![]() Secondly, they sort them on the basis of the “semantic relationships between the supporting tokens” (p.97). Firstly, they sort all possible pairs of the relevant tokens (i.e., in the sequence A 1B 1A 2B 2, they sort A 1B 1, A 1B 2, A 1A 2, B 1B 2, B 1A 1, and B 2A 2) for whether or not their lemmata are identical. Then they filter the candidates by augmenting “the Dubremetz features” in two ways. use broader search criteria than Dubremetz and Nivre, looking only for reverse repetition of part-of-speech, which brings back way more candidates. Their ‘general chiasmus’ includes e.g., utterances like 32 and 33, but they exemplify their expansion of candidates over Dubremetz and Nivre by this utterance (from Schiller), which they can detect but Dubremetz and Nivre cannot: “ Eng ist die Welt/und das Gehirn ist weit ( Narrow is the world,/and the brain is wide”.) This example is a reversal of syntactic structure (the first clause has the structure, NP V cop AP, and the second clause has the structure, AP V cop NP), which is given additional salience by the antithesis of eng and weit. They detect what they call “general chiasmus” rather than antimetabole and report considerable success (improving “the average precision from 17 to 28% and the precision among the top 100 results from 13 to 35%”–p. expand Dubremetz’s work in interesting ways that suggest further approaches to the search for figural collocations. But proximity is a common feature in traditional definitions of many schemes, including chiastic figures (e.g., antimetabole as “Repeating two words in successive phrases, but in reverse order” ). And it may be that she has a similar view of chiastic figures, that proximity is only a methodological convenience not a defining feature certainly it is both reasonable and a long-established practice to rely on proximity in corpus research. (This approach is in contrast with Claus Strommer’s work on epanaphora ). “To keep the problem simple”, she says, “we will consider only epanaphora relying on immediately successive sentences”. But she adopts high proximity methodologically. Dubremetz is clearly aware that high proximity is not criterial for figuration–as in her discussion of epanaphora, the figure in which words repeat at the beginning of clauses or phrases, where she acknowledges that it can occur in non-successive clauses or phrases, with such units as paragraphs or verses intervening. The matter of proximity is more complicated. See, for instance, their talk of “rhetorical purpose” for intentionality as a defining criterion. Rhetorical figure detection is also discussed in some detail with respect to figural and grammatical collocation. The communicative functions explored include Semantic-Feature Promotion, Reciprocal Specification, Reciprocal Energy, Irrelevance of Order/Rank, Subclassification, and Reject-Replace. Examples are drawn from epanalepsis, as well as antimetabole, mesodiplosis, and parison, which collate frequently as a group and also collectively with the trope, antithesis. These collocations, in turn, are coordinated by linguistic features such that the relevant expressions fit the notion of a construction as understood in the Construction Grammar framework. ![]() The communicative functions of rhetorical schemes rely not so much on individual schemes as on certain collocations of schemes (sometimes with tropes or other figures as well). This article illustrates form/function alignments for a small handful of rhetorical schemes, with some examples of how they communicate specific meanings. Rhetorical schemes, however, have been almost universally ignored by linguists, including computational linguists. The figures which are especially valuable in these ways are known as schemes, figures that are defined by their material (phonological, orthographical, morpholexical, or syntactic) form, in distinction particularly from tropes, which are defined by their conceptual (semantic) form. Rhetorical figures are form/function alignments in which the form (1) serves to convey the function(s) and (2) supports their computational detection therefore, (3) they are particularly rich for various text mining activities and other Natural Language Understanding purposes.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |