Working on Сomputer-Assisted Translation platforms: New advantages and new mistakes

Abstract

The paper presents analysis of errors in translation on the CAT platform Smartcat, which accumulates all tools for computer-assisted translation (CAT) including a machine translation (MT) system and translation memory (TM). The research is conducted on the material of the translation on Smartcat platform (a joint project of a tourist guide translation (35,000 words) from Hebrew to Russian, English, and French). The errors on the CAT platform disclose difficulties in mastering text semantic coherence and stylistic features. The influence of English as lingua franca appears in peculiar orthographic and punctuation errors in the target text in Russian. Peculiar errors in translation on the CAT platform reveal the necessity of advanced technological competence in translators. The peculiar errors uncover problems associated with a source text segmentation into sentences. The segmentation can trigger a translator to preserve the sentence boundaries and use a Russian complicated compound sentence that provoke punctuation errors. Difficulties of the anaphora resolution in distant semantically coherent segments are also associated with the source text segmentation and working window formatting. A joint project presupposes different translators to translate different files of the source document. To generate the coherence, contiguity and integrity of the whole document, the files have to be revised by a third-party editor to avoid conflict of interest. The editor-reviser is also responsible for improving the target text pragmatic and genre characteristics while applying top-down strategy to target text analysis. Thus, the translator’s errors while applying CAT tools reveal the effect of bottom-up text processing alongside with cross-language interference.