C. L. Hardin and Luissa Maffi, Editors, Color categories in thought and language. Cambridge: Cambridge University Press, 1997. Pp. x + 404. ISBN 0 521 496934 (hardback) — ISBN 0 521 498007 (paperback).

Robert MacLaury, Color and cognition in Mesoamerica: Constructing categories as vantages. Austin: University of Texas Press, 1997, xxvii + 616. ISBN 0-292-75193 (hardback).

In a message posted to one of the cognitive science discussion groups the author asked, to paraphrase roughly, what should be read to get an up-to-date account of research into color naming? My advice is (and was) to consider the two books under review here: C. L. Hardin and Luisa Maffi’s excellent collection of essays on color language research; Robert MacLaury’s magnum opus on color naming and cognition.

Color language research finds its modern roots in the work of the anthropologist Brent Berlin and the linguist Paul Kay. Basic Color Terms: their universality and evolution is their seminal text which, when published in 1969, created a controversy that continues to this day (N.B. the recent Behavioral and Brain Sciences target article "Are their non-trivial constraints on color categorization," Saunders and van Brakel 1997). At the heart of that controversy is the following claim: some color categories—linguistic classifications such as "red" or "black" in their English versions—are "basic" in the sense of being cultural universals. One might expect this to mean that such words appear in all languages. Not so. A color word such as "pink" is universal according to Berlin and Kay (1969), but there are many languages which do not have an associate term. In what sense, then, is there "universality"? Enter Berlin and Kay's (1969) evolutionary hypothesis. While it is not the case that every language has all of the basic terms, each language does possess some subset of them—a subset which can be ordered with all others—into a developmental sequence that moves from languages with only two basic terms to those with all eleven (i.e., red, yellow, green, blue, orange, purple, pink, brown, grey, black, white). Every language (choosing some of the small set of basic terms) stands at some "stage" in the sequence and adds further color terms in a predictable way.

For thirty years an ongoing and interdisciplinary research tradition has refined and revised these central claims. A major aspect of this tradition has been devoted to their explanation. It is argued, for instance, that the nature and structure of the color vision system is responsible for the high degree of psychological salience which attaches to exemplars of the basic color names. Most researchers designate four "elemental" colors—red, yellow, green, and blue—as especially salient (similar claims are made for black and white which I ignore here). These are said to be elemental because, for each of these four colors or "hues," there are samples which are perceived to contain no other chromatic content. There is a "pure" blue, for example, in a way in which there is not a pure orange. Orange is not elemental but composed of a mixture of red and yellow. What accounts for elementalness? Again, we are propelled to vision science and the widely agreed upon opponency of the color vision system. Articulated psychophysically and physiologically, the idea is that the vision generates the four elemental hues, a fact that color categorizers exploit in their construction of linguistic categories. Taken as a whole, the goal of contemporary color language research is to understand the links and relationships between our vision system (understood psychologically, psychophysically, and physiologically) and the socially embedded practices of color naming.

The essays collected in Color categories in thought and language originate from a conference that the philosopher C. L. Hardin organized and which took place at Asilomar, California in the fall of 1992. Organized into four sections, Color categories in thought and language begins, in Part I, with an essay by Kay, Berlin and their collaborators Luisa Maffi and William Merrifield. This is appropriate not only because Berlin and Kay have founded the contemporary research tradition but also and especially because they are due to publish their long-awaited multi-volume work, theWorld Color Survey (Forthcoming).Their essay "Color naming across languages" is based on the data of that project and "serves as something of a progress report on the current state of analysis of the WCS [World Color Survey] data, as well as a promissory note on the full analysis to come" (p. 21). This article provides terms of reference for those to follow: a summary of cross-linguistic research, a brief presentation of the data and the techniques used to collect it, a survey of the conceptual issues that have been and need to be addressed. Part II, "Visual Psychologists" is devoted to essays by visual scientists. The title of Part III "Anthropologists and Linguists" is self explanatory. Part IV, "Dissenting Voices" provides two articles critical of color naming research in the mold of Berlin and Kay. Bracketing the four parts of the book we find introductory and concluding essays by the editors Hardin and Maffi (Ch. 1 and Ch.16) that put the various contributions into a broader perspective.

