ArticlePDF Available

Abstract

Independent models of basic and nonbasic color terms are combined to explore the domain as a whole. We analyze an elicited set of English color terms, ranked by a combined frequency and order of mention salience index. We find that (1) basic color terms emerge intact as a distinctive category, and (2) the remaining types of terms array themselves in an extended pattern of distribution with respect to the domain's broader scheme of lexical, organizational, semantic, and salience dimensions.
/.
Jerome
Smith
DEPARTMENT OF ANTHROPOLOGY
UNIVERSITY
OF
SOUTH FLORIDA
Louanna
Furbee
Kelly Maynard
Sarah Quick
Larry Ross
DEPARTMENT
OF
ANTHROPOLOGY
UNIVERSITY
OF
MISSOURI, COLUMBIA
Salience Counts: A Domain
Analysis of English Color Terms
Independent models
of basic
and nonbasic color terms are combined to explore
the
domain
as a whole. We analyze an elicited set of
English
color terms,
ranked
by
a
combined frequency
and
order of
mention salience
index.
We find
that (1)
basic color terms emerge
intact
as
a distinctive
category,
and
(2) the
remaining
types of
terms
array
themselves
in an
extended
pattern of
distri-
bution
with
respect to
the
domain's broader
scheme of
lexical,
organizational,
semantic,
and
salience
dimensions.
I
n 1969 Brent Berlin and Paul Kay published Basic Color Terms
(1991,
bibliographically updated from 1969), generally acknowledged as a
pivotal challenge to an unfettered linguistic relativity. Using primary
and secondary data from 98 languages, they argue for a radically revised
perspective on color lexicons and the spectral phenomena they encode.
Berlin and Kay begin by distinguishing between basic and nonbasic color
terms. They then present evidence for constrained patterns of basic term
number, sequential emergence, and color mapping within and among their
sample of languages.
Journal
ofUnguistic
Anthropology
5(2):203-216. Copyright © 1995, American Anthropological
Association.
203
204
Journal
of Linguistic Anthropology
Basic Color Terms
has received much critical attention (see reviews Collier
1973;
Durbin
1972;
Hickerson
1971;
Merrifield
1971;
Mitsunobu
1972;
New-
comer and Fans 1971), and its proposals about the structure and dynamics
of basic color vocabularies have stimulated a vast array of subsequent
research on cultural, linguistic, psychological, neurophysiological, and
other aspects of basic color terminologies. Luisa Maffi's bibliography in the
paperback edition of
Basic Color Terms
(1991:173-189) lists more than 200
publications on color terminology, most postdating Berlin and Kay and
dealing directly with issues raised by them in 1969. While some have
argued for fairly thorough reworkings of Berlin and Kay's central premises
(e.g., Crawford 1982; Hays et al. 1972), the dominant thrust of recent
color-term research has been toward confirmation and expansion into ever
wider areas of inquiry.
The notion of basic color term—and its cognitive, biological and behav-
ioral correlates—has been the focus of most color-term research over the
past 25 years. For a term to be considered unequivocally basic, Berlin and
Kay argue that it must be (1) monomorphemic, (2) taxonomically superor-
dinate, (3) broadly applicable, and (4) salient (1991:6). They invoke addi-
tional criteria of grammatical productivity, recent borrowing, et cetera, to
screen remaining problematic terms (1991:6-7). These
first
four criteria, and
the formal class of basic color terms which they establish, have proven their
utility in the many studies alluded to above. By attending almost exclu-
sively to basic color terms, our understanding of the interplay of language,
culture, and biology has been, and continues to be, much enhanced. Al-
though certainly not taken for granted, however, little explicit attention has
been paid to the ways these criteria and others might structure the larger
domain of color terms in which basic color terms actually reside.
As typically presented, basic color terms tend to be isolated from other
color terms for purposes of particular research programs. Berlin and Kay's
Basic Color Terms
(1991) is, of course, explicitly and exclusively about basic
color terms. Most subsequent research was designed with basic color
terms—or something very much like them—clearly in mind. For instance,
Wattenwyl and Zollinger elicited lists from students speaking German,
French, English, Hebrew, and Japanese "that in their opinions were abso-
lutely necessary for a minimum
color
lexicon"
(1979:281, emphasis added).
And, as mentioned above, Hays et al. (1972) critique Berlin and Kay's
approach to basic color terms and argue for a more rigorous comparative
methodology. Their article is entitled "Color Term
Salience,"
but—in keep-
ing with the momentum generated by Berlin and Kay's study—it is devoted
entirely to the examination of basic color terms.
Other Color Terms
Ronald Casson's recent study (1994) of the growth oi English secondary
(nonbasic) color terms is an excellent departure from this trend. It invokes
a broader model of terminology types that includes basic terms. Having
done that, however, Casson turns his exclusive attention to the dynamics
of
a
particular kind of nonbasic color nomenclature.
Salience Counts
205
Whether basic or nonbasic terms have been isolated for examination in
one study or another is not a subject of criticism in and of
itself.
We only
wish to suggest
that,
given the results achieved by these narrowing efforts,
the time may have arrived to attempt a wider exploration.
We will argue that this more inclusive perspective can be achieved by
viewing Berlin and Kay's four criteria for basic color terms as selected
attributes of dimensions relating to the lexical, hierarchical, semantic, and
salience ordering of the domain as a whole—basic terms, other monomor-
phemics, multimorphemics, and multilexemics taken together. Thus, basic
color terms are monomorphemic, but nonbasic terms may be monomor-
phemic, multimorphemic, or even multilexemic phrases. Also, basic color
terms must be taxonomically superordinate, but even if this were their
exclusive privilege, there remains a potentially rich subordinate organiza-
tion in which nonbasic forms may be distributed. Further, basic color terms
must be abstract and widely applicable across contexts, but many nonbasic
color terms may share these properties, and still others have both abstract
and concrete senses and may be utilized in both general and restricted
contexts.
