Topic versus cohesion in the prediction of causal ordering in English conversation

We tested Schiffrin's (1985) hypothesis that the choice between "CAUSE so RESULT" sequences and "RESULT because CAUSE" sequences is primarily determined by topic continuity (topic being operationally defined as grammatical subject) against British conversational data from the London-Lund Corpus of Spoken English printed in Svartvik and Quirk (1980). Our test produced results similar to those produced by Schiffrin's test of the hypothesis against her U.S. data. We hypothesized that noun phrase cohesion would be a better predictor of causal ordering than subject topicality and achieved a significantly better prediction rate (57.5% vs. 21.5% of 200 causal sequences). This study supports Schlobinski and Schütze-Coburn's (1992) critique of the use of topic as a uniting and distorting metaphor for a number of different formal, semantic, and pragmatic patterns, such as grammatical relation, role, and cohesion. However, we also conclude, stopping short of metaphorizing cohesion as topic, that our study and Schiffrin's study reveal a special cohesive functional load concentrating in grammatical subject.