Home › Archive for Academic publications
Category: Academic publications

Posted on by Michael Farrell Posted in Academic publications

Abstract

As the use of generative artificial intelligence (GenAI) becomes more mainstream, an increasing number of authors may turn to this technology to write directly in a second language, bypassing traditional translation methods. Consequently, professional editors may have to develop new skills: shifting from correcting translation and non-native errors to editing AI-assisted texts. This study includes several stages: participant selection, text planning, prompt engineering, text generation and text editing. The recruited authors provided prompts for GPT-4 to generate texts, edited the output as they desired and then passed them on to professional editors for a final edit. All participants reported their experiences and described the nature of their interactions. The findings reveal that, while GenAI significantly improved the grammatical accuracy of the non-native English texts, it also introduced anomalies. In conclusion, although AI was useful in these two cases, it did not fully replace the human editors, and professional translators-with their language skills-may like to consider offering this additional service. The study also suggests that both authors and editors should be trained in synthetic-text editing to fully harness the benefits of AI-assisted writing, and that further research should be conducted with diverse texts and authors to generalize the findings.

Published in

Translating and the Computer 46: proceedings. Asling: International Society for Advancement in Language Technology, 18-20 November 2024; pp. 35‑46 (ISBN 978-2-9701733-2-8).

Download

Download full paper.

Posted on by Michael Farrell Posted in Academic publications

Abstract

This paper presents the findings of an anonymous online survey conducted in early 2024 on the use of generative artificial intelligence (GenAI) among professional translators. The survey revealed that 29.4% of professional translators incorporate GenAI into their workflow, in line with the results of another recent study. There is a significant association between the use of machine translation (MT) and GenAI, with MT users more likely to also use GenAI. Translators primarily use GenAI for writing-related tasks, such as finding contextual meanings, rephrasing sentences, shortening, summarizing and simplifying, and finding metaphors, synonyms and definitions. This suggests that GenAI enhances translation quality rather than productivity. Only 28.8% of GenAI users use it more than 50% of the time, implying that it is just one of several tools. ChatGPT is the most popular GenAI system, used by 80.8% of GenAI users, followed by Microsoft Copilot at 29.6%. However, only 20% of GenAI users pay for premium services. Many professional translators do not use GenAI (70.6%), often due to strong negative attitudes. GenAI’s role as an alternative to traditional MT followed by post-editing is less common than might be expected.

Published in

Translating and the Computer 46: proceedings. Asling: International Society for Advancement in Language Technology, 18-20 November 2024; pp. 23‑34 (ISBN 978-2-9701733-2-8).

Download

Download full paper.

Posted on by Michael Farrell Posted in Academic publications

Abstract

In the article ‘I, robot’ in the May-June Bulletin, I left you with two puzzles and a provocative question (‘Do these puzzles demonstrate that, given enough examples…, it is always possible to translate accurately between two languages that you do not know without understanding the meaning of the sentence you need to translate?’). The idea behind the puzzles was to get you thinking like machines, and the purpose of the provocative question was, well, to provoke.

Published in

ITI Bulletin, July-August 2024

Download

Download full article.

Posted on by Michael Farrell Posted in Academic publications

Abstract

You’ll have read it all over the place by now: machine translation (MT) and generative artificial intelligence (GenAI) work by identifying patterns and reproducing them. They don’t really understand language. They just replace tokens (words or subwords) with numeric vectors, perform various arithmetic operations on them and calculate probabilities at an amazing speed.

Published in

ITI Bulletin, May-June 2024

Download

Download full article.

Posted on by Michael Farrell Posted in Academic publications

Abstract

There is no hiding it: the original idea of machine translation (MT) was to replace human translators completely. Way back in the 1950s (yes, MT is that old – older actually), when the first computer translation systems came into being, some of the researchers working on them were predicting that translators – or at least technical translators – would be gone within a matter of years.

Published in

ITI Bulletin, September-October 2023

Download

Download full article.

Posted on by Michael Farrell Posted in Academic publications

Abstract

The experiment reported in this paper is a follow-up to one conducted in 2017/2018. The new experiment aimed to establish if the previously observed lexical impoverishment in machine translation post-editing (MTPE) has become more marked as technology has developed or if it has attenuated. This was done by focusing on two n-grams, which had been previously identified as MT markers, i.e., n-grams that give rise to translation solutions that occur with a higher frequency in MTPE than is natural in HT. The new findings suggest that lexical impoverishment in the two short texts examined has indeed diminished with DeepL Translator. The new experiment also considered possible syntactic differences, namely the number of text segments in the target text. However no significant difference waThe experiment reported in this paper is a follow-up to one conducted in 2017/2018. The new experiment aimed to establish if the previously observed lexical impoverishment in machine translation post-editing (MTPE) has become more marked as technology has developed or if it has attenuated. This was done by focusing on two n-grams, which had been previously identified as MT markers, i.e., n-grams that give rise to translation solutions that occur with a higher frequency in MTPE than is natural in HT. The new findings suggest that lexical impoverishment in the two short texts examined has indeed diminished with DeepL Translator. The new experiment also considered possible syntactic differences, namely the number of text segments in the target text. However no significant difference was observed. The participants were asked to complete a short questionnaire on how they went about their tasks. It emerged that it was helpful to consult the source language text while post-editing, and the original unedited raw output while self-revising, suggesting that monolingual MTPE of the two chosen texts would have been unwise. Despite not being given specific guidelines, the productivity of the post-editors increased. If the ISO 18587:2017 recommendation of using as much of the MT output as possible had been strictly followed, the MTPE would have been easier to distinguish from HT. If this can be taken to be generally true, it suggests that it is neither necessary nor advisable to follow this recommendation when lexical diversity is crucial for making the translation more engaging.

