Basic color categories, essentialism and gender-oriented perspectives
Ondřej Beran
Abstract: This paper reflects upon the theory of “basic color terms” (BCT) from the perspective of gender aspects. Some presumptions of BCT theory are thereby questioned. In Part 1, I explain that languages appear to be context-specific and context-conditioned institutions of practice. In Part 2, I introduce BCT theory and show that its conceptual hierarchy of basic and non-basic terms tallies with the pattern of language use commonly attributed to men, as opposed to a different, “female” pattern, which entails different conceptual relations. In Part 3, I relate some evidence indicating i) the historical inadequacy of the evolutionary framework of BCT and ii) the historical origin of the very notion of “basic colors” (questioning the inborn nature of the “basic”). In Part 4, I discuss, in hypothetical terms, possibilities for color nomenclature outside the basic vs. non-basic differences, whereby possible variants of the putative “women’s language” appear to be viable and useful in practice. Finally in Part 5, I refer to some evidence that “women’s language,” more finely distinguishing among colors, is probably an artifact, and thus “men’s language,” displaying the BCT framework, would be an artifact as well.
1. Opening considerations; boundaries among “languages”
In this paper, I would like to confront the theory of “basic color terms” or “categories,” as propounded originally by Brent Berlin and Paul Kay (1969), with empirical and theoretical works concerning gender differences in color recognition and in the use of color vocabulary. I am going to assume, rather briefly now – I will try to document it later in greater detail – that the way people name colors is intrinsically language-related, i.e. the particular language that people speak plays an important role that cannot be overlooked. On the specific elaboration of this intuition, Paul Kay and his successors have devoted much attention; suffice to say here that there are, for instance, languages that lack systematic tools for distinguishing between what English calls “blue” and “green” and that thus operate within significantly different categories. It seems that such a general distinction between blue and green is absent in many Native American languages; tests show that the degree of the “subjectively perceived” difference between blue and green color samples is comparatively lower in the speakers of these languages than in the speakers of English.
See Kempton and Kay (1984).
Though I do not intend to question such findings, mostly based on extensive field research, I will attempt to change the perspective on them. Kay and his associates distinguish the scale of particular empirical languages (the units or points of the scale are English, Chinese, etc.). This view on language is inherently essentialistic: language seen as a system of signs that carry meanings, homogeneous for all competent users in any given language. Typically, all native speakers of a particular language are considered to be the bearers of that language; supplemented by (usually, with the exception of English, for instance) a lesser number of people, comparable and compatible in communicative skills, who are able to adopt some mastery of it as their “second” language. Kay’s view is also inherently naturalistic: the human capacity to see and distinguish color is hard-wired into our brains, and the differences in color vocabularies among particular languages only amount to various half-displays of the full spectrum of our inborn abilities.
This assumption that language is what it is, issuing from the mouths of all its speakers (while there is no substantial problem lurking within the question of who a speaker of a language is) or, perhaps, independently of its speakers, is a fairly typical idea in philosophy; most philosophers speak in an ingenious and profound way about “language” without further qualifications. This perspective is deemed incomplete and inaccurate by the research of languages as spoken in practice. Language can be regarded as a set of skills, and the portfolio of adopted linguistic skills differs from speaker to speaker (practically between every two of them, two native speakers of the same language being no exception). There are a great many transitive positions between the poles of “incompetence” and “true language mastery” (if there is any such thing). But since language is a practice rooted within a particular context, it can be considered not only the skill of the speaker – the individual’s property or quality, so to speak – but also the claim or expectation placed on speakers, depending on the practical contexts within which they act and how they find themselves. In this respect, English and – for instance – my native Czech are not languages in quite the same sense. It makes good sense to adopt functional (professional) sub-segments of English; indeed, it is expected from many speakers of other languages and many of them actually speak English in this particular sense. Here, language acts externally upon the speakers and they stand in front of it as a task or a claim (they are “put into” the language, put into a situation with which they have to cope).
Blommaert, Collins and Slembrouck (2005, p. 198) say that “spaces [are] constitutive and agentive in organizing patterns of multilingualism.” Considering this, it is understandable that, even in one’s “own” language, almost nobody develops the full range of speech skills, not to mention “second” languages. Language is not just a projection of our inborn dispositions, in a Chomskyan sense, but rather, it is a result of the confrontation of these dispositions (and not just, strictly speaking, linguistic dispositions, but also qualities like extroversion, patience or endurance) with the space within which we find ourselves and with the tasks it places on us.
I opened my argument with a consideration, which is perhaps a little too general, in order that I could point out one thing: if language, in whatever general terms we reflect upon it, is so tightly connected to its practice in speech, we cannot simply consider that there are internally homogeneous and mutually discreet (distinctly different) “languages,” such as – in common parlance – English, Chinese, Czech, etc. There seem to be parallel boundaries as well, which do not tally with the “official” boundaries among languages. Among such parallel boundaries, I would like to then focus on the boundaries that separate speakers according to their gender.
2. The theory of basic color terms and “female superiority” in color discrimination
The gender boundary is naturally not the only existing parallel boundary, but it is extremely interesting – among other things – for the way in which it sheds light on the theory of basic color terms. This theory, as promoted by Paul Kay and others,
Apart from Berlin’s and Kay’s (1969) classic work, see e.g. Kay and McDaniel (1978) or Kay and Maffi (1999). presents an evolutionary development of the system of “basic color terms” covering the whole color spectrum. Whereas the least developed languages, which exhibit only two “basic terms” sufficient for all hues (in Kay’s view it is the Dani language from New Guinea),
He relies on the research made by Heider (1972), which is, however, less specific with respect to the BCT framework. the most developed ones (typically the modern IE languages like English, French or Spanish) display 11 “basic terms:” red, green, yellow, blue, black, white, grey, brown, purple, orange and pink. These 11 terms occupy the significant (distinguished) positions in the topology of our concepts, limited only by the physiological limits of human perception. There are, certainly, a lot of additional color names, which coexist with the “basic” names and they fill the gaps within and amongst them. They can, however, be sharply distinguished from basic names; typically, they are non-monolexemic (e.g. blue-green or light blue) or they are clearly derived from a name for a non-color phenomenon (e.g. Burgundy).
