Language, usage and cognition (review)

1. Introduction

There is a long tradition in linguistics and philosophy of analyzing language without reference to usage and experience. This tradition is reflected in Chomsky’s famous division between competence and performance and Saussure’s related distinction between langue and parole, which have influenced linguistic research for many decades. This view of language, however, has been called into question by usage-based linguists who have emphasized the importance of communication, cognition, and processing for the development and organization of grammar. In the usage-based approach, grammar is seen as an emergent system consisting of fluid categories and dynamic constraints that are in principle always changing under the influence of general cognitive and communicative pressures of language use.

Joan Bybee is one of the pioneers of the usage-based approach, which has its roots in functional and cognitive linguistics (e.g. Givón 1979, Hopper 1987, Langacker 1987) and is related to various strands of research in cognitive science (e.g. Bates & MacWhinney 1989, Bod 2003, Clark 1996, Elman et al. 1996, Tomasello 2003). Thus far, B’s work has been mainly concerned with morphology (e.g. Bybee 1985) and phonology (e.g. Bybee 2001); but in the current book the focus is on larger grammatical patterns (i.e. linguistic units that exceed the single lexeme). The book provides an overview of some of B’s earlier research on grammar and grammaticalization and presents some new corpus studies and usage-based analyses from the current literature.

B’s view of grammar rests on central assumptions of construction grammar, which has made important contributions to the usage-based approach (e.g. Goldberg 2006, Langacker 2008). In fact, usage-based linguists have drawn so frequently on concepts of construction grammar that the two approaches are often presented as a unified theory (e.g. Dabrowska 2004, Diessel 2004, Tomasello 2003).¹ In this approach, grammar consists of emergent form-function units, or ‘fluid constructions’ (Goldberg 2006), which are related to each other by probabilistic links that are determined by their similarity and cooccurrence in usage. Since constructions involve the same cross-modal associations of form and meaning as words and morphemes, they are subject to the same cognitive processes as lexical expressions. Thus, many of the usage-based principles that B analyzed in her previous research on morphology and phonology can also be applied to the analysis of larger grammatical patterns, that is, constructions.

The general goal of the book is ‘to derive linguistic structure from the application of domain-general processes’ (1), or, as Lindblom and colleagues (1984:187) put it, ‘derive language from nonlanguage!’. Domain-general processes are cognitive mechanisms that are relevant not only for language but also for other cognitive phenomena such as vision and thought. Drawing on general research in cognitive science, B relates the analysis of language use and change to general processes of cognition.

The discussion in the book is divided into eleven chapters. The introductory chapter presents some central assumptions about the usage-based approach and introduces the domain-general cognitive processes B considers important for the emergence of linguistic structure. The four following chapters are concerned with four of the five cognitive processes presented in the introduction: rich memory (Ch. 2), chunking (Ch. 3), analogy (Ch. 4), and categorization (Ch. 5); but from Ch. 6 on, the structure of the book is difficult to grasp. Some chapters are concerned with particular mechanisms of change (e.g. reanalysis; Ch. 7), others present case studies on English auxiliaries and modals (Chs. 7 and 9), and yet others cover an array of topics that are only loosely related (e.g. Chs. 6 and 11). Since many issues are discussed across several chapters, I summarize the main points of the book in ten theses on the usage-based approach that underlie the entire discussion.

2. Ten theses on the usage-based approach

Thesis 1: Language is grounded in domain-general cognitive processes. This is the most general thesis, providing a foundation for the whole approach. B conceives of...