Cornips & Corrigan's Syntax and variation: Reconciling the biological and the social presupposes that we have a biological and a social dimension to linguistics, that the biological dimension is represented by generative syntax, and that the two dimensions require reconciliation. Cornips & Corrigan (C&C) consider this reconciliation possibly to be “the initial phase in the creation of a mature scientific community” (p. 2). These presuppositions should be questioned. In the first place, it is questionable whether linguistics really has a biological dimension, at least if generative syntax is that dimension. Second, it is far from clear that reconciliation is necessary for linguistics to be considered a mature science.

It actually turns out that at least two other academic disciplines have biological and social dimensions. These are zoology, which has molecular zoology and ethology, and anthropology, which has physical anthropology and sociocultural anthropology. It is fairly clear that neither zoology nor anthropology sees these subfields as the major contrast that many linguists, including C&C, see in the biological and social for linguistics. Nevertheless, I believe the parallels are not entirely spurious. Molecular zoology examines organisms at the cellular level, trying to understand development and evolution within the physical structure of organisms. Ethology is the study of animal populations and their behaviors. The methodologies of the two kinds of research are quite different, one being largely a laboratory science and the other involving fieldwork. Physical anthropology studies the physical structure of humans and other primates. Sociocultural anthropology, as readers of this journal hardly need to be told, is the study of human cultural phenomena around the world. Both disciplines have structural biological branches and social behavioral ones, but neither seems to have the “integrated theory” of both that C&C expect from a mature science (7). In fact, both disciplines do have subfields that combine the biological and the social – neuroetholgy in zoology and sociobiological anthropology in anthropology – but apparently in neither discipline do these subfields seek to replace the subfields that they combine.

These comparisons with other fields should not be pushed too far. The situation in linguistics, as presented in Syntactic variation (SV), is strikingly different in at least one crucial way; that is, the “biological” part of linguistics is not nearly biological enough, if zoology and anthropology are to be taken as models. Generative syntax does not examine actual biological structures because the techniques for doing so noninvasively have only just begun to be available, but rather constructs an abstract formal model to which the biological structures will presumably have to conform, albeit probably very indirectly. The infeasibility of studying language in the brain directly is Chomsky's motivation for referring to the “mind/brain.” Since language in the brain is largely inaccessible, we examine properties of the mind, expecting the relationship to what is in the brain to emerge in due time. Viewed this way, linguistics will not be fully comparable to zoology and anthropology until mind research gives way to brain research (if it ever does).

Meanwhile, it is important to note that generative syntax is a theory of sentences and their substructures. These structures are independent of any use to which they may be put in actual human interaction; they are not utterances. Jenny Cheshire's contribution inadvertently provides an example of what I mean. She cites the following example from Crystal (1976:158):

Cheshire (or perhaps Crystal) suggests that the unbracketed material goes together as one syntactic unit and the material in square brackets is another, even though they are interleaved. Suppose this analysis is correct. This is not a problem for a theory of sentences. The sentences, eliminating the interleaving and removing the self-editing, tag, and hesitation phenomena that I have placed in slanted brackets, are

and

That these two sentences are actually uttered interleaved and the utterance is interspersed with other subsentential constructions is of no concern to the formal syntactician. The editorial and tag material are fully or partially formed grammatical constructions of less than sentence length. There is no reason why a speaker cannot utter less than a sentence, part of a structure, or something without syntactic structure (the hesitation form erm) – even in the middle of a complete structure. A theory of abstract sentences is not required to give an account of all the properties of utterances. It is not a theory of utterances. As long as sociolinguists work with utterances and generative syntacticians work with sentences, as each must do to be consistent with their conception of what is most worth studying in language, it is not to be expected that the two will be combined in a unified theory. A comparison with zoology and anthropology indicates that it is not necessary, and, in my view, it is not particularly desirable. The study of syntax and the study of linguistic variation are complementary endeavors, not opposing pieces that need to be reconciled into one overarching theory.

