic
CHAPA
Mean
Position
Color
Terms
on Listin
SAN JU
Mean
Rank
Mean
Position
BLACK (Negro)
1.5
3.95
4.49
WHITE (Blanco)
1.5
3.78
5.54
RED (Rojo)
2.07
1.98
YELLOW (Amarillo)
4.5
3.45
3.04
GREEN (Verde)
4.5
2.96
2.92
BLUE
4.03
3.37
BROWN(Cafe)
5.27
6.48
PINK (Rosado)
9.5
5.00
6.58
PURPLE (Morado)
9.5
6.13
10
5.93
ORANGE (Anaranjado)
9.5
5.31
5.94
GRAY (Gris)
9.5
7.67
11
10.00
(Azul)
293
TABLE 3
Nonbasic
Color
Terms on the
Listing
Task
FREQUENCIES
Terms
Color
Chapa
Celeste
Blue)
(Light
Plomo
(Gray)
Violeta
San Juan
Total
22
17
39
(Violet)
Marron
(Maroon)
Granate
(Garnet)
(Cinnamon)
(Charcoal)
Mulberry?)
Cherry)
Esmeraldino
(Emerald)
Azul
(Navy
36
35
71
Lila
(Lilac)
Canelo
Carbon
Moral
(Black
Guinda
(Sour
Marino
Blue)
TOTALS
lists (mean ranks). Twenty-four different color terms appear in these 113 lists.
The eleven basic color terms account for 89.6 per cent (668) of all terms
mentioned; thirteen nonbasic terms account for the remaining 10.4 per cent
(7I). (See Table 3 for information on the nonbasic color terms.) With one
exception, the basic color terms are more salient than the nonbasiccolor terms.
Indeed, nine terms were mentioned by fewer than four informants.
Two color categories merit our attention. First, informantslabeled GRAY by
two terms, gris and plomo. Only one informant mentioned both terms; thus, it
would seem that these terms serve as alternativenames for the same focal color.
In my analysis,I have enteredgris as the basic color term becausegris is not also
the name of an object whereasplomo may refer either to GRAY or to lead (the
metal). It would have been possible to use data on plomo instead of gris or data
on both terms combined without altering the results. Second, informantsgave
two terms for BLUE, azul and celeste. On the basis of frequencies and mean
ranks,azul must be consideredmore salient than celeste; azul was mentioned 87
times while celeste was mentioned 39 times, and the mean ranks were 3.66 and
294
ETHNOLOGY
attempt to map these two terms in color space, there are differences between
them. Celeste refers to "light blue," while azul can be used to indicate the entire
range of blue hues or it can be used to refer to a more restrictedrange of darker
blue. At times, though not in these lists, celeste is used as a modifier of the term
azul; i.e., azul celeste. It is conceivablethat celeste should be given the status of
basic color terms; it occurswith greater frequencyin our lists than do BROWN,
PINK, PURPLE, and GRAY. Our data for Peruvian Spanish parallel the
findings reported by Berlin and Kay (I969: 35) for Russian and possibly for
other Slavic languages with respectto BLUE. In Russian, siniy is the term for
"dark blue" and goluboy for "light blue," with the former term being more
salient than the latter. A similar finding is reported by Harkness (I973) for
Guatemalan Spanish in which celeste also occurs as a possible basic color term.
In the following analyses,I have not included celeste as a basic color term, but
the fact that it is so salient in our data should lead to the consideration of
designating it as the twelfth basic color term. If celeste were included with the
basic color terms, then fully 96.7 per cent of the data in our lists would be
accounted for by basic terms.
The results of the tests of the hypothesis of association between the Berlin
and Kay evolutionarysequence and the salience of basic color terms in the data
derived from the listing task are presented in Table 4. It should be noted that
there are statistically significant correlations,as predicted, for all measures of
salience in Table 4. Absolute frequenciesof mention as well as rank position
within the lists correlatewith the Berlin and Kay sequencefor the data analyzed
separatelyfor each researchlocation and for the combined data. An analysisof
the first term in the lists (see Table 5) produced one of the highest levels of
association (rs = 0.77); in fact, 112 of the first color terms were basic color
terms, the only exception being a single occurrenceof celeste at the head of a
list.
