allelism, in much the way that a tailor s swatch of cloth means the colour or material which comprises it.) Style is thus a kind of meaning, holding of a text only as the content of a thought about the text. For example, parallelism holds within a text to the extent that a reader is justified in formulating the thought parallelism holds within this text laquo ;, with the justifications drawn from various stereotyped deductions ( if the first and second lines have the same sequence of word classes, then there is parallelism in the text laquo ;, etc). Or a text is in a specific genre to the extent that we are justified by the text in formulating that assumption about it. Linguistic form offers one of a number of different and potentially competing sources of evidence from which the presence of a style is inferred, and this is the relation - in this theoretical approach, much weakened - between form and style. Style can thus be indeterminate, ambiguous, metaphorical, ironic, strongly implied, weakly implied, and so on - having all the characteristics of a meaning, because style is a meaning. If this is true, it has a consequence which helps us resolve some of the problems for the teaching of Stylistics.The key problem in Stylistics is to work out the causal relation between style and effect, where effect 'includes various cognitive effects such as meanings, emotions, beliefs, etc. My proposal is that style is itself an effect; hence rather than mediating between two quite different kinds of thing (style vs. effect) we are really looking at the relation between effects, with the distinction between style and effect no longer clearly defined. This means that the theory of how style causes effect is now a theory of how thoughts are connected, which comes under the theory of Pragmatics. This suggests a route out of the problem of Stylistics which has been chosen by a number of authors: to assume that Stylistics basically falls under the theory of Pragmatics, and to start from here in the teaching of Stylistics all linguistic levels (eg lexicology , syntax, text linguistics, and intonation), stylisticians analyze both the style of specific texts and stylistic variation across texts. These texts can be literary or nonliterary in nature. Generally speaking, style may be regarded as choice of linguistic means; as deviation from a norm; as recurrence of linguistic forms; and as comparison. Considering style as choice, there are a multitude of stylistic factors that lead the language user to prefer certain linguistic forms to others. These factors can be grouped into two categories: user-bound factors and factors referring to the situation where the language is being used. User-bound factors include, among others, the speaker s or writer s age; gender; idiosyncratic preferences; and regional and social background. Situation-bound stylistic factors depend on the given communication situation, such as medium (spoken vs. written); participation in discourse (monologue vs. dialogue); attitude (level of formality); and field of discourse (eg technical vs. nontechnical fields). With the caveat that such stylistic factors work simultaneously and influence each other, the effect of one, and only one, stylistic factor on language use provides a hypothetical one-dimensional variety. Drawing on this methodological abstraction, stylistic research has identified many correlations between specific stylistic factors and language use. For example, noun phrases tend to be more complex in written than in spoken language in many speech communities, and passive voice occurs much more frequently in technical fields of discourse than in nontechnical ones. Style, as deviation from a norm, is a concept that is used traditionally in literary stylistics, regarding literary language as more deviant than nonliterary language use. This not only pertains to formal structures such as metrics and rhyme in poems but to unusual linguistic preferences in general, which an author s poetic license allows. Dylan Thomas s poetry, for example, is characterized by word combinations that are semantically incompatible at first sight and, thus, clearly deviate from what is perceived as normal (eg a grief ago, once below a time). What actually constitutes the norm is not always explicit in literary stylistics, since this would presuppose the analysis of a large collection of nonliterary texts. However, in the case of authorship identification, statistical approaches were pursued at a relatively early stage. For example, by counting specific lexical features in the political letters written by an anonymous Junius in the 1770s and comparing them with a large collection of texts from the same period, and with samples taken from other possible contemporary authors, the Swedish linguist Elleg? Rd could identify, in the 1960s, the most likelyauthor of those letters. The concept of style as recurrence of linguistic forms is clo...
