This preview shows page 1. Sign up to view the full content.
Unformatted text preview: . Concordancing is a
powerful observational tool, but no more than a tool; a researcher is
needed to decide what to examine and how to interpret the resulting data
(Deignan 1999: 180). On top of that, corpora are of limited usefulness in investigating innovative metaphor and require a bottom-up approach to theory building.
Deignan discusses problems inherent in metaphor theory research
as well. She sees distinguishing metaphor from other related phenomena as
the major problem. There are three aspects to this problem, i.e. a distinction
between dead metaphor and polysemous senses (possibly as in catachresis),
a distinction between metaphor and idiom, and between idiom and metonymy. In many studies, it must be admitted that these distinctions are neglected and all figurative uses are viewed as metaphoric (see for instance
Lakoff – Turner’s (1989) discussion of poetic language and a rather unclear
distinction they make between metaphors, personification and proverbs),
while others make minute distinctions (for example Sadock (1993) in building a speech act theory of metaphor, which distinguishes it painstakingly
from other nonliteral linguistic figures such as metonymy, synecdoche, hyperbole, understatement, irony and euphemism).
One of the most recent approaches to metaphor in combination
with corpus linguistics is Charteris-Black (2004), in which he integrates
the insights from CMT and discourse analysis. His investigations focus on
specific-domain corpora, such as the British party political manifestos,
American presidential speeches, sports reporting, financial reporting and
religious discourse (the Bible and the Koran). But he stresses, like Stubbs
did before him, that it is “beneficial to compare the findings of a domainspecific corpus with those of a general corpus (…). In such cases a gen- Corpus linguistics and the language of mass media 63 eral corpus serves as a control corpus” (Charteris-Black 2004: 31). For
example, in analysing financial reporting he checked the frequency of potentially metaphoric words in The Economist sub-corpus and the entire
Bank of English corpus and found out that words such as attack and haven were far more frequent in the general corpus. This led him to conclude that metaphors including these words are general and not domainspecific. He also remarks on the interaction of researcher’s intuition and
corpus data: “It is not that corpus linguists do not rely on their intuitions
as much as in the traditional approaches, but that their intuitions are
measured against attested linguistic evidence” (Charteris-Black 2004: 31).
Yet another recent investigation of metaphor, which combines
corpus linguistics, discourse analysis and conceptual metaphor theory is
Koller (2004). In her study of metaphors in business media discourse, she
uses two self-compiled, purpose-built corpora from British and American
business magazines and newspapers. Her aim is to study the rhetoric of
the language of business press reports. She finds out that there are three
View Full Document