Additional supportfor the hypothesisis to be found in the analysispresented
in Table 6. By tabulating the number of occurrencesof a given term preceding
another basic color term in these lists (e.g., RED before BLUE) and then
comparing that number with the number of reversalsof those terms (BLUE
before RED), it is possible to calculate whether one term is significantlymore
salient than another term. Of 47 predictions of differences on these paired
comparisonsbased on the Berlin and Kay theory, 25 were confirmedat the 0.05
probability level, and only five were shown to be significantlydifferent in the
opposite direction. This distributionitself (25 significanthits versus five significant misses) is highly significant(Binomial Test, p < o.oooI). Moreover, if the
"wrong" color term comes first, less space exists between it and the paired term
than exists when the "right" color term comes first (Binomial Test, p <
0.0033).
The most prominent deviations from our predictions involve the relative
salience of BLACK and WHITE. In the lists in Tables I and 2 and in the results
reportedin Table 5, it can be seen that BLACK and WHITE are not as salient
as they should be. They tend to be less salient than the four primary colors
(RED, YELLOW, GREEN, and BLUE), whereasaccordingto the theory they
should be the most salient color terms. I shall offer an interpretationof this
TABLE4
Intercorrelation
Matrix of Relative
Salience
Variables
Berlin-Kay
Total
Total
Frequence
Evolutionary
Sequence
Salience
Variables
CORRELA
4
.70
.62
.73
.93
.97
->
_--
Sequ
.96
--_
Ranks
Ranks
Ranks
Color Term
N = 11.
Based on Spearman rank-order correlations.
p < 0.025; 0.709 for p < 0.01; 0.818 for p < 0.005;
Critical
values of r :
(one-tailed
test).
0.5
ETHNOLOGY
296
TABLE 5
Relative
Salience
for
(Frequencies)
the First
of Basic
Color
Terms
Task
Term on Listing
TERMLISTED FIRST
Color
Terms Color
Terms
Frequency
Frequency
Rank
BLACK (Negro)
WHITE (Blanco)
RED (Rojo)
57
YELLOW (Amarillo)
16
GREEN (Verde)
15
BLUE (Azul)
10
BROWN (Cafe)
PINK
9.5
PURPLE (Morado)
9.5
ORANGE (Anaranjado)
9.5
GRAY (Gris)
9.5
(Rosado)
finding below. A second area of discrepancybetween the theory and our data
involves the term for BROWN. BROWN should be more salient than it
appearsto be; in our data BROWN is essentiallyindistinguishablein salience
from the other nonprimarycolors (PINK, PURPLE, ORANGE, and GRAY).
On the whole, I think it is reasonable to conclude that the data from the
listing task provide substantial support for the Berlin and Kay formulation.
Although the fit between theory and data is not perfect, the relative salience of
color terms in these lists does correspond to expectations derived from that
formulation.
Relative Salience in SubsistenceDomains
The listing task generates valuable data for measuring the psychological
salience of color terms, but such data tell us little about the relative salience of
color termsgenerally in a culture. Naturalistic methods of measuringcolor term
salience in terms of normal use patterns are needed to supplement the more
experimentallyoriented approachrepresentedby the listing task. The analyses
that follow attemptto deal with salience in context; they are based on materials
gathered not with the topic of color terms in mind, as was the case with the
listing task, but ratherin connectionwith investigationsfocusing on a numberof
297
TABLE 6
Probabilities
Associated
Terms in Paired-Comparisons
Salience
of Listing
of Basic
Color
Task Data
PROBABILITIES
Color Terms
BLACK (Negro)
WHITE (Blanco)
RED (Rojo)
YELLOW(Amarillo)
GREEN (Verde)
BLUE (Azul)
BROWN
PINK
PURPLE
----
.001
.077
.035
.078
.001
.001
.060
.001
.022
n.s.
