This paper reports on a corpus-driven investigation into sense ambiguity. The work grew out of a curiousity about the mechanics of 'semantic negotiation': that is, structured ambiguity between word senses as resource for meaning-making in text. While corpora have been enthusiastically embraced for the insights they can provide into just about every aspect of lexis, a survey of corpus work on ambiguity in lexical meaning reveals a pre-occupation, for the most part, with eliminating it. The application-driven and formal orientation of much corpus work, and perhaps the exhortation "throw your [messy] data away", has seemed to preclude much interest in ambiguity for its own sake - for what corpora can tell us about word sense ambiguity, and what ambiguity can tell us about meaning-making.

The investigation centres on the polysemous lexeme CARE. It was undertaken in collaboration with an interdisciplinary group researching values about medical treatment - the Centre for Values, Ethics and the Law in Medicine, in the Department of Surgery, University of Sydney. In the interview data collected for that project, as in wider medical discourses, questions of  what 'care' is, or should be, are currently emotive and politicised. One thing which had emerged as a concern for researchers in the group was the "slippage" between meanings of CARE in text. I became interested in how "slippage" might be, rather than a mistake or misunderstanding, a functional resource for speakers/writers to negotiate the possible meanings of CARE. Taking such a viewpoint leads to old questions:  How is it that people are able to re-negotiate meaning? For, if the meaning is 'in' (i.e. both inheres in, and is flagged by) the usage, then how is it that a word can both realise a certain meaning, by virtue of conforming to past instances/current potentials of usage, and simultaneously re-negotiate what that meaning should be? What kind of picture of lexical meaning does that give us, and how might it be crystallised as a tool for interpretation?

This investigation draws on theory & methodology from linguistic traditions following Halliday and Sinclair, both heritors of Firth, but since diverging in their treatment of lexical meaning. From the former, it takes a view of language as socially functional in and through context, realising choices from an ordered set of shared resources. From the latter, a view of the meaning-making process as structured to a greater or lesser extent by defaults of co-occurence, which can best be uncovered by empirical investigation. The work on CARE is a pilot experiment in approaching polysemy as a mappable semantico-grammatical space. The senses of a word emerge from reference corpora as clusters of tokens, with similar semantico-grammatical symptoms of function, form and contextual features. (These were collected manually in 3 reference corpora of written & spoken Australian/NZ English, totalling 2.4 million words, for application to the medical interview corpus of 350,000 words.) These senses and their symptoms can then be seen as both a set of probabilities (defaults) in the language and as a meaning potential (resource). An ambiguous instance of CARE, through less probable combinations of symptoms in text (ambiguity), can be seen as an opportunity to negotiate between senses, while simultaneously 'tweaking' the probabilities of the semantico-grammatical space as a whole. Collecting senses and symptoms from reference corpora provides a portable and non-ad hoc resource for 1) exploring variations in meaning in specific text-types/communities (e.g. interview corpus); and 2) interpreting the function and motivation of individual instances of word sense ambiguity in context.
 

Contact details:

Cassily Charles
Institut für Anglistik und Amerikanistik
Akademiestraße 24
A-5020 Salzburg, Austria

Ph:    + 43 662 8044 4413
Fax:  + 43 662 8044 167

cassily.charles@sbg.ac.at