Thinking Through Language*PAUL BLOOM AND FRANK C. KEIL1. IntroductionWhat would it be like to have never learned English, but instead only to knowHopi, Mandarin Chinese, or American Sign Language? Would that changethe way you think? Imagine entirely losing your language, as the result ofstroke or trauma. You are aphasic, unable to speak or listen, read or write.What would your thoughts now be like? As the most extreme case, imaginehaving been raised without any language at all, as a wild child. What—ifanything—would it be like to be such a person? Could you be smart; couldyou reminisce about the past, plan the future?There is a common sense set of answers to these questions, one that rep-resents the mainstream in many circles of cognitive science (see Pinker, 1994for a lucid exposition). Under this view, the language you speak does notaffect how you think. Rich, powerful and abstract cognition can take placewithin minds that, due to injury or deprivation, have no natural language.Even babies know about the kinds and individuals that occupy their world.They just don’t know their names. Before being exposed to words in alanguage such as English, all humans possess the concepts that these wordscorrespond to, as part of what Jerry Fodor (1975) calls ‘mentalese’ or ‘a langu-age of thought’. Under this view, as Fodor puts it, all language learning isactually second language learning—when a child learns the vocabulary ofEnglish, all that happens is that the child learns the mappings from the Englishwords onto the symbols of this prior language of thought.There is another perspective, however, one that is also rooted in commonsense, and which is popular across many disciplines. Many linguists and anthro-pologists claim that the language one learns has a profound influence on how*This is the fourth in a series of Millennial pieces that have been specially commissioned byMind & Language to appear in this volume and the next. Each contributor has been allowedchoice as to topic and approach, being asked only to present a personal view of issues that heor she sees as being important in the present context of cognitive studies.Some of the article is a modified version of Bloom (2000, Ch. 10). Parts of this paper benefitedfrom support provided by NIH Grant R01-HD23922 to Frank Keil.Address for correspondence: Department of Psychology, P.O. Box 208205, Yale University,New Haven, CT 06520, USA.Email: [email protected] or [email protected] & Language, Vol. 16 No. 4 September 2001, pp. 351–367. Blackwell Publishers Ltd. 2001, 108 Cowley Road, Oxford, OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA.352 P. Bloom and F.C. Keilone thinks. Many developmental psychologists have been struck by thecorrelation between language development and cognitive development—a12-month-old has few words and a limited mental life; a 24-month-old has manywords and a much richer mental life—and see this as showing that languagedevelopment has a profound influence on cognitive development. And manyphilosophers reject the idea that thought, or at least rich and abstract thought,can exist without language. Indeed, it is often claimed that the unique cogni-tive powers of our species are largely by-products of our evolved communicat-ive abilities (see Bloom, 1998 for review and discussion). From this perspective,speakers of English think differently from speakers of Hopi, and people whohave no language will inevitably have an impaired mental life.There are many versions of the language-affects-thought claim. Oneversion, most often associated with Whorf (1956) and Sapir (1921), is thatdifferences between languages, such as between English and Hopi, lead todifferences in thought. Another version, sometimes attributed to Vygotsky(1962), is that cognition is shaped by properties that all languages share. Fromthis second perspective, the interesting contrast isn’t between speakers ofEnglish versus speakers of Hopi; it is between speakers of any language versusthose people or animals who have no language at all.1These two views are not independent. One can sensibly claim that thereexist language-general effects without believing in linguistic-specific effects—universal features of language shape human thought, but differences are irrel-evant. For instance, if one were to hold the view that the acquisition of basicsyntactic structure (phrases and sentences) has a profound effect on how peoplethink about the world, this would be a language-general theory but not alanguage-specific theory—since all languages have this basic syntactic structure.On the other hand, it would be perverse to claim that differences betweenlanguages have an effect on thought, but to deny that there is also some moregeneral effect of knowing a language as opposed to not knowing a language.Even putting aside the more subtle universal properties posited by linguists,nobody would doubt that all languages have words and that all languages havesyntax. And it is words and syntax that are most often argued to have effectson thought.In sum, we see three tenable positions, each of which is very much in playin the cognitive science community. One can believe in language-generaleffects. One can believe in both language-general effects and language-specificeffects. And, of course, one can believe that neither of these effects exists.A second sort of distinction between different theories of how language1There is also the view, defended by Bickerton (1995), that it is not exposure to a naturallanguage such as English that gives rise to cognitive effects—it is the mere possession of alanguage faculty. This view merits discussion, but for reasons of space, we will restrict our-selves to the more standard proposal that exposure to a natural language is a catalyst tohuman thought. Blackwell Publishers Ltd. 2001Thinking Through Language 353affects thought concerns the aspect or aspects of language that are said tomatter. The most obvious cut is between words and syntax. Some scholarsargue that the specific words that a language has determines how our mindsbreak reality into different chunks; others propose that our thoughts coalesceinto larger complexes through the vehicle of syntax. But this contrast isoversimplified. Morphology lies at the interface between words and syntax,and can move from the level of the word to the level of the sentence, especiallywhen polysynthetic languages are considered. Semantics principles of