Kay and his collaborators clearly wanted to eliminate such terms as “vinous” or “Burgundy;” nevertheless, “orange” is no less a derived term than “vinous,” being derived from the name for the fruit. In my native Czech, the word for the color, pink, is “růžová,” which is in turn derived from the Czech name for roses (the flowers) – “růže;” I suppose there have to be a lot of cases such as these. The basic nature of truly basic terms must be testified by experimental practice, where respondents – speakers of various languages – work (or are expected to work) with “basic” terms as well as “focal colors.” These are expected to represent, in their pure form, centers (foci) around which all the other color samples fall under the same basic color concept.
See e.g. Regier et al (2005). Unfortunately, this expectation is neatly fulfilled only by pre-informed respondents, who share the same origins as the respondents (typically, college students volunteering for a test
See e.g. Saunders and van Brakel (1997, p. 215).). The respondents – speakers of “exotic” languages – display much less willingness to cluster the color samples into the groups expected, according to the identification of their focal basic colors; they either fail to divide them at all, or they distinguish among them too finely or in an apparently “illogical” manner.
An almost hilarious exhibition of such cases, usually hidden behind the neat scientific results, is provided by Saunders and van Brakel (1997, p. 174): “If four unique hues were a universal human perceptual grounding, crosscultural research would confirm it. But empirical evidence for a fixed number of primitives (whether four or any other number) is utterly evanescent. Of the Munsell colour chips commonly used in cross-cultural experiments, 60–80% often remain unnamed (see Berlin et al. 1991). When presented with chips people get confused and give inconsistent answers. Tougher-minded subjects find the naming and/or categorization tasks absurd, not self-evident. For example Berlin and Berlin (1975, p. 85, n. 5) recount their difficulties with monolingual Peruvian Aguaruna informants. Many would simply stare at the array. Others who could bring themselves to place the pen to the plastic would begin drawing individual black circles around chip after chip often moving along some level of brightness, completely ignoring hue. Several attempted to provide a different name for each perceptually different chip, employing terms that later proved to be the names of trees, plant dyes, and parrot feathers. One informant, when asked to show where all the red chips were, took the pen and very carefully circled the entire board.”
This alleged, overly refined discrimination does not only concern “savages” or otherwise “retarded” respondents. The fine discrimination of colors and shades, which engage, with equal effectiveness, the basic colors as well as the derived (non-basic) shades, is traditionally attributed to women. The tradition of this “observation” is rather long, reaching back to purely anecdotal remarks and to the first attempts at scholarly research in the early 20th century. Back then, the sample of respondents were students of American public schools and girls scored better in the Wordsworth-Wells test, which determines promptness when recognizing (naming) the five “standard” colors (red, yellow, green, blue, black).
See here e.g. Rich (1977) or Frank (1990). Women were simply expected to have a “better sense” for color recognition than their male counterparts – note that what was tested in this way was a certain linguistic skill. But this did not concern only promptness in recalling a color name, but also the extent of the vocabulary used.
See here Rich (1977) or Nowaczyk (1982).
It was not until the seventies that the assumption of female superiority in color discrimination underwent a more critical consideration. The American linguist, Robin Lakoff, offered an analysis that captures some essential points of this peculiar issue:
... imagine a man and a woman both looking at the same wall, painted a pinkish shade of purple. The woman may say:
(2) The wall is mauve,
with no one consequently forming any special impression of her as a result of the words alone; but if the man should say (2), one might well conclude he was imitating a woman sarcastically or a was homosexual or an interior decorator. Women, then, make far more precise discriminations in naming colors than do men; words like beige, ecru, aquamarine, lavender, and so on, are unremarkable in a women's active vocabulary, but absent from that of most men. I have seen a man helpless with suppressed laughter at a discussion between two other people as to whether a book-jacket was to be described as “lavender” or “mauve”. Men find such discussion amusing because they consider such a question trivial, irrelevant to the real world.
We might ask why fine discrimination of color is relevant for women, but not for men. A clue is contained in the way many men in our society view other “unworldly” topics, e.g. high culture and the Church, as outside the world of men’s work, relegated to women and men whose masculinity is not unquestionable. Men tend to relegate to women things that are not of concern to them, or do not involve their egos. Among these are problems of fine color discrimination. We might rephrase this point by saying that since women are not expected to make decisions on important matters, like what kind of job to hold, they are relegated the non-crucial decisions as a sop. Deciding whether to name a color “lavender” or “mauve” is one such sop.
Lakoff (2004), p. 43.
When we face the issue of “basic” and “derived” color names, some suggestions offer themselves. In the inferences we make, the “derived” names for specific shades are subordinated to their allegedly superior basic names, i.e. we assume that it is possible to infer “X is pink” from “X is mauve,” but not necessarily the other way round. The other suggestion casts a questioning light on the first one: that each of these two exemplary utterances belongs to a different “language game,”
Wittgenstein (2007). or simply a different linguistic context. In a context where speakers are satisfied with the “basic” discrimination among a few colors, the former description (mauve) is superfluous and is replaced by the latter (pink). However, where the purpose is to capture the color nuances as accurately as possible, the latter description is deemed superficial, idle, inaccurate, and is replaced by the former (or by another specified label – magenta, etc.). The question, “When does practice allow “X is mauve” to be asserted and, at the same time, “X is pink” to be inferred?” may be answered: Probably, quite rarely. A possible exception is the reflexive, philosophical or scientific game that endeavors at an explicit systematization of this nomenclature. That, however, is a rather artificial context.
The system of 11 “basic color categories,” as advocated by Kay, admits a certain structure of conceptual relationships. Among other things, by introducing the “topology” of the 11 colors, it opens up the possibility of ascribing two neighboring colors at once (or more precisely, it opens space for a situation where a speaker could refuse to choose only one of the possibilities, e.g. in that it makes equally good sense to say that X is blue as it is to say that it is green). The topology can be represented as follows:
Fig. 1. The topology of color terms; the links highlight the pairs of colors allowing the double ascription (omitting the possible triple ascription, green-yellow-brown). The figure, based on Kay and McDaniel’s spherical scheme (1978, p. 628), is adapted from my article (XY 2012).
However, this system seems to only cover the system of rules governing that segment of language spoken primarily by men, according to the older linguistic or folk-linguistic intuitions. Can it cover the speech specificity of women? – if, indeed, we accept that there really is such a fundamental difference. It appears that women are capable of playing the male language game of colors as competently as men. But, does their game constitute a system in its own right? Although women are said to tend to use more idiosyncratic, specific, or qualified names, this usage does not avoid the “basic” terms; what happens to these terms is that the space for their appropriate use is reduced
This was shown by the research findings of Simpson and Tarrant (1991, p. 59ff). – in comparison with “male” usage, the “basic” terms no longer cover the whole range of colors. The “basic” terms, broadly used by men, now become, in women’s use, reserved for the “foci” of individual “basic” colors – what may be called “pure green” or “pure red” and shades close to them. Thus, the hierarchy of color terms that permitted to express rules like:
(1) X is mauve
X is pink
disappears (or is considerably weakened). On the other hand, rules of the following kind appear:
(2) X is mauve
X is not (pure) pink,
but also the other way round (from “X is pink” or rather “pure pink,” one can infer that “X is not mauve”). In fact, one can substitute any pair of color names into the rule of the kind (2), and the resulting rule will always be acceptable. Even if we use here the names of the so-called “basic” (in this context, “pure”) colors, it does not allow us to form any inference of the kind seen in (1). In the end, the “female usage” of colors is only a highly fine-grain version of the longstanding pre-philosophical intuition (expressed later in the form of the Color Incompatibility Theorem, etc.) that each thing has only one color.
The sense for detail and fine differences, attributed to women in such studies, thus results, somewhat surprisingly, in a certain impoverishment of the hierarchical or “vertical” structure of color names. This complexity – that is, refining of the applied discernment tools – is, effectively, a simplification of the system of “basic,” “special” or “derived” color terms, as suggested by normative color science, perhaps best exemplified in the following shape:
Fig. 2: This possible scheme captures the relationships among “basic” terms that can be ascribed together (marked by the bidirectional arrows) along with “special” terms that “fall under” “basic” terms (according to rule (1)). This sometimes occurs under more than one term (marked by the thin vertical and oblique lines). No special term can be ascribed together with another special term (marked by the thicker vertical lines).
There is an expectation for the putative linguistic usage of women to focus more finely on color discrimination and the conceptual relationships may look rather similar to the following:
Fig. 3: This scheme captures the relationships among “basic” and “special” color terms in an impoverished, leveled form: all the names stand on the same level and there is no relationship of “possible double ascription” or “falling under” between any two of them.
It does not matter much, what actual “juxtapositions” we place into the scheme; the important thing is that the use of the names do not overlap. Still, it is true that some pairs may provide meaningful material for an argument – e.g. mauve vs. magenta – whereas some of them do not – e.g. azure vs. magenta – since we cannot imagine a meaningful situation in which such an agreement takes place. Thus, each of the names is not equally “distant” from each other. These links can be fruitfully used in language games where we consider which colors “go well together,” etc. However, even if we face a pair of names allowing a reasonable argument, it will not end up in an admission that both are equally relevant. What is much more probable, following the logical relationships as suggested by Fig. 3, is that either a final agreement on one option (“magenta then!”) is reached, or no option is chosen. A list of such pairs that allow for argument can be created using the above figures – that would be either the neighboring pairs of names separated by the vertical lines, or the pairs in Fig. 2 linked by the relationship of “falling under,” where the “basic” terms are further qualified by the addition of “(pure).”
All these relationships may be represented by a scheme – some combination of Figures 2 and 3; and actually, these are a variation on Figure 1, where the network of neighboring basic terms would be extremely densified by the addition of “non-basic” terms. The way to mapping the conceptual relationships in “women’s language” relies, however, on some problematic assumptions. First of all, it is as if we are still trying to map the excessive or eccentric “female” linguistic model as an aberration, a secondary derivation from something basic, more general (which tallies with “the male”). It is true that this explanatory sequence is also supported by this paper; what I want to do, however, is to exemplify the way we are accustomed to certain linguistic structures. We can ask questions about rules that hold generally, and only after some time, when we reach a certain level of refinement in our answer, can we mention the exceptional or the peculiar. If we assume that women use language to talk about colors in a systematically different way to this – and if one wants to state such a thing, he/she actually makes a series of questionable points, as we will see – this does not represent a curiosity but a system in its own right, a system that is connected in many respects to the “basic,” default system, but which is not necessarily “subordinated” to it or derived from it. The idea of such a derivation would require, among other things, to show that the semantic system containing a hierarchy of i) a limited number of basic terms, ii) a simple complex of connections among them and iii) a greater number of “secondary” color names is in some sense a primary system. But is it?
