Machine Translation Evaluation Research in Persian-English Language Pair: A Non-Statistical Meta-Analysis

It seems an established fact that machine translation has proved its unique utility capacities for globalcommunication. However, evaluating the quality and performance of MT has been and still is a challenge,especially in Persian-English language pairs. Against this background this article examines the current stateof research on machine translation evaluation of such pairs. The study reviewed research on evaluatingPersian-English MT published since the so-called “neural turn”, searching academic databases usingkeywords, using a coding framework based on strengths, weaknesses, challenges, and limitations of MTsystems, and categorizing information like identified weaknesses and proposed improvements.The results reported in this study include: a) a lack of systematic research due to limited industrialdevelopers in Iran, b) academic disinterest, and c) a need for tailored metrics. Finally, the results will bediscussed and suggestions will be provided and areas for future research will be mentioned.

Keywords:

Neural machine translation , Machine translation evaluation , translation accuracy , translationadequacy , Machine translation evaluation research areas.