The pragmatics of paragraphing English argumentative text

• Use of a computational procedure to predict paragraph breaks in argumentative text.
• Connecting pragmatics of discourse literature to computational findings.
• Noting significant presence of literary rhetorical devices at paragraph juncture affecting procedure results.

Computational linguistic work into the paragraph and paragraphing has highlighted the significant role that intra-paragraph lexical cohesion plays in ‘marking off’ one paragraph unit from another. The goal of the research reported on in this paper is to consider, in some detail, the relationship that exists between the lexical repetition patterns in an argumentative text (as identified by a computational procedure), the genre moves within it, the actual paragraphing of the texts, and the textual colligation features of the paragraphs. The Link Set Median procedure (Berber-Sardinha, 1997, Berber-Sardinha, 2001 and Berber-Sardinha, 2002) is used to document exact, inflectional and derivational lexical repetition usage across 10 short English argumentative texts, and to predict where segmentations originally occurred in the texts. The resulting data are then analyzed in the light of diverse research interests into the paragraph, and classified accordingly. A comparison of these results is made with data where there is either a marginal or no difference in the link set medians of adjacent sentences across paragraph junctures within the same texts. It is suggested that this novel approach of analyzing computational data from multiple paragraph-specific research interests results in a clearer picture of paragraphing practice emerging.