---
.001
.001
.001
.001
NP
.047
NP
---
---
---
10
11
.015
n.s.
.044
n.s.
n.s.
n.s.
.032
n.s.
.001
.001
.001
.061
.001
.001
.001
.001
.061
..038
.001
.001
.001
.001
.061
----
.006
.002
.001
.001
---
(Cafe)
---
(Rosado)
ORANGE (Anaranjado)
11
GRAY (Gris)
NP = No Prediction;
NP
----
(Morado)
10
. n.s.
n.s.
.061
n.s.
NP
NP
NP
NP
---
NP
---
n.s.
= not significant;
underlining
= opposite
predicted
direction.
Binomial
Test.
298
ETHNOLOGY
TABLE 7
Relative
Salience
II,
(Frequencies)
and III
of Basic
Color
Terms on Tasks
I,
POTATOTASKS
Color
Terms
Berlin-Kay
r T nayE
Evolutionary
I
Frequency
II
Frequency
Rank
Frequency
III
Frequency
Rank
Frequency
Frequency
Rank
BLACK
(Yana)
1.5
84
55
48
WHITE
(Yurah)
1.5
106
87
107
RED (Puka)
113
73
54
YELLOW (Q'ellu)
4.5
6.5
15
GREEN
4.5
9.5
13
10.5
(Gomer)
BLUE (Azul)
27
BROWN(Sani)
PINK (Rosado)
9.5
9.5
10
7.5
PURPLE
9.5
9.5
7.5
11
10.5
33
(Morado)
ORANGE [none]
9.5
GRAY (Oqe)
9.5
0
29
9.5
4
6.5
44
terms, again mostly in Quechua basic color terms. Finally, informants were
shown examples of fifteen potato varieties and asked to name them and respond
to a series of questions about them; e.g., color of the skin, color of the pulp,
numberof eyes, size of plant, and optimal productionconditions (Task III). On
this task, responsesincluded 286 color terms, once more, with few exceptions,
Quechua basic color terms. The frequenciesof occurrenceof each basic color
term are given in Table 7 for each of these interview tasks.
Overwhelmingly, BLACK, WHITE, and RED are the most salient colors;
they account for 86.3 per cent, 72.2 per cent, and 73.1 per cent of all color terms
elicited on Tasks I, II, and III respectively.GRAY, BLUE, YELLOW, and
BROWN are of lesser prominence, while GREEN, PINK, PURPLE, and
ORANGE are mentioned rarely,if at all. A combinedsalience score was created
for the domain of potatoes by adding the frequenciesof color terms on all three
tasks (see Table 8). As shown in Table 9, all of these indicators of salience
correlated significantly with the Berlin and Kay evolutionary sequence and
among themselves are highly intercorrelated.There are two major differences
between the predicted ordering and our data. First, GRAY is more salient than
it should be, and second, GREEN is less salient than it should be. I shall discuss
these discrepanciesfrom the theory in the comparativesection of the paper.
