ArticlePDF Available

Color Naming Across Languages

Authors:

Abstract

this paper. (See Hardin 1993 and Stanlaw 1993 for careful appraisals of several points.) 8
2. Color Naming Across Languages
Paul Kay, Brent Berlin, Luisa Maffi and William Merrifield
1. Introduction: Prior cross-linguistic research on color naming
This chapter summarizes some of the research on cross-linguistic color categorization and
naming that has addressed issues raised in Basic Color Terms: Their Universality and
Evolution (Berlin and Kay 1969, hereafter B&K). It then advances some speculations
regarding future developments—especially regarding the analysis, now in progress, of the
data of the World Color Survey (hereafter WCS). In the latter respect the chapter serves as
something of a progress report on the current state of analysis of the WCS data, as well as
a promissory note on the full analysis to come.
B&K proposed two general hypotheses about basic color terms and the categories
they name: (1) there is a restricted universal inventory of such categories; (2) a language
adds basic color terms in a constrained order, interpreted as an evolutionary sequence.
These two hypotheses have been substantially confirmed by subsequent research.
1
There have been changes in the more detailed formulation of the hypotheses, as
well as additional empirical findings and theoretical interpretations since 1969. Rosch’s
experimental work on Dani color (Heider 1972a, 1972b), supplemented by personal
communications from anthropologists and linguists, showed that two-term systems
contain, not terms for dark and light shades regardless of hue—as B&K had inferred—but
rather one term covering white, red and yellow and one term covering black, green and
blue, that is, a category of white plus ‘warm’ colors versus one of black plus ‘cool’ colors.
Rosch reported further that these ‘composite’ categories, as they were later christened by
Kay and McDaniel (1978, hereafter K&McD), tend to be focused not only in white and
black, but sometimes at the foci of red or yellow, on the one hand, and of green or blue on
the other. B&K had conceived basic color categories in terms of foci and extensions and
2-Kay, Berlin, Maffi & Merrifield 2
had expressed the evolutionary sequence of hypothesis (2) as a sequence of constraints on
the successive encoding of foci. Rosch’s finding that composite categories may have
multiple foci was a major reason for the reconception of the evolutionary sequence in terms
of successive divisions of the color space (see e.g., Kay 1975: 258-262)
2
.
K&McD modeled these successive divisions of the color space as fuzzy partitions.
They interpreted individual color categories as fuzzy sets (Zadeh 1965),
3
and defined the
notion of fuzzy partition in terms of a (standard) set of fuzzy sets (K&McD: 641-644).
Accordingly, basic color categories were divided into three types. The first type consists of
the six fundamental categories, corresponding to Hering's primaries (Hering 1964): black,
white, red, yellow, green, blue.
4
The second type, the composites, consists of fuzzy
unions of the fundamentals. These include the ‘white/warm’ and ‘black/cool’ categories of
two-term systems, as well as several categories comprised by unions of pairs of the six
fundamentals (about which more presently). The third type were called ‘derived’ categories
and were defined in terms of the fuzzy intersections of the fundamentals.
Examples of this
type are colors that are seen as mixtures of fundamentals: for example, orange is seen as a
mixture of red and yellow (Sternheim and Boynton 1966).
5
The WCS was begun in 1976.
6
It was designed for two major purposes. The first
was to assess the general hypotheses advanced by B&K against a broader empirical basis.
Methodological objections had been raised to the empirical generalizations of B&K. The
most important of these were that: (1) the twenty languages studied experimentally were not
prima facie sufficiently numerous to justify universal conclusions; (2) the data were
obtained in Berkeley rather than in native communities; (3) most of the speakers
interviewed spoke English as well as their native language; (4) the number of speakers
interviewed for most of the languages was three or fewer; and (5) the interviewers were
not, for the most part, skilled speakers of the languages studied.
7
The second major
purpose of the WCS was to deepen our knowledge regarding universals, variation and
historical development in basic color term systems.
2-Kay, Berlin, Maffi & Merrifield 3
The methods and some initial results of the WCS are reported in Kay, Berlin and
Merrifield (1991, hereafter KBM). With the help of field linguists of the Summer Institute
of Linguistics and using a stimulus array substantially the same as that of B&K,
comparable data on naming ranges and focal choices for basic color terms were collected on
110 languages in situ. In most cases twenty-five speakers were interviewed per language.
Monolingual speakers were sought insofar as possible. A methodological departure of the
WCS from the method of B&K was that chip-naming judgments were obtained on
individual chip presentations, rather than the full array of stimuli. Judgments of best
example (focal judgments) were obtained in the same way as in the original study, by
requesting selection of the chip or chips that best represent each basic color word of the
native language from an array of 330 color patches, representing forty equally spaced
Munsell hues at eight levels of lightness (at maximum saturation) plus ten levels of
lightness of neutral (black, grey, white) shade.
The preliminary results of the WCS, as reported in KBM, were as follows. (1)
B&K had defined evolutionary stages on the assumption that all composite categories are
eliminated in favor of the six fundamentals before any derived categories appear. Kay
(1975) and K&McD had taken over this assumption in their reformulations of the
evolutionary sequence, except for the latter’s making formal provision for the optional early
appearance of grey.
8
KBM report further cases of early grey and point out, more
importantly, that either brown or purple or both not infrequently appear before the
green/blue composite is dissolved. (2) Kay (1975: 260-261) had noted evidence from
several sources
that there might be languages with composite categories comprising yellow
and green. MacLaury (1986, 1987a) was the first to document such categories with
controlled stimuli. Several more have been found in the WCS languages and were reported
in KBM. (3) Prior to the WCS, there had been no rationale offered in the literature for the
restricted inventory of composite categories actually reported, distinctly fewer than the
sixty-three logically possible combinations of the six fundamentals. KBM both extended
2-Kay, Berlin, Maffi & Merrifield 4
the inventory of composite categories empirically attested and provided a partial
explanation, in terms of generally acknowledged properties of the visual system, for the
restricted membership of this inventory (KBM: 15 ff).
2. The current state of analysis of the WCS data
The initial stage of processing of the WCS data converted the hand-collected data for each
collaborating speaker into two arrays, one for naming choices and one for focal choices.
The data for the first five speakers of Buglere are displayed in part 2 of Figure 1, naming
choices to the left and focal choices to the right. Each symbol in these arrays corresponds to
a Buglere color term, as indicated in part 1 of Figure 1.
9
The columns represent the forty
equally spaced Munsell hues mentioned earlier
10
and the rows levels of lightness.
11
2-Kay, Berlin, Maffi & Merrifield 5
Figure 1, part 1: Buglere
Language Country Family Tot. interviewees Fieldworker(s) Date
Buglere Panama Unclassified 25 (15 F; 10 M) K. Fisher and
J. Gunn
1978
Terms Appearing In Aggregate Naming Arrays
Symbol Term Users Symbol Term Users
/
jere/jerere 25
*
moloin/
moloinre
25
-
jutre/jusa 25
@
lere/lerere 25
+
dabe/dabere 25
#
leren 24
Aggregate Naming Arrays
Modal Agreement Level
1 2 3 4
01234567890123456789012345678901234567890
A ----------------------------------------- A
B ----------****--------------------------- B
C -**************@@@@@@@@@@-----#------**** C
D -*************@@@@@@@@@@@@@@#@#----*+**** D
E -+++++******@*@@@@@@@@@@@@@@@@####*#*++** E
F /++++*******@@@@@@@@@@@@@@@@########+++++ F
G /+++++*////@@@@@@@@@@@@@@#@#@@#####*+++++ G
H /++++//////////@@@@@@@@@@#@##########++++ H
I /+++/////////////@@@@@@@/@#/############+ I
J ///////////////////////////////////////// J
30% Agreement Level, 8 of 25 speakers
1 2 3 4
01234567890123456789012345678901234567890
A ----------------------------------------- A
B ----------****--------------------------- B
C -**************@@@@@@@@@@-----#------**** C
D -*************@@@@@@@@@@@@@@#@#----*+**** D
E -+++++******@*@@@@@@@@@@@@@@@@####*#*++** E
F /++++*******@@@@@@@@@@@@@@@@########+++++ F
G /+++++*////@@@@@@@@@@@@@@#@#@@#####*+++++ G
H /++++//////////@@@@@@@@@@#@##########++++ H
I /+++ ////////////@@@@@@@/@#/############+ I
J ///////////////////////////////////////// J
70% Agreement Level, 18 of 25 speakers
1 2 3 4
01234567890123456789012345678901234567890
A ----------------------------------------- A
B --- ----- ** ---- - ------------------ B
C -* ****** ---- C
D - ****** @@@@@@@ D
E + + **** @@@@@@@@ E
F +++ * @@@ @@@@ ## + F
G /+++ @ @@ @ ### ++ G
H /++++ // / @@ ## H
I / ///////// # I
J ///////////////////////////////////////// J
100% Agreement Level, 25 of 25 speakers
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C C
D D
E E
F F
G G
H H
I I
J J
Terms Not Appearing In Aggregate Naming Arrays
Symbol Term Users Symbol Term Users
o
lejre 10
>
mnule 2
=
kwajusa 7
x
dagikwale 2
Speakers (By I.D. Number, Age And Sex)
1 21 F 6 30 F 11 40 F 16 18 M 21 26 M
2 22 F 7 35 F 12 40 F 17 18 M 22 35 M
3 22 F 8 35 F 13 40 F 18 18 M 23 45 M
4 23 F 9 35 F 14 45 F 19 20 M 24 45 M
5 30 F 10 38 F 15 50 F 20 23 M 25 45 M
2-Kay, Berlin, Maffi & Merrifield 6
Figure 1, part 2: Buglere
Individual Naming Arrays
*** Speaker 1 ***
1 2 3 4 1 2 3 4
01234567890123456789012345678901234567890 01234567890123456789012345678901234567890
A ----------------------------------------- A A A
B ---------*****-------#------------------- B B B
C -**--*-********@@@#@@-#@@--##-#------**** C C +++ * C
D -*-***********@@@@@@@@##@####/#---**+**+- D D D
E /*+*********@*@@@@@@@@@@#########/#*@++** E E E
F -+++***-****@@@@@@@@@@@@######@####**++++ F F # F
G /+++*+*////@/@@@@@@@##@############*+#+++ G G @ G
H /++++/////////@@@@@@@@@@##@##########++++ H H H
I ///+///////////@@@o@@@@@@@//##/#####/###/ I I I
J ///////////////////////////////////////// J J J
*** Speaker 2 ***
1 2 3 4 1 2 3 4
01234567890123456789012345678901234567890 01234567890123456789012345678901234567890
A ----------------------------------------- A A ----------------------------------------- A
B ---------*-***@----o@*------------------- B B B
C -*************@@o@ooo@@@-@@@###-#----**** C C C
D /+************oo@@o@o@@@@oo#######o#+***+ D D * D
E /+++*+******@*@o@o@@@@@@@ooo##########*** E E E
F /+++********@@@@o@@@oo@@o#####o#####+++++ F F @ F
G /++++++****@o@oooooo@@o#o###########+#+*+ G G + G
H /+++++++/o/oo@/ooooo@o@#/######o#####++#+ H H # H
I /++++//////////@ooo@oooo/o/##o/#####++o++ I I I
J ///////////////////////////////////////// J J ///////////////////////////////////////// J
*** Speaker 3 ***
1 2 3 4 1 2 3 4
01234567890123456789012345678901234567890 01234567890123456789012345678901234567890
A ----------------------------------------- A A A
B ---*>>-->-****-*---@-@----->>--->--->---> B B B
C -++*+-*********o@@@@@@@@>->------>->-*--+ C C * C
D >*++++********@@@@@@@@@@-@-@-@#-#--##--*+ D D D
E >++++++****@@@@@@@@@@@@@@@@@#@@@>*-##++++ E E @ E
F /+++++++++*@@@@@@@@@@@@@@@#@########-++++ F F + # F
G -++++++/++@@@@@@@@@@@@@@@#############+++ G G G
H /+++++//////@/@@@@@@@@@@######@###+##++++ H H H
I /++++///////////@@@@@@@#//##############+ I I I
J ///////////////////////////////////////// J J J
*** Speaker 4 ***
1 2 3 4 1 2 3 4
01234567890123456789012345678901234567890 01234567890123456789012345678901234567890
A ----------------------------------------- A A ----------------------------------------- A
B ---**-*--*****@--####@------------------- B B B
C +*--***********@@@@@@######-#=#----+-**++ C C * C
D -+-*+++*****@@@@@@@@@@#@##@#@###-#@#+#**+ D D D
E /++-+++*+**#@@@@@@@@@@#@####@####-###**++ E E E
F /++++**++@*@@#@@@@@@@@#@@#@@##########+*+ F F @ F
G /+++++*+@@@@+@#@@@@@#@@@#####@@####@###*+ G G + # G
H /#+++///+@//@/@#@@@@@@#@@@/@##@######*#@+ H H H
I /+++-+@/////////@##@@@/@#/@##########+#/+ I I I
J ///////////////////////////////////////// J J ///////////////////////////////////////// J
*** Speaker 5 ***
1 2 3 4 1 2 3 4
01234567890123456789012345678901234567890 01234567890123456789012345678901234567890
A ----------------------------------------- A A ----------------------------------------- A
B -*+*+--*-**++o-----o-o@o---------------** B B B
C -*+**+*******o*o@o#oo#o@ooooooo----*+*++* C C C
D -+*+*****+**@ooo#oooooooo#oo#oo-**#**++** D D D
E /****+*++**ooooooooooooo##ooooooo****+++* E E ***** E
F /*+*+****/**@#oo#oooooooooo##@ooo*+****+* F F F
G /**+*++/+///o/o#oo#ooooooo##o##o#o**+*+++ G G +++ @ # G
H /+++++/**//////ooo#o@ooooooooo#o#oo*****+ H H H
I /******//////////#ooooo#/##oo##oooo#+**** I I I
J ///////////////////////////////////////// J J ///////////////////////////////////////// J
2-Kay, Berlin, Maffi & Merrifield 7
Numerous recombinations of the data in the individual speaker arrays were
performed, two of which merit particular mention here. First, the reader will note in the
middle of part 1 of Figure 1 four arrays of the same general shape as those in part 2 of
Figure 1. These also refer to Buglere but they characterize the data for the language sample
as a whole, rather than for each speaker individually. They are labeled Modal Agreement
Level, 30% Agreement Level, 70% Agreement Level and 100% Agreement Level. The
Modal Agreement array displays for each stimulus chip the symbol corresponding to the
term most often applied to that chip, regardless of how often that was. The 30% Agreement
array displays for each stimulus chip the symbol corresponding to the term most often
applied to that chip only if that term was used for that chip by at least 30% of the
respondents; otherwise no symbol is recorded for the chip. The 70% Agreement and the
100% Agreement arrays are constructed correspondingly, according to the obvious
substitutions. These displays are called ‘naming arrays’ because they record a mapping
from stimulus colors to the terms assigned to them in the naming task.