Finally, basic color terms should be highly salient, but the salience of all
color terms is measurable and deserving of empirical examination. Berlin
and Kay cite "a tendency to occur at the beginning of elicited
lists"
(1991:6)
as one measure of the relatively greater psychological salience of basic color
terms compared to others. At
a
group level, order of mention and frequency
of mention across individuals' lists may be taken
as
associated measures of
cultural salience (Romney and U Andrade 1964:155-157; Weller and Rom-
ney 1988:10-11). Where listing tasks are employed in color term research,
both kinds of data tend to be reported side by side (e.g., Bolton
1978;
Bolton
et
al.
1980;
Davies et
al.
1992;
Davies and Corbett
1994;
Pollnac
1975).
In the
present study we will make use of
a
single measure (Smith 1993) that takes
into account both order and frequency of mention of a large number of
elicited English color terms—basic and nonbasic alike.
We suggest, then, that when basic color terms are taken for what they
are—as some among many color terms—Berlin and Kay's four criteria
(especially salience) serve not so much as a means to segregate basic terms
from the others as a means to integrate them with the others. In the present
study, this approach is applied to a large set of elicited English color term
lists.
Ranked by a salience measure sensitive to both frequency and order
of occurrence across lists, the lists' color terms are further classified em-
ploying a typology combining Berlin and Kay's criteria with broader
dimensions adapted from Casson's recent work (1994) mentioned above.
A model is then proposed that supports the argument that basic color terms
do indeed constitute a distinctive category, along with other nonbasic
categories, within a well-formed cultural domain of English color term
types.
Specifically, we find that (1) basic color terms emerge intact within
the domain model by means of the matrix of attributes that Berlin and Kay
propose for them, and (2) the remaining types of terms array themselves
in an extended pattern of distribution with respect to the domain's broader
scheme of lexical, organizational, semantic, and salience dimensions.
206
Journal
of Linguistic Anthropology
Data Collection and Description
To examine the domain of English color terms, a corpus was first gath-
ered by means of a free-listing exercise, a simple form of systematic data
collection especially well suited for exploring the contents of cultural
domains (Weller and Romney 1988:9-19). The free-listing task is entirely
cognitive (no color stimuli were present) and can be contextualized (or
decontextualized) through instructions to suit the purposes of the research.
For the present study, 353 natively English-speaking students (by self-re-
port) in undergraduate linguistics and anthropology courses at the Univer-
sity of South Florida (n=78) and the University of Missouri, Columbia
(n=275) were asked—with little preamble—to write down all the color
words that they could think of in a two-minute period. To encourage
spontaneous "top-of-the-cognitive-deck" responses from all individuals,
context was minimized by avoiding any comments about the purpose of
the inquiry, and time was put at a premium. No prior discussion of color
terminology had occurred in
the
classes at the times of the several exercises,
although it cannot be guaranteed that individual students had not antici-
pated coverage of the topic. The average list contained 19 entries. A total
of 487 different terms were gathered and assigned a free-list
salience index—a
measure of their relative overall salience.
Anthropac 4.0 (Borgatti 1992), software for working with systematically
collected data, provides several routines for gathering and analyzing free-
list data, among others. (Specifically, version 4.6 was utilized throughout
the present research.) After student free lists were edited for spelling
consistency, the 487 terms collected were processed through Anthropac,
which established each term's frequency (number of mentions across all
lists) and rank (mean order of mention across lists in which the term was
present at
all).
Additionally, one of us has developed a free-list
salience index
that combines these two traditional measures into a single more compre-
hensive assessment of salience (Smith 1993)—the mean percentile ranking
of
a
term across all lists, whether or not
a
term is mentioned in a given list.1
(Although this was not documented for version 4.x
[1992],
Borgatti has
incorporated this new index into the Anthropac 4.6 software package.)
The free-list salience index ranges from 0 to 1, with higher figures
denoting higher salience. For a given term in a given list in which it occurs
(ranked according to its order of appearance in that list), an index score is
calculated as follows:
term index score = (list-length term rank) x (1 / (list length -1))
The term's index score for each list in which it occurs is thus calculated. A
term's mean score across all lists (even those in which the term does not
appear) is its free-list salience index. Unless specifically noted, all sub-
sequent references to salience and its measurement refer to the free-list
salience index and the rank ordering that it imposes on the color terms that
comprise the data for this study.
Salience Counts 207
A Typology of Color Terms
Color vocabulary research has been focused largely on basic terms since
1969.
As discussed above, in order to examine the domain of color more
generally, we propose to bring Berlin and Kay's restricted set of basic color
terms into a broader scheme employed recently by Ronald Casson (1994).
Combining Berlin and Kay's classification criteria with ideas derived from
Casson's approach to exploring English vocabulary change, color terms in
general may be modeled as a typology with respect to several contrastive
dimensions (see Figure 1): (1) conventional/novel, (2) simplex/com-
plex/residual, (3) basic/secondary, (4) opaque/transparent.
Casson, interested in comparing elicited data with documented etymolo-
gies
(1994:7-9),
describes conventional terms as "words or senses of words
that are fixed in the lexicon of the language" (1994:8). Relying more com-
pletely on elicited (free-list) data alone, we have chosen to emphasize a
sense of convention that emphasizes shared intersubjective knowledge.