Color terms in the potato domain tend to be associatedwith objects in the
following manner:
Color of the skin:
BLACK, WHITE, RED, YELLOW, BROWN,
GRAY, BLUE, PURPLE, PINK, GREEN, GARNET, and MAROON;
Color of the pulp:
WHITE, YELLOW, RED, BLACK, BLUE,
PURPLE, GRAY, GREEN;
299
TABLE 8
Relative
Salience
of Basic
(Frequencies)
Domains:
Potatoes
COMBINED
POTATOTASKS
Color
Terms
Frequency
Color
HAWKES'POTATOLIST
Frequency
Rank
Frequency
GUINEAPIGS
Frequency
Rank
Frequency
Frequency
Rank
BLACK (Yana)
187
46
55
WHITE (Yurah)
300
44
49
RED (Puka)
240
62
30
10
YELLOW (Q'ellu)
GREEN (Gomer)
BLUE (Azul)
45
BROWN (Sani/Chumpi)
13
PINK (Rosado)
9.5
15
14
8.5
10
36
9.5
10.5
PURPLE (Morado)
10
ORANGE [none]
11
10.5
10
106
8.5
23
GRAY (Oqe)
In an effort to verify these results with data in the domain of potatoes from
elsewhere in the Andes, I examined a list of potato names published by Hawkes
TABLE 9
Intercorrelation
Matrix
of Relative
Evolutionary
Salience
Variables
Berlin-Kay
Potato
Task I,
Potato
Task II,
Potato
Task III,
Hawkes' Potato
Combined Potato
Guinea Pigs,
Evolutionary
Ranks
Frequency
Frequency
List,
Tasks,
Frequency
Sequence
Frequency
Salience
for Basic
and Kay
Sequence
CORRELATION
COEFFICIENTS
4
5
.69
.70
.65
---
.94
.85
--
.88
Ranks
---
Ranks
Frequency
Variables
Ranks
Frequency
.83
.69
.60
.81
.94
.66
.86
1.00
.60
.73
.91
.46
---
.85
.42
---
Ranks
.58
-
Ranks
Critical
(one-tailed
values of rs:
test).
0.536
for
p < 0.05;
0.618
for
300
ETHNOLOGY
30I
TABLE 10
Relative
of Basic Color
Salience
(Frequencies)
in Four Domains of Expressive
Culture
RITUALS
Color Terms
BLACK
(Negro)
WHITE (Blanco)
Frequency
FOLKSONGS
FOLKTALES
Frequency
Rank
Frequency
Frequency
Rank
Terms
Frequency
Frequency
Rank
137
TRUCKNAMES
Frequency
22
Frequency
Rank
35
27
39
17
71
26
RED (Rojo)
3.5
3.5
16
YELLOW(Amarillo)
12
5.5
9.5
GREEN (Verde)
3.5
60
BLUE (Azul)
3.5
11
10
BROWN(Cafe/Moreno)
10
8.5
PINK (Rosado)
9.5
8.5
9.5
PURPLE (Morado)
9.5
10
12
5.5
9.5
ORANGE(Anaranjado)
9.5
10
10
9.5
GRAY (Gris)
9.5
11
7.5
7.5
Leaving aside references to gold and silver (qori and qollqe, respectively, in
Quechua) which in some instances labeled precious metals rather than colors,
we find 93 instances of color terms in these texts, all but one referringto basic
color categories (the exception is a single mention of color manteca, "color of
lard"). The frequencies and rank ordering of basic color terms in rituals are
noted in Table io. The association between the Berlin and Kay evolutionary
sequence and the relative salience of basic color terms in rituals is strong (r =
0.95). Furthermore,there are no serious deviations between the ritual salience
rankings and the predicted ordering.
To test the hypothesis on another domain of expressiveculture, I turned to
two published collections of folktales, most of which had been gathered in rural
communities in southern Peru (Arguedas and Stephan 1957: 93-I60; Miranda
I971: I-I36). The 74 folktales in these collections contain approximately60,ooo
words of text. An English edition of the tales collected by Arguedaswas used in
this analysis; the tales collected by Miranda were published in Spanish. References to colors are relatively rare in these materials (only 84 items) and the
proportion of nonbasic terms is higher in this domain than in the others (I7.8
per cent). The nonbasic terms included "golden" (5), "emerald" (6), "ivorycolored" (I), "maroon" (i), "color of vicunia" (i), and "walnut" (i). The
objects referred to by basic color terms in this and other domains of expressive
culture are contained in Table ii, and information on the frequenciesand rank
ordering of basic color terms in folktales is found in Table io. In this domain,
too, we find a high correlation between the Berlin and Kay evolutionary
sequence and the relative salience of basic color terms (rs = o.9I). And again
we find no significant differences between the predicted outcome and the
folktale data, although it should be kept in mind that several nonbasic color
terms in this context are more salient than some basic color terms.
