A Reply to the Critiques of 'Grammar is Grammar and Usage is Usage'

1. Introductory remark

I would like to begin my reply to the three critiques of ‘Grammar is grammar and usage is usage’ (henceforth GGUU) by thanking Clark, Meyer and Tao, and Laury and Ono. I am delighted that they took it seriously enough to devote the time and trouble to produce interesting challenges to the arguments therein. Not only have the replies corrected some factual errors that made it through the reviewing process into publication, but they have led me to clarify my own thinking with respect to a number of issues. I am especially grateful to the five linguists, given that a version of GGUU will be incorporated into Possible and probable languages: A generative perspective on linguistic typology (Newmeyer 2005). It will be a better book, thanks to their input.

The three replies fit into two broad categories, one involving the notion of stochastic grammar and the related issue of the use of corpora in linguistic research, the other involving the question of functional explanation in linguistics. I take these up in turn.

2. Stochastic grammar and the use of corpora: replies to clark and to meyer and tao

GGUU devotes a full section (§10) to arguing against stochastic grammars as models of individual linguistic competence, in particular against those stochastic models that encode variation within a broadly defined speech community. To this point, I observe that ‘No input data that an individual did not experience can be relevant to the nature of his or her grammar’ (p. 696). Clark (henceforth C) defends stochastic grammar, arguing that ‘mental grammar accommodates and generates variation, and includes a quantitative, noncategorical, and nondeterministic component’ (p. 207). As he notes, such a position entails ‘mak[ing] the idealization that speakers share the same mental grammar’ (p. 208).

C’s remarks echo what is perhaps the most famous (some would say ‘notorious’) passage in any of Chomsky’s writings: ‘Linguistic theory is concerned primarily with an ideal speaker-listener, in a completely homogeneous speech-community . . . ’ (Chomsky 1965:3). That passage evoked outrage among the growing community of grammarians interested in describing and explaining linguistic variation and led, among other things, to the development of the ‘variable rules’ of sociolinguistics. C identifies his position as being within the tradition represented by the variable rules approach, but in an important respect he is mistaken.¹ Cedergren & Sankoff 1974 (one of the foundational papers on variable rules) was explicit (see p. 335) that the numerical quantities associated with the features in the environment of a variable rule are not discrete probabilities ‘in the head of the speaker’. But for C and other modern advocates of stochastic grammar, the probabilities are very much in the head of the speaker. Indeed, C’s stated goal in his reply is to defend ‘models of mental grammars that incorporate probabilistic information’ (p. 207).²

C recognizes that ‘one possible objection [to stochastic grammar] is that by mixing data from different individuals together in a large data set, evidence relevant to the investigation of the mental grammar of particular individuals is potentially obscured’ (p. 208). But he then dismisses the importance of the objection. Again, sociolinguists in the 1970s were well aware of the problem, which is why many considered and then rejected the formal representation of variability advocated by C:

One never seems to find the ‘ideal speaker-hearer [sic] in the perfectly homogeneous speech community’ . . . Individuals may and do act linguistically in ways which are not reflected by group data. Berdan (1973) has shown that for a sample of Los Angeles school children, individual behavior and group behavior do not match. Similarly, Levine and Crockett (1967) showed that the high status group of white speakers is disproportionately made up of extremely high and extremely low users of post-vocalic r. When the means for groups are considered, the higher PVR users balance the low PVR users and the mean use of PVR does not differ significantly from other groups.

(Anshen 1975:6)

I had never heard a ‘positive anymore’ sentence like We go there anymore (in the meaning...