Therefore, all terms in our corpus that are mentioned by more than one
student are counted as conventional and culturally salient. As shown in
Figure 1, a total of 209 (43%) of the 487 color terms in the corpus are
conventional by this measure, including 50 terms (e.g.,
aqua
blue,
sapphire,
and
teal green)
noted by only 2 among the
353
students producing lists.
Novel color terms, in contrast, are "unique words or senses of words
... that individuals invoke to meet particular needs and circumstances'
(Casson 1994:8). Complementing our designation of conventional color
terms as those with two or more mentions, all terms mentioned only once
in our data are taken as novel with regard to a minimal level of group use.
Our set of
278
novel terms
(57%
of the corpus described in Figure 1) does
include 10
(buff,
caramel,
chocolate,
fawn, ginger, henna, russet,
sand,
sepia,
f RMdual (U)
Figure 1
A
typology of English color
terms.
(Numbers in parentheses indicate the number
of terms of that type in the database of
487
elicited terms.)
208
Journal
of Linguistic Anthropology
straw) that appear in Casson's list of 92 conventional secondary terms
(1994:10-12).2
Baby
shit brown,
puke
green,
and wispy
blue
were among the
more colorful examples of idiosyncratic terms cited by only one list maker.
As with Berlin and Kay
(1991:6),
Casson (1994:6) takes conventional color
terms to be lexically simplex or complex. Complex terms may be morpho-
logical and/or lexical derivatives of simplex terms, which are themselves
not further analyzable from their internal semantic or lexical structure.
Thus,
green is a simplex color term while greenish is a morphologically
complex lexeme
{green
+ ish) and blue green is lexemically complex (a
phrase).
Among simplex terms, the contrast between basic and secondary terms
is central to Berlin and Kay's
thesis,
pertinent to most subsequent color term
research, and crucial to our consideration of the larger domain of color
terms in which these types are argued to exist. We follow Casson (1994) in
viewing secondary terms as a semantically interesting group of structurally
simplex terms that are not basic by Berlin and Kay's criteria. Berlin and
Kay's 11 English (simplex) basic color terms are white,
black,
red, green,
yellow, blue,
brown,
purple, pink,
orange,
and gray (1991:2). Casson collected
92 simplex secondary color terms—to the exclusion of any other types in
the larger domain—through a combination of elicitation and dictionary
screening methods in order to study the dynamics of color terminology
over time (1994:7-9). Our 487 terms include all kinds of color terms,
consisting of whatever students produced in their lists. Among the 81
simplex terms we classified as secondary following Casson's 1994 model
were 74
(80%)
of the 92 in his data and 7 terms
(alabaster,
canary, copper,
coral,
ebony, melon, neutral) not collected by Casson but meeting his criteria for
simplex secondary status (1994:7-9, and personal communication, 1994).
The 8 terms in Casson's secondary terminology data but not in ours are
cerise, chestnut,
manila,
oatmeal,
orchid,
platinum,
steel,
and
ultramarine.
As
noted above, an additional
10
terms classified as secondary in Casson's data
were only mentioned once in our data and thus were classified by us as
novel.
For
simplex terms, Casson also draws the following distinctions between
transparent and opaque terms:
Transparent conventional terms are "institutionalized" ... forms in the lexicon
that are synchronically analyzable, that is, words or senses of words that are
semantically transparent but not regarded as
figurative
extensions (cf. Bauer
1983:45). Examples from the color realm include, in addition to basic
orange,
secondary
rose,
lilac,
lemon,
raspberry,
emerald,
gold,
copper,
cream,
chocolate, salmon,
ivory,
wine,
and
navy.
Opaque conventional terms are "lexicalized" ... forms in
the lexicon that are synchronically unanalyzable, that is, words and senses of
words that are semantically opaque. Examples of opaque terms include basic
purple, pink, green, and yellow, as well as secondary
russet,
beige,
tan,
puce, crimson,
buff,
sepia, taupe, magenta, ecru,
teal,
scarlet,
and
tan
[sic\. [1994:8]
The emphasis here is not on structural complexity, as in the lexically
simplex/complex dichotomy, but on semantic complexity. Transparent
terms (semantically complex), while having clear color denotations, also
Salience Counts
209
point directly (transparently) to their colorful real-world sources. Opaque
terms (semantically simplex) have emerged over time with denotative
reference to the world of color only and, in this sense, are "more ab-
stract/less complex" than their transparent comrades. Of the 11 English
basic color terms, only
orange
is transparent as defined here. The other 10
are opaque.
Casson does not include complex forms in
his
analysis,
although he does
anticipate future research on " 'complex' secondary color terms" (1994:7).
The task will be formidable. Complex terms are inherently productive, and
the possibilities (both morphological and lexical) are essentially infinite,
involving, for instance, all permutations of various kinds of simplex terms
in two-term
phrases.
