Poetry, stylistics and imaginary grammars

Hendricks (I969) has one substantive criticism to make of my paper (Thorne, I965), which can be summarized as follows: 'Thorne's paper would have been better if, instead of using the grammatical model proposed in Syntactic Structures, he had used that proposed in Aspects of the Theory of Syntax.' This is very true. Unfortunately, I wrote my paper two years before the publication of Aspects of the Theory of Syntax. Nevertheless, the point is an important (if obvious) one, and one that I would want to emphasize. The most important limitations of our analyses of literary language are likely to be the result of limitations in our understanding of natural language structure. An important principle follows from this. Stylistic analysis, if it is to be of any interest at all, must be undertaken in the context of a well-defined linguistic theory. Of course, everybody now knows that the grammar outlined in Syntactic Structures is an inadequate model for natural language. It is hardly surprising, therefore, that my attempt to use it for the analysis of English lyric poetry should now be seen to be deficient in certain respects. But because it was clearly related to a well-defined model it is quite easy (as even Hendricks has to admit) to reformulate this approach in terms of a more sophisticated grammar. Hendricks's proposals, on the other hand, seem to me quite vacuous because they relate to linguistic theories which are not so much 'ad hoc' (as he wishes to term them) as non-existent. Since Hendricks does not give a consistent account of my proposals I shall begin by giving a brief summary of them. Poems tend to contain many sentences which, in the technical sense of the term, are not 'well-formed'. (Presumably this has something to do with the declared inability of many people to understand