Gill Philip
[Studies in Corpus Linguistics 43] 2011
► pp. 143–165
The variant forms of idioms that have so far been encountered represent a broad class of relatively unmarked variation types in that despite the variety of their instantiations, none except red rag to a bull has involved the substitution of an idiomatic key word. This opens up marked variation, which will be discussed from this chapter forward. While idioms and their unmarked variants can be quite easily located in corpus data, the same cannot be said for marked variants. A study of variations made to the most salient constituents of an idiom, its key words or core collocation, necessarily involves the retrieval of versions which do not contain those words. This is exceptionally problematic in a corpus study: corpus query software is designed to retrieve specified words, but if those words cannot be identified before the data is analysed, how can they be extracted in the first place? This chapter starts with an overview of the paradox of finding the unfindable in corpus data, followed by an explanation of how it can be overcome. Later sections then deal with the variation tendencies that can be observed when colour words are substituted by replacement colour words, finding connections between referential, metaphorical and propositional meaning.