Arrays of the other type to be considered here are called ‘term maps’. The term
maps are illustrated for Candoshi in Figure 2, part 2. There is a separate map for each term.
In the map for a given term, each chip c receives a typographical symbol (including blank)
of visual ‘density’ intuitively commensurate with the frequency with which speakers named
c with that term, this frequency expressed as a proportion of the number of speakers
naming any stimulus with that term.
12
Term maps give a graphic portrayal of the meaning
of each term. High-agreement symbols tend to occur in the interior of categories and lower
agreement symbols at the edges. Term maps also give a quick but accurate insight into the
degree of consensus of speakers regarding the reference of a term. Compare the very high
agreement (‘#’) throughout the blue region in the application of the traditional Candoshi
green/blue term and the lower consensus on the emergent green term and the purple term.
Note finally that the two phonologically similar
13
Candoshi words for yellow have similar
term maps. Term maps provide important information, beyond that given in the naming
2-Kay, Berlin, Maffi & Merrifield 8
arrays, for characterizing an internally variable speech community with respect to its degree
of basic color term development.
2-Kay, Berlin, Maffi & Merrifield 9
Figure 2, part 1: Candoshi (Peru; Jivaroan; Map: x;
14
B. Hinson
15
)
Basic stage IV.
G/Bu
V
Derived categories purple (weak)
Heterogeneous categories desaturated (weak)
Candoshi is transitional between stages IV and V. An original green/blue composite category
(kavabana ‘=’) appears to have recently split and a new term for green (kamachpa ‘*’) has emerged.
Kavabana extends at 30 % agreement to unique green and the green term is almost exclusively confined to
yellowish and brownish greens. Nonetheless, all 11 speakers used kamachpa ‘green’. These facts suggest
that kavabana was originally focused in blue and denoted all of blue or green and that this term is currently
retracting from green. Two similar expressions are found for ‘yellow’ (ptsiyaro and ptsiyaromashi), the
second of which is treated here as a morphological variant of the first.
16
A weak term for purple (tarika ‘P’) has begun to emerge: four speakers have a well-established
word for this category and two show incipient purple.
Finally, a desaturated term (pozani ‘x’) occurs with a discontinuous distribution. It is also weak
and displays low consensus in the term map.
Basic Color Terms
Term Gloss Symbol
kantsirpi ‘black’
borshi ‘white’
o
chobiapi ‘red’
+
ptsiyaro(mashi) ‘yellow’
|
kavabana ‘green/blue (blue-focused)' ‘blue’
=
kamachpa '(emergent) green’
*
tarika ‘purple’
P
pozani ‘desaturated’
x
Aggregate Naming Arrays
Modal Agreement Level
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B oooooooooo|||||oooo=o=ooo==oooooooooooooo B
C o++++++||||||||**===*==*=======oooo++++++ C
D o+++++|||||||||**===*============xx++++++ D
E x++++++|||||*******=================+++++ E
F •++++++|||||*****===================+++++ F
G •+++++x++x*x*****==================++++++ G
H •+++++++x•••••***=***================++++ H
I •++++••••••••••**=**==*=============P++++ I
J ••••••••••••••••••••••••••••••••••••••••• J
30% Agreement Level, 4 of 11 speakers
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B oooooooooo|||||oooo=o=ooo==oooooooooooooo B
C o++++++||||||||**===*==*=======oooo++++++ C
D o+++++|||||||||**===*============xx++++++ D
E x++++++|||||*******=================+++++ E
F •++++++||||| ****===================+++++ F
G •+++++x+ x*****==================++++++ G
H •+++++++ •••••***=***================++++ H
I •++++••••••••••**=**==*=============P++++ I
J ••••••••••••••••••••••••••••••••••••••••• J
2-Kay, Berlin, Maffi & Merrifield 10
70% Agreement Level, 8 of 11 speakers
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B oo oooooo | ooo o ooooo ooooo o B
C o + | = = = = o C
D o+ + * ========== + D
E +++++ | ** =========== +++ E
F +++++ =========== +++ F
G •+++++ * * = = ========== ++++ G
H •++++ ** * ========== ++ H
I • ••••••••* ======== I
J ••••••••••••••••••••••••••••••••••••••••• J
100% Agreement Level, 11 of 11 speakers
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B o o ooo o oo o B
C o C
D = = D
E ++++ = ==== E
F +++ = ====== F
G + ++ === == + G
H • + = ===== H
I • •• ••• ===== I
J ••••••••••••••••••••••••••••••••••••••••• J
2-Kay, Berlin, Maffi & Merrifield 11
Figure 2, part 2: Candoshi
Term Maps
(# = 81-100% agreement, + = 61-80 % agreement, - = 41-60% agreement, . = 21-40% agreement)
•: kantsirpi 'black'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C C
D D
E . E
F - F
G # G
H # .. .-.-- H
I # --+####### I
J ######################################### J
11 of 11 speakers searched; 11 used term
o: borshi 'white'
1 2 3 4
01234567890123456789012345678901234567890
A ######################################### A
B ##+###+##- ##+- - +-#. +#####+###+#++ B
C # ++#- . C
D + . . D
E . E
F F
G G
H H
I I
J J
11 of 11 speakers searched; 11 used term
+: chobiapi 'red'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C --++-. ..-+++ C
D ++++-. --++-+ D
E #####+ .-+### E
F #####-. ..-### F
G #####... .+##+# G
H ####+.. . .++++ H
I ++++ . . .-+- I
J J
11 of 11 speakers searched; 11 used term
|: ptsiyaro 'yellow 1'
1 2 3 4
01234567890123456789012345678901234567890
A A
B .++++- B
C . ..+###+###. . . C
D . .--+###+##-- . . .. D
E . -#+#+#-- . E
F .. .+#+#-. F
G .-.-. G
H . . .. H
I . I
J J
11 of 11 speakers searched; 5 used term: 1..3,5,11
|: ptsiyaromashi 'yellow 2'
1 2 3 4
01234567890123456789012345678901234567890
A A
B .#+##- B
C .+##+##+# C
D -++++##+- D
E .+++#+.. E
F -+-+-. F
G .. G
H H
I I
J J
11 of 11 speakers searched; 8 used term: 1,4..10
=: kavabana 'green/blue'
1 2 3 4
01234567890123456789012345678901234567890
A A
B -.+.. -+. B
C .---.+-.++-+#+# C
D .--+.#+##+#####-- . D
E ..---#########++-+-.. E
F ...-+++###########+-+. F
G ...-++#++#########++-. G
H ..+-..-+##########+.-. H
I -.--+.++#######+--... I
J J
11 of 11 speakers searched; 11 used term
*: kamachpa '(emergent) green'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C --.-.- .- C
D .+---.- . D
E .-+++---- E
F .++++-..- F
G . ++-++--- .. G
H ...+++.-++-. H
I #+.+-. + . I
J J
11 of 11 speakers searched; 11 used term
P: tarika 'purple'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C C
D D
E E
F . -++ F
G . ---. . G
H -.. ------. H
I ++++ --.-#---- I
J J
11 of 11 speakers searched; 6 used term: 1,3,4,6,8,9
x: pozani 'desaturated'
1 2 3 4
01234567890123456789012345678901234567890
A A
B .. .. . . ..... . . . .- B
C --.--.. .. ...... .... -..-.-- C
D . - .. . ..----..-- D
E - .-... E
F . .. . F
G -. ..- . G
H ..... H
I I
J J
11 of 11 speakers searched; 10 used term: 1..4,6..11
2-Kay, Berlin, Maffi & Merrifield 12
3. Recent conceptual developments
Analysis of the WCS data is currently being conducted within the following conceptual
framework, based on our provisional examination of the data (and therefore subject to
revision as the analysis proceeds).
A. Ever since B&K (41-45) discussed the ‘premature’ appearance of grey, evidence
has accumulated suggesting that the temporal development of basic color term systems
should be seen, not as a single process, but as two partially independent processes: (1) the
division of composite categories into the six fundamentals and (2) the combination of
fundamental categories into derived categories.
17
(Recall KBM’s report that purple or
brown or both frequently appear before green and blue separate.) Consequently, the
developmental status of a system is now expressed in terms of a ‘basic stage’, which
characterizes the system with respect to its composite and fundamental categories, plus a
list (often very short) of the derived and heterogeneous
18
categories which correspond to
basic color terms in this system. For example, we might have a system characterized as
“Stage V; purple, pink”, which would be a system containing basic color terms
corresponding to black, white, red, yellow, green, blue, purple and pink. There are just
five basic stages, corresponding to systems containing two to six composite or fundamental
categories. This conceptual simplification leads to a more perspicuous notation for the
sequence of stages, which will be described presently.
B. The categories spanning yellow and green remain a problem, as discussed in
KBM. They are few in number, but they unquestionably exist and cannot be dismissed as
ethnographic or experimental error. A special study of systems containing categories of this
kind is planned. For the moment, systems with a category spanning yellow and green are
set aside. (They are taken up again in connection with Figure 4.)
C. Composite category reduction is itself profitably viewed as consisting of two
partially independent processes: dissolution of the white/warm channel (w) and dissolution
2-Kay, Berlin, Maffi & Merrifield 13
of the black/cool channel (c)
19
. From this perspective, composite category reduction is the
same thing as basic stage evolution, that is, the progressive division of the two original
composite categories into their six constituent fundamentals, representing the sequence of
basic stages I through V. Progress from Stage I (two composite categories comprising
three fundamentals each) to Stage V (six fundamental categories) requires two divisions in
each of the w and c channels.
D. Although w-division and c-division are partially independent processes, they
interact. In our model, the first of the four divisions is always in the w channel, with the
result that Stage II systems retain the 3-fundamental c-composite category (Bk/G/Bu).
Also, the fourth and final division is always in the c channel, entailing that Stage IV
systems always retain a c-composite (and, of course, no w-composite). (See Figure 3
below.)
E. In addition to such constraints on the interaction between the w and c channels,
our model also sets constraints on the process of division within each channel. The w
channel is more tightly constrained than the c channel. These intra-channel constraints are
presented in Table 1 in both words and symbols.
2-Kay, Berlin, Maffi & Merrifield 14
Table 1
w(arm)1:
...
W/R/Y
...
W
R/Y
...
A W/R/Y category
(always) divides
into a W category
and a R/Y category.
w(arm)2:
...
W
R/Y
...
W
R
Y
...
A R/Y category divides
into a R category and a
Y category.
c(ool)1:
...
Bk/G/Bu
...
G/Bu
Bk
...
or
G
Bk/Bu
...
A Bk/G/Bu category
divides either into a
G/Bu category and a
Bk category or into a
G category and a
Bk/Bu category.
c(ool)2:
G/Bu
Bk
...
or
G
Bk/Bu
...
G
Bu
Bk
...
A two-component
cool category (either
G/Bu or Bk/Bu)
divides into its
components.
F. The between-channel and intra-channel constraints introduced by our model
restrict basic stage evolution to the system types and developmental trajectories portrayed in
Figure 3. Within this framework, there are just eight basic system types possible, with
three possibilities at Stage III and two possibilities at Stage IV.
20
G. The limitation to basic stages and to just the types shown in Figure 3 allows a
more transparent notation for types than was previously available. Each of the five types
constituting Stages III and IV is unambiguously represented by subscripting to the roman
numeral denoting the stage an indication of the composite category representing the c
channel, as shown in boldface in Figure 3.
2-Kay, Berlin, Maffi & Merrifield 15
Figure 3
W
R/Y
G/Bu
Bk
III.
G/Bu
w2
W
R
Y
G/Bu
Bk
IV.
G/Bu
c2
W/R/Y
Bk/G/Bu
w1
W
R/Y
Bk/G/Bu
c1
c1
w2
W
R/Y
G
Bk/Bu
III.