The first three essays in Part II are primers on different aspects of vision science and their relevance to color naming research. In Chapter 3, "The psychophysics of color," Bill Wooten and David Miller provide an account of the opponent processes theory of color perception and its experimental and conceptual development. In Chapter 4, "Physiological mechanisms of color vision," Israel Abramov addresses the encoding of color stimuli in the nervous system, offering fact and speculation as to the relationship between the functional, psychophysical models described in the previous chapter and contemporary physiological accounts of vision. Chapter 5, by the neuropsychologist Jules Davidoff, deals with color vision deficits and the light they cast on the mechanisms of color vision and on linguistic color classification itself. There is much overlap in these three articles (throughout the book in fact) and this is a good thing. One comes away with a clear sense that there is no ultimate separation of, say, the psychophysics and the physiology of color, two sub-disciplines of vision science that are continually posing questions to one another. Abramov, for example, makes it clear that there is no simple fit between the psychophysics of opponency and its physiology as they are presently understood (p. 107). That said, he speculates about what we should expect from physiology given the constraints of psychophysics: "At higher, presumably cortical, levels of the nervous system the responses of LGN [Lateral Geniculate Nuclei] cells must be combined and recombined in many ways in order to disambiguate their responses, to extract the information corresponding to each sensory function" (p. 108, my italics). Other contributions to Part II focus on more specific areas of interest. In Chapter 6, Robert Boynton ("Insights gained from the naming of OSA colors") discusses his own history of color naming research and concludes that too much emphasis has been placed on opponent models when it comes to discussion of Berlin and Kay’s work. His research indicates that the original eleven basic color terms name colors that are equally salient. Elementalness may be a fact, but it does not matter much when it comes to hue-naming experiments. Boynton—and again we see the constraining role of psychophysics—proposes that there may be eleven neural processes, each corresponding to a Berlin-Kay color (p. 148).

The hierarchical distinction returns in Chapter 9, the first essay of Part III. The linguist Greville Corbett and the psychologist R. L. Davies ("Establishing basic color terms: measures and techniques") argue that a hierarchy with the elemental colors as most significant appears when one utilizes appropriate techniques for identifying basic terms but is submerged if other tests are used. The elicitation of lists from subjects is good at identifying basic terms, not good at revealing hierarchy. Evaluation of the frequency with which basic color names appear produces an opposite result: hierarchy emerges but one does not get an adequate reading of the range of basic terms in a language. As the authors note, these two types of test (and others are considered as well) are "complementary" (p. 218). On the other hand, their evaluation of the various tests for basicness underlines an important and sometimes overlooked home-truth. The "results" one gets often depend on the nature of the task and it is unwise to hang too much generalization on any result or type of result. This theme threads its way through many of the papers in this volume—see, in particular, the contributions by Lars Sivik ("Color systems for cognitive research," Ch. 8) and David Miller ("Beyond the elements: investigations of hue," Ch. 7)—and it is one which we might apply to Boynton's claims, based as they are on hue-naming tasks.