302
ETHNOLOGY
TABLE 11
Basic
Color
Colors
and Their
in Domains of Expressive
Referents
with
Sample of Objects Associated
in Domains of Expressive
Categories
BLACK
scarf,
sheep, suit,
young girl,
forest,
tarantula,
eyes,
angel,
soul, ear of corn,
fare, eagle,
mountain, homespun, bundle
WHITE
girl,
jacket,
pants,
snow, mountain, fog,
cloud,
message,
feather,
flamingo,
paper,
peach,
Culture
Color Categories
Culture
homespun,
poncho,
field-
ash, eagle,
heart, beer,
bear,
sugar
shade, hair,
eyes, poncho, hornet, whip, laurel,
rose, horse, truck, covering
flamingo,
eagle,
devil,
diamond,
arrow,
RED
YELLOW
hot pepper,
GREEN
bird, arrow,
bowman, beans, eagle,
water, cloak, blanket,
sea, hornet,
handkerchief,.
collar,
onion stem, willow tree, wings of a bird, lemon, papaya,
sweater,
eyes, pointed hood, kindling,
ear of corn
totora reed, grasslands,
automobile,
verbena,
field,
rose, cloak, skirt,
BLUE
light,
BROWN
virgin,
PINK
ribbon,
PURPLE
dress,
ORANGE
(none)
GRAY
suit,
sky,
dress,
flowers,
sea gull,
goose,
hawk, siren,
rose,
flower,
cloud,
lips,
blanket,
bird,
fruit,
rose,
swan, prince,
skirt,
eyes,
automobile,
beard,
lake,
barley
paper
person
stockings,
chicha,
carnation
carnation
hat
303
TABLE 12
Intercorrelation
Matrix
of Relative
Salience
Salience
Variables
Berlin-Kay
Ritual
Evolutionary
Frequency
Sequence
---
Folktale
Frequency
Ranks
Folksong
Frequency
Ranks
for Basic
Color
Sequence
Ranks
Variables
Culture
CORRELATION
COEFFICIENTS
4
3
.95
.91
.88
.72
--
.85
.79
.78
---
.83
.69
---
.57
--
Ranks
N = 11.
Based on Spearman rank-order
correlations.
p < 0.025; 0.709 for p < 0.01; 0.818 for p < 0.005;
Critical
(one-tailed
values of r :
test).
0.536
0.618
for
304
ETHNOLOGY
TABLE 13
Indicators
Berlin
Color Terms
of Basic
Salience
Relative
Berlin-Kay
Evolutionary
Rank
Aggregated
Salience
Score
to the
Sequence
Average
Mean
Score
Aggregated
Salience
Rank
Average
Mean
Rank
BLACK
1.5
1576
2.14
WHITE
1.5
1507
2.71
RED
928
3.14
YELLOW
4.5
362
5.43
GREEN
4.5
568
5.00
BLUE
420
5.50
BROWN
335
7.07
PINK
9.5
98
8.79
PURPLE
9.5
71
10
8.93
10
ORANGE
9.5
60
11
9.58
11
GRAY
9.5
268
7.71
Salience
Evolutionary
Rank versus
Aggregated
Evolutionary
Rank versus
Rank
= .96
p<0.0001
= .97
p<0.0001
salience scores,I gave equal weight to each of the seven domains for which there
was information. These two overall salience scores, one based on absolute
frequenciesand the other on mean ranks, were then correlatedwith the Berlin
and Kay sequence. The results suggest an almost perfect fit between basic color
term salience and the predictions (rs = 0.96 and rs = 0.97, respectively). I
believe that these findings must be interpretedas overwhelmingconfirmationof
the Berlin and Kay theory. Although I did not attempt to measure basic color
term salience in all of the cultural domains in which colors are certain to be
important, I did include a set of extremely diverse domains. It seems unlikely
that the addition of data from other domains (e.g., clothing or other areas of
material culture) could alter these findings to any significant degree. These
results, moreover, underscorethe inadequacyof any approachto salience that is
based on only one or two limited indicatorsof salience. Obviously, effortsshould
be made to replicatethese findings using a similar approachin other cultures.
