Machine Translation and Translator Training

In this era of globalisation, translation technologies have become more popular in daily communication, the education sector, and the translation industry. It is observed that there is a prevalent use of machine translation (MT) among translation learners. The proper use versus abuse of MT can be a critical issue regarding its role in and impact on translation teaching. This exploratory study aims at investigating learners’ and instructors’ knowledge of MT, experience in MT use, perceived MT quality, ethics of MT use, and the perceived relationship between MT and translation training, in order to figure out the usefulness of MT in translation competence acquisition and the necessity of MT training. To this end, we conducted surveys and semi-structured interviews and found that the influence of MT in translation competence acquisition is determined by the properties of MT and learners’ quality. MT is particularly helpful in gaining lexical knowledge and knowledge to ensure translation efficiency, but not in bicultural knowledge. However, such usefulness builds on learners’ language proficiency, analytic ability, and learning motivation. In light of the findings, issues including the sustainability of MT from ethical and linguistic perspectives, and the potential and proper use of MT to inform translator training, are discussed.

Social justice refers to the equitable distribution of resources, opportunities, and rights in society. Social justice frameworks acknowledge that structural inequalities can hinder accessibility to education and that the use of and recourse to technology is not always ethical or equitable. As translator trainers with more than two decades of experience in higher education, we reflect on the nexus between technology, translator training, ethics, and social justice, and put forward a list of strategies with which to humanize translator training or education and professional practice. We focus on Canada and draw from both a literature review using the Translation Studies Bibliography which shows that research on the subject of translator training or education and social justice is currently underdeveloped. We also draw from two media scans (conducted from November 2022 to October 2023) on the topics of higher education, translator training, social justice, and technological disruption and related digital divides, with a specific focus on machine translation and artificial intelligence. Along with other demolinguistic data from the 2021 Canadian Census, the review and scans contextualize the list of pedagogical recommendations and strategies we propose. We adopt the position that social justice should take precedence in the way we think about translator training and develop curricula rather than allowing market forces and the tech industry to determine training priorities and objectives. Artificial intelligence and other new(er) technologies can have pedagogical merit in higher education and in translator training, but it is imperative that we consider how and when to use the tools and to focus on issues that go beyond plagiarism and student surveillance. We therefore argue in favour of human, humane, and humanising translation in and beyond Canada and this means advocating and developing pedagogical strategies and curricula that align with the ethos of social justice.

With the advent of the neural paradigm, machine translation has made another leap in quality. As a result, its use by trainee translators has increased considerably, which cannot be disregarded in translation pedagogy. However, since legal texts have features that pose major challenges to machine translation, the question arises as to what extent machine translation is now capable of translating legal texts or at least certain types of legal text into another legal language well enough so that the post-editing effort is limited, and, consequently, whether a targeted use in translation pedagogy can be considered. In order to answer this question, DeepL Translator, a machine translation system, and MateCat, a CAT system that integrates machine translation, were tested. The test, undertaken at different times and without specific translation memories, provided for the translation of several legal texts of different types utilising both systems, and was followed by systematisation of errors and evaluation of translation results. The evaluation was carried out according to the following criteria: 1) comprehensibility and meaningfulness of the target text; and 2) correspondence between source and target text in consideration of the specific translation situation. Overall, the results are considered insufficient to give post-editing of machine-translated legal texts a bigger place in translation pedagogy. As the evaluation of the correspondence between source and target text was fundamentally worse than with regard to the meaningfulness of the target text, translation pedagogy should respond by raising awareness about differences between machine translation output and human translation in this field, and by improving translation approach and strengthening legal expertise.

This paper examines the transformation of translation and translator training within the field of Buddhist Studies amid the rapid proliferation of Artificial Intelligence (AI) and Machine Translation (MT) technologies. While Neural Machine Translation systems offer unprecedented efficiency and accessibility for multilingual Buddhist textual corpora, they frequently fall short in capturing the intricate philosophical nuances, doctrinal subtleties, and culturally embedded meanings that characterise Buddhist philosophical literature. The distinctive challenges of Buddhist translation – with its specialised terminology, complex ontological frameworks, and hermeneutic traditions – reveal both the limitations of current AI approaches and the enduring necessity of human interpretive expertise. This study proposes a framework that neither uncritically embraces technological advances nor retreats into traditionalist resistance but instead cultivates a reflective integration of AI tools while reinforcing essential human competencies. By prioritising sustained engagement with untranslated source materials and developing distinctly human capacities such as textual interpretation and ethical discernment, this framework contributes to the training of translators who are simultaneously technologically adept and critically grounded. The paper ultimately advocates for an integrative approach that ensures Buddhist translation remains a deeply contemplative and humanistic practice – one that preserves the wisdom traditions it transmits – even as it navigates an increasingly digitally mediated landscape.