As befits its devotion to the work of anthropologists and linguists Part III is richer in cultural and linguistic detail. In the wonderful Chapter 10—"Color shift: evolution of English color terms from brightness to hue"—the anthropologist Ronald Casson reconstructs the history of English color words and their development from essentially brightness classifiers to hue terms (see also MacLaury, Ch. 12, "Skewing and darkening: dynamics of the Cool category"). James Stanlaw (Chapter 11, "Two observations on culture contact and the Japanese color nomenclature system") discusses the vexing problem of loan words and the results of different cultures coming into contact with each other. Stanlaw is also an anthropologist and, like Casson, he is circumspect when it comes to universalism: "the universalist arguments of Berlin and Kay do not necessarily refute all Whorfian [i.e. relativistic] considerations under all conditions. Languages can certainly vary semantically, but obviously not without constraint. . . . These constraints, however, are often a complex interface of both human and cognitive universals, and the particulars of cultures and languages in contact" (p. 259). This is a modest form of cultural relativism. As the quotation from Casson indicates, he is not denying the legitimacy of the Berlin-Kay schema. It is, rather, a matter of locating them in cultural context.

In the first of the two essays that appear in Part IV, "Dissenting Voices," Kimberly Jameson and Roy D'Andrade (a cognitive psychologist and a cognitive anthropologist respectively) take direct aim at the relevance of opponent colors theory (cf. Boynton, Ch. 6). The standard explanatory paradigm supposes a psychophysical color space organized in terms of two axes: one from red to green, the other from yellow to blue. The high degree of salience which attaches to these colors—a result that shows up in many (though not all) measures of significance—is held to derive from physiological organization (though the physiological story is not complete: see the comments on and by Abramov, above). Jameson and D'Andrade reject this paradigm, arguing that, on certain psychological and physiological measures, a different set of axes is specified (pp.300-301). This is well and good but what justifies their choice of measures over the numerous tasks and techniques that converge on the elementalness of red, yellow, green, and blue? In their concluding remarks Hardin and Maffi discuss this issue (pp. 356-58) and a similar one (p. 350) that arises from the anthropologist John Lucy's essay "The linguistics of color" (Ch. 15). Lucy has long been a critic of Berlin and Kay's work on color (Lucy and Schweder 1979). In this, the final contribution to the volume, he argues that color words do not form a legitimate linguistic category: "we have the extraction of a set of individual lexical items from the grammar primarily on the basis of their capacity to refer to a fixed stimulus array, and then the reduction of that list in terms of the item's denotational potential and internal relations with one another" (p.326). This continues a time-honored line of attack on universalist research: universalist and evolutionary claims are "reductionist," eliminating important social and linguistic dimensions. Color words do not, Lucy argues, constitute a well-formed syntactic class, being interesting to universalists for essentially semantic reasons. Further, universalists fail to do justice the semantic dimension. It is hard be unsympathetic to Lucy and his plea for detailed studies. It is also difficult to see the supposed chasm between the local, specific, ethnographical studies of color language, such as Lucy prefers, and those that are explicitly simplifying, general, and cross-cultural. It is an old debate. Levi-Strauss, with some exasperation, had this to say about an earlier relativist, Franz Boas: "Are we compelled to carry Boasian nominalism to its limit and study each of the cases observed as so many individual entities?…. to reject institutions exclusively in favor of societies?" (quoted in Sahlins 1976, pp. 68-69 ). Levi-Strauss' general point is that whether it makes sense to construct theories of practices which extend beyond a given society is partly determined by what those theories reveal and cannot be settled by appeal to traditional methods, approaches, and assumptions. This is a useful idea to keep in mind when evaluating claims about color naming—claims which straddle the boundaries between traditional disciplines.

If Color categories in thought and language has a weakness, it is that many though certainly not all of its authors are unfamiliar with research conducted in related areas. Hardin and Maffi fill in many gaps in their "Introduction" and "Closing thoughts" but it remains true that, as the book-jacket text informs us, the Asilomar conference is where "visual scientists and psychologists met with linguists and anthropologists for the first time…." Given the length of the research tradition, and the fact that vision science has been in on the story from the start, this is surprising. What is not is the fact that there is some insularity in the otherwise excellent essays of this volume.