Pollnac (I975:94) has observed that "different measures of salience may
result in orderingswhich do not exactly parallel Berlin and Kay's evolutionary
sequence for color terms." Our work clearly supportshis observationsin this
regard. Although salience orderingstend to be highly correlated(see Table 14),
TABLE 14
Intercorrelation
Matrix
Terms in Seven
Berlin
Salience
of Relative
Domains
Peruvian
Variable
Berlin-Kay
San Juan/Chapa
Combined Potato
Guinea
Pig Frequency
Ritual
Frequency
Folktale
Frequency
Ranks
Folksong
Frequency
Ranks
Evolutionary
Sequence
Sample Frequency
Tasks
Frequency
Ranks
for
Culture
an
Sequence
Variables
Salience
of Southern
.73
CORRELATIO
3
.69
.5
.31
.0
-
Ranks
.5
Ranks
Ranks
Ranks
correlations.
Based on Spearman rank-order
0.818 for
0.709 for p < 0.01;
p < 0.025;
N = 11.
p < 0.005;
values
of r :
Critical
(one-tailed
test).
3.536
306
ETHNOLOGY
they do differ from one domain to the next, and this raisesfascinatingquestions.
For example, why does the salience of color terms vary according to domain,
and why is salience in some domains more closely associatedwith the Berlin and
Kay sequence than in other domains? In this section I would like to present
some hypotheses and comment on these mattersin relation to the results of the
studies described above.
Part of the answercan be found, I believe, in the constraintsimposed by the
fact that certain colors may be virtually absent in the objectsassociatedwith any
given domain. And, therefore, those colors are not likely to be salient in that
domain. By way of illustrationof this point, I would remind the reader of the
data in the domain of guinea pigs. Color is importantin that domain particularly with respectto the animals themselves, and the set of colors of guinea pigs
is limited. Can our findings on color term salience in the guinea pig domain be
explained by the relative frequencies of guinea pigs of various colors? The
answer is "Yes, to a large extent." As one part of our investigation of this
domain, we (Bolton and Calvin I976) examined the guinea pigs in eight flocks,
recording their colors, and we obtained information on the actual colors of
guinea pigs in several other flocks not personally observed. The correlation
between those figures and the salienceof basiccolor terms in this domain is high
(rs = 0.84); indeed, that correlationis higher than the one with the Berlin and
Kay theory (rs = 0.59). I pointed out above that GRAY and BROWN are
more salient in this domain than they should be; in view of the frequency of
occurrence of guinea pigs with these colors, that discrepancyis easily understood. RED, GREEN, and BLUE were said to be less salient than they should
have been, and this, too, is understandablesince these colors have little or no
importance in relation to the raising and consuming of guinea pigs. This same
interpretation can be applied to the problems encountered in the domain of
potatoes, where GRAY occurredmore frequentlythan expected.Many potatoes
do exist with gray pulp or a gray skin, and these attributes are important in
classifying tubers. But I also noted that GREEN was less salient in this domain
than it should have been. Surely, green is a prominentcolor in plant cultivation.
GREEN may be deviant for several reasons: first, we did not specificallyraise
questions about the colors of plants in our interviews; we did not do so because
this matter had not appearedin preliminary,nondirectiveinterviewsconcerning
the potato domain; second, it may be that green, while conspicuousin nature, is
not significant for the human activitiescarried out in this domain; and third, it
may be that where there is invariancewith regard to a given color, that color
does not become salient (all leaves are green) in spite of its physicalpresencein
objects of a domain. In short, it could be argued that both the environmentand
the use-significanceof objects in a domain may lead to deviations from the
predicted Berlin-Kay sequencein that domain.
