Logistic calibrated items (LCI) method: does it solve subjectivity in translation evaluation and assessment?

Abstract

This research paper introduces a model of translation evaluation called Logistic Calibrated Items method. The aim of LCI method is to maximize a translators’ performance and to identify top competent translators through detecting all parsing items within a source text. Parsing items are extracted by the application of Brat software. The purpose of LCI was to identify parsing items having optimal item difficulty and item discrimination values. LCI method involves six stages: (1) holistic scoring; (2) the application of Brat software to extract all parsing items; (3) the calculation of item difficulty; (4) the calculation of item discrimination; (5) the identification of items with optimal item difficulty and item discrimination values; and (6) the recalculation of scores. 125 translation students and 4 professional translation evaluators took part in this research. The final results showed that LCI method was more consistent when compared to holistic method. Limitations and implications were also discussed.