Bk/Bu
w2
W
R
Y
G
Bu
Bk
W
R
Y
Bk/G/Bu
III.
Bk/G/Bu
c1
c1
W
R
Y
G
Bk/Bu
IV.
Bk/Bu
c2
I II III IV V
H. Our initial screening of the data indicates that the vast majority of the languages
in the WCS sample fit the model set out in Table 1 and Figure 3 and thus correspond to one
of the eight basic system types shown in Figure 3. One important aspect of the ongoing
analysis of the WCS materials is to evaluate this claim on a careful language-by-language
basis and to establish the extent to which every language in the sample can be revealingly
characterized in terms of this model. It should be noted that, according to the model, a
given stage subtype may be reached by more than a single route. Type IV.
G/Bu
can develop
either from III.
G/Bu
via w2 or from III.
Bk/G/Bu
via c1. Type IV.
Bk/Bu
may develop either
from III.
Bk/Bu
via w2 of from III.
Bk/G/Bu
via c1. Type V may develop, via c2, from either
IV.
G/Bu
or IV.
Bk/Bu
. It is clear from our preliminary analysis that some languages are better
characterized as transitional between subtypes (according to a specific transition; see Table
1) than as belonging to a single stage or type. Also, while some languages seem to be best
characterized as recently emerged instances of their type, others appear to be on the verge
of evolving into a new type. Related to the goal of discovering whether the data of every
language are naturally organized by the model is the converse goal of checking the extent to
2-Kay, Berlin, Maffi & Merrifield 16
which every subtype and transition generated by the model is realized in attested languages
(cf. note 20).
I. When applied to the data from individual speakers, it appears that the specific
inter-category transitions proposed in Table 1 and displayed for full systems in Figure 3,
will go a long way toward ordering language-internal variation as well.
Evaluating this
preliminary generalization constitutes another current research activity.
J. Systems containing yellow/green composites can now be added to the picture, as
shown in Figure 4. Extension of the model to yellow/green systems requires us to add
significant complexity of an ad hoc kind
21
to cover a small amount of data. Yellow/green
systems remain an area that needs careful additional work.
Figure 4
W
R/Y
G/Bu
Bk
III.
G/Bu
w2
W
R
Y
G/Bu
Bk
IV.
G/Bu
c2
W/R/Y
Bk/G/Bu
w1
W
R/Y
Bk/G/Bu
c1
c1
w2
W
R/Y
G
Bk/Bu
III.
Bk/Bu
w2
W
R
Y
G
Bu
Bk
W
R
Y
Bk/G/Bu
III.
Bk/G/Bu
c1
c1
W
R
Y
G
Bk/Bu
IV.
Bk/Bu
c2
W
R
Y/G/Bu
Bk
III.
Y/G/Bu
W
R
Y/G
Bu
Bk
IV.
Y/G
W
R
Y/G
Bk/Bu
III.
Y/G
I II III IV V
2-Kay, Berlin, Maffi & Merrifield 17
K. Two categories have turned up in the preliminary analysis that do not fit any of
the generalizations mentioned so far (see also Greenfield 1987). One is a category of
desaturated, non-vivid or ‘bad’ color. Usually this category contains grey and a diverse
collection of hues that never attain high saturation. An example is Candoshi pozani. Note in
the aggregate naming arrays (Figure 2) that at modal and 30% agreement pozani (‘x’) has a
scattered distribution and that this term does not occur at all in the 70% agreement array.
The term map for this term (Figure 2) shows a wide range, with no chip attaining a high
level of consensus. Compare the lack of ‘#’ and ‘+’ here to the maps for the other terms
(emergent purple being a partial exception). Lack of focus appears to be characteristic of
desaturated terms, and probably of heterogeneous terms generally. Since the WCS data
contain only hues at maximum available saturation, careful study will be required to decide
if and when a ‘desaturated’ term may name an unbroken volume of the color solid. Another
problematical category for which there appears to be some evidence is a category one is
tempted to gloss ‘peripheral red’. Several languages have a term that includes colors on the
long wavelength border of red, such as parts of pink, orange, maroon, or brown, and also
colors on the other, purple, side of red, including a variety of red-purples and lavenders of
different lightness levels. We characterize categories which do not name a continuous area
of the surface of the color solid as heterogeneous.
22
4. Current and future activities of the WCS
The research activities currently underway are conveniently described within the framework
of a planned publication.
23
This is to be a two-volume monograph of which the first
volume is devoted to analysis and the second to presentation of the WCS data in a format
that will make them readily available to all scholars.
It is convenient to describe the second volume first. This volume will present the
full WCS data for each speaker of each language along with some summary information for
that language. A prototype Volume 2 entry for one language, Buglere, is given in Figure 1.
2-Kay, Berlin, Maffi & Merrifield 18
In Figure 1, the initial table gives language name, country, language family (if
known), number of interviewees, name(s) of fieldworker(s), and date of data collection.
The second table lists the terms that occur in the aggregate naming arrays, each term
preceded by the typographical symbol representing it in the arrays which follow. This list
contains every term which was the most popular name given to any chip. It will always
include all the basic terms, and sometimes include one or two non-basic terms as well.
The four aggregate naming arrays at modal, 30%, 70% and 100% agreement appear
as the third item of Figure 1. These have already been discussed.
Following the aggregate naming arrays are a table listing the remaining terms for the
language (those not appearing in the aggregate arrays), and a table representing each native
collaborator by an identifying number, followed by corresponding age and sex
information.
24
On the second page of Figure 1, the individual naming arrays for each collaborator
are given, with naming data on the left and focus data on the right. Figure 1 shows only the
first five of the twenty-five Buglere speakers participating in the study. The full Volume 2
entries will of course include the data from all participating speakers of the languages in
question. Thus, each Volume 2 entry presents the full WCS data on chip naming and focus
identification, arranged in such a way as to maximize their utility to other researchers.
Volume 1 of the proposed monograph will present the analysis of the WCS data.
There will be chapters on a number of theoretical topics, several of which were touched on
above. Chief among these are the accuracy and generality of the hypotheses embodied in
Table 1 and Figures 3 and 4. Also, the nature and extent of heterogeneous categories, the
prevalence in the data (or lack thereof) of the phenomenon of coextension (MacLaury 1986,
1987b,1991,1992), and the special problems posed by yellow/green categories must be
considered.
25
A number of other general issues have not been mentioned. Notable among
these is the treatment of purple in languages which lack a basic term for purple. This
question is important because of the apparently privileged position purple holds
2-Kay, Berlin, Maffi & Merrifield 19
perceptually in ‘closing the hue circle’, that is, shading into short wavelength blue on one
side and long wavelength red on the other just as green and yellow each shades into the
adjacent shorter and longer wavelength colors (see, e.g., J&D: 000). Preliminary
screening of the 110 WCS languages reveals sixteen with a basic term for purple and at
least one undivided composite. No other non-fundamental hue comes close to this number,
suggesting independently a special status for purple.
In addition to chapters, or sections, devoted to the topics sketched in the preceding
paragraph, a significant portion of Volume 1 will be devoted to an analysis of each
language in the sample on the model of the prototype entry for Candoshi given in Figure 2.
In Figure 2, the title line gives the name of the language, the country in which the
data were gathered, the genetic affiliation and an indication of which map the language is
marked on, there being a section with maps indicating the location of each language
elsewhere in the volume.
The table just below the title line gives the evolutionary stage coordinates of the
language in terms of (1) basic stage, (2) derived categories and (3) heterogeneous
categories. The notation ‘IV.
G/Bu
V’ in Figure 2 indicates that Candoshi is classified as
transitional between stages IV.
G/Bu
and V. The full range of possibilities envisaged for
basic stage characterizations are as follows.
X Y in transition from X to Y
X stable X
X entering (‘early’) X
X exiting (‘late’) X
Below the table characterizing the evolutionary stage, there is a portion of text
which reports the analysis, based on the aggregate naming arrays and the term maps (both
shown further on in the figure), which underlies the classification assigned.
26
Candoshi
represents an interesting example partly because it demonstrates how the distinctions
established in connection with Table 1 and Figure 3 can order what might otherwise be
2-Kay, Berlin, Maffi & Merrifield 20
confusing data. It is projected that the set of analytical distinctions proposed here will
permit stage characterizations and brief analyses which capture the main features of internal
variation of each color term system while simultaneously placing it in the developmental
sequence with some finesse. Analysis of the stage status of the language concludes with a
discussion of the derived and heterogeneous categories, if any.
The table in the middle of the first page of Figure 2 presents the basic color terms of
Candoshi. As mentioned, this set of terms will normally coincide with the set of terms
represented in the aggregate naming arrays, although additional criteria are used to
determine the basic color terms. These include all the criteria of B&K: 5-7, especially as
these have been evaluated by the field linguist (in response to instructions accompanying
the field kit).
The list of basic color terms is followed by the four aggregate naming arrays.
27
The second page of Figure 2 presents the term maps for Candoshi.
To summarize, Volume 1 will consist of a number of chapters dealing with
theoretical topics as indicated above, plus a long section containing an analysis of each
language in the WCS sample in the format of the analysis of Candoshi constituting Figure
2.
5. Examples of Individual Color Naming Systems
In this section, we apply the conceptual framework developed above. Here we present
analyses of WCS languages that are representative of the basic stage types predicted by the
theoretical scheme embodied in Table 1 and Figure 3, following the format envisaged for
Volume 1 entries.
Stage I
W/R/Y
Bk/G/Bu
2-Kay, Berlin, Maffi & Merrifield 21
As indicated in footnote 20, the World Color Survey sample includes no languages
exhibiting a Stage I color system, although earlier field research by Rosch on the Dani
shows that such systems do exist and that they conform to the typology suggested here.
Furthermore, while the WCS files contain no single language whose basic stage could be
classified as Stage I, numerous individual speakers in several languages, e.g., Martu-
Wangka of Australia, show Stage I systems of color naming.
Stage II
W
R/Y
Bk/G/Bu
Ejagham (Nigeria, Cameroon; Niger-Congo; J. Watters)
Basic stage II
Derived categories none
Heterogeneous categories none
Ejagham is a Niger-Congo language spoken by 80,000 people in Nigeria (45,000) and Cameroon
(35,000).
28
Its color classification illustrates a typical Stage II system, with terms for Bk/G/Bu (ényàgà),
W (ébáré), and R/Y (ébí) These categories are strongly established at high levels of consensus (80-100 %
agreement in the term maps). In the WCS sample, Stage II systems are found predominantly in Africa.
Basic Color Terms
Term Gloss Symbol
ényàgà ‘black/green/blue’
ébáré ‘white’
o
ébí ‘red/yellow’
+
2-Kay, Berlin, Maffi & Merrifield 22
Aggregate Naming Arrays
Modal Agreement Level
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B ooooooooooooooooooooooooooooooooooooooooo B
C o++++++++++++••••••oooooooooooooooo+o++++ C
D o+++++++++++++•••••••••••••••••oooo++++++ D
E •+++++++++•••••••••••••••••••••••••++++++ E
F •+++++++•••••••••••••••••••••••••••++++++ F
G •+++++•+•••••••••••••••••••••••••••++++++ G
H •+++++•••••••••••••••••••••••••••••++++++ H
I •+++•••••••••••••••••••••••••••••••••++++ I
J ••••••••••••••••••••••••••••••••••••••••• J
30% Agreement Level, 8 of 25 speakers
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B ooooooooooooooooooooooooooooooooooooooooo B
C o++++++++++++••••••oooooooooooooooo+o++++ C
D o+++++++++++++•••••••••••••••••oooo++++++ D
E •+++++++++•••••••••••••••••••••••••++++++ E
F •+++++++•••••••••••••••••••••••••••++++++ F
G •+++++•+•••••••••••••••••••••••••••++++++ G
H •+++++•••••••••••••••••••••••••••••++++++ H
I •+++•••••••••••••••••••••••••••••••••++++ I
J ••••••••••••••••••••••••••••••••••••••••• J
70% Agreement Level, 18 of 25 speakers
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B ooooooooooo ooooooooooooooooooooooooooo B
C o+++ + +++++ oooooooooo ++++ C
D o++++++++++ •••••• • • • ++++++ D
E +++++++++ • •••••••• •••••••• • ++++++ E
F •+++++++ •••••••••••••••••••••••••++++++ F
G •+++++• •••••••••••••••••••••••••• ++++++ G
H •++++ •••••••••••••••••••••••••••• +++++ H
I • + •••••••••••••••••••••••••••••••• + I
J ••••••••••••••••••••••••••••••••••••••••• J
100% Agreement Level, 25 of 25 speakers
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B oo oo o ooo o oo oo oooo o oooooo B
C o o C
D + +++ ++ D
E ++++++ ++++ E
F +++++ •• • • +++ F
G •++ • ••• ••• • • ••• + G
H •++++ • • ••••••••••••••••••••• H
I • • •••••••• ••••••••••••••• I
J ••••••••••••••••••••••••••••••••••••••••• J
Term Maps
(# = 81-100% agreement, + = 61-80 % agreement, - = 41-60% agreement, . = 21-40% agreement)
•: ényàgà 'black/green/blue'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C -.---+---.-... ... C
D . ..++#+###+#+-++++++.... D
E + --+++#######+########+#-- E
F # --+#++#####################. F
G # +-++########################-... G
H # -+###########################+-. H
I #-..-################################--.. I
J ######################################### J
25 of 25 speakers searched; 25 used term
o: ébáré 'white'
1 2 3 4
01234567890123456789012345678901234567890
A ######################################### A
B ##########++++########################### B
C #...-.. ....--.-+-+-+++##+#####-+ . C
D # . . . ...--.....++-+ D
E - . . . E
F F
G G
H H
I I
J J
25 of 25 speakers searched; 25 used term
+: ébí 'red/yellow'
1 2 3 4
01234567890123456789012345678901234567890
A A
B .... B
C +##-#-#####+.. --#### C
D ##########++-. . ###### D
E #########.-.. .-+##### E
F #######-.. .. .+##### F
G #####.-.. -++#### G
H ####-. --+#### H
I -++- . --++ I
J J
25 of 25 speakers searched; 25 used term
2-Kay, Berlin, Maffi & Merrifield 23
Stage III.
