Cognitive Grammar (Oxford Textbooks in Linguistics) [ペーパーバック]

As a theory of linguistics, cognitive grammar is usually presented as an alternative to the generative theory of grammar and this book is no exception. Indeed, who would have thought that idioms, which in generative theories are considered to be "exceptions" to the general rules of grammar, might in fact form the basis of a correct theory of grammar. Cognitive grammar asserts that this is the case, and the author explains how in this highly detailed and fascinating book. One has to wait until nearly the end of the book to encounter this discussion of idiomatic structure, but along the way one will encounter most of the basic theory behind cognitive grammar in a way that makes it much easier to understand than many of the standard works on the subject (for example the works of R.W. Langacker).

As in any subject that one is attempting to learn, one can expect that a fair amount of cognitive perspiration will be required to gain and understanding of it, even if one does not intend to reach total mastery. This reviewer approached this book as one who is interested in computational linguistics and its connection with artificial intelligence. One encounters examples of this connection throughout the book. One that stands out in this regard is the author's discussion of how conceptual structures need to `accommodate' each other when they combine, so as to ensure their coherent participation in the resulting complex conceptualization. This blending of conceptual structures has important ramifications in resolving the problem of `domain specificity' in artificial intelligence.

Although many readers may think so, this work should not be viewed as an anti-Chomsky polemic, for it offers a constructive viewpoint of cognitive grammar and explains rather than just expounds. If one is encountering linguistic theory by first examining the theory of cognitive grammar, then obviously when one not have any biases towards the generative theory, but one will also not be able to appreciate various discussions in the book that compare the two theories. One strategy for such a reader might be to read a monograph on the generative theory in conjunction with this one.

The author defines language as a "structured inventory" of linguistic units. The designation as "structured" refers to the complexity in which the units can be interrelated, and the units can be phonological, semantic, and symbolic. Symbolic structures are viewed as relations between the phonological and semantic structures. For some reason the author refers to cognitive grammar as being "minimalist" or "sparse" because of its assumption that only these three structures are needed to build human language, but a priori there is no reason to believe that these structures are simple or that they cannot be reduced to even more elementary units. Research in cognitive neuroscience may shed light on this, but in the book the author expresses caution in relying on a theory of what he calls `neurocognitive linguistics', the main reason being that neuroscientists study more of the generalities of language structure and omit the details that linguists are interested in.

But it is the symbolic unit that the author spends much of his time articulating on, with the first part of the book discussing the historical origins of the "symbolic" thesis, the justification for the labeling of this linguistic theory as being "cognitive," and the contrast between cognitive grammar and the formalist and behavioral approaches. Philosophical musings are present but luckily are kept at abeyance. This makes the book more readable and results in an economy of thought that would otherwise make the learning of cognitive grammar very awkward. The differences between the formalist/generative schools and cognitive grammar are made crystal clear in these beginning discussions, for it is asserted that the knowledge of a language is based on the knowledge of usage, and the acquisition of language is done by linguistic experience. The immediately implies that the acquisition is a dynamic process, and that linguistic systems do not have to be identical between speakers. The author spends an entire chapter answering possible objections to the `symbolic thesis.' But the dynamism or "adjustment" of language knowledge almost has a Bayesian flair to it, for the author speaks of it as the updating of a hypothesis when a particular communication between two parties fails (an interesting sidebar in the book along these lines is a possible connection with autism). And most interestingly, cognitive grammar admits without hesitation that linguistic expressions only occur in `mental space', i.e. they only refer to things as represented in and by the mind. The author discusses the interesting consequences of this assumption.

The most interesting discussion in the book concerns the combination of semantic units, or what the author refers to as `syntagmatic relations.' The notion of the `disposition to combine' of a phonological or semantic unit is a fascinating one in that it illustrates to what degree such a unit needs to be combined with others in order for it to be meaningful. The author thus refers to the degree of `autonomy' or `dependence' of a unit, and `valence' is a measure of the combinatorial possibilities of a unit. One could generalize the idea of the valence of a conceptual unit to be a measure of how one concept, or collection of concepts is "entangled" with each member of the collection, in the sense that each unit requires the others for its meaningful elaboration.