Even if this is so, there is one critical issue that should be faced. Obviously, variation exists and varying syntactic structures have to come from somewhere. There are two possible sources. One is that different static grammars each make different alternative constructions available. It is up to grammar-external principles to explain which of these grammars a speaker chooses to activate in a given instance. The other is that variation is grammar-internal, and alternatives are produced by a single grammar while it is running, so to speak.

The first of these alternatives is the equivalent of Anthony Kroch's (1989, 1994) often-cited notion of competing grammars. C&C (10) take it that Kroch's proposal implies that individual speakers have a large series of different grammars, even if they are monolingual, and Sjef Barbiers (234) assumes that Kroch's idea entails a potential “explosion of grammars.” These assumptions are the result of an incomplete appreciation of generative syntax as an abstract theory. When a generative syntactician speaks of different grammars, there is no implication that speakers have dozens of nearly identical grammars clogging their brains. Rather, a small difference changes the grammar conceptually, so that one can speak of the same system, with and without the change, as different systems. An analogy might make this clearer. It has been reported that Microsoft Corporation has decided that if a computer owner replaces the central processor in his or her computer, a “new machine” is created. Therefore, the owner is obliged to purchase a new license for the Windows operating system, since each license is valid on only one machine. A computer owner who has replaced the processor, in Microsoft's view, now owns two machines. If you were to look on this power user's workbench, you would not see anything that looks like two machines; you would see one computer that now has the new processor installed, and the old processor lying on the bench beside it. The same is true of “competing grammars,” one, say, with verb seconding and the other without. We speak of two grammars: One has whatever produces verb seconding “plugged in,” and the other is the very same grammar, without a different “plug-in.”

So how can we understand the nature of “competing grammars” precisely? One way is to assume that there is one grammatical system, but there are different items selectable from the lexicon, including functional elements with features that produce relatively dramatic syntactic effects. In the verb seconding case, for example, we can assume that the lexicon contains two versions of C(omplementizer), one of which projects a syntactic feature that must be satisfied by the movement of the verb to C and the other of which lacks that feature. There are two grammars, in a sense, but really only one, with a lexical choice accounting for the difference.

On my reading of the contributions to SV, all but three explicitly or implicitly assume a competing grammars approach, or at least provide analyses that are consistent with it. This includes David Adger & Jennifer Smith, who explicitly reject the “multiple grammars” account (162–63) but adopt an analysis consistent with the version of competing grammars I have just described. One of the exceptions is the contribution by Jenny Cheshire, who calls for generative syntacticians to help develop a grammar of spoken language, without seeming to realize that such a device would be an oxymoron for them, since it would be responsible for utterances. The other two contributors explicitly propose grammar-internal variation. One of these is Alison Henry, who presents numerous examples of apparent I(nternal)-language constraints on the frequency of alternate agreement patterns in Northern Ireland English, and calls for a reinstatement of optional or variable rules in generative grammar. It seems that some of Henry's examples can be understood in ways that would not require optionality, but others stand as serious challenges to the notion that syntax must be categorical, up to lexical selection or parameterization. The other is Sjef Barbiers, who provides an analysis of verbal word order in Dutch that allows for limited optional movement. He himself mentions a way around making his movements to the specifier position of VP optional: allowing the class of verbs to have alternative features, one projecting an E(xtended) P(rojection) P(rinciple) feature on the specifier of VP that forces movement and the other not. He rightly points out that this solution is no more insightful than allowing the movement to be optional, but it utilizes an established mechanism and avoids a serious reconceptualization of the nature of formal syntactic systems.

I have used my allotted space to discuss what I take to be critical issues in understanding syntactic variation but that are not confronted directly in SV. This leaves no room for comment on the individual contributions to a fascinating collection of analyses of variation in the realm of syntax, with a refreshing combination of sociolinguistic and syntactic analysis and methodological and theoretical innovation. C&C have done a great service to the field in providing a high-quality collection of research that combines approaches that are usually self-isolating.