G/Bu
W
R/Y
G/Bu
Bk
Múra-Pirahã (Brazil; Unclassified; S. Sheldon)
Basic stage III.
G/Bu
Derived categories none
Heterogeneous categories none
Múra-Pirahã is an unclassified language spoken by a small group of foragers residing in four
villages along the Maici River in west-central Brazil. The language exemplifies a Stage III.
G/Bu
system,
with color terms for four basic color categories (W, R/Y, G/Bu, Bk). All four are well established at the
70% level of agreement in the aggregate naming arrays. Múra-Pirahã naming responses suggest that the
focus of the composite R/Y category is in red while that of G/Bu is in green.
Basic Color Terms*
Term Gloss Symbol Term Gloss Symbol
bio
3
pai
2
ai
3
‘black’
bi
3
i
1
sai
3
‘red/yellow’
+
ko
3
biai
3
‘white’
o
a
3
hoa
3
saa
3
ga
1
‘green/blue’
=
* Raised numerals represent phonemic tones.
Aggregate Naming Arrays
Modal Agreement Level
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B ooooooooooooooooooooooooooooooooooooooooo B
C o++++ooo+++oo=o==o=o=o=oooooooooooooo++++ C
D o++++++++o+ooo================ooooo++++++ D
E •++++++++oo===================oooo+++++++ E
F •+++++++o•o•===================•••+++++++ F
G •+++++••••••••===========•=====••++++++++ G
H •++++•••••••••••===========•==•=•++++++++ H
I •+++•••••••••••••••=••••••••=••=•+•+•++++ I
J ••••••••••••••••••••••••••••••••••••••••• J
30% Agreement Level, 8 of 25 speakers
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B ooooooooooooooooooooooooooooooooooooooooo B
C o++++ooo+++oo=o==o=o=o=oooooooooooooo++++ C
D o++++++++o+ooo================ooooo++++++ D
E •++++++++oo===================oooo+++++++ E
F •+++++++o•o•===================•••+++++++ F
G •+++++••••••••===========•=====••++++++++ G
H •++++•••••••••••===========•==•=•++++++++ H
I •+++•••••••••••••••=••••••••=••=•+•+•++++ I
J ••••••••••••••••••••••••••••••••••••••••• J
70% Agreement Level, 18 of 25 speakers
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B ooooooooooooooooooooooooooooooooooooooooo B
C o+++ oo o =o oooooooooo ++++ C
D o++++++++ ============= = oooo +++++ D
E •++++++ ================ ++++++ E
F •++++++ ============== == ++++++ F
G •+++++• •••••• ========== = == ++++++ G
H •++++ ••••••••• ======== = +++++++ H
I •++ •••••••••••• •• ••• • • • +++ I
J ••••••••••••••••••••••••••••••••••••••••• J
100% Agreement Level, 25 of 25 speakers
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B o ooooooo o oo o o o oooooooo oooooo B
C = o C
D + + + D
E +++ = == = = ++++ E
F ++ ==== +++++ F
G •++ === +++ G
H +++ • ++++ H
I ••• I
J J
2-Kay, Berlin, Maffi & Merrifield 24
Term Maps
(# = 81-100% agreement, + = 61-80 % agreement, - = 41-60% agreement, . = 21-40% agreement)
•: bio
3
pai
2
ai
3
'black'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C C
D . D
E # .. . .. E
F # .+-+.. - -+.. F
G # ++####+#- - . ..--.. G
H # +#########-. . . -..-..-.-. H
I #..-+##########+++--+++++##--#--+-+--. I
J ######################################### J
25 of 25 speakers searched; 25 used term
o: ko
3
biai
3
'white'
1 2 3 4
01234567890123456789012345678901234567890
A ######################################### A
B ######################################### B
C #. .-+#+.. -+-+-.- +-+-+++#########-+ C
D + .. -.++- . ... +####-. D
E . ..-+... . --+.. E
F ...- .. F
G G
H H
I I
J J
25 of 25 speakers searched; 25 used term
+: bi
3
i
1
sai
3
'red/yellow'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C +##-. .--+- -.#### C
D ######+#-+. -+#### D
E ######+-. .-###### E
F ######- -###### F
G ##### . --###### G
H ####. -####### H
I #+-. - -.+### I
J J
25 of 25 speakers searched; 25 used term
=: a
3
hoa
3
saa
3
ga
1
'green/blue'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C . .-.++-#.--+... C
D .-############++++. D
E . ..--##########++####.. . E
F . .++##############-##-. F
G . -##########-#+#++-.. G
H -####+#+#+++--+.+-. H
I ...-+. .-. -+.-- I
J J
25 of 25 speakers searched; 25 used term
Stage III.
Bk/Bu
W
R/Y
G
Bk/Bu
As previously mentioned (cf. note 20), no unequivocal Stage III.
Bk/Bu
language has
yet been attested in the preliminary analysis of the WCS data sets. However, the presence
at Stage IV of four languages with W, R, Y, G, and Bk/Bu and at Stage III of three
yellow/green languages with W, R, Y/G and Bk/Bu indicates that Stage III.
Bk/Bu
systems
are likely to be discovered. Furthermore, Konkomba shows several Stage III.
Bk/Bu
features
and is worthy of discussion here.
Konkomba (Ghana, Togo; Niger-Congo; M.A. Langdon)
Basic stage ΙΙ→ ?III.
Bk/Bu
?IV.
Bk/Bu
Derived categories none
Heterogeneous categories none
2-Kay, Berlin, Maffi & Merrifield 25
Konkomba is a Niger-Congo language spoken in northeastern Ghana (220,000 speakers) and Togo
(50,000 speakers). The aggregate naming arrays for this language suggest that, like many other African
languages, it originally exhibited a Stage II color system but is moving toward a Stage III.
Bk/Bu
system
(developing a term for G), or, alternatively, may be in rapid transition toward IV.
Bk/Bu
(developing in
addition a term for Y). The data suggest that the terms bòmbòn, pipi(i)n, and maman at one time marked
the categories Bk/G/Bu, W, and R/Y, respectively. These terms are used by all 25 speakers in the sample.
29
A new term, Ñaankal, used by 19 speakers, is emerging at 30% agreement level as the name of the category
G (primarily in the light greens, while its full range appears to be that of a G-G/Bu term), leaving bòmbòn
to cover the category Bk/Bu. Finally, the term diyun, used by a small number of speakers (9) and emerging
at modal agreement level, appears to be developing as a term for Y. While maman remains largely the most
popular term for the yellow area of the spectrum, including focal yellow, diyun is a well-established Y for a
majority of its users, as seen in its term map. Its full range indicates that some users extend it to other light
colors in the warm area.
Basic Color Terms
Term Gloss Symbol Term Gloss Symbol
bòmbòn ‘black/blue’
Ñaankal ‘green'
*
pipi(i)n ‘white’
o
diyun 'yellow'
|
maman ‘red/?yellow’
+
Aggregate Naming Arrays
Modal Agreement Level
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B ooooooooooooooooooooooooooooooooooooooooo B
C o++++++++++||******o*oooooooooooooo+o++++ C
D o++++++++++|******•*•*•o*ooooo•oooo++++++ D
E •++++++++++••***•****•••*•••••••o++++++++ E
F •+++++++++•••••••••**••••••••••••••++++++ F
G •+++++++••••••••••••*•••••••••••••+++++++ G
H •+++++++••••••••••••••••••••••••••+++++++ H
I •++++••••••••••••••••••••••••••••••••++++ I
J ••••••••••••••••••••••••••••••••••••••••• J
30% Agreement Level, 8 of 25 Speakers
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B oooooooooo oooooooooooooooooooooooooooooo B
C o++++++++++ * * *o oooooooooooooo+o++++ C
D o++++++++++ *****•*•* *oooo •oo o++++++ D
E •+++++++++ • **•****•••*•••••• +++++++ E
F •+++++++++ ••••••••**••••••••••••••++++++ F
G •+++++++••••••••••••*•••••••••••••+++++++ G
H •+++++++••••••••••••••••••••••••••+++++++ H
I •++++••••••••••••••••••••••••••••••••++++ I
J ••••••••••••••••••••••••••••••••••••••••• J
70% Agreement Level, 18 of 25 Speakers
1 2 3 4
01234567890123456789012345678901234567890
A A
B o o B
C + C
D ++++++++ ++++ D
E ++++++ • +++++ E
F ++++++ • ++++++ F
G •+++++ • •• • +++++ G
H •++++ ••••••• •• •••••••••• +++++ H
I •+ •••••••••••••••••••••••••• I
J ••••••••••••••••••••••••••••••••••••••••• J
100% Agreement Level, 25 of 25 Speakers
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C C
D D
E +++++ E
F ++++ +++ F
G ++ ++ G
H ++ • • + H
I I
J J
2-Kay, Berlin, Maffi & Merrifield 26
Term Maps
(# = 81-100% agreement, + = 61-80 % agreement, - = 41-60% agreement, . = 21-40% agreement)
•: bòmbòn 'black/blue'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C C
D . ..... . . D
E - . .-.. .-...-.--+-.. E
F - ........--..-.--++-+---.-- F
G # -.--+-.-----.+++-#++-++--.. G
H # ...-+#+#++++-++-+#++######+--. H
I #....+++###+#######+#+++#########-+--.... I
J ######################################### J
25 of 25 speakers searched; 25 used term
o: pipin 'white'
1 2 3 4
01234567890123456789012345678901234567890
A +++++++++++++++++++++++++++++++++++++++++ A
B +++-----+-....++#+++--++++++++++++++----+ B
C + .. -..-...++-.-++-+.... C
D + . .....-...-... D
E . . . .. E
F F
G G
H H
I I
J J
25 of 25 speakers searched; 23 used term:
2,4..25
o; pipiin 'white'
1 2 3 4
01234567890123456789012345678901234567890
A ----------------------------------------- A
B --.--....... ..- ------..-.----.-..----.. B
C .. . ... ...--.-.. C
D . . .. .. D
E . E
F F
G G
H H
I I
J J
25 of 25 speakers searched; 21 used term:
1..3,6..10,12,14..25
+: maman 'red/?yellow'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C -+++--.--- . . ---- C
D #+#++##+.-. +-+#++ D
E ######-+-. ..+##### E
F ######+-. . . +##### F
G #####--. .-+##### G
H ####---. +++#### H
I +++-.. ....---- I
J J
25 of 25 speakers searched; 25 used term
*: Ñaankal 'green'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C . -.-..+.. . . C
D ..++-#---.-..-.. D
E ....-+--+++--.+.... . .. E
F . --.--------..... .. F
G .. ---.--+.... ...... . G
H ....... . H
I I
J J
25 of 25 speakers searched; 19 used term:
2..10,12..15,17,20..24
|: diyun 'yellow'
1 2 3 4
01234567890123456789012345678901234567890
A A
B ..- . B
C . . -.++-+-.. .. ....-. C
D .--+... . . D
E . ...- . E
F F
G G
H H
I I
J J
25 of 25 speakers searched; 9 used term:
5..8,10,11,19,21,24
Stage III.
Bk/G/Bu
W
R
Y
Bk/G/Bu
Kwerba (Irian Jaya, Indonesia; Trans-New Guinea; J. and S. De Vries)
Basic stage III.
Bk/G/Bu
Derived categories none
Heterogeneous categories none
Kwerba, a Trans-New Guinea language, is spoken by some 1500 people in the Upper Tor River
area of Irian Jaya, Indonesia (western half of the island of New Guinea). It typifies an early Stage
III.
Bk/G/Bu
system. In this language, an expression is attested for the composite category Bk/G/Bu, words
2-Kay, Berlin, Maffi & Merrifield 27
for W and R are well established, and a term for Y, kainanesènum, has begun to emerge. Sixteen of the
twenty-five speakers interviewed use this term, and the category appears to be well on its way to becoming
fully established for the language as a whole.
Basic Color Terms
Term Gloss Symbol Term Gloss Symbol
icèm ‘black/green/blue’
nokonim ‘red’
+
ësiram
(ëhèrèm, ërèm)*
‘white’
o
kainanesènum ‘yellow’
|
* Terms in parentheses are synonyms for ësiram .