As to a more detailed typology of complex terms that
might be employed (in Figure
1
or elsewhere), it is not clear to us (or to
Casson, with whom we have discussed
this
issue) that it could successfully
invoke
a
secondary/basic level of contrast, because the notion of a complex
basic color term
(reddish,
blue green?)
is a contradiction in terms. We could
reinterpret the organization of Figure 1 and create a simplex/complex
contrast subordinate to the secondary node of the typology. But the appli-
cation of such a distinction to complex terms seems unmotivated and
would then invite a consideration of opaque
(blue
green,
lime green?)
and
transparent
(silvery
pearl,
reddish copper?)
complex forms that would not
appear to be useful. Other typologies are certainly possible, but they would
take us well beyond the scope of the present study, and so the complex
color terms in our data will remain without further internal analysis for
now.3
Figure
1
shows our conventional color-term data grouped into simplex,
complex, and residual categories. The group that we have designated as
residual is displayed in the interest of completeness. It consists of 53
simplex forms presented by more than one student (and thus at least
nominally conventional) that appear in neither Berlin and Kay's basic
English terms nor Casson's set of secondary English terms. Some of these
terms are
blood,
cool,
dull,
loud,
paisley,
plaid,
pigment,
and rainbow. They
suggest various sources of color-related vocabulary and are probably
deserving of future examination in their own right. We retain them here to
give the fullest possible account of terms elicited from two or more students
participating in the present study.
In the treatments of salience that follow, all terms in
all
lists were retained
in order to establish their independent free-list salience indexes within and
across all lists, without regard to their essential status as color terms. For
purposes of the present study, however, novel color terms will be excluded
from further analysis, and—as noted above—conventional complex and
residual terms will be treated as separate categories, without additional
comment as to their possible internal structural or semantic properties.
Conventional Color-Term Salience
With
salience indexes assigned and novel terms removed, we may order
the remaining 209 conventional terms from most salient to least. Five
210
Journal
of Linguistic Anthropology
250
<D
O
8
To
100
50
11
Basic29 a
Sec. Opaque Sec. Transparent
Color Term Type
ComplexS3
Residual
Figure
2
Box-and-whisker
plots
of free-list salience orderings of English color term types.
(N
=
the number of terms of each type in the data.)
groups (Figure 1) are examined as to their members' distributions among
the ranked
list:
(1)
simplex
basic,
(2)
simplex secondary opaque,
(3)
simplex
secondary transparent, (4) complex, and (5) residual. Box-and-whisker
plots (Figure 2) allow us to visually inspect both their means and their
internal distributions. Each box represents an upper and lower quartile
of
ranks adjacent to the mean (the horizontal line). The upper and lower
whiskers indicate the lowest and highest quartiles, respectively, with any
extreme outliers shown as points beyond the whiskers (as is needed to
show two extremely low-ranked secondary opaque terms in the figure). In
Figure 2 the 11 basic color terms are distributed in a narrow range at the
highest end of salience
ordering.
Secondary opaque and transparent terms
show the next highest means, with the 29 opaque terms more tightly
distributed than the 52 terms of the transparent subgroup. The conven-
tional complex and residual terms make up the bulk of lower ranks, with
equivalent means and only slightly different distributions.
The place of the 11 basic English color terms at the top of the free-list
salience index order is seen in Figure
3,
where indexes of the
21
most salient
terms (index 0.1) are displayed While there
is
no fixed method of making
"cuts"
in a range of free list data (Weller and Romney 1988:16), the basic
Salience Counts 211
E
o
o
o
RED(B) 1
BLUE(B) 2
GREEN(B) 3
YELLOW(B) 4
BLACK(B) 5
ORANGE(B) 6
PURPLE(B) 7
WHITE(B) 8
BROWN(B) 9
PINK(B)10
VIOLET(T)11
GREY(B)12
MAGENTA(O)13
GOLD(T)14
TAN(O)15
TURQUOISE(T)16
SILVER(T)17
MAROON(O) 18
INDIGO(O) 19
MAUVE(O)20 '
FUCHSIA(O)21
=
=
MM
••
HMI
HMI
=
=
1
1
1
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Free List Salience Index0.9
Figure 3
The salience of English color terms. (Terms included are those with a free-list
salience index greater than
0.1.
Letters in parentheses indicate color-term types:
(B)
basic, (O) secondary opaque, and (T) secondary transparent.)
terms shown in Figure 3 (along with violet) form
a
grouping at the
top.4
We
note that the four fundamental hues form a subgroup at the very top (see
Kay and McDaniel 1978; Kay et al. 1991). The finding is far from unique.
For instance, in a similar (but not identical) study of elicited Luganda (East
Africa) color terms (a three-basic-term
system),
Pollnac comments that "the
twelve most salient color terms, which are determined by frequency and
mean rank of recall, refer to the eleven hues which Berlin and Kay argue
are the only ones to be named with basic color terms (1975:93). Terms
translated as
yellow,
green,
red,
and blue are at the top of their salience
ordering. More recently, Davies and Corbett (1994:183) provide fre-
quency/mean ranked data on elicited Xhosa color terms (a five basic term
system) showing terms translated as
white
and
black
at the top, followed by
red,
yellow,
and grue. But our intent in the present study is not directly
comparative. For our more limited purposes, finding the
11
basic English
color terms in such
a
distinctive distribution only confirms that they retain
their identity as a distinctive type when we turn our attention
to
the domain
as a whole.
A clustered bar chart of salience orderings of the five types of conven-
tional color terms under review shows how distinctive their distribution is
(Figure
4).
Again,
all
of the basic terms place in the upper half of the salience
indexes. Some 90 percent of the secondary opaque terms are in the upper
half,
while less than 60 percent of the transparent terms fall there. Little
12
Journal
of
Linguistic Anthropology
100
80
60
5
40
Q)
Q.
20
Upper Half Lower Half
Free List Salience Order
Basic
ComplexSecondary Opaque
ResidualSecondary Transparent
Figure
4
Percentage frequencies
of
English color types in upper- and lower-half free-list
salience index orderings.
distinguishes the complex and residual color terms, although we see again
their predominant location in the lower half of salience figures.
