Dependency Distance Differences Across Interpreting Types

Interpreting is generally recognized as a particularly demanding language processing task on cognitive system.Previous studies have proved an increment in the overall cognitive load during interpreting.These studies mainly focus on theoretical and behavioral studies,and few researches have paid attention to quantifying working memory load during interpreting and comparing working memory differences across various types of interpreting.Dependency distance,the linear distance between two syntactically related words in a sentence,is an index of sentence complexity,and is also able to reflect the cognitive constraints during language processing.The current research adopts a treebank-based quantitative approach with the index of dependency distance to investigate the working memory load during interpreting.Here,we examine the difference in dependency distance among three interpreting types,namely,simultaneous interpreting,consecutive interpreting and read-out translated speech,based on a treebank comprising these types of interpreting output texts with dependency annotation.The data of this study comes from thirty speeches and reports in political and economic fields,which are transcribed manually and put into computer for further processing and tagging.Our study finds that different interpreting renditions yield different dependency distances,among which consecutive interpreting entails the smallest dependency distance and read-out translated speech the largest.There are two major reasons for the largest dependency distance of read-out translated speech:firstly,previous preparation can reduce some processes that may tax working memory and thus removes part of pressure on working memory;secondly,read-out translated speech usually requires higher textual density and linguistic acceptability,such that the speeches could be made with much polish and sophistication,which consequently gives rise to a higher mean dependency distance.For simultaneous interpreting,the sentence-by-sentence pattern leads to the strong interference in syntactic structures between two languages,mainly from source language to target language,which may have an essential impact on the mean dependency distance for the output speech and thus contributes to a larger dependency distance of SI.In consecutive interpreting,interpreters are more "self-paced" because of the separation of listening phase and the reformulation phase.In order to deal with the high working memory burden generated by the temporal constraint as well as the insufficient note-taking information,consecutive interpreters may have a strong preference for syntactic structures with a smaller DD in order to lessen the working memory burden and processing difficulty.This preference is consistent with the universal preference for dependency distance minimization for human languages and the principle of least effort.The current study also rules out some non-cognitive factors that might influence the mean dependency distance of interpreting texts,including the output text size,MDD of input texts and interpreters’ individual styles.The results show that consecutive interpreting bears heavier cognitive demands than simultaneous interpreting and read-out translated speech,which leads to the smallest dependency distance.The current study applies a treebank-based approach to interpreting research for the first time and reveals quantitatively different working memory loads among simultaneous interpreting,consecutive interpreting and read-out translated speech.It is hoped that the method as well as the findings of this study will shed some light on future quantitative study on working memory demands during interpreting.