Profiling effects of syntactic complexity in translation : a multi-method approach

Heilmann, Arndt; Neumann, Stella (Thesis advisor); Alves, Fabio (Thesis advisor)

Aachen : RWTH Aachen University (2020, 2021)
Dissertation / PhD Thesis

Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2020

Abstract

Translation is one of the more complex text production tasks a writer can engage in and linguistic complexity is one of the likely determinants for the features of the translation product and process. This thesis set out to better understand how the syntactic complexity of the source text affects translation properties. My account of complexity describes complexity in ontological/structural terms rather in agent related complexity i.e. difficulty. On the basis of annotations, I operationalized syntactic complexity and a number of control variables that could be used to statistically control against a variety of linguistic and non-linguistic confounders. I used linear mixed regression modelling to analyse my process-product corpus. Counter to my expectations, I could show that the translations were actually longer than the source texts. Thus, explicitation seemed to have a stronger effect than simplification – at least for texts with the properties of my sample. I also found that complexity reductions at one syntactic level usually encompassed complexity increases at another and vice versa. These shifts tended to happen when one syntactic level of complexity was abundant. From all investigated variables (nr. of clauses per sentence, nr. of chunks per clause and average group length) only the first showed a small but statistically significant reduction. I could show that this reduction was not due to a loss of overall complexity but that virtually all of the lost clauses could be accounted for by means of complexity shifts and not reductions. So many clauses were in fact shifted, that there was a notable over-compensation at the level of chunks per clause. Interestingly, the reduction of complexity at the clause level was not only caused by an abundance of subordinate structures, but by attempts to normalize the language of the translations with respect to non-finite clauses. I found that the reason for these complexity shifts are likely due to the translators' considerations of processing effort and/or stylistic preferences by the target audience because translation quality ratings showed more dissatisfaction with more complexity at virtually all studied levels. This likely makes the translation of syntactically complex source texts more difficult as syntactic complexity could be shown to affect measures of reception, production or both. Indirectely embedded clauses could be shown to require substantial effort during reading and writing. In terms of reading effort subordinate clauses turned out to be the most demanding during translation and they also caused the translators to write less efficiently. This thesis could demonstrate that effects of syntactic complexity are small but pervasive and it could demonstrate the usefulness of linguistically enriched product-process corpora to reduce the gap between process and product based research in translation studies.

Identifier

Downloads