Effect of word order asymmetry on cognitive process of English-Chinese sight translation by interpreting trainees : evidence from eye-tracking

Abstract

The current study examines effects of word order asymmetry between source language and target language on cognitive processes of English-Chinese sight translation (STR) performed by interpreting trainees. Word order asymmetry as one indicator of language-pair specificity is a widely debated topic in interpreting literature (Gile, 2002, 2011; Wang & Zou, 2018), but its effect on cognitive behaviour during interpreting has received virtually no systematic investigation (Wang & Gu, 2016). A set of studies have been conducted to examine how word order differences influence interpreting product but few approach the issue from a cognitive perspective: whether and how does word order asymmetry affect on-line processing during interpreting remains an underexplored area. It is well-documented that English and Chinese are strikingly different in grammatical rules: the two languages conform to distinctive branching directions. For instance, Chinese is a typical left-branching language in which long strings of modifiers are always located before the nouns while English is mostly right-branching which puts nouns before the modifiers. Differences in word orders may impose extra cognitive burden on interpreting process, which requires more in-depth explorations to reveal mental activities associated with addressing structural incongruences. This study uses eye-tracking as the primary data collection method to observe and analyze in what way and to what extent word order asymmetry as indicated by relative clauses and passive constructions affects on-line processing during English-Chinese sight translation and to examine the potential role of contextual information and working memory span in offsetting the asymmetry-induced disruptions. In addition, the study also attempts to explore: 1) How do interpreting trainees that have been trained in basic sight translation skills cope with word order asymmetry in English-Chinese sight translation and will their choices of strategy for processing asymmetric structures affect the cognitive load and on-line reading patterns? 2) Does word order asymmetry negatively impact interpreting performance in English-Chinese sight translation in terms of error frequency and distribution pattern? And is the interpreting trainees' self-perceived degree of structural difficulty related to the actual cognitive load involved? To answer the above questions, a group of interpreter trainees who had been trained in basic interpreting skills were recruited as the participants. A 2 x 2 quasi-experimental design for sight translation tasks was adopted with sentence type (asymmetric sentences vs symmetric sentences) and task condition (interpreting under single sentence context vs interpreting under discourse context) as independent variables. A set of English sentences containing relative clauses or passives were employed as the asymmetric sentences (critical sentences) and an equal number of comparable sentences that resembled Chinese in word orders were used as the symmetric sentences (control sentences). In Condition 1 (single sentence context), the participants were required to sight translate 12 critical sentences and 12 control sentences which were presented in an individually randomized order; In Condition 2 (discourse context), the participants sight translated 12 critical sentences and 12 control sentences which were embedded in coherent discourses. Each participant sight translated 24 critical sentences and 24 control sentences for the task in both conditions and eye movements during the task were recorded as indicators of cognitive load during on-line processing. To measure the cognitive load in sight translation, eye metrics at both global level (dwell time and fixation count) and local level (first fixation duration, first pass reading time, second pass reading time and regression path duration) were employed; to access the impact of word order asymmetry on interpreting product, an error analysis based on frequency and distribution of errors in the target output was conducted. In addition to the sight translation task, the participants filled in the post-experimental questionnaires to indicate their perception of asymmetry-induced difficulties during sight translation. They also undertook an English reading span test which was designed to test their working memory capacity, in particular the ability to coordinate input storage and information processing. Findings of the study were triangulated based upon both on-line and off-line data. Major results can be summarized as follows: 1) Word order asymmetry exerted a highly significant effect on cognitive process during sight translation. There was significantly greater cognitive load in the critical sentences than in the control sentences at both sentence-level and word-level. Local eye measures indicated that as compared with the single sentence context, a wider context in discourse condition imposed extra contextual constraints which resulted in considerably higher cognitive load. Besides, the on-line processing of the critical sentences was influenced by contextual information to a greater degree. 2) Segmentation was the primary strategy adopted by the trainees to address asymmetric structures and it did relieve the cognitive burden to some degree as compared with restructuring, the other applied strategy. Different reading patterns in form of the rereading rate were observed between processing through segmentation and processing through restructuring, indicating relations between strategy choice, corresponding cognitive load and reading behaviour. 3) Interpreting quality was negatively affected by word order asymmetry as indicated by significantly more frequent errors made in asymmetric sentences, but there was a mismatch between participants' self-perception of syntactic difficulty and the actual cognitive load devoted. Theoretically, results of the study enrich the existing knowledge on cognitive constraints during interpreting and on sentence processing in a cross-linguistic scenario. Observation based on eye-tracking data deepens the understanding of mental workings in sight translation, a special hybrid of interpreting and translation. Methodologically, eye-tracking data in the current study provides empirical evidence for a cognitive-taxing effect of language-pair specific factors on interpreting and also complements previous researches based on product-oriented approaches. A distinction of early and late eye measures in data analysis instead of relying merely on overall measures provides a more complete picture of on-line processing at different stages. In terms of pedagogical implications, it emphasizes the role of language pair specificity in shaping interpreting process and product and the need to develop awareness of structural differences between source and target language in daily training.