Quality Evaluation of Machine-Aided Translation Programs for Business Correspondence

Abstract

This Master’s Thesis is devoted to the quality evaluation of machine-aided translation programs when working towards business correspondence materials in order to use them more effectively in the translation process. The study purpose is to analyze typical errors that occur in automated translation of business correspondence in order to develop translation quality evaluation parameters. The development of such parameters makes it possible to identify the key problems of machine-aided translation systems and provide tools to improve their quality, supporting the human factor in translation. As a result, the research aims to optimize translation, improve human-machine interaction, and increase the business communication efficiency. The introduction of the Master’s Thesis includes background information on the topic, establishes the research rationale, purpose, objectives, problem, scope and unit of analysis. The issue of application of automated translation programs in Kazakhstan is also considered in the Introduction of the work, which allows to assess the degree of their integration into professional translation activities at the local level. The Literature Review reveals the key concept of this thesis, namely – machine aided translation program. The issue of reliability of automated translation programs and quality assessment of their final output based on existing research is studied in this section. The relevant classifications of errors made by machine translation programs is also considered in Literature Review. On the basis of such categorization a table is compiled, bringing together the types of errors identified by the authors previously engaged in the study of the issue under consideration. The derived classification has been proposed to two experts with experience in the field of translation in order to supplement and update it in accordance with modern requirements to translation quality. The methodological part defines the nature of the work, namely, empirical product based research. This section also reveals the applicable research paradigm, within which the applicable qualitative research methods are defined. Results chapter presents the main results of the research, outlines the corpus of the data under study, performs comparison and collation of the obtained results with the Literature Review section outcomes, and compiles an actual categorization of errors made by machine-aided translation programs when processing business correspondence materials. This chapter also presents the results of the quantitative analysis of the identified errors, establishes the most frequent errors and identifies the automated translation program that distorts the final product to the least extent and, as a consequence, can be used by a translator when working towards business correspondence materials subject to further post-editing. In the Discussion chapter, examples of typical machine-aided translation errors identified in the process of analysis are considered, and their interpretation in terms of linguistic and contextual features is given. In the Conclusion section, the results of this work are summarized, the limitations of the study are identified, and recommendations for further research in this area are given.