TRENDS in Cognitive Sciences Vol.5 No.3 March 2001 http://tics.trends.com 1364-6613/01/$ – see front matter © 2001 Elsevier Science Ltd. All rights reserved. PII: S1364-6613(00)01594-1 102 Opinion Opinion ‘Twas brillig and the slithy toves/Did gyre and gimble in the wabe;/All mimsy were the borogoves,/And the mome raths outgrabe’ 1 . The nouns, verbs and adjectives in this famous piece of verse are all inventions, products of Lewis Carroll’s febrile imagination. And yet as readers we have a sense – imprecise, perhaps, but real nonetheless – that we know what is going on in this poem. Take the first two lines. It is plain that two or more creatures of some kind (the slithy toves) are engaged in a couple of different activities (gyring and gimbling), while remaining in a fixed location (in the wabe). Why is Carroll’s prose not utter gibberish? How are we able to extract meaning from it? The answer is that meaning can be inferred from the syntax of these lines, independent of the words that appear in those structures. The past 15 years have seen a blossoming of interest in the semantics of syntactic structures. This interest has taken two related forms: among linguists, as an intensified examination of the role that syntactic structures play in the composition of sentence-level meaning; and among psycholinguists, as a proposal for how children might overcome some of the serious problems associated with word learning. This article aims to review these two lines of work and show how they relate. Although we are not the first to note this connection, we believe this article offers the first synthesis of this work grounded in both linguistics and psycholinguistics. What are the semantics of syntactic structures (TSOSS)? By ‘syntactic structures’, we mean large units of syntax, in particular noun phrases (NPs) and verb phrases (VPs). By ‘semantics’ we mean abstract notions such as objecthood, substancehood, causation, motion and mental activity. TSOSS are carried independently of the open-class content words (nouns, verbs and adjectives) that inhabit these structures (although certain closed-class words such as prepositions and the plural marker ‘s’ are also integral to TSOSS). For now, we offer some concrete examples of these structures and their semantics, deferring more-detailed discussion until the next section. Mass nouns and count nouns, for example, characteristically appear in different NPs. The mass noun milk generally appears as a bare noun, as in ‘Mary bought milk’. The count noun glass, by contrast, typically appears with a determiner of some kind, as in ‘Mary fetched three glasses’. Critically, count nouns and mass nouns are not prohibited from syntactic cross-dressing; when such cross-dressing occurs, however, the resulting semantic interpretation is a function of the dress, not of the wearer. Compare ‘Mary bought glass’ with ‘Mary fetched three milks’. Here, the structure of the NP acts as the engine of construal, guiding one to imagine either a non-individuated substance (milk, glass) or a countable entity (three glasses, three milks). Much as NPs can guide the construal of objects, VPs can also guide the construal of events. Consider the effects of placing the verb kick in a variety of syntactic environments. In the sentence ‘Susan kicked the ball’, the verb denotes an event involving contact or causation; in ‘Susan kicked the ball to Bob’, it denotes an event of transfer; in ‘Susan kicked her way out of the locked closet’, it denotes an event of motion by kicking. Where do these different construals of the kicking event come from? One answer is that each use of these uses of kick represents a different sense of the verb. Another answer – the one that we pursue in this article – is that the VP itself carries semantic content, and it is these semantics that effect the appropriate construal. TSOSS and linguistic theory Modern notions of syntactic structure date back to Noam Chomsky’s early work 2 . Within the framework of his Aspects of the Theory of Syntax 3 , transformations served to map deep structures generated by the rules in the base of the grammar to surface structures that could then be pronounced by the phonetic–phonological system. To take an example, the basic structure ‘Elvis shot the television’ served as the deep structure for both the simple active sentence ‘Elvis shot the television’ and the passive sentence ‘The television was shot by Elvis’. According to Chomsky, semantic interpretation took place on the configuration of ‘deep structure’. The subject of a verb was interpreted as the ‘doer’ (later called the Agent), while its direct object was interpreted as the ‘done-to’ (later called the Patient). Transformations re-ordered elements and made essential structural modifications (for example, adding the preposition by The semantics of syntactic structures Edward Kako and Laura Wagner Over the past 15 years, both linguists and psycholinguists have shown a growing interest in the idea that syntactic structures can carry meanings that are independent of the meanings of specific words. This article considers how this idea relates to traditional notions of compositionality in generative grammar, and examines tw o modern theories that, although based on different starting assumptions, both readily allow syntactic structures to bear independent meaning. We review work from psycholinguistics suggesting that observation alone is often insufficient to support the efficient learning of word meanings, and that some of the ‘slack’ left by observation can be picked up by the semantics of the syntactic structures in which words appear. We argue that this convergence between linguistic theory and psycholinguistic experimentation should be no surprise, because language must be learnable. Edward Kako* Dept of Psychology, Swarthmore College, 500 College Avenue, Swarthmore, PA 19081-1390, USA. * e-mail: ekako1@ swarthmore.edu Laura Wagner Dept of Psychology, New York University, 6 Washington Place, New York, NY 10003, USA.