3. Some empirical and conceptual shortcomings in the Berlin-Kay framework
The evolutionary scheme promoted by Berlin and Kay nevertheless cannot be proven to be such a primary point of departure. It surprisingly ignores the fact that the evolutionary ancestors of some modern European languages are ancient Greek and Latin, languages that we have had the opportunity to study, in the form of rich textual material. Here, we may also cite Indian and Iranian languages and their present progenies. In ancient Greek or Latin, we do not meet the simple stratification of color vocabulary into the 11 (or, expectably, less) basic terms, supplemented occasionally by some special terms for intermediate or context-specific shades. The color-naming repertoire in Greek and Latin is quite extensive, but does not own a clear hierarchy; however, it possesses a rich synonymy; what we translate today as the single word “red” used to assume the guises of ruber, rubeus, russus, rufus, purpureus. It is characteristic for Latin to use various terms for the same shade in combination with various stuff (materials), which is fairly typical of white color. And an issue on its own is – in these nautical cultures – represented by terms used for expressing marine colors – kyanos and caeruleus – which consequently cannot be translated into modern languages, though they occupy an important (not marginal) place in the color pragmatics of these languages.
For a more detailed account of this topic, see John Lyons (1999). Kay (1999) devotes a reply to Lyons’ criticisms of his theory, opposing it from his anti-relativistic position.,
From another point of view, there is also research that indicates the inaccuracy of Kay’s evolutionary hypothesis by tracking and describing cases of regression (Kristol 1980).
This historical excursus is not without interest for us; it shows as that the existence of a rich structure of color names (in a language of a highly developed literary culture) is relatively independent of the emancipation of a small number of terms that would acquire the status of “basic.” The study of a certain few colors as basic, i.e. placing the other names at a secondary position, is typical of modern science, in which Isaac Newton plays a distinguished role. The existence of a rich – though not hierarchically structured – vocabulary of colors in European antiquity does not prove that Newton’s works are incorrect; it rather shows that Newton, along with other modern and later scientists discovered an interesting and fruitful viewpoint, from which the topic of colors can be investigated, but which intentionally (from some good reasons) omits a part of the pragmatics of colors altogether.
The interesting point is that it is the optical studies of Newton and others that established the central positions of basic colors, whose interrelationships became a subject of study for the new, successful science. The earliest – at first, quite amateurish – treatises on the systematic differences in language used by men and women originated at roughly the same time; they attributed to women a greater amount of the rough language “production” and considered women stuck to the ornamental and superficial level of language
For a more detailed account of this history, see Jennifer Coates (2004, chapters 2 a 3). (let us recall, in this context, the modern tradition that distinguishes primary and secondary qualities). Then, at the turn of the 19th and 20th centuries, this questionable intuition was continued in a more scholarly manner –studies explicitly thematized the greater linguistic skills of women (concerning colors), especially the sharp distinguishing between “non-basic” shades. From this perspective, it is not “women’s language” alone which appears as an eccentric, “different” opposition to the “norm” – a non-marked, actually male language. Both can be artifacts equally, with a tradition reaching back to the same time, and as it is no coincidence that they are “opposite” to each other, they actually share the same history. It is within the framework of this history, where the structure of “female” concepts proves to be secondary. But this attribution is far from a matter-of-course; considering that the “natural” development of color terms in the pre-modern age seemed to proceed (which is suggested by what we know about ancient Greek or Latin) by way of “female” division, as sketched in Fig. 3. Again, this is rather hypothetical; but not more than Kay’s evolutionary hypothesis.
Unfortunately, the scheme sketched in Fig. 3, unlike Fig. 2, is not derived from a wealth of empirical research; rather, it relies upon a combination of conceptual reflection, introspection and anecdotal evidence.
Lakoff has been justly criticized for this methodological untidiness; many authors have hesitated to take seriously what she has written about the institution of the female language. On the other hand, a lot of work, time, energy and funding have been invested into the research and support of the thesis that there is a certain “basic” level of color names in language, to which a secondary, non-basic hierarchy is related. However, a proposal of an independent procedure for researching such a parallel “women’s” structure would have to face certain problems. First of all, it is not at all clear whether it should be tested in all the possible languages, just as the evolutionary hypothesis of basic color terms had been. Lakoff – as well as other feminist critics of the institution of “female language” – supposes that this phenomenon is actually culturally conditioned and, as such, it is imposed on women (rather than originating spontaneously), and can hardly be expected to be imposed independently of a social and cultural milieu. The gender connotations of the objective (scientific) vs. subjective opposition equally suggest cultural conditioning. This opposition, along with these gender connotations, is linked to the history of the institution of modern science, closely related to the cultural space of European languages, but absented (at least in the past) from the space dominated by other culturally important languages, such as Chinese. Therefore, what should be tested is whether such a finer and non-hierarchical discernment is typical for female speakers, when compared to male ones, in all languages equally, or whether there are differences, and what the nature and cause of these differences are.
4. Some speculations and conceptual reflections on distinct, “women’s language”
I cannot actually test all these hypotheses. What I can do is consider some assumptions concerning the 11 basic terms and their relation to the specific “female” view. Such an undertaking means that I must “bracket” some of the instruments and strategies regularly used by the field’s researchers – e.g. their endeavors to eliminate the answers of those respondents who try to divide the color chips into “inappropriately many” groups (anything that excessively outnumbers the admitted 11, or 12 groups is a priori approached as suspicious) or in a way that appears to be illogical “in general.” Let us suppose that there is no such thing as a wrong answer (a division that could be a priori ruled out as nonsensical, impossible). That would mean, among other things, that we systematically disregard the very concept of “basic” or “focal” colors in the framework of the task – all the names are equally basic, until proven otherwise.
