Desenvolvimento e aplicação do Litterae: um sistema para marcação, etiquetação, estatística e buscade unidades textuais e processuais

Abstract

This thesis, which was developed within the scope of SEGTRAD Project (Cognitive Segmentation in Translation with and without the aid of Translation Memory Systems) and is related to other researches carried out at LETRA (Laboratory for Experimentation in Translation), Faculdade de Letras, Federal University of Minas Gerais, proposes a newannotation method for Translog® keylogs using the new Litterae tool. Drawing on the concept of translation unit of Carl (2009) expanded by Alves and Vale (2009) and relying on SEGTRAD keylogs analyzed by Renata R. Rodrigues from 12 professional translatorsin the pairs German-Portuguese and English-Portuguese, we carried a corpus research in which we coud demonstrate how the new annotation method helps in identifying distinct patterns in the process segmentation by language stratum (HALLIDAY & MATTHIESSEN, 2004), occurence sort, language pair and translation phase (JAKOBSEN, 2002; 2003; ALVES, 2003; 2005). In this work, we found out that dactilography and semantics are strata whose patterns depend on the translators and thatlexicogrammar is a stratum whose patterns depend on the language pair. In the lexicogrammar, the English-Portuguese pair showed relatively more revisions at the lexical pole while the German-Portuguese pair showed relatively more revisions at the grammatical pole. The occurrence sort had three well-defined patterns: reducing TT equivalents during the writing phase, iteratively substituting TT segments during therevision phase and reducing the number of TT words and structures during the revision phase. We demonstrated that the stratum categories and its subtypes were adequate to identify patterns of translation process and product. The practical consequences of using Litterae tool consisted of reducing the analysis time and therefore increasing the laboratorial research productivity. The theoretical consequence of using this tool consists of being able to analyse the product during the translation process. This type of analysis leads us to interpreting the translation recursivity as a textual revision process and understanding how these revisions affect the translation product. With this research methodology, we were able to represent a macro-UT, which is a unit of translation process, as a series of products (interim versions and a final one), thus creating a strong connection between the process- and product-oriented studies.