Letters to Language

Language accepts letters from readers that briefly and succinctly respond to or comment upon either material published previously in the journal or issues deemed of importance to the field. The editor reserves the right to edit letters as needed. Brief replies from relevant parties are included as warranted.

Grammar and usage: A response to Gregory R. Guy

August 14, 2006

To the Editor:

In two recent contributions to Language—Grammar is grammar and usage is usage’,79.4.682–707, 2003 (GGUU) and ‘A reply to the critiques of “Grammar is grammar and usage is usage” ’, 81.1.229–36, 2005 (RCG)—I critiqued stochastic approaches to grammar, namely those in which grammatical constructs are linked directly to probabilities of occurrence in discourse. My arguments included the point that probabilistic generalizations derived from a sociologically and dialectally heterogeneous group of speakers are irrelevant to determining the grammatical competence of any particular member of that group; that as far as syntactic variation is concerned, since variants typically differ in meaning, the probabilities are likely to be more a function of the meaning to be conveyed than a characteristic inherent to the structure itself; and that the multiplicity of genres confounds the simplistic treatment that one finds in much of the literature defending stochastic approaches to grammar.

Gregory R. Guy (G), in a letter to Language (81.3.561–63, 2005), has taken issue with my position. Distressingly, from my point of view, he sees my remarks as exhibiting ignorance of the fact that ‘speakers manifestly are sensitive to frequencies and manipulate them systematically’ (G, 561). But in fact, I recognized explicitly that ‘language users and hence their grammars are sensitive to frequency’ (GGUU, 697). I did stress, however, that ‘from the fact that Y is sensitive to X, it does not follow that X is part of the same system that characterizes Y’. Far from rejecting the essentials of the variationist project, I described it in a supplement to the journal The Sciences (‘Benchmarks: 35 years of linguistics’, 36.6.13, 1996) as a major breakthrough of modern linguistics, where I commented that ‘[t] he work of William Labov of the University of Pennsylvania and his colleagues has demonstrated that much of the everyday variability in speech is systematic, showing both social and linguistic regularities’.

I don’t deny that ‘variability in language is highly structured’ (G, 561), nor that ‘individuals are grammatically similar as a function of social proximity’ (G, 562). But I do deny that the grammar of any particular individual can be derived from probabilities gleaned from corpora full of utterances where social proximity among speakers is absent. Surprisingly to me, given his other remarks, G agrees, writing that ‘a corpus drawn from the New York Times does not offer a very close approximation to the linguistic input and personal grammars of Carl Pollard or Ivan Sag’ (G, 562). But that’s just my point. The corpora used in most stochastic approaches to grammar (though perhaps not in the more careful sociolinguistically informed ones) conflate utterances from speakers from a wide variety of regions and social classes.

G provides one example to rebut my ‘hoary argument that perhaps syntax is different from phonology’ (G, 562), namely that in Brazilian Portuguese (BP) subject-verb agreement is disfavored (though not categorically banned) with postposed subjects. While G does not provide a literature reference to track down, my immediate reaction was that he put forward a particularly problematic case in support of his position. In the first place, G ignored my point about syntactic variation and meaning. Since in BP, as in all Romance languages, preverbal and postverbal subjects differ dramatically in their discourse properties (Anthony J. Naro and Sebastião J. Votre, ‘Discourse motivations for linguistic regularities: Verb/subject order in spoken Brazilian Portuguese’, Probus 11.1.75–100, 1999), a complicating factor that G’s approach fails to deal with enters the picture. Second, if the analysis referred to by G is correct, in which postverbal subjects originate in object position, then we have an independent explanation for the agreement facts. Verb-object agreement is crosslinguistically significantly less common than...