Evaluating Sight Translation A Corpusbased Approach Wallace Chen

Evaluating Sight Translation: A Corpus-based Approach Wallace Chen, Ph. D Assistant Professor Graduate School of Translation, Interpretation, and Language Education (GSTILE) Monterey Institute of International Studies Monterey, California, USA wallace. chen@miis. edu 28 July 2010 UCCTS 2010

Outline n n n Sight Translation Current assessment approach Corpus-based assessment approach Data Applications & findings Implications & conclusions 28 July 2010 UCCTS 2010 2

Sight Translation n an integral part of 1 st-year translation courses skim a text and verbally translate at the same time Skills required analytical skills cohesion / transition vocabulary time management summarizing skills pressure / stress management 28 July 2010 UCCTS 2010 3

Current assessment approach n n n assess performance based on verbal output length of text: less than 1, 000 words assessment criteria: meaning (accuracy), completeness, naturalness, backtracking, pauses, time management, grammar, pronunciation, numbers, terminology, register, cohesion/coherence, sentence structure…… n Class review and introspection 28 July 2010 UCCTS 2010 4

Corpus-based assessment approach n n Corpus: a systematic collection of texts on computer for purposes of linguistic research quantitative analysis in support of qualitative introspection reference (parallel) texts for improving translation quality data retrieval for the analysis of translation behaviors 28 July 2010 UCCTS 2010 5

Data n Reference Corpus (RC): 28, 000 words Subjects: rainforest conservation, computing, telecommunications n Translation Learners’ Corpus (TLC) - 2 sight translation exams (1 with reference official translation) - 32 students - 64 transcripts of sight translation into English n Corpus of Comment Tags (CCT) - 10 texts with comment tags 28 July 2010 UCCTS 2010 6

Three corpora: applications and findings (I) n Reference Corpus (RC) - collocation - terminology idiomatic expressions confirmation of intuition translation equivalent target language patterns new expressions 28 July 2010 UCCTS 2010 #some examples 7

Three corpora: applications and findings (II) n Translation Learners’ Corpus (TLC) - compare translation behaviors among students - compare between student translation and reference translation - time management: students finishing earlier have a tendency to score higher - lengthy translations suggest lexical overuse (39 vs. 59 for and) 28 July 2010 UCCTS 2010 8

Three corpora: applications and findings (III) n Corpus of Comment Tags (CCT) - fewer tags mean higher scores - fast translators receive fewer tags - repeated tags suggest common textual difficulties - comment tags reveal room for improvement for individual translators 28 July 2010 UCCTS 2010 9

Implications & conclusions n n n Supplement current approaches to assessment Identify and visualize common issues Understand translation patterns and suggest room for being flexible Learn from examples Corpus construction for research and teaching purposes, e. g. text selection 28 July 2010 UCCTS 2010 10