Speakers of Native American languages are often heard to say, "When I talk Indian, I think differently". Linguists who work with these languages are likely to be sympathetic to this statement, and to have a gut feeling that it is correct. At the same time, having puzzled over the complex issues surrounding the relation of language and thought, they may find it difficult to decide whether the statement has real substance, or whether people who say this are being misled by superficial differences between their native languages and English. On the one hand, in the current climate that favors a possibly exaggerated view of the universality of both language and thought (e.g., Pinker 1994 and Jackendoff 1994), it is all too easy to conclude that the statement does not really capture anything very profound. On the other hand, those who would like to take it more seriously may be encouraged by the fact that the idea refuses to go away in spite of repeated attacks, as well as by the recent resurgence of interest from a more positive point of view, represented, for example, by Lucy (1992a, 1992b), Lee (1996), Gumperz and Levinson (1996), and the present volume. My plan here is to begin by taking a broad look at the role of thought in language, and to suggest that something analogous to the constrained diversity that is obviously present in the linguistic organization of sounds might be expected in the linguistic organization of thoughts as well. I will then isolate several distinct components of thought, emphasizing that total thought is more encompassing than its organization in language. I will cite evidence from repeated verbalizations of the same experience showing unconformities between thoughts and the language used to express them. I will then turn to translation for evidence that each language organizes thoughts in its own way. I will suggest that ideas of events, states, and referents constitute a minimal skeleton of thought that is susceptible to transfer across languages, but that categorizations, inflections, and the constructions by which thoughts are combined show qualities that are unique to particular languages. In the end I hope to explain why it is possible for well-meaning scholars to hold opposite opinions on this major issue, some investigators being impressed by cross-language resemblances in the skeletal elements of thought organization, others focusing more on the diverse ways those elements are fleshed out.
