Since its existence 10 years ago, the Center for Research and Innovation in Translation and Translation Technology (CRITT) at the Copenhagen Business School has been involved in Translation Process Research (TPR). TPR data was initially collected by the Translog tool and released in 2012 as a Translation Process Research Database (TPR-DB). Since 2012 many more experiments have been conducted and more data has been added to the TPR-DB. In particular, within the CASMACAT (Sanchis-Trilles et al. 2014) project a large amount of TPR data for post-editing machine translation was recorded and the TPR-DB has been made publicly available under a creative commons license. At the time of this writing, the TPR-DB contains almost 30 studies of translation, post-editing, revision, authoring and copying tasks, recorded with Translog and with the CASMACAT workbench. Each study consists of between 8 and more than 100 recording sessions, involving more than 300 translators. Currently, the data amounts to more than 500 h of text production time gathered in more than 1400 sessions with more than 600,000 translated words in more than 10 different target languages. This chapter describes the features and visualization options of the TPR-DB. This database contains recorded logging data, as well as derived and annotated information assembled in seven kinds of simple and compound process—and product units which are suited to investigate human and computer-assisted translation processes and advanced user modelling.
This study aims to analyse how translation experts from the German department of the European Commission’s Directorate-General for Translation (DGT) identify and correct different error categories in neural machine translated texts (NMT) and their post-edited versions (NMTPE). The term translation expert encompasses translator, post-editor as well as revisor. Even though we focus on neural machine-translated segments, translator and post-editor are used synonymously because of the combined workflow using CAT-Tools as well as machine translation. Only the distinction between post-editor, which refers to a DGT translation expert correcting the neural machine translation output, and revisor, which refers to a DGT translation expert correcting the post-edited version of the neural machine translation output, is important and made clear whenever relevant. Using an automatic error annotation tool and the more fine-grained manual error annotation framework to identify characteristic error categories in the DGT texts, a corpus analysis revealed that quality assurance measures by post-editors and revisors of the DGT are most often necessary for lexical errors. More specifically, the corpus analysis showed that, if post-editors correct mistranslations, terminology or stylistic errors in an NMT sentence, revisors are likely to correct the same error type in the same post-edited sentence, suggesting that the DGT experts were being primed by the NMT output. Subsequently, we designed a controlled eye-tracking and key-logging experiment to compare participants’ eye movements for test sentences containing the three identified error categories (mistranslations, terminology or stylistic errors) and for control sentences without errors. We examined the three error types’ effect on early (first fixation durations, first pass durations) and late eye movement measures (e.g., total reading time and regression path durations). Linear mixed-effects regression models predict what kind of behaviour of the DGT experts is associated with the correction of different error types during the post-editing process.
This chapter describes an eye tracking study which compared eye movements during reading for comprehension with reading for translation. In addition, the number of target words likely to be used for the translation of a single source word was manipulated. Results showed large significant task effect on Total Reading Time, Average Fixation Duration, Fixation Count, Regression Count and Progressive Saccade Amplitude and the manipulation had a 23ms effect on first fixation durations. This is the first eye tracking study which charts the time course of the effect of translation on source text reading in detail. The results are interpreted in terms of co-activation of the two linguistic systems during translation, in line with previous studies which used different methods.
The manuscript provides readers with a basic methodological toolset for experimental psycholinguistic studies on translation. Following a description of key methodological concepts and the rationale behind experimental designs in psycholinguistics, we discuss experimental paradigms adopted from bilingualism research, which potentially constitute a methodological foundation for studies investigating the psycholinguistics of translation. Specifically, we show that priming paradigms possess several inherent advantages which make them particularly suitable for research on translation. The manuscript critically discusses key methodological problems associated with such paradigms and illustrates the opportunities they may offer for translation research, concludes with a review of past and current translation process research highlighting ways in which these can contribute to the issues raised by cross-linguistic priming studies.