Discussion
We have seen
how
Berlin
and
Kay's four criteria
for
determining
the
status
of
a color term as basic may be applied
to a
wider domain
of
color
terms,
at
least
as
wide—for English—as
to
include basic, secondary, and
complex terms (as defined by Casson 1994 and represented
by
209 terms
in
our
particular data set). How might these criteria be viewed as dimen-
sions and attributes of meaning encompassing these types in this domain?
Figure
5
is an attempt to conceptualize what our analyses suggest.
The first thing
to
note
is the
treatment
of
opaque
and
transparent
as
attributes
of
a
semanhcally simplex
dimension, rather than
as
terminology
types in their own right. We take this approach in order to highlight some
parallels between semantic and lexical structuring across all three domain
categories. Even though,
for
instance, only the basic term
orange
is
trans-
parent,
it
is
clear that the presence or absence of a real-world referent
is
not
the exclusive property of secondary terms, as opposed to basic ones.
The parallels with Casson's lexical dimension also seem relevant. Lexical
complexity is a quantitative notion here. From monomorpherruc through
m u I h morphemic tomultilexemic, lexical structures acquire more and more
units
In the
same way, somanhcally simplex (opaque) color terms refer
Salience Counts
213
Lexical Structure
simplex | complex
Taxonomic Position
superordinate | subordinate
BASIC
TERMSSECONDARY
TERMSCOMPLEX
TERMS
< Salience > -
opaquetransparent
Semantic Structure
Figure 5
Domain analysis of conventional English color-term types.
only to the quality of color, while semantically complex (transparent) terms
also continue to have real-world-object referents.
This is a different slant on the semanticity of basic color terms, but it fits
nicely, for instance, with the idea that basic color terms are more abstract
than others. Indirectly, it also captures Berlin and Kay's "wide application"
criterion in that color terms with real world referents will typically (though
not always) be constrained in their range of reference by that connection.
Salience, of course, has been shown here
to
order clearly the types of color
terms, with the other dimensions under discussion providing means of
segregation. While we expected to find basic terms more salient than
secondary or complex ones, we did not anticipate the degree of difference
revealed or its stability among, for instance, men and women.6
Lexical structure and taxonomic location emerge as the minimal formal
dimensions required to establish color term categories. Basic terms are
taxonomically superordinate, complex terms are lexically (and semanti-
cally) complex, and secondary terms are lexically simplex forms subsumed
by other "kinds of colors. Salience is a continuum, however skewed in
favor of basic terms, that binds the categories within the domain. Semantic
structuring appears more dynamic than the other dimensions. One can
214 Journal of Linguistic Anthropology
imagine secondary terms as a source of both basic terms by the attenuation
of object reference and of complex terms through the exploitation of
morphological and lexical productivity (see Casson 1994:17-19 on meton-
ymy).
Color terminology has been the subject of intense scrutiny and debate
over the years—all the more
so
since Berlin and Kay's
1969
landmark
work.
We have employed their criteria for isolating basic color
terms,
in conjunc-
tion with Casson's 1994 model of secondary terminology, in an effort to
explore the benefits of a broadened perspective on the domain of color
terms as a
whole.
While what we present here
is
certainly only one of many
views, the good fit of the data to both Berlin and Kay's basic-color-term
model and to Casson's treatment of secondary terms provides convincing
evidence of a robust underlying organization to these terms.
Notes
Acknowledgments.
This article
is
an expanded treatment of
a
presentation, "Basic
Color
Terms:
The Salience Criterion/' made by the senior author
at
a volunteered
session, Light
on
Color Ethnography, chaired
by
Robert E. MacLaury
at
the 91st
Annual Meeting
of the
American Anthropological Association
in
1992.
It has
benefited from comments by session participants, especially MacLaury, and by the
anonymous reviewers
who
labored with early drafts. Helpful suggestions from
Ronald Casson and Luisa Maff i are also acknowledged.
We,
of course, remain solely
responsible
for
its contents.
1.
Smith (1993:2-3) discusses some
of
the problems
in
interpreting the relative
salience of free-list items using measures of frequency and rank separately. Briefly,
frequency across
lists
is
an easily calculated measure of salience, but it incorporates
no information about order
of
mention,
and
nondiscriminating
tie
scores
are
common—especially in the zone
of
high-frequency/salience where our interest is
usually focused. (All items mentioned by everyone receive the same
score.)
Simple
(average) rank order
of
mention,
on
the other hand, while intuitively compelling,
is
a
poor measure
of
free-list data, because
it
completely disregards the distorting
effects
of
rarely mentioned items that happen
to
appear early
in
the few lists that
do include them. (An item mentioned only once-—but
first—receives
the
highest
score available, because nonmentions are not calculated.) Finally, neither measure
directly considers variable list length,
an
inherent feature
of the
open-ended,
unstructured free listing such
as
generated
the
data that we
are
examining here.
The free-list salience index takes frequency, order,
and
variable list length into
account in a single number—a mean percentile ranking.
2.
We note that
russet[t]—one
of
Casson's conventional secondary terms men-
tioned by only one of our students and thus classified as novel in our analysis—ap-
pears in the title of his 1994 article "Russett, Rose, and Raspberry: The Development
of English Secondary Color Terms." These and other "rough
edges"
of our ongoing
research remain to be resolved, or at least better understood.
3.
What could
be
of interest regarding complex color terms is more research into
similarities and differences between morphological and lexical complexity.