Aggregate Naming Arrays
Modal Agreement Level
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B ooooooooooooooooooooooooooooooooooooooooo B
C o+++o+oo|||||o|oooooooooooooooooooooo++++ C
D o+++++++|||||o|•••••••ooo•••ooooooo++++++ D
E •+++++++|•||•|oo•••o••••••o••••ooo+++++++ E
F •+++++++••••••••o••••••••••••••••••++++++ F
G •+++++••••••••••••••••••••••••••••+++++++ G
H •+++++••••••••••••••••••••••••••••+++++++ H
I •+++•••••••••••••••••••••••••••••••••++++ I
J ••••••••••••••••••••••••••••••••••••••••• J
30% Agreement Level, 8 of 25 speakers
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B ooooooooooooooooooooooooooooooooooooooooo B
C o+++o+o ||||| |o oooooooooooooooooooo++++ C
D o+++++++|||||o| • •• oo•••ooooooo++++++ D
E •+++++++ ||•|o •••o••• •• ••••ooo+++++++ E
F •+++++++••••••••o••••••••••••••••••++++++ F
G •+++++••••••••••••••••••••••••••••+++++++ G
H •+++++••••••••••••••••••••••••••••+++++++ H
I •+++•••••••••••••••••••••••••••••••••++++ I
J ••••••••••••••••••••••••••••••••••••••••• J
70% Agreement Level, 18 of 25 speakers
1 2 3 4
01234567890123456789012345678901234567890
A A
B o o o oo oooooo o ooo o oooo B
C o o C
D +++++ +++++ D
E ++++++ ++++++ E
F •++++++ • • ++++++ F
G •+++++ •• •• ••• ••• ++++++ G
H •+++++ ••••••••••• •••••••••••••• +++++++ H
I •+++ •••••• •••••• •••••••••••••• ++++ I
J ••••••••••••••••••••••••••••••••••••••••• J
100% Agreement Level, 25 of 25 speakers
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C C
D + D
E + + + E
F + ++ + F
G ++ + + G
H +++ + H
I I
J J
Term Maps
(# = 81-100% agreement, + = 61-80 % agreement, - = 41-60% agreement, . = 21-40% agreement)
•: icèm 'black/green/blue'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C . . . .. C
D . ..-.-......... . D
E - . . .-.-.---.--.---+... E
F + .+.-+---.-+-++++-+-+++#+--- F
G # --++++#++-++---+++++#++++#++ G
H # .+++#+####+#+++#+#++##+##+++- H
I # -######+##++#++#+#+##+##+##+#+--- . I
J ######################################### J
25 of 25 speakers searched; 25 used term
o: ësiram 'white'
1 2 3 4
01234567890123456789012345678901234567890
A +++++++++++++++++++++++++++++++++++++++++ A
B ++++++++++-----++++++++#+++++++++#+++#+++ B
C +.....-. ......-.-.--+--++++--+++++-+-... C
D - . .. ... ............--.---+. D
E . ........ . ... . ...-... E
F . . F
G . . . G
H . H
I . I
J J
|
25 of 25 speakers searched; 24 used term:
1..9,11..25
2-Kay, Berlin, Maffi & Merrifield 28
+: nokonim 'red'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C -++.-. . . . ---+ C
D ++##++- -####+ D
E ######-. .-+##### E
F #####+- .-###### F
G ##### . .+###### G
H ####+ .+###### H
I ###- ..- #+## I
J J
25 of 25 speakers searched; 25 used term
|: kainanesènum 'yellow'
1 2 3 4
01234567890123456789012345678901234567890
A A
B .... B
C . ..+++#+--. C
D . --++++--.. D
E --.+-.+. E
F .- .. F
G G
H H
I I
J J
25 of 25 speakers searched;16 used term:
1,2,4..6,8..13,15,17,18,23,24
Stage IV.
G/Bu
W
R
Y
G/Bu
Bk
Sirionó (Bolivia; Tupí; P. and A. Priest)
Basic stage IV.
G/Bu
Derived categories none
Heterogeneous categories none
Sirionó is a Tupian language spoken by approximately 500 individuals in the eastern Beni and
northwestern Santa Cruz departments of the Bolivian lowlands. It is classified as Stage IV.
G/Bu
.
The
language shows well established terms for W, R ,Y, G/Bu, Bk. The G/Bu composite category is focused in
blue.
Basic Color Terms
Term Gloss Symbol Term Gloss Symbol
erondeì ‘black’
echo ‘yellow’
|
eshiÕ ‘white’
o
eruba ‘green/blue’
=
eìreõIõ ‘red’
+
2-Kay, Berlin, Maffi & Merrifield 29
Aggregate Naming Arrays
Modal Agreement Level
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B oooooooooo|||||oooooooooooooooooooooooooo B
C o|++++|||||||||||o=o==o=o=oo=oooooo|o++|+ C
D o++++||||||||||||==============oooo++++++ D
E •+++++|||||||||===================+++++++ E
F •++++++|||||||||===================++++++ F
G •+++++•••••••||====================++++++ G
H •++++••••••••••===================+++++++ H
I •+++••••••••••••••=•••••••••====•=•••+++• I
J ••••••••••••••••••••••••••••••••••••••••• J
30% Agreement Level, 8 of 24 speakers
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B oooooooooo|||||oooooooooooooooooooooooooo B
C o|++++||||||||||| =o= o o=oo=oooooo|o++|+ C
D o++++||||||||||||==============oooo +++++ D
E •+++++|||||||||===================+++++++ E
F •++++++||||| |||================== ++++++ F
G •+++++ ••••••||====================++++++ G
H •++++••••••••••===================+++++++ H
I •+++••••••••••••••=•••••••••====• • +++ I
J ••••••••••••••••••••••••••••••••••••••••• J
70% Agreement Level, 17 of 24 speakers
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B ooooooooo oooo o ooooooooooooooooooo B
C o |||||||| ooo C
D o+ + |||||||| == = = + + D
E +++++ |||| || ==== ==== +++++ E
F •+++++ = ==== ++++ F
G •++++ = = == === +++++ G
H •++++ • ••••• ++++ H
I •+ •••••••••• + I
J ••••••••••••••••••••••••••••••••••••••••• J
100% Agreement Level, 24 of 24 informants
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B o o B
C C
D D
E E
F + ++ F
G ++ ++ G
H + + H
I I
J J
Term Maps
(# = 81-100% agreement, + = 61-80 % agreement, - = 41-60% agreement, . = 21-40% agreement)
•: erondeì 'black'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C C
D D
E - . . E
F + .. ... F
G # .++----. . . . G
H # .++-#####+-. .... .. . H
I # ..-+###+#####+-- -.+-++-++-- .. ... . I
J ######################################### J
24 of 24 speakers searched; 24 used term
o: eshiÕ white'
1 2 3 4
01234567890123456789012345678901234567890
A ######################################### A
B ##+##+###+... .+#++++-#+##+############## B
C # . . .- ---+-+-###+ . C
D + . .. --.. D
E . E
F F
G G
H H
I I
J J
24 of 24 speakers searched; 24 used term
+: eìreõIõ 'red'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C .--.- ..--.- C
D ++#-.. .++++- D
E +####. ..-+#### E
F ####+-. -+#### F
G ####- .++#### G
H ####. ---#### H
I +-+. . .--+. I
J J
24 of 24 speakers searched; 24 used term
|: echo 'yellow'
1 2 3 4
01234567890123456789012345678901234567890
A A
B . .+-++- B
C .. ..-#######+--. .. . . ... C
D .-+#####+++--.. . . . D
E .+++#-.++... E
F ...----..... F
G . .. .- G
H H
I I
J J
24 of 24 speakers searched; 24 used term
=: eruba 'green/blue'
1 2 3 4
01234567890123456789012345678901234567890
A A
B .. B
C ...-.-....-..- . C
D ..-.+-++--#-+++--... D
E .. ----++++#+-++##+--. E
F .....-+---+++#++#+#+--.. F
G .----+---+++++##+++.. G
H ..---+--.++-+-++++--... H
I .. ..-... ....--+-..... . . I
J J
24 of 24 speakers searched; 24 used term
2-Kay, Berlin, Maffi & Merrifield 30
Stage IV.
Bk/Bu
W
R
Y
G
Bk/Bu
Martu-Wangka (Australia; Pama-Nyungan; J. and M. Marsh)
Basic stage IV.
Bk/Bu
Derived categories none
Heterogeneous categories peripheral red?
Martu-Wangka is an Australian (Pama-Nyungan) language spoken by about 820 people in the
Jigalong area of Western Australia. It is classifed in the WCS as Stage IV.
Bk/Bu
, with terms for W, R, Y,
G, and a composite color category encompassing black and blue. At the 30% level of agreement, parnaly-
parnaly, a term restricted in its distribution to brownish reds, is used by 15 speakers. According to the field
linguists for Martu-Wanka, parna is the word for ‘earth, ground, sand’ and “probably does not qualify as a
‘basic’ color term, as it would appear to mean ‘earth-like’.” The term map for parnaly-parnaly, however,
suggests a possible meaning of ‘peripheral red’. Although the judgment of the field linguists that parnaly-
parnaly should not be considered a basic color term is probably deserving of acceptance, we have included
the discussion of this term for completeness and to illustrate the kind of borderline cases that can arise in
analyzing the WCS data.
Martu-Wangka color terms commonly exhibit reduplication and appear to be derived from verbs or
nouns, e.g., maru-maru ‘black, blue’ < maru ‘to darken or become black’; miji-miji ‘red’ < miji ‘blood’;
yukuri-yukuri ‘green’ < yukuri ‘grass’. The unreduplicated term karntawarra is the word for ‘yellow ochre’.
Basic Color Terms
Term Gloss Symbol Term Gloss Symbol
maru-maru ‘black/blue’
karntawarra ‘yellow’
|
piila-piila/piily-piily/
pily-pily/pilya-pilya/
pira-pira/piirl-piirl/
piily/pilya/pirilypa/
pirilyi/pirly
‘white’
o
yukuri-yukuri,
yakuripiti
‘green’
*
miji-miji ‘red’
+
parnaly-parnaly,
parna
‘peripheral red’?
~
2-Kay, Berlin, Maffi & Merrifield 31
Aggregate Naming Arrays
Modal Agreement Level
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B oooooooooo||**ooooooooooooooooooooooooooo B
C o++++++|||||************ooooooooooo++++++ C
D o+++++|||||*************•••••••ooo+++++++ D
E •++++++||||*************••••••••+++++++++ E
F •+++++~~~•~************•••••••••+++++++++ F
G •+++++~+~•••*••*******••••••••••+++++++++ G
H •++++~•••••••••••***•••••••••••••++++++++ H
I •++++••••••••••••••••••••••••••••+••+++++ I
J ••••••••••••••••••••••••••••••••••••••••• J
30% Agreement Level, 8 of 25 speakers
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B oooooooooo|| *ooooooooooooooooooooooooooo B
C o+++ + |||||************ ooooooooo +++++ C
D o+++++|||||*************•• •••• oo+++++++ D
E •++++++ | |*************•••••••• ++++++++ E
F •+++++~~ ************•••••••••+++++++++ F
G •+++++~ ~•••*••*******•••••••••• ++++++++ G
H •++++~•••••••••••***•••••••••••••++++++++ H
I •++++••••••••••••••••••••••••••••+••+++++ I
J ••••••••••••••••••••••••••••••••••••••••• J
70% Agreement Level, 18 of 25 speakers
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C * **** C
D ********** ++ D
E +++ ********* + +++ E
F ++++ ********* •• +++++ F
G •+++ ****** • • ++++++ G
H •+ ++ ••••• ••••••• • ++++ H
I • ••••••••••••••••••••••••• I
J ••••••••••••••••••••••••••••••••••••••••• J
100% Agreement Level, 25 of 25 speakers
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C C
D D
E E
F F
G + G
H H
I I
J J
Term Maps*
(# = 81-100% agreement, + = 61-80 % agreement, - = 41-60% agreement, . = 21-40% agreement)
* A single linguistic variant for each basic category is shown in the arrays
•: maru-maru 'black/blue'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C . ... .. C
D ..-.-..-. D
E + ...--.--++- E
F - . .. . . ...-+++++++-. F
G + ...--.-- ......-+++++++++. G
H # .--#+###+-+-.--+++##++#+++-. H
I # .-+##################+####+-+.-.. ... I
J ######################################### J
25 of 25 speakers searched; 25 used term
o: piila-piila 'white'
1 2 3 4
01234567890123456789012345678901234567890
A ######################################### A
B +++++-++-+ .. -+-+-++-+-++-+++++++-++-+++ B
C -. ... . .- .------+-+... . C
D - . . .... .... D
E . . E
F . F
G G
H H
I I
J J
25 of 25 speakers searched; 21 used term:
1..5,7..9,11..18,20..23,25
+: miji-miji 'red'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C -..... ...---- C
D +----. ..--+++- D
E #++++- .+-+++#++ E
F ####- . ---+++### F
G ###+- . .-++#+### G
H #+#+. .--+++##+ H
I ----. . -...---. I
J J
25 of 25 speakers searched; 25 used term
|: karntawarra 'yellow'
1 2 3 4
01234567890123456789012345678901234567890
A A
B ---. B
C -+#--. C
D -#+++- D
E ---- E
F . . F
G G
H H
I I
J J
25 of 25 speakers searched; 17 used term:
1,3..5,7..9,12,13,15,16,18..20,22,24,25
*: yukuri-yukuri 'green'
1 2 3 4
01234567890123456789012345678901234567890
A A
B ..-. B
C ..-#+####++-.- C
D .+##########. . D
E .-+########++-.. E
F --+######+#-.. F
G ....-###++++.. G
H -.++-... H
I I
J J
25 of 25 speakers searched; 23 used term:
1..5,7..13,15..25
~: parnaly-parnaly 'peripheral red'?
