Source language difficulties in learner translation: Evidence from an error-annotated corpus

Abstract

This study uses an error-annotated, mass-media subset of a sentence-aligned, multi-parallel learner translator corpus, to reveal source language items that are challenging in English-to-Russian translation. Our data includes multiple translations to most challenging source sentences, distilled from a large collection of student translations on the basis of error statistics. This sample was subjected to manual contrastive-comparative analysis, which resulted in a list of English items that were difficult to students. The outcome of the analysis was compared to the topics discussed in dozens of translation textbooks that are recommended to BA and specialist-degree students in Russia at the initial stage of professional education. We discuss items that deserve more prominence in training as well as items that call for improvements to traditional learning activities. This study presents evidence that a more empirically-motivated design of practical translation syllabus as part of translator education is required.