We
are
grateful
to one of the
anonymous reviewers
for
bringing
the
following
to our
attention:
Greenish
(a
degree
of
membership in the category termed
green)
is
dearly
secondary
and
morphemically complex,
but it is not
included
in
(taxonomically
subordinate to)
green
in
the same way that
blue green
(a
lexically complex kind
of
green)
is. We were somewhat surprised
to
find that,
at
least
for
English speakers
engaged
in
free listing tasks, the former type of term is virtually excluded in favor
Salience Counts
215
of the latter. Only 6 morphemically complex terms are present among the 64
complex terms
(glossy,
golden,
metallic,
monochromatic,
sandy,
and
shaded).
All the
other complex terms are phrases of two or more
lexemes,
in spite of the potentially
restrictive free list instruction to write down color words (but not kinds of words).
Reddish,
a morphemically complex term used as an example of a nonbasic term by
Berlin and Kay (1991:5-6) is not even a conventional term in our data, being
mentioned by only one of the
353
students. We are not in a position to discuss this
fascinating issue in the present article, but it would appear that these two kinds of
complex terms are clearly regarded by speakers of English as semantically, as well
as structurally, distinctive—perhaps in just the way suggested by our reviewer.
4.
The 12 most salient color terms in the data, as measured by the
free-list
salience
index, are the
11
basic
terms,
including
purple—and
violet.
This is a matter that must
be left to future research, but we do note that, in keeping with the domain model
developed in this article, basic/opaque
purple
has a higher salience index (0.4844)
than secondary/transparent
violet
(02631).
5.
A 2x2 cross-tab analysis of the distributions of secondary opaque and trans-
parent terms in the upper and lower half of their ordering yields a significant
chi-square of
9.586
(df
=
1,
a < 0.01).
6. Exploring differences between the free-listing patterns of female and male
students revealed a somewhat unexpected pattern of
similarities.
There is little or
no difference between their lists. For instance, the two groups ranked basic color
terms in almost identical fashion. While this may run counter to notions about
gender differences in the area of color cognition, it is in hindsight not surprising,
given the minimal contextual ization of the task. What is of interest is the degree to
which highly shared (culturally salient) formal domain dimensions are apparently
called upon by everyone when a search of domain contents is not driven by any
specific real-world circumstances, such as buying clothes or repainting a car.
References Cited
Bauer, Laurie
1983 English Word Formation. New
York:
Cambridge University Press.
Berlin, Brent, and Paul Kay
1991 [1969]
Basic
Color
Terms.
Berkeley: University of California Press.
Bolton, Ralph
1978 Black, White, and Red All Over: The Riddle of Color Term Salience.
Ethnology 17(3):287-311.
Bolton, Ralph, Anne
Curtis,
and Lynn
L.
Thomas
1980 Nepali Color Terms: Salience on a Listing Task. Journal of the Steward
Anthropological Society 12:309-322.
Borgatti, Stephen P.
1992 Anthropac
4.0.
Columbia,
SC:
Analytical Technologies.
Casson, Ronald W.
1994 Russett, Rose, and Raspberry: The Development of English Secondary
Color
Terms.
Journal of Linguistic Anthropology 4(1)5-22.
Collier, George A.
1973 Review of
Basic
Color
Terms.
Language 49(1)245-248.
Crawford, T.D.
1982 Defining "Basic Color Term." Anthropological Linguistics 24338-343.
Davies, Ian R.
L.,
and Greville
G.
Corbett
1994 A Statistical Approach to Determining Basic Color
Terms:
An Account of
Xhosa.
Journal of Linguistic Anthropology
4(2):175-193.
216 Journal of Linguistic Anthropology
Davies, Ian R. L., Catriona MacDermid, Greville G. Corbett, Harry McGurk,
David Jerrett, Tiny Jerrett, and Paul Sowden
1992 Color Terms in Setswana: A Linguistic and Perceptual Approach, linguis-
tics 30(6) :1065-l
103.
Durbin, Marshall
1972 Review of Basic Color Terms. Semiotica 6(3):257-278.
Hays,
David G., Enid Margolis, Raoul Naroll, and Dale Revere Perkins
1972 Color Term Salience. American Anthropologist 74:1107-1121.
Hickerson, Nancy P.
1971 Review of Basic Color Terms. International Journal of American Linguis-
tics 37(4) :257-270.
Kay, Paul, and Chad K. McDaniel
1978 The Linguistic Significance of the Meanings of Basic Color Terms. Lan-
guage 54:610-646.
Kay, Paul, Brent Berlin, and William Merrifield
1991 Biocultural Implications of Systems of Color Naming. Journal of Linguistic
Anthropology l(l):12-25.
Maffi, Luisa
1991 A Bibliography of Color Categorization Research, 1970-1990. In Basic
Color Terms. Brent Berlin and Paul Kay. Pp. 173-189. Berkeley: University of
California Press.
Merrifield, William R.
1971
Review
of
Basic
Color Terms. Journal of Linguistics 7(2)259-268.
Mitsunobu, Meiyo
1972 Review of Basic Color
Terms.
Studies in English
Literature:
English Literary
Society of
Japan
(March 1972):170-184.
Newcomer, Peter, and James Faris
1971 Review of Basic Color
Terms.
International Journal of American Linguistics
37(4):270-275.
Pollnac, Richard B.
1975 Intra-cultural Variability in the Structure of the Subjective Color Lexicon
in Buganda. American Ethnologist 2(l):89-110.
Romney, A. Kimball, and Roy Goodwin EKAndrade
1964 Cognitive Aspects of English Kin Terms. American Anthropologist 66(3,
pt.2):146-170.
Smith,
J.
Jerome
1993 Using ANTHROPAC 3.5 and a Spreadsheet to Compute a Free List Sali-
ence Index. Cultural Anthropology Methods Newsletter 5(3):l-3.
