Introduction

The linguistic relativity hypothesis, that is, the possibility that language can influence thought, has constituted an important part of Western intellectual and philosophical discussions over the centuries (see Gumperz & Levinson, 1996; Lucy, 1992a, for some historical background on the hypothesis). The hypothesis, however, is best associated with the work of Benjamin Whorf (1956), whose ideas can be traced back to Sapir and Boas, as well as to the German linguist and philosopher von Humboldt. Even though the hypothesis fell into disfavor in the 1960s with the predominant view in linguistics of language as separate from cognition, and the emphasis in cognitive psychology on the universality of human conceptual structures, the last decade and a half has witnessed a renewed interest in this important issue, as evidenced by the methodologically rigorous empirical work developed by Lucy (1992a, 1992b) and the influential volumes edited by Gumperz and Levinson (1996) and Gentner and Goldin-Meadow (2003).

One important outcome of this renewed interest in the linguistic relativity hypothesis has been Slobin's (1996a) thinking-for-speaking hypothesis, which focuses on the influence of language on the kind of thinking that goes on, online, while we are using the language. According to this hypothesis, the language that we speak directs our attention, while speaking, to particular ways of filtering our experiences of the world. In other words, our experiences of the world are filtered into verbalized events not only through the choice of the individual speaker's perspective, but critically through the particular set of options provided by the language that we speak as well. Thinking for speaking thus entails 'picking those characteristics of objects and events that (a) fit some conceptualization of the event, and (b) are readily encodable in the language' (Slobin, 1996a: 76).