Support for this hypothesis can be gained by comparing the closeness of fit
between salience and the Berlin-Kay sequence in the various domains investigated. In descending order of agreement, our findings were as follows:
DOMAIN
rs
Ritual
Folktales
Folksongs
0.95
0.9I
0.88
THE
RIDDLE
OF COLOR TERM
Listing task
SALIENCE
307
0.73
Truck Names
0.72
Potatoes
0.69
Guinea Pigs
o.59
The lowest correlations are found in relation to subsistence domains where
environmentaland use-significanceconstraintsare obvious. The highest correlations are found in the domains of expressiveculture where fewer constraintsof
an environmental or use-significance exist. Apparently where the choice of
objects is relatively free, as in the case of ritual, the evolutionary sequence is
most closely approximated.Recall that only a single deviation from the predicted ordering was found in the domains of ritual, folktales, and folksongs.
Truck names, though, present another problem. In that domain, it was
discovered that WHITE and YELLOW were less salient and BLUE was
slightly more salient than expected. An analysis of truck names (Bolton and
Stuart I975) led us to believe that "power" is the dominant attribute that an
owner attemptsto convey when he names his truck. Therefore, I hypothesized
that colors which expresspowerfulnessare more likely to be chosen as elements
in truck names. Osgood and his collaborators (Adams and Osgood I973;
Osgood, May and Miron I975) have studied the affective meanings of colors.
They found widespreadcross-culturalsimilaritiesin feelings about colors. Their
rank-orderingof eight colors on the potency dimension is given in Table 15. To
TABLE 15
Relative
Salience
of Basic
Osgood's Mean
Potency Score
Ters
Color Color
Terms
of Potency
Potency
Rank
Truck Names
Frequency
Truck Names
Rank
RED
2.2
26
BLACK
2.8
22
BLUE
3.7
GREEN
4.0
WHITE
6.0
GREY
6.1
YELLOW
6.9
rs
= 1.0
p < .008
Note:
Low score equals high potency,
and high score equals low potency.
Potency
scores for four basic color terms are not given by Osgood, May, and Miron (1975),
and therefore
those terms are excluded from this analysis
(purple,
pink, brown,
and orange are the missing colors).
308
ETHNOLOGY
309
within a culture: i) the presence and ubiquity of the actual color in objects
associated with the domain; 2) the significance of a color in providing useful
information for decision-making processes and behavior in a domain; 3) the
nature of the task used to elicit color terms regardinga domain; 4) the affective
meanings being conveyed concerning objects and activities in a domain; and 5)
semantic links between a basic color term and specificsignificant objects. These
factors operate differentiallyto produce distortionsfrom the evolutionaryorder
when individual domains are considered separately, but they seem to have a
minimal impacton the overall assessmentof the relativesalience of color terms.
CONCLUSION
3Io
ETHNOLOGY
NOTES
BIBLIOGRAPHY
Adams, F. M., and C. E. Osgood. I973. A Cross-CulturalStudy of the Affective Meanings of
Color. Journal of Cross-CulturalPsychology 4: I35-I36.
Arguedas, J. M., and R. Stephan. I957. The Singing Mountaineers: Songs and Tales of the
Quechua People. Austin.
Berlin, B., and E. A. Berlin. 1975. Aguaruna Color Categories. American Ethnologist 2: 6i88.
Berlin, B., and P. Kay. I969. Basic Color Terms. Berkeley.
Bolton, C., and R. Bolton. I976. Rites of Retribution and Restoration in Canchis.Journal of
Latin American Lore 2: 97-II4.
Bolton, R., and L. Calvin. I976. Guinea Pigs, Protein, and Ritual. Paper presented at the
annual meeting of the Southwestern Anthropological Association, San Francisco.
Bolton, R., and M. Stuart. I975. A Lover in Every Town: A Study of Truck Names and
Mottoes in Peru. Paper presented at the annual meeting of the American Name Society,
San Francisco.
