Towards a more comprehensive usage-based account
Willem B. Hollmann
Structuralists and generativists define word classes distributionally (Palmer 1971, Baker 2003, Aarts 2007), while cognitive linguists take a semantic (Langacker 1987a) or semantic-pragmatic approach (Croft 1991, 2001). Psycholinguistic research, by contrast, has shown that phonological properties also play a role (Kelly 1992, Monaghan et al. 2005). This study reports on a production experiment involving English nonce nouns and verbs. The data confirm the importance of phonology, whilst also suggesting that distributional facts are involved in lexical categorisation. Together with the existing psycholinguistic evidence, the results show that both the generative and cognitive models of word classes are too restricted. However, the usage-based model can accommodate the facts straightforwardly. This was already anticipated by Taylor (2002) but is worked out in more detail here by elaborating on his notion of phonological "subschemas" and by bringing together insights from Croft (1991, 2001) related to discourse propositional act constructions and recent suggestions by Langacker (2008b) concerning "summary scanning" and "sequential scanning".
This chapter sets out to outline an account of word classes that is responsible to evidence from psycholinguistic research. With their focus on distributional criteria, structuralist and generative linguists have ignored evidence concerning the importance of semantic and phonological properties. Cognitive linguists, by contrast, have only emphasised semantics (see Langacker's Cognitive Grammar, e.g. 1987a) or semantics-pragmatics (see Croft's Radical Construction Grammar, e.g. 2001). This study will draw on existing psycholinguistic literature that discusses semantic, distributional, and phonological cues in lexical categorisation, and provide additional empirical evidence concerning phonology and distribution from a questionnaire study involving the production of novel nouns and verbs in English. I will show how the results may be naturally incorporated into a usage-based theory of language. The focus here is on English, but there is no reason to assume that these types of cues do not play a role in other languages. However, the properties themselves will of course be subject to crosslinguistic variation (see also Berman 1988, Cassidy & Kelly 1991: 348).
The chapter is structured as follows. Section 2.1 summarises structuralist-generative work on word classes, with Section 2.2 dealing with Cognitive Grammar (Langacker 1987a, 1987b, 1991, 2000, 2002, 2008a) and Radical Construction Grammar and earlier work by Croft (1990, 1991, 2001). The psycholinguistic literature is surveyed in Section 2.3. Section 3 describes the methodology of the production experiment. Section 4 discusses the conclusions concerning the role of phonological properties. This part of the study is discussed more elaborately by Hollmann (submitted), but the present study goes further than that, by addressing the importance of distributional facts as well (Section 5). This section relies on Croft's (1990, 1991, 2001) suggestion that a crosslinguistically valid characterisation of word classes needs to incorporate Searle's major (1969) propositional act functions. This claim is combined with Langacker's recent (2008b) revision of summary and sequential scanning to yield an interpretation of the propositional acts as gradient phenomena. By operationalising the propositional act functions in terms of actual constructions, I also draw on some aspects of the structuralist-generative view of word classes, in that the constructions in question correspond to some of the traditional distributional criteria. Section 6 wraps up the discussion with some conclusions and a few suggestions for further work in this area.
-  I am grateful to Nikolas B. Gisborne, Ekkehard König, and two anonymous reviewers for their comments on this chapter.