Paper: Automatically Predicting Sentence Translation Difficulty

ACL ID P13-2062
Title Automatically Predicting Sentence Translation Difficulty
Venue Annual Meeting of the Association of Computational Linguistics
Session Short Paper
Year 2013

In this paper we introduce Translation Dif- ficulty Index (TDI), a measure of diffi- culty in text translation. We first de- fine and quantify translation difficulty in terms of TDI. We realize that any mea- sure of TDI based on direct input by trans- lators is fraught with subjectivity and ad- hocism. We, rather, rely on cognitive ev- idences from eye tracking. TDI is mea- sured as the sum of fixation (gaze) and saccade (rapid eye movement) times of the eye. We then establish that TDI is correlated with three properties of the in- put sentence, viz. length (L), degree of polysemy (DP) and structural complexity (SC). We train a Support Vector Regres- sion (SVR) system to predict TDIs for new sentences using these features as in- put. The prediction done by our frame- work is well correlate...