Wattenwyl, Andre von, and Heinrich Zollinger
1979 Color-term Salience and Neurophysiology of Color Vision. American An-
thropologist 81:279-288.
Weller, Susan C, and A. Kimball Romney
1988 Systematic Data Collection. Newbury Park,
CA:
Sage Publications.
... In this task, C.A. asked a subset of participants-who expressed belief in Haine and Ishoko-to list things that pleased and angered these deities. 6 Using this data, we can get a single index of ubiquity and cognitive salience (i.e., item placement in lists; Smith et al., 1995). Data items were coded using a twelve-category coding system by two coders ; see Supplementary Section S4.1). ...
... Romney and D'Andrade (1964) instead employed only order of mention as an indicator of salience. A consistent development is represented by Smith (1993), who combines the two elements to create the Salience Index, an easily calculated algorithm, which was refined later by Smith et al. (1995). Although it combines the different variables, it was shown that Smith's Salience Index was influenced by the length of individual lists. ...
Book
Taste is considered one of the lowest sensory modalities, and the most difficult to express in language. Recently, an increasing body of research in perception language and in Food Studies has been sparkling new interest and new perspectives on the importance of this sense. Merging anthropology, evolutionary physiology and philosophy, this book investigates the language of Taste in English, and its relationship with our embodied minds. In the first part of the book, the author explores the semantic dimensions of Taste terms with a usage-based approach. With the application of experimental protocols, Bagli enquires their possible organization in a radial network and calculates the Salience index of gustatory terms in both American and British English. The second part of the book is an overview of the metaphorical extensions that motivate the polysemy of Taste terms, with the aid of corpus analysis methods and various texts. This book is the first to review systematically and in a usage-based perspective the role of the sensory domain of Taste in English, showing a more complicated picture and suggesting that its under-representation and difficulty of encoding does not correspond to lack of importance.
... An important advance would be the ability to evolve our measure to a dimension of greater quantitative representation, such as that found in interval measures. Also, if we considered the quantitative approach of the Cognitive Salience Index we still face the problem of determining the mean point without considering the spreading of the term on the list, reducing its accuracy (Smith and Borgatti, 1998;Smith et al., 1995). ...
Article
Full-text available
Our study aimed to present a new method to generate a scale for color concept distances using the Law of Comparative Judgment and derivate an interval scale for the amount of color concept salience. A hundred thirty-eight participants (mean age = 22.7 yrs; SD = 1.9) with normal visual acuity and color vision were evaluated. Additional 225 children were also evaluated (mean age = 8,9yrs; SD = 2.3). The task consisted of writing a list of the one-term word for colors and data were analyzed based on Thurstone's rank order scaling method. For adults, the most salient concept was blue and the less was violet with a distance for concepts of 3.51 units between them, whereas for children the most salient concept was blue and the less was violet with a distance for concepts of 3.67 units between them. Significant differences between adults and children occurred for white, brown, pink, orange, gray, and violet. The internal consistency of our measurement was lower than 0.1% for almost all colors which suggests high reliability for both age groups, except for brown and anil in children. Thus, our innovative approach produces quantitative interval class data rather than categorical or ordinal data allowing quantitative measurement of the amount of the concept salience on a one-dimensional scale, feasible to be used for adults and children.
... We analyze adult interview data using the AnthroTools package (Purzycki and Jamieson-Lane 2016) for R (R Development Core Team 2008). The cognitive salience of listed items is calculated as a mean salience score, reported as Smith's S (Smith 2016;Smith and Borgatti 1997;Smith et al. 1995). Individual item salience is calculated by taking into consideration both the total number of items an individual listed and the order of the items. ...
Article
Full-text available
Recent work with infants suggests that plant foraging throughout evolutionary history has shaped the design of the human mind. Infants in Germany and the US avoid touching plants and engage in more social looking toward adults before touching them. This combination of behavioral avoidance and social looking strategies enables safe and rapid social learning about plant properties within the first two years of life. Here, we explore how growing up in a context that requires frequent interaction with plants shapes children’s responses with the participation of communities in rural Fiji. We conducted two interviews with adults and a behavioral study with children. The adult interviews map the plant learning landscape in these communities and provide context for the child study. The child study used a time-to-touch paradigm to examine whether 6- to 48-month-olds (N = 33) in participating communities exhibit avoidance behaviors and social looking patterns that are similar to, or different from, those of German and American infants. Our adult interview results confirmed that knowledge about daily and medicinal uses of plants is widely known throughout the communities, and children are given many opportunities to informally learn about plants. The results of the child behavioral study suggest that young Fijian children, like German and American infants, are reluctant to reach for novel artificial plants and are fastest to interact with familiar household items and shells. In contrast to German and American infants, Fijian children also quickly reached for familiar real plants and did not engage in differential social looking before touching them. These results suggest that cultural contexts flexibly shape the development of plant-relevant cognitive design.
... Using only average rank would result in an item mentioned only once, but first in a list receiving the highest salience score (Smith et al. 1995, 214). Instead, there are two commonly-used indices that combine these two parameters into more reliable and comprehensive measures of salience: the Smith free-list salience index (Smith 1993;Smith et al. 1995;Smith and Borgatti 1998) and the Sutrop cognitive salience index (Sutrop 2001). ...
... Freelisted data allows the researcher to discover the relative salience of items across all respondents within a given domain. Salience is a statistic accounting for rank and frequency (e.g., in the domain of English color terms, "red" is more salientit appears more often and earlier in freeliststhan "maroon" [Smith et al. 1995]). Researchers can calculate the mean salience value for all listed items, to reveal the intracultural salience of each term (below). ...
Chapter
Full-text available
A freelist is a mental inventory of items an individual thinks of within a given category. Freelists reveal cultural “salience” of particular notions within groups, and variation in individuals’ topical knowledge across groups. The ease and accuracy of freelist interviewing, or freelisting, makes it ideal for collecting data on health knowledge and beliefs from relatively large samples. Successful freelisting requires researchers to break the research topic into honed categories. Research participants presented with broad prompts tend to “unpack” mental subcategories and may omit (forget) common items or categories. Researchers should find subdomains to present individually for participants to unpack in separate smaller freelists. Researchers may focus the freelist prompts through successive freelisting, pile sorts, or focus group-interviews. Written freelisting among literate populations allows for rapid data collection, possibly from multiple individuals simultaneously. Among nonliterate peoples, using oral freelists remains a relatively rapid method; however, interviewers must prevent bystanders from “contaminating” individual interviewees’ lists. Researchers should cross-check freelist responses with informal methods as much as practicable to contextualize and understand the references therein. With proper attention to detail, freelisting can amass high-quality data on people’s medical understanding, attitudes, and behaviors. Keywords: Free list - Free recall - Salience analysis - Systematic data collection - Domain analysis - Rapid Ethnographic Assessment (REA) approaches
Article
Full-text available
Agricultural entrepreneurship is imperative for the enhancement of the livelihood and living condition of the people in rural areas. Transformation of farmers into entrepreneurs is often emphasised as a strategy for development of agricultural as well as rural sector in developing countries. Assam is one of the agricultural backward states of India and yet to experience agricultural advancement. The paper attempts to answer two specific research questions. Why is agriculture not a remunerative livelihood option in lower Brahmaputra valley? What are the strategies used by the farmers to make farming a profitable avenue? It is based on key informant interviews with 30 agricultural entrepreneurs in the lower Brahmaputra Valley, Assam. Free listing technique of cultural domain analysis was used to collect data. A multi-stage sampling procedure was used to select districts, block, villages, and agripreneurs. Amongst the ten districts from lower Brahmaputra valley, two districts have been chosen, on basis of the intensity of agricultural activity. The qualitative data was analysed with the help of Anthro tools an R package. Smith's salience was to analyze the free list responses. In the perception of the agricultural entrepreneurs, inadequate financial capital, farmer's non-willingness to work hard, and non-remunerative prices were the three main reasons for farming remaining unprofitable livelihood option in the lower Brahmaputra valley. The entrepreneurs use a number of strategies to make agricultural as profitable as a business venture. The most prominent among them were undergoing training, farm mechanisation, availing bank loans, adoption of modern agricultural practices and networking with officials so as to avail the benefits of government programmes.
Article
The erosion of local/indigenous farming knowledge in the face of HIV/AIDS deaths in Africa has been noted as a point of concern in the literature and by organizations such as the Food and Agriculture Organization of the United Nations. These concerns are about a break in the transmission of knowledge from adults (deceased parents) to children (orphans). Ultimately, erosion of farming knowledge is implied. This paper examines one aspect of knowledge, using an ethnobiological approach that is language based. Free-listing elicitation of pests in maize fields was conducted with 45 child orphans, 15 non-orphan children, and 30 adults in rural Benin. A cognitive salience index (CSI) was developed and an advanced analysis of the CSI scores was conducted examining the score differences between child orphans and non-orphan children and adults. The results indicate that orphaned children were more knowledgeable than non-orphaned children. One-parent orphans residing with the surviving parent are more knowledgeable than double orphans farming on their own. Non-affected adults and their children scored significantly lower than AIDS-affected adults and children. Other variables including gender and age were further examined to explain some of the observed differences. The findings indicate that there is a need for rethinking the implications of HIV/AIDS on farming knowledge.
Article
Full-text available
There are semantic universals in the domain of color; i.e. there are constraints on the types of possible basic color lexicons. These constraints arise from the structure and function of the visual system. Thus in the case of color at least, rather than language determining perception (cf. Sapir and Whorf), it is perception that determines language. In deriving the semantic universals from properties of the visual system, one must employ a continuous rather than discrete mathematics, in particular the theory of fuzzy sets. The resulting model of color semantics thus conflicts with the discrete-feature concept of semantic primes shared by structuralists and generativists. It is argued on this basis that discrete-feature semantic theories are of limited accuracy.
Book
Preface 1. Introduction 2. Some basic concepts 3. Lexicalization 4. Productivity 5. Phonological issues in word-formation 6. Syntactic and semantic issues in word-formation 7. An outline of English word-formation 8. Theory and practice 9. Conclusion Bibliography and citation index Index.
Article
Eleven focal colors are named by basic color terms in many languages. The most salient colors (black, white, and perhaps red) are named in all languages; the least salient of the set are named in fewer languages. Salience correlates with earliness of introduction, as measured by a scale of social evolution; with brevity of expression, as measured by phonemic length of basic color terms; with frequency of use, as measured by frequency of basic color terms in literary languages; and with frequency of mention in ethnographic literature. None of these correlations are established in the pioneer study of Berlin and Kay (1969), a study whose defects are well exposed by Durbin (1972) and Wescott (1970). The first two were documented respectively in Naroll (1970) and Durbin (1972); the last two are documented here. These four correlations independently support the Berlin-Kay color salience theory. They furnish a sound basis for further research on color term salience in particular and indeed on salience phenomena in general. We speculate that salience may be an important general principle of cultural evolution.