Post-editing of Technical Terms based on Bilingual Example Sentences

Proceedings of Machine Translation Summit XIX, Vol. 1: Research Track, pages 385–392 September 4–8, 2023, Macau SAR, China. ©2023 The authors. This article is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0). 385 Post-editing of Technical Terms based on Bilingual Example Sentences Elsie K. Y. Chan elsie.chan@mail.com John S. Y. Lee jsylee@cityu.edu.hk Department of Linguistics and Translation, City University of Hong Kong Chester Cheng chester.cheng@gmail.com Department of Translation, The Chinese University of Hong Kong Benjamin K. Tsou rlbtsou@cityu.edu.hk Department of Linguistics and Translation, City University of Hong Kong Chilin (HK), Ltd. Abstract As technical fields become ever more specialized, and with continuous emergence of novel technical terms, it may not be always possible to avail of bilingual experts in the field to perform translation. This paper investigates the performance of bilingual non-experts in Computer- Assisted Translation. The translators were asked to identify and correct errors in MT output of technical terms in patent materials, aided only by example bilingual sentences. Targeting English-to-Chinese translation, we automatically extract the example sentences from a bilin- gual corpus of English and Chinese patents. We identify the most frequent translation candi- dates of a term, and then select the most relevant example sentences for each candidate accord- ing to semantic similarity. Even when given only two example sentences for each translation candidate, the non-expert translators were able to post-edit effectively, correcting 67.2% of the MT errors while mistakenly revising correct MT output in only 17% of the cases. 1 Introduction Post-editing of machine translation (MT) system output is now commonly incorporated as part of the workflow in the translation industry, since it can produce higher quality texts than manual translation (Garcia, 2011; Green et al., 2013). For texts in scientific or technical domains, it would be ideal to have bilingual domain experts to perform the post-editing. Given the large number of specialized domains and language pairs, however, translators with the required skills are unfortunately not always available. It is therefore important to understand whether those without the full linguistic or technical background could still perform post-editing adequately. While previous research has explored the feasibility of monolingual post-editing (Mitchell et al., 2013), few studies have investigated how well bilingual non-experts can post-edit MT output of technical texts. This paper evaluates the performance of bilingual novice translators in identifying and cor- recting MT errors in technical term translation. To simulate a realistic scenario with time con- straints, the translators are aided with only a small number of bilingual example sentences from a database of patents. These example sentences are automatically retrieved from PatentLex, a 386 large corpus of English and Chinese patents (Lu et al., 2009; Tsou et al., 2019), to illustrate the most likely translation candidates. Results show that, despite their unfamiliarity with the domain, the translators managed to correct a majority (67.2%) of the MT errors, and mistakenly revised correct MT output in less than 17% of the cases. The rest of the paper is organized as follows. After sketching the …

Content truncated to 500 words