Cognitive approaches to grammar
Introduction
The term grammar is polysemous. In descriptive approaches, grammar is usually taken to include both syntax (i.e. word order patterns) and (inflectional) morphology (see e.g. Quirk et al. 1985: 12). Within the generative tradition, grammar is viewed as the set of principles by which a language works (see Radford 1997: 2–4 for discussion). This definition is based on the conceptualization (or construal) of language as a system made up of unpredictable forms (i.e. the lexicon) and productive rules (i.e. syntax). Syntactic rules specify how lexical items can be combined in a principled and regular way. Cognitive linguistics does not accept the sharp distinction made in generative grammar between syntax and the lexicon. In fact, it does not subscribe to the existence of generative rules either. Rather, it contends that syntax and the lexicon form a continuum of constructions ranging from very specific elements (e.g. cat, kick the bucket) to increasingly more general patterns (e.g. noun, transitive construction). General patterns (or schemas) capture the commonality of various specific instantiations (e.g. kick the bucket and stroke the cat are both instantiations of the transitive construction) and thus serve the same purpose as rules in generative grammars, i.e. they account for our ability to combine linguistic elements in a regular way. Despite the agreement on the non-derivational (or monostratal) nature of grammar– since rules are not posited, constructions are not taken to be derived by manipulating underlying forms– cognitive linguists do not always view grammar identically, as will be shown below.
The present overview begins by presenting some of the core notions of Cognitive Grammar (see Langacker 1987, 1990, 1991, 1999), which offers the clearest definition of what is meant by grammar within a cognitive theory of language. It should become apparent from the discussion that grammar can (and will) be used in two main senses: a narrow sense, corresponding to the descriptive definition (i.e. syntax plus morphology), and a broad sense, as in generative and cognitive approaches. In the latter sense, grammar refers generally to how language works and is not restricted to syntax and morphology (so one can speak of a Cognitive Grammar approach to English phonology, for example). The reader should also be aware that a similar distinction applies to the notion of construction. In a narrow (or traditional) sense, a construction refers to the combination of syntactic phrases but, in a broad (or cognitive) sense, construction is a synonym of (any kind of) linguistic unit.
The title (and hence the topic) refers mainly to grammar in the narrow sense. That is, my primary concern will be problems which are part and parcel of traditional grammar, e.g. the definition of word classes and grammatical functions, noun phrases and change constructions (i.e. caused motion constructions and resultative constructions). Nevertheless, in order to deal with such problems in a principled way, I will of course have to discuss cognitive approaches to grammar in the broad sense, i.e. grammar as a theory of language. But I will do so only insofar as they are relevant to the problems discussed here.
After having examined Cognitive Grammar, I will illustrate and discuss the basic tenets of Goldberg’s Construction Grammar (see Goldberg 1995). We will summarize the fundamental assumptions behind Croft’s Radical Construction Grammar (see Croft 2001). We will briefly review how grammar (in the traditional sense) is handled in Fauconnier and Turner’s Blending Theory (see Fauconnier and Turner 2002). We will draw the main conclusions. In particular, I will point out that, despite some (important) differences, a coherent picture emerges from the various cognitive approaches examined. The importance of the various theories lies in the fact that they tend to highlight different (although related) facets of the shared conceptualization of language as a taxonomic and diffuse network.
