Portal de Periódicos da CAPES

The differential linguistic realization of comparative and additive coherence relations

1999; De Gruyter Mouton; Volume: 10; Issue: 2 Linguagem: Inglês

10.1515/cogl.1999.006

1613-3641

Henk Pander Maat,

linguistics and terminology studies

It is commonly assumed that comparative coherence relations (e.g., Peter is tall but his brother is short) can be analyzed in terms of semantic contrast. In this article it is claimed that comparative relations need to be understood in a context containing a similarity assumption. This assumption may be con®rmed (positive polarity) or denied (negative polarity) by the comparative text passage. When two entities are characterized in terms of a common variable without a similarity assumption being present (e. g., Peter's favorite color is red. His brother prefers green), the coherence relation is merely additive. Additive relations are not de®ned for polarity. While comparative relations may be linguistically marked by connectives like but, by contrast, or and in English or maar, daarentegen, or en in Dutch, additive relations remain unmarked. The distinction between comparative and additive relations is empirically supported by a corpus-based study of the linguistic marking of similarities and di€erences between the price movements of shares, using 400 fragments from stock-market reports in a Dutch daily newspaper. Some of these fragments clearly invoke similarity assumptions, for instance because of the fact that two companies are in the same branch of industry or because the fragment is preceded by an announcement of the general trend in share prices on a particular day. It was found that the overwhelming majority of linguistic markers of similarities and di€erences in share price movements occurred in contexts containing similarity assumptions. For instance, similarities that were to be expected may be marked by en `and' (e.g., Elsevier gained 1.10 [en] Kluwer went up forty cents), while unexpected di€erences are often marked by daarentegen `by contrast' (e.g., Elsevier gained 1.10 (new price 67.80). [Daarentegen] Kluwer lost forty cents (new price 50.70)). Some apparent counterexamples, in which two companies from the same branch of industry show di€erent share price movements while this di€erence remains unmarked, in fact do not invalidate the similarity assumption framework but Cognitive Linguistics 10±2 (1999), 147±184 0936±5907/99/0010±0147 # Walter de Gruyter support it. In the majority of these cases the similarity assumption has been eliminated or weakened by earlier announcements of di€erences within a certain branch. A closer analysis of ®ve Dutch comparative connectives reveals that two dimensions of comparative polarity need to be distinguished: a propositional polarity parameter with the values ``di€erence'' and ``similarity'', and an assumptional polarity parameter distinguishing between con®rmations and denials of the relevant assumption. Di€erent linguistic elements may be characterized in terms of di€erent polarity dimensions. For instance, ook `also' expresses positive propositional polarity, while maar `but' is an indication of negative assumptional polarity.