The emergence of syntax, the combination of meaningful elements into propositions, is developmentally preceded by the solitary appearance of those elements themselves. In the child's development, as in human evolution, the advent of words – culturally forged markers for shared notions – confers the ability to convey the conceptions of one's mind to that of another, albeit imperfectly. In a very immediate way, the word allows us to control other people's minds, which provides a perverse but perhaps useful way of understanding the hackneyed observation that culture is based upon language. While nonverbal forms of communication, in our own and other species, can also be said to consist in this transfer of mental contents, words allow much greater precision in this, sentences an exquisite degree.

The first words, which seem to emerge out of nothing around the child's first birthday, do not, of course, appear ex nihilo. There is a developmental history to these earliest utterances, one that has been charted in great detail by Bates (1976,1979), Bloom and Lahey (1978), and Bruner (1983), to name just the Bs.

PRECURSORS OF THE WORD

Utter neonates are communicative to the extent that their vocalizations, gestures, and expressions convey information about their condition to caretakers. Though this communication may be unintended on the infant's part, it is effective in eliciting the kinds of nurturance that undoubtedly account for its having been naturally selected. Such controlled utterances are soon supplemented by still unconventional but intentional vocalizations and gestures. These articulations, which tend to accompany the child's taking notice of something, acquire a certain consistency in sound or appearance, although they are quite variable from one child to the next.

A minimal general definition of syntax would be a rule or set of rules dictating the relationships permissible among the elements of a set. These elements, in addition, must be grouped into categories. The syntactic rules are formulated as permissible relations among those categories, rather than as configurations of the elements themselves. Otherwise, the rules would consist merely of an enumeration of all well-formed expressions (whether sentences, propositions, equations, or designs) in the representational system. In that case, one would have specified not the grammar of the system but the system itself, that is, all of its component expressions. Such an approach is not only ungainly in general but impossible as well for natural languages and all other representational systems that have among their rules recursive functions, since, as discussed in chapter 4, such rules provide for an infinite number of expressions.

It is beyond dispute that all natural languages are predicated on rule systems – grammars – involving elements grouped into nested categories of increasing abstraction. Thus “bird,” an element of English, is a member of the noun category, which in turn is a constituent of the noun phrase, which may be part of a sentence subject or sentence object, components of the category “sentence.” The production and understanding of sentences in a natural language entails the speaker's processing of low-level elements, such as “bird,” but involves also the mental manipulation of the more abstract grammatical constituents.

The rules for forming questions in English are an often-cited and clear example of this.

The objective of this chapter is to illustrate an approach capable of addressing whether or not different linguistic configurations of meaning produce cognitive consequences. This will be accomplished by seeking evidence that the linguistic patterns described in chapter 2 have correlates in cognition. If cognitive performance corresponds with linguistic structure in a distinctive way, the proposal that the specific structure of a language is related to thought gains support. As discussed in Language diversity and thought, chapter 7, it is not possible with correlational techniques to establish unequivocally that language is the shaping factor in such a relationship; the language patterns may in fact derive from culturally specific thought patterns, or both patterns may derive from some third cultural factor. But correlational evidence can be extremely suggestive of a causal role for language if the relationships are strong and distinctive and if no other explanation for the contrasting cognitive patterns seems plausible. It will be difficult to achieve conclusive results in a single study – especially when it is exploratory, as the present one is – and further work using additional assessment procedures and other linguistic and cultural groups will be needed to establish the reliability and generality of the present findings and to rule out plausible alternative explanations. Nonetheless, the results reported here are suggestive of some interesting and reliable relationships between linguistic patterns and habitual thought.

General approach

To find connections between language and thought, specific hypotheses concerning potential linkages must be formulated. The move from the linguistic forms of English and Yucatec to an assessment of possible psychological connections requires three analytic steps.

Subject characteristics and task sequence

A general description of subjects engaging in the tasks and the sequence of tasks was given in chapters 1 and 3. Tables 39 and 40 present additional specific details on these topics. Only tasks 6–12 in these tables have been reported here. The series 1 experiments concerned other issues and only served in the present context to familiarize the men with the task procedures. Unnumbered items represent supplementary procedures used with the preceding numbered task.