Conklin, H. C. 1973. Color Categorization. American Anthropologist 75: 93I-942.
Dougherty, J. W. D. I974. Color Categorization in West Futunese: Variability and Change.
Paper presented at the annual meeting of the American Anthropological Association,
Mexico, D.F.
Durbin, M. I972. Review of'Basic Color Terms,' B. Berlin and P. Kay. Semiotica 6: 257-278.
Farfan,J. M. B. I957. Onomastica de Vehiculos: El Chofer en los Nombres de Sus Vehiculos.
Folklore Americano
5: 140-154; 285-286.
Hage, P., and K. Hawkes. I975. Binumarien Color Categories. Ethnology 14: 287-300.
Harkness, S. 1973. Universal Aspects of Learning Color Codes: A Study in Two Cultures.
Ethos i: 175-200.
Hawkes, J. G. I944. Potato Collecting Expeditions in Mexico and South America. II.
Systematic Classification of the Collections. Bulletin of the Imperial Bureau of Plant
Breeding and Genetics. Cambridge, England.
Hayes, D. G., E. Margolis, R. Naroll, and D. R. Perkins. 1972. Color Term Salience.
American Anthropologist 74: 1107-1121.
3"I
Heider, E. R.
97I. 'Focal' Color Areas and the Development of Color Names. Developmental
Psychology 4: 447-455.
i972a. Probabilities, Sampling, and EthnographicMethod: The Case of Dani Colour
Names. Man 7: 448-466.
1972b. The Structure of the Color Space in Naming and Memory for Two Languages. Cognitive Psychology 3: 337-354.
Hickerson, N. P. 1971. Review of 'Basic Color Terms,' B. Berlin and P. Kay. International
Journal of American Linguistics 37: 257-270.
Hill, J. H., and K. C. Hill. 1970. A Note on Uto-Aztecan Color Terminologies. Anthropological Linguistics 12: 231-238.
Kay, P. I975. SynchronicVariability and Diachronic Change in Basic Color Terms. Language
in Society 4: 257-270.
Lenneberg, E. H., and J. M. Roberts. 1956. The Language of Experience: A Study in
Methodology. Memoir 13 of InternationalJournal of American Linguistics.
Magnus, H. 1877. Die geschichtliche Entwickelung des Farbensinnes. Leipzig. Viet.
I880. Untersuchunger iber den Farbensinn der Naturvolker. Jena. Fraher.
Merrifield, W. R. 1971. Review of Basic Color Terms by B. Berlin and P. Kay. Journal of
Linguistics 7: 259-268.
Naroll, R. 1970. What Have We Learned from Cross-CulturalSurveys? American Anthropologist 72: 1227-1288.
Osgood, C. E., W. H. May, and M. S. Miron. I975. Cross-CulturalUniversals of Affective
Meaning. Urbana.
Pollnac, R. B. 1975. IntraculturalVariability in the Structureof the Subjective Color Lexicon
of Buganda. American Ethnologist 2: 89-109.
Rivers, W. H. R. I9oIa. Introduction and Vision. Reports on the Cambridge Anthropological
Expedition to the Torres Straits, ed. A. C. Haddon, Physiology and Psychology 2: I-140.
Igoib. Primitive Color Vision. Popular Science Monthly 59: 44-58.
Observations on the Senses of the Toda. British Journal of Psychology i:
I905.
321-396.
Rosch, E. I973. Universals and Cultural Specifics in Human Categorization. Cognitive Development and the Acquisition of Language, ed. I. E. Moore, p. 341. New York
1974. Linguistic Relativity. Human Communication: Theoretical Explorations, ed.
A. Silverstein, pp. 95-121. New York.
1975. The Nature of Mental Codes for Color Categories. Journal of Experimental
Psychology: Human Perception and Performance I: 302-322.
Snow, D. L. I971. Samoan Color Terminology: A Note on the Universality and Evolutionary
Ordering of Color Terms. Anthropological Linguistics I3: 385-390.