Log in / Register
Home arrow Language & Literature arrow Theory and Data in Cognitive Linguistics


In the debate between structuralist-generative theory on the one hand and the cognitive, usage-based approach on the other, word classes represent one of the principal arenas. Members of the former camp criticise notional definitions of categories, although the definitions they criticise are sometimes older ones (from traditional descriptive grammars such as Nesfield 1908), which functional-typological and cognitive linguists do not actually subscribe to either. Instead of semantic criteria, structuralists and generativists emphasise distributional facts. One recent derivative of generative theory, Distributed Morphology, takes a different view, arguing that word classes are merely linguists' generalisations over patterns of usage. I argued in Section 2.1 that this model must be rejected in light of psycho- and neurolinguistic evidence that the categories are somehow real. I also suggested that its proponents' claim that Distributed Morphology is "parsimonious" only works if we disregard the so-called Encyclopedia.

Cognitive, usage-based theories take semantics, and in the case of Croft (1990, 1991, 2001), semantics-pragmatics, to be fundamental. However, the usage-based perspective allows in principle for the possibility that other aspects of usage are also stored as part of the representation of lexical categories.

The psycholinguistic literature has demonstrated in various ways that distributional, semantic, and also phonological properties play a role in lexical categorisation. The main contribution of the present study is that this evidence has been incorporated into a usage-based model of nouns and verbs, and that the database has been expanded by means of a study on production, which had not been attempted by psycholinguists.

In order to arrive at the more comprehensive usage-based view of nouns and verbs proposed here, certain traditional assumptions of and boundaries between theories were broken down.

In both generative and cognitive theories of grammar, phonology often does not play a very prominent role. The evidence, however, suggests that it contributes significantly to our knowledge of nouns and verbs a point anticipated by Taylor (2002: 181-185) but worked out here in more detail and using more original data.

Taylor (2002: 563) also seems to acknowledge the role of distributional data, although he does not draw on any concrete evidence to support his argument. My starting point in investigating distribution was Croft's (1990, 1991, 2001) crosslinguistically valid account of word classes, combined with recent work by Langacker (2008b) on summary and sequential scanning. Langacker's revised account of the scanning modes as gradient phenomena was seen to dovetail with earlier hints by Croft (1991) that reference and predication are matters of degree. But in operationalising these propositional act functions, which constitute the pragmatic dimension of Croft's theory, specific syntactic constructions were analysed constructions which are part and parcel of generative definitions of nouns and verbs.

I analysed the [Det X] construction as less clearly referential than subject or object position (see Section 3.3.2). To the extent that this is correct, the data from this study suggest that co-occurrence with a determiner may be a more salient distributional property of nouns than their occurrence in the most highly referential constructions, viz. those in which they appear as subject or object. This runs counter to what cognitive linguistic theory might lead one to expect.

This issue clearly merits further research, as do a number of other points raised in this relatively small-scale study, e.g. the presence vs. absence of final obstruents as a possible phonological distinction between English nouns and verbs, the degree of salience of the noun and verb super-schemas as compared to lower-level sub-schemas, the exact relation between syntax and different degrees of referentiality and predication, and of course the representation of these and other word classes in languages other than English. But if this study has contributed something beyond a more complete model of nouns and verbs, it should be the notion that such further research is not well served by rigid theoretical assumptions or distinctions, but should be driven first and foremost by empirical evidence.

Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >
Business & Finance
Computer Science
Language & Literature
Political science