초록 열기/닫기 버튼
This study investigates ways to improve the validity of automatic translation assessment tools. Comparison of assessment criteria used by human raters against those employed by automatic assessment tools show that automatic assesment tools are geared towards evaluating ‘meaning’ and not so much towards ‘expression’. Lexical diversity, as measured by type-token ratio, and lexical sophistication, as measured by lexical frequency, are identified as potential criteria that can improve validity of automatic translation assessment when added to automatic assessment tools. To empirically test the effects of lexical diversity and sophistication on translation scores, Korean-English translation of 18 applicants for an interpreting and translation graduate school in Korea were collected and analyzed using an online lexical profiling tool followed by multiple regression analyses designed to validate their significance. Analyses show that lexical diversity has statistically significant correlation with translation scores, explaining 25% of variance in translation scores. On the other hand, lexical sophistication showed no statistically significant correlation with translation scores. Implications and limitations of the study are discussed.
키워드열기/닫기 버튼
lexical diversity, lexical sophistication, automatic translation assessment, translation evaluation criteria