Even if we could proceed in such an unprejudiced way – how should we evaluate the results, with respect to observed overall gender differences? The speakers can differ within the boundaries of particular languages as well as across them, in the number of the sorted groups as well as in the pattern of their division. The difference between men and women, if any, can take shape of gradient, rather than by of a differential leap; the gradient can differ in its slope among languages; the difference can be altogether absent somewhere, or combined with other factors such as age,
Simpson and Tarrant (1991) show that the ability to distinguish is directly proportional to age in both men and women; while according to Rich (1977) there is no marked difference between age groups in women and in the case of male respondents the finery of distinguishing negatively depends on their age. Simpson and Tarrant (p. 61f) just admit this discrepancy, but they do not attempt any explanation.
There is no obvious essential difference in the test design between the two experiments; hence, there is open space for speculation. I do not consider it improbable that the low number of tested subjects is to blame (Simpson and Tarrant worked with 50 respondents; Rich does not state the exact overall number, she only says that she had five groups varying in number, in the range from 7 to 24 respondents in a group). The age differences – unlike gender differences various studies agree on as more or less salient – may be rather unspecific; in order to capture a more general trend, a more systematic work that contains substantially more respondents might be needed. etc. (Or the division can look altogether differently – this is all simply speculation.) Let us assume that we find that women, on average, really divide the chips into a greater number of groups than men, and that this effect is more clearly visible in “civilized,” Western (mostly IE) languages. Let us also assume that this higher level of divisional refinement is not in itself “fractal,” so that the terms are further internally dissected, but that many “female” color names really stand at the boundaries between the “male” or “basic” color terms, as suggested by the relationships of double “falling under” in Fig. 2. At any rate, the evaluation that would aspire to present a schematic structure of “female” color terms will have to also present the decision of the number. To say that women, on average, divide the chips into more groups that men does not mean that every female respondent ends up with the same number of groups. Should we attach more importance to the division into 16 groups, or into 25? On principle, we should not eliminate any excessive dissection, as soon as we have given up the a priori claim of a limited set of basic names. Does any female respondent not distinguish between “pure pink” and “mauve?” That would not be a reason to let the distinction fail, as long as any other respondent consequently keeps it. What if we meet an emancipated group in only one case: if there is only one respondent who separates “honeysuckle” (honeysuckle pink) or “sweet blue?” Is it not justified to ignore such idiosyncratic distinctions? And if so, which of them do we ignore? Is a purely quantitative criterion (the number of users reporting such a distinction) sufficient?
Whether such an idiosyncrasy is acceptable, is one question; whether it can be deemed a name, i.e. something that has a more or less established position within our shared discursive practices, is another. The first question is in fact the question of whether there is an observable and learnable rule attached to the label that is, within the observed sample of people, idiosyncratic – because we have not seen, for some reason, more that just one person using it. Let us consider the abovementioned example: let us say that honeysuckle is a color shade that, in a certain company, only a devoted gardener is able to identify and recognize (a devoted flower-cultivating gardener is mostly – and again, stereotypically – a woman). If this is, however, a shade exemplified by the typical color in which honeysuckle flowers, then comparative objects and comparative procedures are intersubjectively accessible; and so – though with difficulty – the rules needed to distinguish between honeysuckle and, say, mauve can be explained and taught/learned. The important thing is that there is a rule; meaning, that the speaker using the name, “honeysuckle” is under public control, i.e. the meaning of the name is liable to institutionalization and each speaker has therefore, against the background of the institution, the option to make mistakes. In the opposite case, the name would be constituted a term in a “private language,” as discussed by Wittgenstein; hence, it would be dysfunctional.
For a detailed discussion of this topic, see Wittgenstein (2007, §§ 198ff). This suspicion can prove justified in the case of “sweet blue,” if the speaker accords, as the determining criterion, a particular feeling of sweetness, accessible as an “inner experience” only to himself/herself, or to any speaker who knows exactly this inner experience. Essentially, a user of a distinct, color name undertakes a commitment to explain and justify a normatively traceable distinction between this term and other familiar terms, with which others may be prone to confuse. By this commitment, he/she also delegates to the others the entitlement to refer to him/her. (“This is what Julie calls a ‘honeysuckle’ – or ‘sweet blue’ – color. She says she can tell it from mauve – I cannot – so ask her wherein it consists.”)
It was Robert Brandom (2000) who broadly elaborated on the structures of commitments and entitlements.
Certainly, that any such label can be introduced in a consequent manner – or that a few speakers can actually use it – does not mean that it can be called established. The first is a question for a philosopher of language who, using an analysis of the justifying procedures of the concerned speakers, should be able to tell whether it is not just a conceptual confusion or contradiction. But whether “honeysuckle” really is a piece of the actual repertoire of color names in a language, is really a question for the linguistics field – that is, the question whether there is a non-trivial number of speakers of a language able to use the term competently. A “non-trivial number” need not be a majority; and if we are specifically interested in the way women speak, it cannot be a majority. However, the context in which the concerned usage should be established must be institutionally open. Participation in it should not be prevented by explicit conditions – language is, in principle, democratic in this sense. This condition is met also by the particular linguistic patterns described in gender linguistics: though the particular phraseology is specifically allocated by gender, the competent speakers of the opposite gender are always invited to understand it, describe it or even practically adopt it. The determination of when a term is not just idiosyncratic anymore and can be considered established (to a certain extent) will always be somewhat approximate.
The existing dictionaries are, after all, artifacts too: what is included in them and in what structure does not always tally with what can be met in linguistic practice. To this issue, see Sally Rice (2011). The “female,” color vocabularies are mostly pretty well institutionalized in this respect: the existing dictionaries typically include most of the names that are met in the research field and the names can be learned.