Having perused the contributions of distinguished visual scientists, anthropologists, and linguists to Color categories in thought and language one is advised to turn to Robert MacLaury's Color and cognition in Mesoamerica: constructing categories as vantages. This book provides a unified treatment of color naming research by one of the field's preeminent investigators. MacLaury, a cognitive anthropologist and former student of Berlin and Kay at Berkeley has undertaken color naming fieldwork in various areas of the world and he has published numerous articles on the topic. Color and cognition in Mesoamerica: constructing categories as vantages is his magnum opus. It brings together a startling and impressive body of history, data, theory, and speculation based on (but not limited to) fieldwork by MacLaury and others in Mesoamerica.

MacLaury is something of a heretic in the color naming trade. He disagrees with Berlin and Kay's concentration on hue as the attribute of visual sensation that speakers use to categorize color. There is, he argues, a very complex brightness sequence which may run parallel to the hue sequence, though it will always, eventually, merge with the hue sequence (pp. 44-46). This definitely complicates the Berlin-Kay account which was first developed in 1969 and subsequently revised (pp. 23-30) and there has been some resistance to it in the Berlin and Kay camp. My own view is that MacLaury is dead-on about this. There is a legitimate hue sequence, as MacLaury agrees (pp. 44-49), but it is a sequence that occurs only if speakers have latched on to hue. If they have not, then brightness classifications may be operative. Since hue is not the only dimension in which color samples can be scaled by subjects—brightness and saturation are also standard—it would not be surprising to find these attributes operative in classification. MacLaury's data suggests that speakers have to find their way to hue as the primary classificatory dimension, and this is a more realistic view than accounts which take its dominance for granted.

There are other respects in which MacLaury's work differentiates itself from that of Berlin and Kay and the universalist tradition in general. Here I note some significant differences. In the first place, MacLaury takes individual data very seriously. Most color naming research is based on results for linguistic groups. One interviews a number of informants from a given language group, normalizes their responses so that it makes sense to speak of the basic color terms in their language, assigns the language to a stage of development. Where is the individual or social mind in this? Nowhere. The idea is that color language development is like embryological development: an epigenetic process that unfolds as a consequence of human biology under tight constraints as to variation. The basic color categories are physiologically determined and any deviation from the universal sequence—given the robustness of that sequence—is noise. But there are, MacLaury argues, individual differences in color naming that do not fit any standard model and which cannot be chalked up to mistake or noise (p.111). They exhibit, instead, a range of strategies that require articulation and explanation.

The attention to individual data has lead MacLaury to an explicitly cognitive account of color naming. His essential principle is that "Color categorization is both perceptual and cognitive" (p. 86). You cannot categorize in terms of color unless you have conceptualized that domain. More to the point, you cannot have a theory of color categorization that is not, in part, a cognitive theory. While virtually all color naming research pays lip service to a "cognitive component" in categorization (it must be there, after all)—just as it pays lip service to a "cultural component" (it must be there too!)—MacLaury is the only person to have worked out, in detail, such an account (for another discussion of why this is required and what it might be like see Dedrick 1998). Differences that are glossed over by other researchers are grist for MacLaury's "vantage theory" which, he hopes, has the resources to account for a wide range of differences in individual color naming, cross culturally (pp.392-3). One is tempted to the thought that this is the route to madness and, indeed, the complexity of vantage theory—which MacLaury hopes will have general significance as a theory of categorization— is remarkable. Its articulation takes up the middle two parts of this four-part book: 266 pages. That said, the basic idea is simple and good. In a domain with no natural boundaries—that of the color continuum, say—one needs cognitive landmarks. The salience of the elemental hues can serve this function: one can exploit it in one's partition of the sensory color space into categories. This is not a new idea. Eleanor Rosch (1975) utilized it in her early and influential work on color "prototypes." But there are, as MacLaury notes, problems with a prototype theory of color categorization (p. 182) and with its main rival, the "fuzzy set" interpretation of color categorization later introduced by Kay and McDaniel (pp. 30-36). Vantage theory is intended to provide a more adequate account of color categorization. Speakers/cognizers construct complex "vantages" that consist of a hierarchy of fixed and mobile coordinates—analogous to space-time coordinates—in relation to which the individual can "zoom" (a technical term, p. 538 ). This involves the shifting from one sort of coordinate as fixed to another as fixed, collapsing (or inflating) the hierarchy as one goes, and utilizing mobile coordinates in relation to those that are, at a level, fixed. The theory proposes that elemental hues typically serve as the fixed coordinates from which color categories are constructed by subjects: their fixed position plus an attention to similarity and/or an attention to difference driving the boundaries to different positions in color space. Crucially, and for a variety of reasons, the coordinates may shift and this allows MacLaury to explain what is happening when an informant alters his or her categorizational practice, or to explain why it is that the speakers of a single language may differ in systematic ways. This later feature of the theory is especially important for the Mesoamerican data. Mesoamerican speakers use a range of categorizational strategies—especially in their categorization of green and blue—the so-called "Cool" colors (Ch. 8, "Skewing and darkening").