Counterbalancing of stimulus presentation

The picture stimuli were not easy to construct. Thus, it was desirable to use each picture in more than one task and to have every subject deal with each picture at some point, but not necessarily always in the same tasks.

Given five tasks (I–V), three sets of pictures (labeled 1, 2, and 3), and the requirement that a common picture set be used in Tasks I and V, and in Tasks II and III, six possible presentation orders (i–vi) are possible as indicated in table 41. On the basis of this, it made sense for subjects to be sought in groups of six since all possible orders could be equally represented.

Finally, since each picture set contained three originals (for example, 1.1, 1.2, 1.3), it was possible to order the presentation within each set in six possible ways as well. As there were not enough subjects to completely counterbalance this factor, the six possible combinations were assigned randomly (without replacement) to the six occurrences of each set associated with each six subjects.

Introduction

The relationship between language diversity and thought did not receive a significant amount of research attention from psychologists until the 1950s, with the emergence of the subdiscipline of psycnolinguistics. Although this increased interest in language within psychology can be traced in large part to Whorf's work, actual studies of the cognitive significance of linguistic diversity have represented a relatively small portion of the total research effort. This is consistent both with a general tendency on the part of psychologists to avoid intercultural comparison and with a specific reluctance to cope with the full complexity of linguistic structure and diversity. The tradition is also marked by a strong concern with experimental assessment of individual cognition, and all the studies in this tradition compare some feature of one or more languages with the behavioral patterns of a sample of speakers on tasks designed to reveal cognitive processes.

These psycholinguistic studies can be divided into two broad groups: those involving the significance of lexical codability – to be dealt with in this chapter – and those involving the significance of some aspect of grammar such as form classes or logical relators – to be dealt with in the next. By far the majority of the studies fall into the lexical group, and the majority of these, in turn, concern the significance of color terms for cognition. These studies of the lexical encoding of color will be the focus of this chapter.

Basis of the decision to use line drawings

Stimuli used in the picture task series consisted of line drawings of typical Mayan village scenes and included specific numbers and types of objects relevant to the hypothesis in question. Constructed stimuli were used rather than actual village settings because it is very difficult to control the occurrence of objects (especially animate ones) in natural settings and because it is difficult to replicate the same setting across subjects and cultures. Artificial arrangements of objects were not used because it was difficult to include complex contextual features (for example, forest, house interiors), because it was difficult to include effectively many types of objects (for example, smoke, clouds), because pretesting revealed that small replicas of objects evoked unusual responses among the Yucatec subjects (for example, replicas of animals elicited great amusement), and because the practical difficulty of rapidly manipulating arrangements of objects made them difficult to use in some of the nonverbal tasks (for example, recognition memory).

Photographs of artificial arrays of objects would have solved this last problem, but not the others. Photographs of natural settings were rejected in part because they presented some of the same difficulties as the natural settings, especially the difficulty of setting up the situations to photograph so as to include the appropriate objects of interest and to exclude unwanted elements. More importantly, however, in pretesting photographs inevitably evoked too great an interest on the part of the Maya as to the particulars of who was involved, what they were doing, where the photograph was taken, etc. It can be argued that this was a desirable aspect of photographs, better reflecting speakers' actual responses to situations.

Introduction

Benjamin Lee Whorf (1897–1941) was trained as a chemical engineer at the Massachusetts Institute of Technology and worked as a fire prevention engineer for the Hartford Insurance Company for his entire professional career. Avocationally, however, he pursued a wide variety of interests, centering for the most part on a deep concern for the apparent conflict between science and religion. This general interest eventually became focused on linguistic problems, and it is in the area of language-related studies that he made his most important scholarly contributions. (Biographical accounts of Whorf's life and his interests can be found in Carroll, 1956, and Rollins, 1972 and 1980. Whorf, 1956a, also contains a bibliography of most of Whorf's published and unpublished works.)

Whorf was initially self-taught in linguistics, but later (after 1931) benefited significantly from interaction with Sapir and his circle of students at nearby Yale. His interest in and formulation of the specifically linguistic relativity principle probably stemmed in large part from this contact with Sapir. It is important to realize that despite his “amateur” status, Whorf's work in linguistics was and still is recognized as being of superb professional quality by linguists. He produced general descriptive works on the modern Nahuatl (Aztec) and Hopi languages, partial descriptive studies of a variety of other languages contemporary and ancient, historical reconstructions of the Uto-Aztecan and adjacent language families, epigraphic studies of Mayan and central Mexican hieroglyphic writings, and a number of theoretical articles.