Published in

International Conference on Human-Informed Translation and Interpreting Technology (HiT-IT 2023): proceedings. Naples, Italy, 7-9 July 2023.

Download

Download full paper.

Posted on by Michael Farrell Posted in Academic publications

Abstract

This preliminary study consisted of two experiments. The first aimed to gauge the translation quality obtained from the free-plan version of ChatGPT in comparison with the free versions of DeepL Translator and Google Translate through human evaluation, and the second consisted of using the free-plan version of ChatGPT as an automatic post-editor of raw output from the pay-for version of DeepL Translator (both monolingual and bilingual full machine translation post-editing). The experiments were limited to a single language pair (from English to Italian) and only one text genre (Wikipedia articles). In the first experiment, DeepL Translator was judged to have performed best, Google Translate came second, and ChatGPT, last. In the second experiment, the free-plan version of ChatGPT equalled average human translation (HT) levels of lexical variety in automatic monolingual machine translation post-editing (MTPE) and exceeded average HT lexical variety levels in automatic bilingual MTPE. However, only one MT marker was considered, and the results of the post-editing were not quality-assessed for other features of MTPE that distinguish it from HT. It would therefore be unadvisable to generalize these findings at present. The author intends to carry out new translation experiments during the next academic year with ChatGPT Plus, instead of the free-plan version, both as an MT engine and as an automatic post-editor. The plan is to continue to evaluate the results manually and not automatically..

Published in

International Conference on Human-Informed Translation and Interpreting Technology (HiT-IT 2023): proceedings. Naples, Italy, 7-9 July 2023.

Download

Download full paper.

Posted on by Michael Farrell Posted in Academic publications

Abstract

This book looks at various aspects of machine translation, including the history of its technological advancement, quality evaluation, typical errors, techniques for improving its output, and how human translators can transform machine translation into a tool that can take some of the grind out of their work.

Published by

Amazon Digital Services LLC – KDP, 2023

Buy

On Amazon.

Posted on by Michael Farrell Posted in Academic publications

Abstract

The author conducted an anonymous online survey between 23 July and 21 October 2022 to gain insight into the proportion of translators that use machine translation (MT) in their translation workflow and the various ways they do. The results show that translators with more experience are less likely to accept MT post-editing (MTPE) assignments than their less experienced colleagues but are equally likely to use MT themselves in their translation work. Translators who deal with lower-resource languages are also less likely to accept MTPE jobs, but there is no such relationship regarding the use of MT in their own workflow. When left to their own devices, only 18.57% of the 69.54% of respondents that declared that they use MT while translating always or usually use it in the way the pioneers of MT envisaged, i.e., MTPE. Most either usually or always prefer to use MT in a whole range of other ways, including enabling MT functions in CAT tools and doing hybrid post-editing; using MT engines as if they were dictionaries; and using MT for inspiration. The vast majority of MT-users see MT as just another tool that their clients do not necessarily need to be informed about.

Published in

Translating and the Computer 44: proceedings. Asling: International Society for Advancement in Language Technology, 24-25 November 2022; pp. 49‑60 (ISBN 978-2-9701733-0-4).

Download

Download full paper.

Posted on by Michael Farrell Posted in Academic publications

Abstract

The author has conducted an experiment for two consecutive years with postgraduate university students in which half do an unaided human translation (HT) and the other half post-edit machine translation output (PEMT). Comparison of the texts produced shows – rather unsurprisingly – that post-editors faced with an acceptable solution tend not to edit it, even when often more than 60% of translators tackling the same text prefer an array of other different solutions. As a consequence, certain turns of phrase, expressions and choices of words occur with greater frequency in PEMT than in HT, making it theoretically possible to design tests to tell them apart. To verify this, the author successfully carried out
one such test on a small group of professional translators. This implies that PEMT may lack the variety and inventiveness of HT, and consequently may not actually reach the same standard. It is evident that the additional post-editing effort required to eliminate what are effectively MT markers is likely to nullify a great deal, if not all, of the time and cost-saving advantages of PEMT. However, the author argues that failure to eradicate these markers may eventually lead to lexical impoverishment of the target language.

Published in

Translating and the Computer 40: proceedings. Asling: International Society for Advancement in Language Technology, 15-16 November 2018; pp. 50‑59 (ISBN 978-2-9701095-5-6).

Download

Download full paper.
Alternative download.