From my vantage the most impressive thing about this book is the range of issues it covers. Aside from the fifty-page bibliography and the useful (if somewhat idiosyncratic) thirty-page glossary, this book includes interesting discussions on everything from the history of color naming research, to methods and standards, to an excellent meditation on why it is that color categorization develops at all. For the data-junkie this book has scores of naming maps distributed throughout the text, many of them for individuals. Of the eight appendices, mostly concerned with technical information, there are two of particular interest. An "Inventory of observations" (Appendix VII) which lists 100 (!) observations any theory of color categorization ought to account for, and a good discussion of whether there could be more than 11 basic color terms (Appendix IV, "A cognitive ceiling of eleven basic color terms").

Taken together Color categories in thought and language and Color and cognition in Mesoamerica add up to more than 1000 pages of writing on color naming research. Who would want to read this much? Is it worth it? The first thing to note is that there is a lot more where these books came from. The literature is quite large (NB MacLaury's bibliography) and is likely to expand further. These books are good guides to that literature, though neither is designed for that purpose. Secondly, these two books are complementary. Color categories in thought and language offers multiple, often disciplinary, perspectives on color naming while Color and cognition in Mesoamerica presents a more unified treatment. One can, if one wishes, "zoom in" on technical vision-scientific issues that MacLaury deals with perfunctorily, finding them in all their messy glory in an essay from Color categories in thought and language. MacLaury, on the other hand, gives us a bigger if in some respects sketchier picture of how all the material that forms the inter-discipline of color naming research might fit together and this, after all is said and done, is what the research tradition must be about. Finally, as both of these books make clear, color naming research is one of the few perceptual/cognitive areas of study where there actually are workable accounts at a variety of levels: physiological, psychophysical, cognitive-psychological, linguistic, ethnographic. If this is the territory that cognitive science must traverse, the reader of these books will have found a locale where the journey is well underway.

Berlin, B. and Kay, P. (1969)Basic Color Terms: Their universality and evolution. University of California Press.

Dedrick, D. (1998) Naming the rainbow: Colour language, colour science, and culture. Kluwer Academic Publishers.

Kay, P., Berlin, B., Maffi, L. and Merrifield, W. (Forthcoming) World Color Survey.

Lucy, J. and Schweder, R. A. (1979) "Whorf and his critics: Lingusitic and non-linguistic influences on color memory." American Anthropologist 81 : 581-615.

Rosch, E. (1975) "Cognitive reference points." Cognitive Psychology 7 : 532-547.

Sahlins, M. (1976) Culture and practical reason. University of Chicago Press.

Saunders, B. A. C. and van Brakel, J. (1997) "Are their non-trivial constraints on color categorization." Behavioral and Brain Sciences 20 (2) : 167-179.

Don Dedrick

Department of Philosophy

University of Guelph

Guelph, Ontario, Canada N1G 2W1

Email: ddedrick@uoguelph.ca