Evaluating lexical cohesion in telephone conversations

Ever since the publication of Halliday and Hasan (1976), cohesion analysis has received much attention in several branches of linguistics. Lexical cohesion in particular has been shown to contribute to the coherence of discourse in a number of ways (Ferstl and von Cramon, 2001; Hellman, 1995; Hoey, 1991b; Sanders and Pander Maat, 2006), and specific patterns of lexical cohesion have emerged as relevant for the description of different registers and genres (Louwerse et al ., 2004; Taboada, 2004; Tanskanen, 2006; Thompson, 1994). In the present article I challenge existing models of lexical cohesion and offer a revised one which affords particular attention to what I call ‘associative cohesion’. I go on to test the adequacy of the proposed framework in a corpus of 15 telephone conversations (20,043 words) extracted from the International Corpus of English-Great Britain. Using both quantitative and qualitative methods the article analyses 3480 ties and demonstrates that telephone conversations are lexically cohesive mostly due to repetitions (52.6%), associative cohesion (24%) and inclusive relations (10.5%), which overwhelmingly occur across the turns (87.6%) and over remote-mediated spans (71.9%). In addition, lexical patterns are shown to collaborate in topic management processes and to be sensitive to genre-specific factors, thereby demonstrating the descriptive potential and applicability of the framework.