Shifts in the Segmentation of Sentences in Translation : Reasons and consequences Corpus-based Research in Contrastive Linguistics and in Translation Studies
Abstract
Splitting and joining of sentences in translation, i.e. shifts in segmentation, may have several reasons; on the one hand, structural differences between the languages involved, on the other hand, specific features of translation (co-called translation universals, cf. Baker 1996, Blum-Kulka 1986, etc.).
As for the first reason, according to Fabricius-Hansen (1996 and 1999) and Solfjeld (1996) languages differ in their (relative) information density. High information density languages (such as German or French) encode the discourse information in complex, hierarchical sentences, whereas low information density ones (e.g.
Norwegian or Czech) prefer a more incremental, paratactic style. Thus, while translating from a high information density language to a low information density one, splitting of sentences occurs more often than in the other direction of translation (cf.
Nádvorníková 2017). More importantly, shifts in segmentation involve other changes: adding of connectives, moving non-finite constructions (gerunds, participial adjuncts, etc.) to the sentence level, changing textual coreference relations, etc.
Nevertheless, shifts in segmentation of sentences occur also independently of the level of information density of the source/target languages. This may be explained by the influence of a general tendency of translations to the explicitation, simplification or normalization of the source text (so-called translation universals, Baker 1996, Blum-Kulka 1986).
In this paper, we will explore these reasons for and consequences of the shifts in segmentation of sentences on the basis of data from the core of the Czech-French-English part of the InterCorp parallel corpus (www.korpus.cz/intercorp, Čermák and Rosen 2012), which contains mainly narrative texts.