Do bilinguals have different concepts? The case of shape and material in Japanese L2 users of English.

Abstract

An experiment investigated whether Japanese speakers' categorization of objects and substances by shape or material is influenced by acquiring English. Based on Imai and Gentner (1997), subjects were presented with an item such as a cork pyramid and asked to choose between two other items that matched it for shape (plastic pyramid) or for material (piece of cork). The hypotheses were that for simple objects the number of shape-based categorizations would increase according to experience of English and that the preference for shape- and material-based categorizations of Japanese speakers of English would differ from monolingual speakers of both languages. Subjects were 18 adult Japanese users of English who had lived in English-speaking countries between six months and three years (short-stay group), and 18 who had lived in English-speaking countries for three years or more (long-stay group). Both groups achieved above criterion on an English vocabulary test. Results were: both groups preferred material responses for simple objects and substances but not for complex objects, in line with Japanese monolinguals, but the long-stay group showed more shape preference than the short-stay group and also were less different from American monolinguals. These effects of acquiring a second language on categorization have implications for conceptual representation and methodology.

Key words

bilingualism

categorization

Japanese ESL users

linguistic relativity

objects and substances

1 Introduction

Since at least Weinreich (1953), research into second language (L2) acquisition and bilingualism has concentrated on the relationship between the first language (L1) and the second language or languages in the mind of the same person but has paid little attention to the concepts present in the same mind. This paper opens up the discussion of whether the minds of people who speak two languages differ from those of monolinguals in concepts as well as language by reporting an experiment that investigated whether the concepts of Japanese speakers are influenced by acquiring English.

The starting point must be to show that speakers of different languages have different concepts. Recent years have seen a renewed interest in linguistic relativity, a regeneration of the Whorf-Sapir hypothesis. To take three examples relevant to the later discussion, people express locative deixis either relative to their own orientation or as absolute orientation (Levinson, 1996); speakers of Berinmo and English have different boundaries between colors (Davidoff, Davies, & Roberson, 1999); those who speak languages that mark gender perceive inanimate objects as having characteristics typical of their grammatical gender (Boroditsky, Schmidt, & Phillips, 2003).

This is not to say that people have not argued strongly against such variation in concepts between cultures, for example the claim by Li and Gleitman (2002) that Levinson's results were experimental artifacts, based inter alia on whether the experiment took place out of doors or indoors, refuted in Levinson, Kita, Haun, and Rasch (2002). In the light of this recent wave of research, at least it cannot be taken for granted that language varies between people but concepts do not, even if it is unclear how important such differences may be. Knowing a particular language goes with knowing one set of concepts rather than another.

If this is indeed the case, it raises interesting questions about the minds of people who know more than one language. The possibilities might be:

1. L2 concepts are not acquired. The L2 user acquires the language, not the conceptual system, and effectively uses L1 concepts with the L2. For instance, the Italian blu corresponds to a darker shade than the English blue. An English-Italian bilingual might therefore talk about blue and blu and always refer to the (English) lighter shade.

2. The two sets of concepts exist in separate compartments. The L2 user effectively thought-switches between the two concept-systems when appropriate. The English-Italian bilingual might think about a darker shade of blue when speaking Italian, and about a lighter shade when speaking English.

3. The two sets of concepts are integrated to some extent. The L2 user has partially overlapping concept-systems. An English-Italian bilingual might think of an intermediate shade in between the English dark one and the Italian light one when speaking both languages.

4. A new conceptual system has been created. The L2 user thinks neither in the same way as the native speaker of the first language nor in that of the native speaker of the second language but in a distinctive way that differs from both. An English-Italian bilingual might think of a new shade, perhaps closer to violet or to green, than any of the shades monolingual English and Italian speakers call blue or blu.

These four possibilities represent different points on the integrative continuum for relating the languages in the L2 user's mind (Cook, 2003), which have been substantiated for syntax, the lexicon, and phonology. The novelty is applying the approach to the domain of concepts rather than to language itself.

Inevitably the discussion of the relationship between language and cognition is bedevilled by pitfalls. One is the extent to which language can be divorced from cognition. Chomsky for instance totally separates language from cognition, which consists of a set of innate universal concepts "essentially available prior to experience" (Chomsky, 1991, p. 29). In this model the linguistic and conceptual systems are partitioned from each other and do not contribute to each other's development. Another difficulty is the relationship between the acquisition of concepts and their existence in the mind. Language and concept might be tied together in the child's development, as Piagetians have always claimed, but separated in language use (Sinclair-de-Zwart, 1967). The use of concepts once acquired has been clarified by the useful distinction made by Slobin (1987; 2003) between thinking and thinking for speaking.

One approach has then been to assume that, as far as possible, concepts should be studied in a "pure" form; as Boroditsky et al. (2003, p. 62) put it, "Does thinking for speaking a particular language have an effect on how people think when not thinking for speaking that language?" The aim is to minimize the language element so that concepts can be studied independently. However, it may be virtually impossible to eliminate language altogether since any experimental task is always set in a situational context involving language and the instructions for carrying it out have to be conveyed through a particular language. Obviously, the very idea …