Why bother with Corpus Evidence?
Metaphor is a semantic, not a syntagmatic notion. Corpus-based semantic tasks such as finding all metaphors would require syntagmatic criteria—but such criteria do not yet exist. So if corpora do not contain explicit semantic statements, why bother with them? There are two answers. Firstly, a large corpus reveals systematic syntagmatic regularities (norms) of linguistic behaviour. Semantic interpretations are associated with syntagmatic regularities, rather than with words in isolation. Secondly, a corpus shows how norms are exploited.
Quite unconsciously, all users of a language conform on most occasions, to the same general patterns of syntagmatic usage for each word. Using the verb backfire, I show how syntagmatic patterns are associated with senses, and I discuss the concept of metaphor. I argue (e.g., Hanks 2004) that the distinction between norms of usage and exploitations of norms is more important than the distinction between literal and metaphorical meaning.
It is now generally accepted that human beings have a predisposition from birth to behave linguistically (just as horses have a predisposition to trot and gallop). The actual details of language acquisition depend on the triggering effect of exposure to a particular language. Corpus data can help with the understanding of the triggering effect, as explored in Hoey (2005). By studying a corpus, we can identify probabilities regarding the words and contexts to which a child will be exposed. A corpus tool such as the Word Sketch Engine (Kilgarriff et al. 2004) prompts hypotheses about statistically significant likelihoods of priming. The relationship between statistically significant collocates and concrete experiences of the world remain to be studied.
It is not possible to find all metaphors in a corpus, but it is easy to find all similes. Creativity is demonstrably more active in similes than metaphors. Similes exploit semantic norms.
1. The loud, coarse voice ripped through the air with shocking force, like a dagger through silk.
2. He looked like a broiled frog, hunched over his desk, grinning and satisfied.
3. A single woman in their midst acts like a demented lighthouse: enticing hapless travellers, by its safe and steady beam, onto the rocks below.
These similes are triggers for the reader’s imagination. They are not conventional, and they are not an experiential Gestalt (cf. Lakoff and Johnson 1980). It is not necessary that you should have experienced cutting silk with a dagger before you can understand 1. Similes also license syntactic displacement and other kinds of linguistic mayhem, as in 2 and 3.
Hanks, P. 2004. ‘Syntagmatics of Metaphor’ in International Journal of Lexicography 17:3.
Hanks, P. 2005. ‘Similes and Sets’ in R. Blatná and V. Petkevič (eds.): Jazyky a jazykovĕda (Festschrift for Prof. Fr. Čermák). Prague: Charles University, Philosophy Faculty.
Hoey, M. 2005. Lexical Priming: a New Theory of Words and Language. Routledge.
Kilgarriff, A., P. Rychly, P. Smrž, and D. Tugwell. 2004. ‘The Sketch Engine’ in Proceedings of Euralex. Lorient, France.
Lakoff, G., and M. Johnson. 1980. Metaphors we live by. University of Chicago Press.