The objections to the “horizontal,” non-hierarchic dissection of vocabulary are plain enough, e.g. the objection that the essential differences between names that play a systematically important role and the marginal ones are wiped out – that, so to speak, such a pattern of distinction testifies to finer sight and taste, but to a lazier intellect. The problem is that establishing a hierarchical dissection lacks a foundation that would be clearly intuitive. The field tests vary in the refinement of discriminative categories, of which the following is a possible example:
“basic” terms (blue)
“combined” terms (blue-green)
“qualified, basic” terms (light blue)
“elaborated” terms (azure blue)
“special” terms (azure)
“idiosyncratic, elaborated” terms (basin blue).
This is a variation of not-quite-identical categorizations in Rich (1977) and Simpson and Tarrant (1991).
There are some obvious formal criteria. We can make the point that at least formally, the “basic” names play the role of those that are further specified by other terms, but do not themselves specify others, i.e. we are accustomed to coming across “azure blue,” but not “blue azure.” How important is this formal observation? Should we perhaps eliminate such divisions of color chips where only some of the groups (that contain all of the relatively similar numbers of chips) are labeled “basic” terms, or some non-basic? Should we say such divisions are confused or flawed? I do not think so. As far as we are concerned with the pragmatics of language, with linguistic practice, we can also trace different criteria within the “basicality” of terms. We could equally well label as “basic” those terms that are more frequently used, the terms with which speakers work with a greater certainty (that are quicker or easier to identify, to recall, to attach with precision, etc.).
All these criteria are perfectly empirical, that is, can be, and actually have been tested. Lucy and Shweder (1979, 1988) show that from the viewpoint of this testing, the more basic terms are actually the more specified ones. What makes Berlin and Kay’s 11 the “basic” set is the fact that if the respondents are equipped with the instruction that they have only these names available to dissect the entire Munsell table, that should be enough to complete the task (irrespective of whether they are men or women). With a set of non-basic terms, they might quite easily fail in this task. At least because it would be difficult to compile a “complete” set of all these terms – the “Berlin and Kay” team has a head-start here, having investigated (or constructed) their framework for many years and having quite a clear idea of what is meant by a “complete” framework. This head-start allows them to present their set as the set, which fulfills the abovementioned condition (a set sufficient for an exhaustive division of color samples).
Anyway, we can easily imagine – there can be little doubt that it has actually happened somewhere – that the respondents lacking the introductory instruction about the “basic” nature of the 11 terms refuse to divide the chips set without residue, because they will consider the set to be insufficiently telling. On the other hand, a different design of the test can be imagined: for example, we could extract from an extensive authoritative dictionary of a language (say, English) all color names, except for the 11 basic terms, and make these available to the respondents for their chip-division task, with the proviso that they are not obliged to use all of them (but, on the other hand, without the explicit instruction that they are obliged to divide the whole set without residue). As far as I know, nobody has ever run such a test in reality. But it does not strike me as improbable that many respondents (perhaps most or all of them) would be able, without much difficulty, to divide the color chips without residue. (This design could be further elaborated by gradual reduction of the set of available expressions, until we – theoretically – reach the boundary determining the putative “minimum necessary” set of non-basic color terms covering the whole color spectrum.)
But even if we accept that the 11 “basic” terms would be enough for most (instructed) speakers to divide the chips, it is not clear whether this is due to their truly specific, privileged nature, or just because we have grown accustomed, since early childhood, to this specified set of basic colors. When we
Here, “we” is far from being, “we the people, in general.” learn, in childhood, the first names for colors, we meet, at first, a rather limited repertoire of a few names: red, blue, green, yellow, white, black. This basic repertoire is placed in front of us by the milieu into which we are born, and its constituents are formed by the language spoken within it. (And languages are not impartial systems of communication signals conveying information, but they contain a lot of sedimentary features of a “mythological” nature, so to speak.
See here, for instance, Wittgenstein’s “Bemerkungen über Frazers The Golden Bough” (1967). That we speak of “soul” or “mind”, just as we speak of “body”, promotes an unreflected assumption that “mind” and “soul” are objects just as bodies are – objects that can be “somewhere” localized.) Our ability to make do with the 11 terms need not be due to their systematically superior or privileged status among all the color terms, but due to the fact that we have merely learnt this as a linguistic skill (we have learnt to speak this way).
From the other point of view – by investigating the real linguistic practice of speakers – we could well observe that many speakers are accustomed to divide the known color spectrum by means of a set of expressions that does not tally with the official 11 and is often wider. Only some of the commonly used terms come from the authoritative “basic” set and there might be no clear hierarch, e.g. “azure” can be used along with “red” and “light-green,” without reflecting that one uses terms that are “actually” of three “different” categories. This promiscue usage of different kinds of names here is not typical for women, I think. What makes this non-discriminative pattern “female” is a rather weak (and not at all striking) trend shown by quite a lot of empirical research, along with, more importantly, prejudices induced by cultural and linguistic stereotypes.
As we have seen, the tendency not to divide the chips set in a “structurally orderly” way and to mix the basic terms with non-basic ones is typical of the “uninstructed” respondents of the “exotic” languages that the field researchers map. Some of them may try to find a special name for each of the chips (the Munsell set contains 330 chips) and this endeavor can be successful in some cases. The researchers in the World Color Survey project admit themselves that though some language has perhaps a narrower set of “basic terms,” its repertoire of expressions used for describing colors can be quite wide (save that they would not be “basic”).
Witkowski and Brown (1977). The “female” way of structuring color terms and their use strongly resembles the “savage” way – in opposition to the language of typical male speakers, from the perspective of how the research is conducted. The unwillingness or inability to work with the conceptual hierarchy and the tendency not to confuse the terms that are essentially more general with more specific ones (a lack of abstractional skill), is one of the features traditionally ascribed to little children, women or “savages” – all the kinds of people stereotypically considered as, say, less rational.
Our aim here is not a thorough critique of this stereotype. To check whether and to what extent it is a stereotype would probably be quite difficult in and of itself. We do not know how generally this difference holds – for all men and all women? – and the conducted experiments are not very significant, due to the differences in initial settings and designs and the relatively low number and diversity of respondents. What we can do is reflect upon these – hypothetical or not – different repertoires of color terms and subject them to some conceptual analysis. Let us note some interesting features of this suggested linguistic difference, linked to its stereotypical assumption and interpretation.
However irrational and unsystematic the strategy of color naming is (that which uses a higher number of terms without a hierarchy) – let us call it “female,” with the awareness of the conventional nature of this label – it has one great advantage. It allows colors to be unambiguously discerned: those who promote two different names for the same color chip either come to no agreement, or choose one of them. It cannot – or should not – happen that these two names can be ascribed at the same time, equally well. As far as we do not endow terms like “blue” or “green” a specifically privileged position, but admit to them a similarly limited place like all the other terms have, they should not be broad, general names with a blurred, boundary zone where they merge with their similarly broad neighbors. (“Broad” and “blurred boundary” are not optical terms characterizing a perception, but characteristics of the manner in which a language term is used.) A narrow and specific (as opposed to broad and general) use of the term “blue” means that we can decide, in all cases, when to use “azure,” when “aquamarine” and when just “blue” (or perhaps we avoid all three terms).
Sticking with the limited set of privileged, general – “basic” – names seems to be, on the other hand, a source of problems in practice, especially with respect to vagueness. Again, vagueness is not an optical problem, but a normative one: we can find extensive “subdomains” of the “male” linguistic mode where we cannot apply a rule that could be unambiguously followed, meaning, under certain circumstances various linguistic moves are deemed proper, but incompatible under different circumstances. It is clearly impractical to use such vague linguistic tools; if these problems could be bypassed by accepting a finer (or not that hierarchically rigid) scale of terms, and if a number of speakers could obviously be able to do this and actually do it in practice, then sticking with the limited scheme could be viewed as a mark of idleness or incapability, to distinguish it in a more adult way. If there are speakers for whom it is a common and practical thing to distinguish (in a normatively regular manner) between sky-blue, azure and cobalt, are they not then justified in seeing the linguistic practice of those who can make an assertion no more qualified than “blue” as a case of over-inclusion?
Is there anything that distinguishes, in practice, the “male” pattern of expressions from the “female” one? According to Paul Kay, the relatively high, yet specifically limited number of “basic” terms among modern Euro-American languages reflects the claims of a technologically developed civilization. Languages of these cultures, says Kay, need to have an operational set of sufficiently general terms, but ones that cover the whole color spectrum. 2 – in the opposition to 11 – terms are not practical enough to describe the surrounding world that captures the important differences within it in an appropriate way.
Kay and Maffi (1999).
5. The roots of the artifact
The unwillingness (or incapability) to abstract, typical of the “female” distinguishing pattern, marks, as it were, the boundary of “practical needs.” Let us remember the point Lakoff makes at the end of her reflections: the finery in color discrimination is left, or attributed to women, as a sop. It is a game men don’t have to play, since for the needs and tasks they are expected to cope with in practice, it is not required. The difference between the “male” and “female” patterns of color discrimination thus corresponds to the traditional difference between the domains of self-realization for men and women: the public, political domain of men and the private, domestic domain of women. On the one hand, there is the ability to express and utter statements that thematize differences (oppositions) understandable to the audience; on the other, there is the ability to lead conversation, continue and elaborate subtopics, without the imperative of arriving at a clear and sharp endpoint (contrary to the clash/refinement of practical stances, conversation should not have an endpoint at all, if possible). We can thus say that to distinguish a certain small, but not unsubstantial number of basic colors (about ten) is practical – which is also applicable to technology (color marking of different kinds of tools or gadgets, such as fuses in cars, operates at a scale approximate to this degree of finery). On the other hand, fine, “female” discrimination is somewhat too fine for the sake of practice and, moreover, relatively many speakers who can competently distinguish between red, yellow and green easily make mistakes when faced with the “ochre-beige-mustard yellow” scale. Making these kinds of mistakes is traditionally attributed to men.
There is various evidence which suggests that the fine, “female” pattern of discrimination might be an artifact. Kira Hall, in her linguistic research conducted among fantasy line workers, shows that these workers had to adopt a specifically “female” speech pattern, expected of, and demanded by their customers, as a role, despite it not having any natural or familiar associations for them. Developing this professional skill also entailed using a wide scale of terms to describe clothing, fabrics and colors. The breadth of the vocabulary mastered and used in their work was even a source of professional pride for some. Yet paradoxically, this manner of speech was expected of them, but their customers were ultimately unable to fully and properly appreciate this learned finesse. Hence, it is expected that women develop a particular manner of speech, typical of those who are conversant with colors, fabrics etc., but their skill is, at the same time, pushed to certain proper limits (what they can effectively display is not worth noticing, in the eyes of someone who understands it).
Hall (1995).
On the other side, the specifically “male” pattern may also appear to be an artifact. Its relative poverty – impoverishment in comparison with the markedly more varied usage of linguistic culture in the age before modern science – is considered to be a certain virtue in its own right (since it endows the individual with the capacity for abstraction, generalization, inspection). It is, in this form, established, conserved and cultivated in speakers, especially if their practice embodies, as an important principle, the instruction of not wasting time on impractical nonsense. This standard typically accompanies many “male” roles too. It is not that unusual to encounter the opinion that, for the true man, few colors – the exact number can vary, but it does not even reach Kay’s full eleven – are enough.
To summarize, there is some, but relatively poor, empirical evidence confirming (not in a dramatically salient manner) a wider repertoire of color names used by women. On the other hand, there are works that reflect critically upon “female superiority” in color naming as a symptomatic feature of the artifact of “women’s language.” To weigh these two interpretations against each other, one has to consider several things. First, language is not just an individual inborn capacity, but it is, to an important extent, an institution. This institution is sustained through individuals and their skills, but the set of opportunities, tasks and demands it sets is locally diversified (situation-specific). Therefore, it is not appropriate to speak of language as such, but rather of many different languages; where what we mean by this multitude is not just the multitude of languages, in the sense of English vs. Chinese vs. Latin etc., but also the multitude of context-dependent patterns of language performance that we can observe within individual languages as well as across them.
One of the prominent examples of this situational diversity is the stereotypical attribution of different speech manners and patterns to men and women. Whatever the origin of this difference is (much suggests it to be an artifact), we can study systematically different patterns that put structure on color vocabularies. We must not forget that a salient difference between these patterns is 1) more constructed than actually observed, and 2) held also as one of the tools that preserve the systematically different (inferior), social position of women. Regarding this, one should be rather suspicious about the claims of female superiority in the context of color discrimination. The trend of linking different social positions to different, gendered patterns of language use has been observed and described in a number of more general works that critically investigate the institutions of the distinct “male” and “female” languages; empirical studies that focus specifically on colors have been conducted only to a limited degree, and almost exclusively with English speakers.
The method of identifying and interpreting the “basic” status of BCT appears to be tied to several presumptive features of objectivity or rationality – stereotypically, characteristics of the male kind of cognition. In as far as we can prove that the opposition of the male and female frameworks in the context of color discrimination have artificial roots, so we can also question the opposition of basic versus non-basic frameworks.
References
Berlin, Brent and Kay, Paul. 1969. Basic Color Terms. Berkeley: University of California Press.
Blommaert, Jan, Collins, James, and Slembrouck, Stef. 2005. “Spaces of Multilingualism.” Language & Communication 25: 197–216.
Brandom, Robert. 2000. Articulating Reasons. Cambridge, MA: Harvard University Press.
Coates, Jennifer. 2005. Women, Men and Language. London: Pearson Education Ltd.
Frank, Jane 1990. “Gender Differences in Color Naming: Direct Mail Order Advertisements.” American Speech 65: 114-126.
Hall, Kira. 1995. “Lip Service on the Fantasy Lines.” In Kira Hall and Mary Bucholtz (eds.), Gender Articulated: Language and the Socially Constructed Self. New York: Routledge, pp. 183-216.
Heider, Eleanor. 1972. “Probabilities, Sampling, and Ethnographic Method: The Case of Dani Colour Names.” Man, New Series, 7: 448-466.
Kay, Paul. 1999. “The Emergence of Basic Color Lexicons Hypothesis: A Comment on ‘The Vocabulary of Color with Particular Reference to Ancient Greek and Classical Latin,’ by John Lyons.” In Alexander Borg (ed.), The Language of Color in the Mediterranean. An Anthology on Linguistic and Ethnographic Aspects of Color Terms. Stockholm: Almquist & Wiksell International, pp. 77-99.
Kay, Paul and Kempton, Willett. 1984. “What Is the Sapir-Whorf Hypothesis?” American Anthropologist, New Series, 86: 65-79.
Kay, Paul and Maffi, Luisa. 1999. “Color Appearance and the Emergence and Evolution of Basic Color Lexicons.” American Anthropologist, New Series, 101: 743-760.
Kay, Paul and McDaniel, Chad K. 1978. “The Linguistic Significance of the Meanings of Basic Color Terms.” Language 54: 610-646.
Kristol, Andres M. 1980. “Color Systems in Southern Italy: A Case of Regression.” Language 56: 137-145.
Lakoff, Robin Tolmach. 2004. Language and Woman’s Place. Revised and Expanded Edition. Oxford: Oxford University Press.
Lucy, John A. and Shweder, Richard A. 1979. “Whorf and His Critics: Linguistic and Nonlinguistic Influences on Color Memory.” American Anthropologist, New Series, 81: 581-615.
Lucy, John A. and Shweder, Richard A. 1988. “The Effect of Incidental Conversation on Memory for Focal Colors.” American Anthropologist, New Series, 90: 923-931.
Lyons, John. 1999. “The Vocabulary of Color with Particular Reference to Ancient Greek and Classical Latin.” In: Alexander Borg (ed.), The Language of Color in the Mediterranean. An Anthology on Linguistic and Ethnographic Aspects of Color Terms. Stockholm: Almquist & Wiksell International, pp. 38-75.
Nowaczyk, Ronald. 1982. Sex-related Differences in the Color Lexicon. Language and Speech 25: 257-265.
Regier, Terry, Kay, Paul and Cook, Richard S. 2005. “Focal Colors Are Universal after All.” Proceedings of the National Academy of Sciences of the United States of America 102: 8386-8391.
Rice, Sally. 2011. “Unlikely Lexical Entries.” Proceedings of the Fourteenth Annual Meeting of the Berkeley Linguistics Society: 202-212.
Rich, Elaine. 1977. “Sex-related Differences in Colour Vocabulary.” Language and Speech 20: 404-409.
Saunders, Barbara and van Brakel, Jaap [et al.]. 1997. “Are there nontrivial constraints on colour categorization?” [+ Commentaries & Response] Behavioral and Brain Sciences 20: 167-228.
Simpson, Jean and Tarrant, Arthur W. S. 1991. “Sex- and Age-related differences in Colour Vocabulary.” Language and Speech 34: 57-62.
Witkowski, Stanley R. and Brown, Cecil H.. 1977. “An Explanation of Color Nomenclature Universals.” American Anthropologist, New Series, 79: 50-57.
Wittgenstein, Ludwig. 1967. “Bemerkungen über Frazers The Golden Bough.” Synthese 17: 233-253.
Wittgenstein, Ludwig. 2007. Philosophical Investigations. 4th Revised Editions. Oxford: Wiley.