The present study forms part of a two volume work that re-examines empirical research on the linguistic relativity hypothesis, that is, the hypothesis that differences among languages in the grammatical structuring of meaning influence habitual thought.

The companion work, Language diversity and thought: a reformulation of the linguistic relativity hypothesis, presents an analytic review of the historical development of the linguistic relativity hypothesis and various past attempts to provide empirical evidence directly relevant to it. The review is concerned with methodology in the broad sense, that is, with identifying the general requirements of adequate empirical work on this problem. The review analyzes in detail the flaws and the achievements of existing studies with the aim of formulating an improved approach to such empirical research. The present work provides a concrete case study which utilizes this improved approach.

General goals

The global aim of the project described in the present volume is to demonstrate how the linguistic relativity hypothesis as traditionally conceived can be investigated empirically. Taking Whorf's formulation and subsequent empirical research as the point of departure, four components of adequate empirical research on the linguistic relativity hypothesis were described in the concluding chapter of Language diversity and thought. The discussion also indicated which approaches were most likely to lead to significant progress at the present time. These components and the proposed approaches to them can be distilled as follows.

First, such research must be comparative, that is, it must present contrastive data on two or more language communities.

General orientation

This study examines the linguistic relativity hypothesis, that is, the proposal that diverse languages influence the thought of those who speak them. This has been an important issue in our intellectual tradition for several hundred years. How far back the concern can be traced is largely a function of how one understands the terms language and thought and how explicit an hypothesis one requires. Relatively contemporary formulations begin to appear in eighteenth-century Germany with the work of Machaelis, Hamann, and Herder (R. L. Brown, 1967; Koerner, 1977; Penn, 1972; Stam, 1980; cf. Aarsleff, 1982; Friedrich, 1986). Interest continues in the nineteenth century, again particularly in Germany, and begins to separate from speculative philosophy, especially in the work of Wm. von Humboldt. In the twentieth century the main line of empirical research moves to America with the development of Boasian anthropology. There are undoubtedly important connections between this new work in America and the earlier nineteenth-century German work of Humboldt, Müller, and Steinthal (Haugen, 1977; Hymes, 1963; Stam, 1980), but there was fresh impetus too deriving from firsthand contact with the incredible diversity of native American linguistic forms and from the emergence in contemporary anthropology of an anti-evolutionary project with an emphasis on the functionally and temporally equivalent value of these diverse languages (Stocking, 1974).

The general roots of our interest in the problem of linguistic relativity have never been extensively examined, although this is an important problem (Lucy, 1985a; cf. Friedrich, 1986).

Introduction

An examination of the relationship of linguistic diversity to thought requires the identification and analysis of a specific case of such diversity. Ideally, following Whorf, the contrast would be of large-scale “fashions of speaking.” This ideal guides the approach taken here in two respects. First, the focus is on a structurally and semantically significant aspect of the grammar rather than on a little used or structurally marginal pattern. Second, some preliminary indications are given as to how these patterns fit together into larger semantic structures in each language.

It would also be desirable to investigate a substantial sample of languages. This is approximated here in that the linguistic characterizations and comparisons explicitly take account of typological data on an array of languages. Further, the analysis has been constructed, both methodologically and substantively, to facilitate extension to additional languages. However, detailed linguistic discussion will be provided only for American English and Yucatec Maya – the two languages for which cognitive assessments will be undertaken.

The phenomenon of concern in this chapter can be loosely labeled in traditional terms as nominal number marking. Notionally, it involves various indications of the multiplicity, number, quantity, or amount of some object of noun phrase reference relative to a predication. Formally, it includes such things as plural inflection, plural concord, and indication of singular or plural by modification of the lexical head of a noun phrase with a numeral or other adjective indicating quantity or specificity of amount. Nominal number has been chosen for several reasons. First, it is one of the most commonly encountered and most central noun phrase categories; as such it is ideal for comparative purposes.