1 2 3 4
01234567890123456789012345678901234567890
A A
B . B
C . .-..-. ... C
D ..... . D
E . .-... . E
F .#+--.. F
G .-#.+-... G
H .. ---+.. . H
I ... I
J J
25 of 25 speakers searched; 15 used term:
1..3,5,7,9,11..13,15,16,21,23..25
2-Kay, Berlin, Maffi & Merrifield 32
Stage V
W
R
Y
G
Bu
Bk
Kalam (Papua New Guinea; Trans-New Guinea) [L. Scholz]
Basic stage V
Derived categories none
Heterogeneous categories none
Kalam is a Trans-New Guinea language spoken by 15,000 people in the Hagen district of the
Western New Guinea Highlands. It is classified as a Stage V system, with distinct terms for each of the six
fundamental categories, W, R, Y, G, Bu, and Bk. At the modal level of agreement, a single purple chip is
given the name anjeÑ-ay. Eight of the twenty-five Kalam collaborators use this term; in all cases it has an
uneven, roughly ‘peripheral red’ distribution. AnjeÑ-ay is probably best not considered a basic color term of
Kalam.
Basic Color Terms
Term Gloss Symbol Term Gloss Symbol
mosimb ‘black’
walin ‘yellow’
|
tund ‘white’
o
minj-kimemb ‘green’
*
likañ ‘red’
+
muk ‘blue’
-
Aggregate Naming Arrays
Modal Agreement Level
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B oooooooooo|||||oooooooooooooooooooooooooo B
C o|||||||||||||||*********oooooooooooo+oo+ C
D o+|+|||||||||||********-**-*--ooooo*+++++ D
E •+++++||||||||**********---------o-++++++ E
F o++++++|||||||**********-----------++++++ F
G •+++++||•••||•*********-----------*++++++ G
H •++++•••••••••********------------+++++++ H
I •++++•••••••••••*****•*-•---------++G++++ I
J ••••••••••••••••••••••••••••••••••••••••• J
30% Agreement Level, 8 of 25 speakers
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B oooooooooo|||||oooooooooooooooooooooooooo B
C o| |||||||||||*********oooooooooooo C
D o+ +|||||||||||********-**-*--ooooo ++++ D
E +++++||||||||**********--------- ++++++ E
F o+++++ |||||||**********--------- -++++++ F
G •+++++ | •| *********---------- ++++++ G
H •++++••• • •••********------------+++++++ H
I •++++ ••••••••••*** * *-•--------- + ++++ I
J ••••••••••••••••••••••••••••••••••••••••• J
2-Kay, Berlin, Maffi & Merrifield 33
70% Agreement Level, 18 of 25 speakers
1 2 3 4
01234567890123456789012345678901234567890
A ooooooooooooooooooooooooooooooooooooooooo A
B oooooooooo|||| ooo oo ooooooooooooooooooo B
C o ||||||| * o oooooo o C
D o ||||||| ** o D
E ++ + ||||| * - ++ E
F +++ * - ---- +++ F
G + ++ ** * ----- +++ G
H ++++ - ---- +++++ H
I +++ ---- ++ I
J J
100% Agreement Level, 25 of 25 speakers
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C o C
D D
E E
F F
G G
H H
I I
J J
Term Maps
(# = 81-100% agreement, + = 61-80 % agreement, - = 41-60% agreement, . = 21-40% agreement)
•: mosimb 'black'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C C
D . D
E . E
F . .. . F
G + ...-... G
H - ---.-.----. H
I - .--+--+-+--.- ... .-.. . . I
J ######################################### J
25 of 25 speakers searched; 18 used term:
1..6,8..11,14,16..19,21,23,24
o: tund 'white'
1 2 3 4
01234567890123456789012345678901234567890
A ######################################### A
B #####+##++ -+##++++################### B
C #. ... . .-#++#+####-+ .. C
D + -+-#+ D
E . . . E
F - F
G G
H H
I I
J J
25 of 25 speakers searched; 25 used term
+: likañ 'red'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C .. . . C
D - -. .-.-. D
E #++#+. -.+++# E
F ###++. -++##+ F
G #+##+ ..+++#+# G
H ####. --+###+ H
I #+#- .. ++## I
J J
25 of 25 speakers searched; 23 used term:
1..18,20..23,25
|: walin 'yellow'
1 2 3 4
01234567890123456789012345678901234567890
A A
B ..####-. B
C ....--+#####++- . .. C
D . .-#######+- D
E .#####+-. E
F .---+-.. F
G . ... . G
H H
I I
J J
25 of 25 speakers searched; 24 used term:
1..6,8..25
*: minj-kimemb 'green'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C ---++---+. C
D ..-++++----.-.. D
E .-+++-++.--.... . E
F ...++#++-+-+-. . F
G .-+++#+++-... . G
H ..-++-+--.-.. H
I .-.-.- ... I
J J
25 of 25 speakers searched; 25 used term
-: muk 'blue'
1 2 3 4
01234567890123456789012345678901234567890
A A
B B
C . C
D ..-.-+.+-. D
E ......---+#++.. . E
F ......-++++###+-.- F
G ... ----++###++.. G
H .. ..--++#++#+#+- H
I . .. ...-.++++##+-.. I
J J
25 of 25 speakers searched; 25 used term
Notes
1
Maffi (1991) provides relevant bibliography.
2-Kay, Berlin, Maffi & Merrifield 34
2
Several studies in addition to Rosch's are cited in Kay (1975) as supporting the idea of
color term evolution involving category boundaries as well as foci.
3
A fuzzy set is a function from a (standard) set of objects to a real interval, conventionally
the interval between zero and unity inclusive.
4
These are called ‘fundamental neural response categories’ in the K&McD model. At the
time, the opponent hue primaries of Hering (red, yellow, green, blue) were considered by
vision researchers to benefit from direct neurophysiological confirmation in the response
characteristics of certain LGN cells of the rhesus macaque (De Valois, et al. 1966). This
simple model of the neurological substrate for the perceptual phenomena of color
categorization (e.g., Sternheim and Boynton 1966, Wooten 1970) has more recently been
replaced by more complex models, based on spatial as well as spectral opponency, by
interaction within cells of color and luminance information, and on the behavior of a wider
range of neural structures, including the cones, the horizontal cells, and the bi-polar and
ganglion cells (e.g., De Valois and De Valois 1993). It is now recognized that the 1966
model of De Valois et al. failed to account for as wide a range of the perceptual phenomena
of color as was originally thought. The validity of the six perceptually salient Hering
primaries retains broad consensus in the vision research community, as does the conviction
that a fully satisfying neurophysiological derivation must eventually be forthcoming
(Abramov this volume).
Jameson and D'Andrade (this volume, hereafter J&D) propose to drive a further
wedge between the Hering primaries and their neurophysiological substrate. J&D argue,
with regard to the psychophysical level, that neither cancellation experiments nor after-
image facts support perceptually unique red, green, yellow and blue as determinants of the
axes of chromatic opponency. At the physiological level, they point to Abramov and
2-Kay, Berlin, Maffi & Merrifield 35
Gordon's (1994) observation that the crossover points of recorded LGN cells do not
correspond well to the phenomenal unique hue points, but rather suggest axes like bright
red/greenish blue and yellow-green/dark purple (although this latter point is effectively
answered by De Valois and De Valois 1993, which J&D also cite).
J&D locate the Hering primaries at the level of a conceptual (or semantic) ‘color
space’, a higher-level cognitive object, whose properties are inferred from the application
of multidimensional scaling techniques to judged similarities among pairs of colors and
whose relation to the psychophysics of color J&D acknowledge to be uncertain. Assuming,
for the sake of brevity, that J&D are correct in all of this, the Hering primaries are deprived
of significant psychophysical support.
Nevertheless, J&D accept the phenomenal reality of the Hering primaries. Indeed,
they attempt to provide for them a different psychological substrate than that of standard
opponent theory. The Hering primaries, whatever perceptual rationale they are finally
accorded, remain a major interface between color vision and the semantics of color.
In unpublished work, Kemmerer (1995) argues that color categories cannot be
represented at the ganglion/LGN level. Color constancy effects (modeled by Land, e.g.
1974) show that color categorization requires comparison of signals arising at points in the
retina further separated than the diameter of the largest area represented by a ganglion or
LGN cell. Based on the work of Zeki (most recently, 1993) and others, Kemmerer
proposes a cortical model of the B&K findings. This model posits comparison and
recoding of color signals at a series of cortical levels, including the V1, V2, V4 and TO
areas.
5
A derived category is defined as twice the fuzzy intersection (that is, twice the minimum)
of the fuzzy categories from which it is derived. Consider the case of red, yellow and the
derived category orange. It follows from the fuzzy set implementation of the opponent
2-Kay, Berlin, Maffi & Merrifield 36
process model that the red function and the yellow function sum to unity throughout the
range in which they are both non-zero. The derived category orange has a fuzzy set
function that rises from zero at the red and yellow unique hue points to unity at the point at
which the red and yellow fuzzy categories both have ordinates of .5. This procedure
expresses formally the observations (i) that the more equal the mixture of red and yellow
perceived in a color, the more orange that color appears and (ii) that a color that consists
perceptually of an equal (non-zero) mixture of red and yellow is as orange as you can get.
6
Funded by NSF grants BNS 76-14153, BNS 78-18303, BNS 80-06802, and SBR
9419702; also supported by the Summer Institute of Linguistics (SIL), the Anthropology
and Linguistics Departments and the Institute of Cognitive Studies at University of
California at Berkeley, and the International Computer Science Institute, Berkeley, CA. All
these sources of support are gratefully acknowledged.
7
Scientific challenges such as these merit serious attention. Epistemological and/or
deontological critiques have also appeared, which empirical research is not equipped to
address. For example, Saunders and van Brakel consider K&McD's “reductionist
argument” according to which “six basic or atomic color categories ... can be reduced to
Fundamental Neural Response categories,” as invalidated by the prior epistemological tenet
that “... there is no privileged discourse in which what is true is independent of our
choices, hopes and fears” (Saunders and van Brakel 1994: 8). The Western scientific
tradition presupposes the existence of an objective world independent of human choices,
hopes or fears. To suppose that the world exists independent of human sentiments is not,
of course, to conclude that unbiased construals of that world are easy to achieve or that
science provides a magic formula for avoiding bias. The empirical researcher believes that
one can reduce (not eliminate) bias through the disciplined application of procedures of
2-Kay, Berlin, Maffi & Merrifield 37
observation and inference designed specifically with the reduction of bias as their object
and trusts that the exercise of this discipline can sometimes result in one type of
understanding: scientific understanding. Saunders and van Brakel are not atypical of those
post-modernists who leap from the observation that the attainment of scientific
understanding is not trivial to the conclusion that it is not possible. We assume,
contrariwise, that the existence of science provides strong evidence of its possibility. A
comprehensive evaluation of the post-modern critique of research on color naming is
beyond the scope of this paper. (See Hardin 1993 and Stanlaw 1993 for careful appraisals
of several points.)
8
See K&McD: 639. B&K and Kay (1975) had earlier noted a few exceptions to this rule
with regard to grey and brown, but these cases were left unexplained by the generalizations
embodied in the evolutionary sequence.
9
Figures 1 and 2 contain more information than is conveniently explained at their first
introduction. All the features in the figures will be fully explained in due course. The
reader's patience is requested for the moment.
10
Column 0 presents the neutral white-to-black sequence.
11
All forty entries of row A denote a single pure white chip and all of row J a single pure
black chip.
12
For example, if at least 81% of the speakers who name any chip with the term being
mapped name chip c with that term, then c receives '#'. If 61 - 80% of the speakers who
name any chip with the term being mapped name chip c with that term, then c receives ‘+’.
2-Kay, Berlin, Maffi & Merrifield 38
If 41 - 60% of the speakers who name any chip with the term being mapped name chip c
with that term, then c receives ‘–’. And so on, as indicated in the legend above the term
maps in part 2 of Figure 2.
13
And almost certainly morphologically related.
14
See Section 4 for details on maps.
15
Names in square brackets following a language name indicate the field linguist(s) who
gathered data on that language for the WCS. We acknowledge with gratitude the work, not
only of the SIL field linguists whose names appear here, but also of each of the over 100
such persons contributing to this study.
16
This is supported by their almost identical distributions in their respective term maps.
Therefore, in the aggregate naming arrays, they are assigned the same symbol '|'.
17
MacLaury (1986) was the first to suggest this.
18
Heterogeneous categories are discussed below.
19
The use of the word ‘channel’ here is motivated by the fact that the grammar of English
requires that some noun be employed and a choice like ‘what's-its-name’ or ‘thingamabob’
could be distracting. In particular, no pretense of denoting a neurological entity is intended.
20
Some caveats apply. First, the reader is reminded that we have not yet introduced
consideration of languages containing words for yellow/green categories (cf. Fig. 4).
2-Kay, Berlin, Maffi & Merrifield 39
Secondly, the WCS sample contains no examples of Stage I systems, although their
existence is documented elsewhere, and therefore noted in Fig. 3. Thirdly, our initial
screening of the WCS sample discloses no unequivocal example of type III.
