Field: Corpus: Typical strings
Speech acts are functional categories. They are defined on the basis of their function, not their form, but many speech acts exhibit typical surface patterns or they occur regularly with a small range of lexical items. Questions are often realized in interrogative form. In written form they end with a question mark and in spoken form with a typical interrogative intonation. Apologies often use words like sorry, pardon, or excuse. Such formal features have been called “explicit illocutionary force indicator” (Searle 1969: 30) or “illocutionary force indicating devices”, “IFIDs” (Levinson 1983: 238) because they indicate the illocutionary force of the speech act in which they occur. IFIDs, therefore, can be used to automatically locate relevant speech acts. But speech acts do not always include IFIDs. Questions, for instance, can appear in different syntactic forms as well, and apologies do not always include the words sorry, pardon or excuse.
Compliments do not have conventionalized IFIDs. According to Manes and Wolfson (1981), however, American English compliments show a remarkable lack of originality. A small range of syntactic strings and a small range of positive adjectives are regularly used. Such strings and such adjectives are potential candidates for corpus searches of compliments (see Jucker et al. 2008).
It is still an open question whether compliments are sufficiently standardized to allow such search techniques. Any research trying to use this kind of approach must first identify typical surface strings and typical lexical items before they can be used in corpus-based searches for compliments.
As in all other corpus-based searches, the researcher will have to balance the precision and the recall of his or her search patterns. The search may either retrieve many hits that on manual inspection turn out not to be compliments (the problem of precision), or it may fail to retrieve actual compliments (the problem of recall).
See also:
For a discussion of this method in connection with a particular set of research questions see: