validation of automated scores of human translation of legal texts against non-test indicators of translation competence

This study tries to explore the validity of lexical automated translation quality evaluation understudy metrics (atqeum) in scoring certified translator accreditation tests in islamic republic of iran as an instance of legal texts. this is conducted against non-test indicators of translation competence, including the participants records of their scores in “the translation of documents and deeds i and ii” courses, their average score of all courses of practical translation nature, and their ba gpa score in translation studies major. although, they have not revealed a significant correlation with the nontest indicators of translation competence, they have had a highly significant correlation with the scores granted by human expert scorers on the two sample tests. therefore, according to all the data collected and analyzed, it has ultimately been concluded that a collection of “-per, -terp-a, bleu-1, nist-1, rouge-1, gtm-1” lexical atqeums (including all various lexical similarity-based techniques of edit distance, precision, recall, and f-measure) can be considered as the optimal translation meta-evaluation set to score certified translator accreditation tests.