Bk/Bu
(but cf.
description of Konkomba in Section 5), although it indicates four languages at Stage IV
with W, R, Y, G and Bk/Bu and three yellow/green languages at Stage III with W, R, Y/G
and Bk/Bu (see Figure 4).
21
That is, there are no evident generalizations, comparable to those summarized in Table
1, regarding yellow/green transitions.
22
Further analysis may show the peripheral red category to describe an ‘unbroken’ region
of the surface of the color solid in the sense that the surface of a lake with an island may be
said to present an unbroken expanse of water. Thus the heterogeneous categories may turn
out to be less bizarre than they appear at first sight.
23
The format described here reflects our current thinking on the monograph. These
decisions are subject to revision as the work proceeds.
24
More information on speakers than this was gathered. The decision to restrict published
information on individual speakers to age and sex stems from our initial evaluation of space
constraints.
25
These are spelled out in some detail in KBM.
26
Consideration of the individual speaker data may also enter into this analysis.
2-Kay, Berlin, Maffi & Merrifield 40
27
The aggregate naming arrays also appear in the Volume 2 entries, as illustrated in Figure
1 for Buglere. This redundancy has been thought desirable to make each volume relatively
self-contained.
28
Here and in the following examples, geo-demographic data on individual languages, as
well as their language family ascriptions, are derived from Grimes (1992).
29
In the case of the W category, this applies to the combined data for the terms pipin and
pipiin, which are analyzed as variants and given the same symbol 'o' in the aggregate
naming arrays. The separate term maps for pipin and pipiin support this analysis by
showing their overlapping distribution.
2-Kay, Berlin, Maffi & Merrifield
References
Abramov, I. (this volume). Physiological mechanisms of color vision. 000.
Abramov, I. and J. Gordon 1994. Color appearance: On seeing red--or yellow, or green,
or blue. Annual Review of Psychology 45: 451-485.
Berlin, B. and P. Kay 1969. Basic Color Terms: Their Universality and Evolution.
Berkeley and Los Angeles: University of California Press. First paperback edition
1991, with a bibliography by Luisa Maffi.
De Valois, R.L. and K.K. De Valois 1993. A multi-stage color model. Vision Research
33(8): 1053-1065.
De Valois, R.L., I. Abramov, and G.H. Jacobs 1966. Analysis of response patterns of
LGN cells. Journal of the Optical Society of America 56: 966-977.
Greenfield, Ph.J. 1987. What is grey, brown, pink,and sometimes purple: The range of
“wild-card” color terms. American Anthropologist 88: 908-916.
Grimes, B. (ed.) 1992. Ethnologue. Languages of the World. 12th ed. Dallas: Summer
Institute of Linguistics.
Hardin, C.L. 1993. Van Brakel and the not-so-naked emperor. British Journal for the
Philosophy of Science 44(1): 137-150; 44(2): 377 [correction].
2-Kay, Berlin, Maffi & Merrifield
Heider [Rosch], E. 1972a. Universals in color naming and memory. Journal of
Experimental Psychology 93: 10-20.
Heider [Rosch], E. 1972b. Probabilities, sampling and ethnographic method: The case of
Dani colour names. Man 7: 448-466.
Hering, Ewald 1964 [1920]. Outlines of a Theory of the Light Sense. Cambridge, Mass.:
Harvard University Press.
Jameson, K. and R.G. D’Andrade (this volume). It’s not really red, green, yellow, blue:
An inquiry into perceptual color space. 000.
Kay, P. 1975. Synchronic variability and diachronic change in basic color terms. Language
and Society 4: 257-270.
Kay, P., B. Berlin, and W.R. Merrifield 1991. Biocultural implications of systems of
color naming. Journal of Linguistic Anthropology 1: 12-25.
Kay, P. and Ch.K. McDaniel 1978. The linguistic significance of the meanings of basic
color terms. Language 54(3): 610-646.
Kemmerer, D.L. 1995. Towards a cognitive neuroscience analysis of ther meanings of
basic color terms. Cognitive Science Technical Report # 95-7. SUNY, Buffalo.
Land, E. 1974. The retinex theory of color vision. Proceedings of the Royal Institution of
Great Britain 47: 23-58.
2-Kay, Berlin, Maffi & Merrifield
MacLaury, R.E. 1986. Color in Mesoamerica: A theory of Composite Categorization.
Doctoral dissertation. University of California at Berkeley.
MacLaury, R.E. 1987a. Color-category evolution and Shuswap yellow-with-green.
American Anthropologist 89: 107-124.
MacLaury, R.E. 1987b. Coextensive semantic ranges: Different names for distinct
vantages of one category. Papers from the 23rd Annual Regional Meeting of the
Chicago Linguistic Society Part I: 268-282.
MacLaury, R.E. 1991. Exotic color categories: Linguistic relativity to what extent? Journal
of Linguistic Anthropology 1: 26-51.
MacLaury, R.E. 1992. From brightness to hue: An explanatory model of color category
evolution. Current Anthropology 33: 137-186.
Maffi, L. 1991. A bibliography of color categorization research 1970-1990. In Basic Color
Terms: Their Universality and Evolution, by B. Berlin and P. Kay, 1st paperback
edition. Berkeley: University of California Press. Pp. 173-189.
Saunders, B.A.C. and J. van Brakel 1994. Translating the World Color Survey. ms. 18
pp.
Stanlaw, J. 1993. Review of B. A. C. Saunders: The invention of basic colour terms.
American Anthropologist 95(1): 183-184.
2-Kay, Berlin, Maffi & Merrifield
Sternheim, C.E. and R.M. Boynton 1966. Uniqueness of perceived hues investigated with
a continuous judgemental technique. Journal of Experimental Psychology 72: 770-776.
Wooten, B.R. 1970. The Effects of Smultaneous and Successive Chromatic Constraint on
Spectral Hue. Doctoral dissertation. Brown University, Providence, RI.
Zadeh, L.A. 1965. Fuzzy sets. Information and Control 8: 338-353.
Zeki, S. 1993. A Vision of the Brain. Cambridge, Mass.: Blackwell Scientific.
... As mentioned previously, one of the main draws of Bayesian nonparametric models is the ability to let the data determine the complexity of the model. The most prominent work to date which uses parametric ML algorithms on the WCS [33] identified 3-6 motifs but pointed to the need for a different approach to capture "minority motifs" that are rare but resemble evolutionary stages proposed by [10,44,47] . Using BBDP yields 18 prominent clusters with some having near equivalence to the motifs, with the remaining clusters possibly being these "minority motifs". ...
... This study was met with numerous criticisms. The critiques of the empirical study were that (i) the sample was too small (20 total participants with three or fewer speakers per language), (ii) the speakers were bilingual (spoke their native tongue plus English), (iii) the data was collected in Northern California instead of their native communities, and (iv) the languages represented by the sample were mostly from industrialized societies [47] . In response to these critics, Berlin and Kay embarked on a project called the World Color Survey. ...
Article
Variational Inference for the Beta-Bernoulli Dirichlet Process Mixture Model is employed to uncover universal patterns in color naming systems. The data used consist of 2552 participants from 106 World Color Survey languages. To study these languages collectively, the model is informed by universal biological, linguistic, and topological features of the task. We find that the majority of the naming systems are represented by eighteen clusters, each constituting a universal pattern. Novel mathematical techniques are developed to study the levels of similarity, underlying consensus, and diversity among these patterns. This implementation of nonparametric models demonstrates how machine learning methods can be tailored for behavioral science applications.
... e., de etiquetas que designam categorias, dentro das quais os sujeitos englobam estímulos muito diversos). 8 A natureza categorial ou contínua da perceção das cores -assim como a existência (ou não) de uma relação entre a categorização percetiva das cores e o léxico específico dos sujeitos -alimentou durante bastante tempo um debate muito alargado, de certa forma inaugurado pelo trabalho de Berlin & Kay (1969) (Newcomer & Faris 1971;Collier 1973;Bornstein 1987;Harnad 1987a: 6;Regan 1987;Pinto 1988;Kay, Berlin, Maffi & Merrifield 1997;Dowman 2007;McIntyre 2009;Huette & McMurray 2010;Lindsey & Brown 2010;Hoonhorst et al. 2011;Hardin 2013): se, por um lado, os sujeitos conseguem, em situações experimentalmente controladas, agrupar objetos por cores nomeando-as ("estes lápis são amarelos", "estes são vermelhos") -o que indicia a natureza categorial do processo -, é possível também que os mesmos sujeitos consigam, dentro de objetos processados como pertencentes à mesma categoria, estabelecer distinções e efetuar comparações (do tipo de "este lápis parece mais verde do que este", "este laranja parece mais vermelho do que amarelo", "este azul é mais escuro do que este"). Este tipo de respostas percetivas indiciaria alguma capacidade de discriminação intracategorial que, como vimos, não existe, em princípio, na perceção categorial. ...
... Este tipo de respostas percetivas indiciaria alguma capacidade de discriminação intracategorial que, como vimos, não existe, em princípio, na perceção categorial. Por esta razão, autores como Kay, Berlin, Maffi & Merrifield (1997) e Huette & McMurray (2010) defendem, com base em evidências desta natureza, que a perceção das cores é parcialmente contínua e parcialmente categorial. Analisando a complexidade da perceção das cores -e enfatizando que, na verdade, a "perceção da cor" consiste no processamento não de uma mas de várias estimulações sensoriais, ou de uma estimulação sensorial multidimensional -, a homenageada desta publicação sublinha, em concordância com Berlin & Kay (1969): "A relação existente entre os termos das cores e o seu sentido revela-se, contudo, mais complexa do que se poderia imaginar à primeira vista. ...
Article
Full-text available
RESUMO As investigações experimentais no campo da perceção da fala demonstraram, a partir da segunda metade do século XX, a natureza categorial do processamento dos sinais acústicos da fala. O fonema passou então a ser concebido não só como uma unidade teórica da descrição das línguas, mas também como uma categoria percetiva. Neste estudo, veremos como, sem a fundamentação metodológica dos estudos experimentais, correntes de pensamento mais tradicionais em linguística-como a fonética tradicional britânica e a fonologia distribucionalista norte-americana da primeira metade do século XX-haviam adiantado já algumas propriedades desta conceção categorialista do fonema. Defenderemos que uma visão mais abrangente desta unidade descritiva e de processamento, combinando informação oriunda de diversas disciplinas e de quadros teóricos alternativos, permite uma melhor explicação não só dos sistemas fonológicos e do seu processamento, mas também da inter-relação entre estes dois objetos de estudo. PALAVRAS-CHAVE Fonema. Categoria. Perceção da fala. Perceção categorial. Fonética e Fonologia. ABSTRACT Experimental studies in the field of speech perception have demonstrated, since the 1950s, that speech perception is, fundamentally, an instance of categorical processing. Accordingly, the phoneme has been conceived of as a theoretical-descriptive linguistic unit and, simultaneously, as a perceptual category. In this paper, we shall see that, even before the more recent experimental methods, traditional views-such as the British school of descriptive phonetics and the American first structuralists-had anticipated some important aspects of the categorical perspectives on the phoneme. We shall defend that a broader view on this unit, based on the combination from different subdisciplines and theoretical frameworks, could offer us stronger explanations of phonological systems, phonetic processing and the interrelation between both objects of study. (*) Unidade de investigação financiada pela FCT-Fundação para a Ciência e a Tecnologia (FCT-UIDB/00022/2020).
... The remaining four termsreferring to purple, pink, orange and greyjoin the colour lexicon in any order. We should note here that the evolutionary sequence has been modified, with new findings and assumptions taken into account (Kay, Berlin, Maffi and Merrifield, 1997;Maffi, 1999, 2005). ...
Article
Full-text available
Colour is a common physical phenomenon involving selective absorption, reflection or transmission of certain wavelengths of light. It is one of the fundamental properties of chemical compounds, which is particularly evident in the field of inorganic chemistry. The aim of this paper is to present colour terms used by scientists in the field of inorganic chemistry. We concentrate both on basic and non-basic colour terms – as understood by Berlin and Kay (1969). The research material is extracted from a corpus compiled with AntConc, consisting of abstracts published in Inorganic Chimica Acta, Inorganic Chemistry Communications, Journal of Inorganic Biochemistry and ten other chemical journals. The size of the corpus is 1,626,380 words. The most frequent basic colour terms found in the abstracts are blue (370 occurrences), green (302) and red (222), whereas the non-basic terms include, among others, violet (46) and cyan (4). In this paper, we investigate the uses of these and other colour words, focusing mainly on their occurrence in names of chemical compounds and other terms as well as in descriptions.
... Examples of cross-classifications or simply mismatched linguistic partitions abound. The domain of color has been particularly and extensively studied (Davidoff, et al., 1999;Kay et al., 1997;Regier & Kay, 2009). The domain of motion events has also received a lot of attention since Talmy (1985) distinguished two fundamental types of languages: those that encode manner of motion (for example, English) and those that focus on trajectories (for example, Spanish). ...
