SearchPublisher and Institutes
You are here: Home :: Area CULI :: Linguistics and literature :: Communication science
Richard Watson Todd
A fuzzy approach to discourse topics
Discourse-level topics are one of the most elusive and intractable notions in semantics, because they are inherently subjective and because more than one topic may be identi?able for a given stretch of discourse. Despite these characteristics of topics, much previous research into topics has attempted to identify a single topic for a stretch of discourse. In the present study, however, based on fuzzy logic, multiple topics are identi?ed. Using data from an English language classroom at a Thai university, the identi?cation of multiple topics involves, ?rst, using six different approaches to analyse the data: theme-rheme progression, given-new progression, the speech act analysis of Sinclair and Coulthard (1975), Hoey’s (1991) lexical analysis, association-based networks of concepts, and topic-based analysis. These methods were then validated through comparison with a control method of analysis, allowing weightings of importance to be assigned to each of the six approaches. The various topics identi?ed by each approach, taking the weighting of the approach into account, were combined to identify a range of topics with values indicating prominence for each point in the discourse. The extent to which a given point in the discourse is centred on a potential topic was calculated using the concept of fuzzy entropy. Although complex, it is argued that this fuzzy approach to discourse topics provides greater precision in describing and identifying topics.
Semiotica, Walter de Gruyter
Print ISSN: 0037-1998
Volume: 2005, 06/2005
Pages: 93 - 123