Direcionalidade em tradução: uma investigação do processamento cognitivo de tradutores profissionais em tradução direta e inversa no par lingüístico inglês-português

Abstract

This thesis is part of the EXPERT@ project (CNPq 479340/2006-4) and builds on a series of empirical-experimental studies previously carried out at the Laboratory for Experimentation inTranslation (LETRA) at Faculdade de Letras of Universidade Federal de Minas Gerais, Brazil,researching expert performance in translation. It attempts to investigate cognitive aspects observed in the translation process of ten professional translators while translating two correlated texts from a foreign language (English) into a native language (Portuguese) direct translation and from the same mother tongue (Portuguese) into a foreign language (English) inverse translation. Data collection was based on the triangulation of translation process data, a methodology proposed byJakobsen (1999) and Alves (2001, 2003), combining different tools of data elicitation to investigate the translation process. The analysis of translation process data drew on linear representations generated by means of keylogged files recorded with the software Translog and analysed in conjunction with retrospective protocols and questionnaires. Directionality of the translation task (direct or inverse translation) was the independent variable in the study which partially replicatesthe work of Buchweitz and Alves (2006). Those authors analysed the impact of directionality on three variables: (i) time spent on the task, (ii) segmentation, and (iii) recursiveness. Buchweitz and Alves (2006) carried out a comparative analysis of the translation processes of translation students and novice translators and showed that inverse translation increases the amount of time spent on the task. In their study, the number of segments, measured by pause intervals, also increased for both groups when performing the inverse translation task. The group of novice translators showed a higher number of revisions during inverse translation, thus indicating that recursiveness may be a representative trait of this groups processing pattern. The present work expands Buchweitz and Alvess study by focusing on the translation process of professional translators. Directionality is also the independent variable in the study. The dependent variables are: (i) time spent on each phaseof the translation process (orientation, drafting, and revision) and on the two tasks as a whole (JAKOBSEN, 2002; ALVES, 2005), (ii) pause patterns, (iii) recursiveness (BUCHWEITZ & ALVES, 2006), and (iv) segmentation (DRAGSTED, 2004; RODRIGUES, 2009). The analysisreveals that only five out of the ten subjects spent more time on the inverse translation task. Task order seems to have played an important role in the allocation of time, the first task always taking longer to complete regardless of directionality. As far as the analysis of the phases of the translation process is concerned, the first phase orientation does not show significant variation among the subjects, who allocated little time to it. The drafting phase, in turn, demanded more time for allsubjects, thus corroborating findings in previous studies (JENSEN, 2001; JAKOBSEN, 2002 e 2003; LIPARINI CAMPOS, 2005, among others). Although, on the whole, subjects spent more time on the drafting phase in both directions, six of them actually spent more time to carry out theinverse translation task, thus corroborating Buchweitz and Alvess results. As to the revision phase, subjects allocated more time to the inverse translation task independently of task order. This result contradicts Buchweitz and Alvess findings, whose subjects spent more time revising the direct translation task. Pause analysis shows that task order also had an influence on the number of segments measured by pause intervals during each task. It points out that, while performing the second task, subjects make use of previously acquired knowledge, thus reducing the time allocated to problem solving. The results also show an increase in recursive movements in the inverse translation task. With respect to segmentation, no specific pattern could be identified among the ten subjects. However, there was an increase in the number of segments in the inverse translation task. Finally, no correlation was found between task order and the number of segments, thus pointing toidiosyncratic traits observable in the output of each subject who, drawing on a complex set of knowledge and abilities, strives to produce target texts in compliance with the specificities of the translation task.