The web as corpus is considered helpful in translation training, although the literature argues that Internet data might be too chaotic and badly organised.For this reason, web concordancers are often resorted to in order to carry out web searches as they are more linguistically oriented and user-friendly.At the same time, machine translation (MT) has made huge strides forward in the last decades, and scholars have increasingly focused on MT in technical translation training.However, so far little has been investigated on the quality of MT vis--vis translations via web concordancers.This paper wishes to bridge this gap and aims at exploring whether and to what extent the web-based translation of a medical article carried out by consulting the Leeds web-concordancer is qualitatively more satisfactory than a translation using MT, or vice-versa.To this aim, a classroom observation study is undertaken with 9 Italian final-year bachelor's students.The students' Leeds-driven translations are compared with the automatic translations performed by two MT tools (DeepL and MateCaT).The paper findings highlight the overall correctness of the machine-translated sentence structures as well as of the related grammar.However, MT presents lexical and collocational issues mostly due to infrequent word usages.By contrast, the web-driven students' translations present a wider variety of lexical and collocational choices, despite some grammatical shortcomings.The paper suggests further research to investigate whether more training in translations based on web concordancers can help students feel at ease with online searches and reduce inaccuracies.

The aim of this article is to investigate translation trainees’ perceived difficulties and benefits of a post-editing oriented neural machine translation (NMT) error annotation and quality evaluation task which was carried out for the language pair English-Turkish and for two separate text types and domains, i.e., environmental blogs and movie/TV reviews, within the scope of an MA course on translation quality standards. The data to be analyzed were collected through a semi-structured questionnaire which was given to the trainees attending the course after the completion of the task. The questionnaire was prepared with the aim of understanding perceived difficulties and benefits of the task. Analysis of the answers revealed that most of the trainees were in the opinion that the task was difficult. Majority of the trainees also believed that the task was beneficial and enabled them to feel empowered to make decisions on human translation quality evaluation of machine translation (MT) and to carry out error annotation and post-editing activities in the future. According to a significant number of trainees, error annotation facilitated their post-editing process and reduced the effort in post-editing. Enhanced understanding of MT error annotation and enhanced ability to perform post-editing were the significant benefits stated by the trainees. The difficulties were associated with being introduced and assigned to perform tasks they were not familiar with. Yet, as displayed in the answers to the questionnaire, a considerable majority of the trainees were positive about the learning experience. The results have also shown that integration of MT-related activities into translator training with a focus on the empowerment of the human translator has its difficulties and benefits also for translation trainers. While the difficulties for the trainers concern the decisions on the design, implementation, and planning of the task and the responsibility to carry out the task in constant interaction and collaboration with the trainees, the benefits are the sense of fulfillment and enrichment brought by positive feedback from the trainees and the discovery of the fact that the so-called ‘teaching experience’ becomes a ‘learning experience’ for trainers as well as trainees.

This article addresses the challenges and opportunities posed by the use of machine translation (MT) and post-editing (PE) in the training of legal translators in the Spanish-Italian combination. The aim is to find out whether students who do not have specific training in MT and PE, but a sufficiently developed translation competence, are able to successfully apply the MT and PE concepts learned during a short workshop in their Specialized Translation course and, as a result, produce a satisfactory post-edited translation of acceptable quality. The results show that students have an inward critical attitude towards the use of MT, so they are not influenced by the MT output. Moreover, they have adequate skills to immediately detect not only most of the errors related to concordance and lack of logic, but also the more hidden ones that require a subsequent revision.

The article analyzes errors of web translators Google Translate and Yandex Translate in the YakutRussian language pair in three texts of different genres. The Yakut language was introduced to web translators several years ago, and their mainstream use is at an early stage. Analyzing machine translation in the Yakut-Russian language pair can contribute to the development of recommendations for the effective training of both machine systems and human translators, primarily in machine translation post-editing. Thus, the classification of machine translation errors in this article is relevant in the current development stage of the Yakut-Russian theory and practice. For the experiment’s purposes, the following texts were translated via the specified services: a) an excerpt from a scientific paper in Russian; b) an excerpt from a translation of the Bible into Yakut (reverse translation into Russian); c) a poem in Yakut on morality. The classification of machine translation errors in this language pair is based on typologies of other language pairs: a) lexicosemantic and lexico-stylistic errors; b) errors in standards and usage; c) grammar errors; d) spelling errors; e) punctuation errors. The nature of errors is described and the reasons for their occurrence are explained. The main causes of errors are: a) the difference in the volume of initial data in the machine systems for analyzing the language of the text being translated; b) the special nature of similarities and differences of contacting languages, associated with polysemanticity and grammar gaps. Upon comparing the quality of translations of both services, their strengths and weaknesses in translation with the Yakut language are noted. The article concludes that the linguistic databases of these services need improvements, and human translators who work with the Yakut-Russian language pair need to acquire corresponding expertise.