Chapter
Full-text available
Two connected questions that arise for anyone interested in inner speech are whether we tell ourselves something that we have already thought; and, if so, why we would tell ourselves something that we have already thought. In this contribution I focus on the first question, which is about the nature and the production of inner speech. While it is usually assumed that the content of what we tell ourselves is exactly the content of a non-linguistic thought, I argue that there can be a lot of transformation in the process of converting a thought into words. Thus, the content of what we tell ourselves, being intrinsically linguistic, is different from the content of the thought our speech transmits. Fleshing out this kind of approach implies dealing with complicated questions which we lack enough knowledge about: the nature of non-linguistic thinking, and how speech (inner and overt) is produced; i.e. how the speaker goes from format a (format of thought) to format b (language). I show that these are pressing issues for any other position, but also suggest ways in which we could tackle such complicated issues.
... (4) It must be psychologically salient for informants, as evidenced, for instance by having high frequency in elicited lists. However, Kay, Berlin, Maffi and Merrifield (1997) state that these criteria for basicness were more a set of guidelines than a formal definition and in practice, the criteria reduce to "simple and salient" (Hardin C. L. & Maffi L., 1997). ...
Article
Full-text available
The aims of this study were to describe the basic color terms (BCTs) of Arabic and, in particular, clarify the relationship among the three Arabic terms for blue: azrock, samawee, and khuhlie. Data were collected from child and adult native Arabic speakers from schools and universities in Riyadh using a list task and a naming task. In the list task, the children’s sample included 113 boys and 140 girls, aged 8 to 12 years, while the adult sample (N = 200) was made up of equal numbers of men and women, aged 18 to 25 years (mean = 20 years). The task involved writing down as many color words as they knew, in four minutes for the children, and one minute for the adults. The pattern of results from the two samples was essentially the same: the terms ordered by their frequency from 100 to 38 percent were ahmar “red”, azrock “blue”, akhdar “green”, asfer “yellow”, asswed “black”, abiyadh “white” boartoogaalee “orange”, bonee “brown”, wardee “pink”, banafsagee “purple”, and rassasee “grey”. In addition, samawee “light blue” was provided by 40 percent of the sample. This is followed by a drop in the frequency scores, with khuhlie “dark blue” being included in 38.6 percent of the responses. In regard to the color naming task, the child sample included 30 boys and 30 girls, aged 8 to 12 years, while the adult sample (N = 60) include equal numbers of men and women, aged 18 to 28 years. The task involved naming each example of a set of 65 colors representing the whole color palette. The two samples again performed similarly. The terms with the highest measures of usage and consensus were ahmar, azrock, akhdar, asfer, asswed, abiyadh, boartoogaalee, bonee, wardee, banafsagee, and rassasee. Based on these results, it appears that Arabic has 11 basic color terms that correspond to Berlin and Kay’s (1969) universal terms. In addition, the terms of particular interest — samawee (“light blue”) and khuhlie (“dark blue”) — are not basic Arabic color terms.
... underlying conceptual information [36][37][38] . Despite documented evidence of this variation across circumscribed knowledge domains with small language samples 7,[9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27]30,35,39 , and even recent larger-scale analyses that broaden the domains, number of languages or both 5,40 , analyses of linguistic relativity have tended to ignore the degree to which the words of one language encode more or less conceptual information than another. We know little about cross-linguistic information density and even less about its effect on human communication and knowledge creation. ...
Preprint
Full-text available
Language is the primary medium through which human information is communicated and coordination is achieved. One of the most important language functions is to categorize the world so messages can be communicated through conversation. While we know a great deal about how human languages vary in their encoding of information within semantic domains such as color, sound, number, locomotion, time, space, human activities, gender, body parts and biology, little is known about the global structure of semantic information and its effect on human communication. Using large-scale computation, artificial intelligence techniques, and massive, parallel corpora across 15 subject areas--including religion, economics, medicine, entertainment, politics, and technology--in 999 languages, here we show substantial variation in the information and semantic density of languages and their consequences for human communication and coordination. In contrast to prior work, we demonstrate that higher density languages communicate information much more quickly relative to lower density languages. Then, using over 9,000 real-life conversations across 14 languages and 90,000 Wikipedia articles across 140 languages, we show that because there are more ways to discuss any given topic in denser languages, conversations and articles retrace and cycle over a narrower conceptual terrain. These results demonstrate an important source of variation across the human communicative channel, suggesting that the structure of language shapes the nature and texture of conversation, with important consequences for the behavior of groups, organizations, markets, and societies.
... However, we can question whether concepts should be represented as convex regions. If convexity is defined on a perceptual space, then the claim is empirically false -even in the lowdimensional space of colour, McMahan and Stone (2015) find expressions like greenish whose meanings are nonconvex, and Kay et al. (1997) similarly describe how some languages have terms which could be glossed as "peripheral red". Sidestepping such objections, Gärdenfors (2014, §2.5) describes how "higher-level" dimensions can be defined in terms of more basic ones, and suggests that the convexity requirement can apply in an abstract space defined by such higher-level dimensions. ...
Thesis
The aim of distributional semantics is to design computational techniques that can automatically learn the meanings of words from a body of text. The twin challenges are: how do we represent meaning, and how do we learn these representations? The current state of the art is to represent meanings as vectors – but vectors do not correspond to any traditional notion of meaning. In particular, there is no way to talk about truth, a crucial concept in logic and formal semantics. In this thesis, I develop a framework for distributional semantics which answers this challenge. The meaning of a word is not represented as a vector, but as a function, mapping entities (objects in the world) to probabilities of truth (the probability that the word is true of the entity). Such a function can be interpreted both in the machine learning sense of a classifier, and in the formal semantic sense of a truth-conditional function. This simultaneously allows both the use of machine learning techniques to exploit large datasets, and also the use of formal semantic techniques to manipulate the learnt representations. I define a probabilistic graphical model, which incorporates a probabilistic generalisation of model theory (allowing a strong connection with formal semantics), and which generates semantic dependency graphs (allowing it to be trained on a corpus). This graphical model provides a natural way to model logical inference, semantic composition, and context-dependent meanings, where Bayesian inference plays a crucial role. I demonstrate the feasibility of this approach by training a model on WikiWoods, a parsed version of the English Wikipedia, and evaluating it on three tasks. The results indicate that the model can learn information not captured by vector space models.
... The cardinal number 'one', indefinite quantifiers like 'few' and 'many' and the focus exclusive operator 'only' will constitute the second category that show up before the noun, examples are provided in (45).Although some adjectives precede the head noun, as further shown in (47) below, others have to be after the noun. The adjectives that follow the noun often show up with an associative adjectives that occur after, and are linked to the noun with the AM, cannot be used in the pre-nominal position, we will return to this in the next section.It is also important to mention that Awing has just three main colours: black, white and red; that is why yellow in (48e) is described as red (see, e.g.,Kay et al. 1997 for cross-linguistic differences). Moreover, as I early noted, the examples in (48) may be considered as some kind of description that has been standardized. ...
Thesis
Full-text available
This project describes the nominal, verbal and ‘truncation’ systems of Awing and explains the syntactic and semantic functions of the multifunctional l<-><-> (LE) morpheme in copular and wh-focused constructions. Awing is a Bantu Grassfields language spoken in the North West region of Cameroon. The work begins with morphological processes viz. deverbals, compounding, reduplication, borrowing and a thorough presentation of the pronominal system and takes on verbal categories viz. tense, aspect, mood, verbal extensions, negation, adverbs and triggers of a homorganic N(asal)-prefix that attaches to the verb and other verbal categories. Awing grammar also has a very unusual phenomenon whereby nouns and verbs take long and short forms. A chapter entitled truncation is dedicated to the phenomenon. It is observed that the truncation process does not apply to bare singular NPs, proper names and nouns derived via morphological processes. On the other hand, with the exception of the 1st person non-emphatic possessive determiner and the class 7 noun prefix, nouns generally take the truncated form with modifiers (i.e., articles, demonstratives and other possessives). It is concluded that nominal truncation depicts movement within the DP system (Abney 1987). Truncation of the verb occurs in three contexts: a mass/plurality conspiracy (or lattice structuring in terms of Link 1983) between the verb and its internal argument (i.e., direct object); a means to align (exhaustive) focus (in terms of Fery’s 2013), and a means to form polar questions. The second part of the work focuses on the role of the LE morpheme in copular and wh-focused clauses. Firstly, the syntax of the Awing copular clause is presented and it is shown that copular clauses in Awing have ‘subject-focus’ vs ‘topic-focus’ partitions and that the LE morpheme indirectly relates such functions. Semantically, it is shown that LE does not express contrast or exhaustivity in copular clauses. Turning to wh-constructions, the work adheres to Hamblin’s (1973) idea that the meaning of a question is the set of its possible answers and based on Rooth’s (1985) underspecified semantic notion of alternative focus, concludes that the LE morpheme is not a Focus Marker (FM) in Awing: LE does not generate or indicate the presence of alternatives (Krifka 2007); The LE morpheme can associate with wh-elements as a focus-sensitive operator with semantic import that operates on the focus alternatives by presupposing an exhaustive answer, among other notions. With focalized categories, the project further substantiates the claim in Fominyam & Šimík (2017), namely that exhaustivity is part of the semantics of the LE morpheme and not derived via contextual implicature, via a number of diagnostics. Hence, unlike in copular clauses, the LE morpheme with wh-focused categories is analysed as a morphological exponent of a functional head Exh corresponding to Horvath's (2010) EI (Exhaustive Identification). The work ends with the syntax of verb focus and negation and modifies the idea in Fominyam & Šimík (2017), namely that the focalized verb that associates with the exhaustive (LE) particle is a lower copy of the finite verb that has been moved to Agr. It is argued that the LE-focused verb ‘cluster’ is an instantiation of adjunction. The conclusion is that verb doubling with verb focus in Awing is neither a realization of two copies of one and the same verb (Fominyam and Šimík 2017), nor a result of a copy triggered by a focus marker (Aboh and Dyakonova 2009). Rather, the focalized copy is said to be merged directly as the complement of LE forming a type of adjoining cluster.
... Abundan los ejemplos de particiones lingüísticas entrecruzadas o simplemente no coincidentes. El dominio del color ha sido particular y profusamente estudiado (Kay et al. 1997, Davidoff et al. 1999, Regier y Kay 2009). También el dominio de los eventos de movimiento ha recibido mucha atención desde que Talmy (1985) distinguiera dos tipos fundamentales de lenguas: las que codifican modo de movimiento (p.e. ...
Article
Full-text available
En este artículo proponemos que existe un problema de traducción del sistema representacional conceptual al lingüístico que tradicionalmente no se ha tenido en cuenta. Tal problema ha pasado desapercibido posiblemente porque se ha echado mano de un sistema representacional intermedio adaptado a las necesidades expresivas del lenguaje. Sin embargo, explicamos que, lejos de resolver el problema, postular este sistema intermedio hace que el problema de traducción sea más difícil de resolver. En consecuencia, proponemos un modelo de producción lingüística que no recurre al nivel de representación conocido como "pensar para hablar" (Slobin 1996). Abstract: In this paper we argue that there is a translation problem from the conceptual representational system to the linguistic one that has not traditionally been taken into account. Such a problem has gone unnoticed possibly because an intermediate representational system adapted to the expressive needs of language has been postulated to exist. However, we explain that, far from solving the problem, postulating this intermediate system makes the translation problem more difficult to solve. Consequently, we propose a model of speech production that does not resort to the representation level known as "thinking for speaking" (Slobin 1996).
Article
Full-text available
Two experiments compared “Red-Green” (R-G) dichromats’ empirical and metacognized capacities to discriminate basic color categories (BCCs) and to use the corresponding basic color terms (BCTs). A first experiment used a 102-related-colors set for a pointing task to identify all the stimuli that could be named with each BCT by each R-G dichromat type (8 protanopes and 9 deuteranopes). In a second experiment, a group of R-G dichromats (15 protanopes and 16 deuteranopes) estimated their difficulty discriminating BCCs-BCTs in a verbal task. The strong coincidences between the results derived from the pointing and the verbal tasks indicated that R-G dichromats have very accurate metacognition about their capacities (they only had considerable difficulty discriminating 13 out of the total of 55 possible BCT pairs) and limitations (Brown-Green and Blue-Purple pairs were rated especially difficult to differentiate) in the use of BCTs. Multidimensional scaling (MDS) solutions derived from both tasks were very similar: BCTs in R-G dichromats were properly represented in 2D MDS solutions that clearly show one chromatic dimension and one achromatic dimension. Important concordances were found between protanopes and deuteranopes. None of these dichromats showed substantial difficulty discriminating the Red-Green pair. So, to name them “R-G” dichromats is misleading considering their empirical capacities and their metacognition. Further reasons to propose the use of the alternative denomination “Brown-Green” dichromats are also discussed. We found some relevant differences between the “Brown-Green” dichromats’ empirical and self-reported difficulties using BCTs. Their metacognition can be considered a “caricature” of their practical difficulties. This caricature omits some difficulties including their problems differentiating “white” and “black” from other BCTs, while they overestimate their limitations in differentiating the most difficult pairs (Brown-Green and Blue-Purple). Individual differences scaling (INDSCAL) analyses indicated that the metacognition regarding the use of BCTs in “Brown-Green” dichromats, especially deuteranopes, is driven slightly more by the chromatic dimension and driven slightly less by the achromatic dimension, than their practical use of BCTs. We discuss the relevance of our results in the framework of the debate between the linguistic relativity hypothesis (LRH) and the universal evolution (UE) theories.