Review of Pym & Hao (2025): How to Augment Language Skills: Generative AI and Machine Translation in Language Learning and Translator Training

En los últimos cinco años, la incorporación de la traducción automática como una herramienta más en el proceso de traducción ha llevado a la comunidad docente a replantearse las competencias y habilidades que tradicionalmente han formado parte del currículum formativo del traductor. En este artículo se afronta este desafío desde la concepción del proceso de traducción como un ciclo iterativo en el que la persona que traduce administra el funcionamiento de la máquina que traduce.

The active development of international tourism contributes to a steady increase in demand for high-quality translation of tourist texts explaining the importance of research on the common problems of their translation into foreign languages. The research subject is the specific features of translating tourist texts from Russian into English. The research objective is to identify the key features of tourist text translation. The article examines the stylistic features of tourism discourse texts which are the most difficult for translation from Russian into English. A definition of the international tourism discourse is given, popular types of tourist texts are listed, and common ways of translating these texts into English are pointed out: creative translation, descriptive translation and translation by searching for functional analogues. The author used the methods of generalization, content analysis, systematization of data from recent studies on this topic and comparative analysis of Russian and English tourist texts. The research novelty lies in the systematization of the works of Russian linguists devoted to the problems of translating tourist texts and in the proposed ways to eliminate possible mistakes. The key problems of translating tourist texts include underestimation of translation in tourism and translation by non-specialists often using machine translation, mistakes in translating terminology, ignoring the genre of the source text and insufficient adaptation of the translation text to the target audience of another culture. It is proposed to create new Russian-English and English-Russian dictionaries of tourist terms, train translators in special courses for translating tourist texts, providing them with knowledge of key concepts in tourism at least in their native language and experience in translating these concepts into a foreign language, and teach students majoring in linguistics to translate tourist texts. It is noted that translation methods and strategies depend on the text genre.

The efficient use of machine translation (MT) training data is being revolutionized on account of the application of advanced data selection techniques. These techniques involve sentence extraction from broad domains and adaption for MTs of in-domain data. In this research, we attempt to improve in-domain data adaptation methodologies. We focus on three techniques to select sentences for analysis. The first technique is term frequency–inverse document frequency, which originated from information retrieval (IR). The second method, cited in language modeling literature, is a perplexity-based approach. The third method is a unique concept, the Levenshtein distance, which we discuss herein. We propose an effective combination of the three data selection techniques that are applied at the corpus level. The results of this study revealed that the individual techniques are not particularly successful in practical applications. However, multilingual resources and a combination-based IR methodology were found to be an effective approach.

Machine translation (MT) technology has developed for many decades and can provide high quality reference translation for certain fields. The main MT methods, on-line MT systems and their defects are introduced by analyzing the necessity of combing modern information technology with translation teaching. The application of the online MT systems in translation teaching is discussed in the aspect of the objectives and patterns. To apply computer-aided techniques into translation teaching will promote the reform of translation teaching and meet the social requirements for the training of translators.

When training multilingual machine translation (MT) models that can translate to/from multiple languages, we are faced with imbalanced training sets: some languages have much more training data than others. Standard practice is to up-sample less resourced languages to increase representation, and the degree of up-sampling has a large effect on the overall performance. In this paper, we propose a method that instead automatically learns how to weight training data through a data scorer that is optimized to maximize performance on all test languages. Experiments on two sets of languages under both one-to-many and manyto-one MT settings show our method not only consistently outperforms heuristic baselines in terms of average performance, but also offers flexible control over the performance of which languages are optimized. 1

We present a framework for statistical machine translation of natural languages based on direct maximum entropy models, which contains the widely used source-channel approach as a special case. All knowledge sources are treated as feature functions, which depend on the source language sentence, the target language sentence and possible hidden variables. This approach allows a baseline machine translation system to be extended easily by adding new feature functions. We show that a baseline statistical machine translation system is significantly improved using this approach.