MT Technology Research Articles

How external and internal factors affect translators’ source preferences – survey results

ABSTRACT Existing studies on translators’ information-seeking behavior (ISB) have either investigated translators in general or a selected group of translators (e.g. legal translators). The aim of this study was to investigate whether there were differences in the source preferences of translators according to selected factors, such as age, experience, direction of translation, use of CAT tools, post-editing practice, working language, and translation speed. The data were collected through a survey of 226 Polish professional translators. In order to examine the possible differences, we compared the answers of the selected translator subgroups (divided according to the above factors) to one question: ‘On a scale of 1–5, how often do you use the following sources when translating?’ The results were tested for statistical significance. It was found that the authorless online sources were used more often by younger and less experienced translators. MT and CAT technologies do not relieve translators of information research (no drastic differences in the frequency of source use). Translation into the mother tongue generally requires more source consultation than translation into the L2. Translators using technology rely more on non-traditional information sources.

College Translation Teaching in the Era of Artificial Intelligence: Challenges and Solutions

AI-powered translation technology has greatly improved machine translation quality, posing three major challenges to college translation teaching: students’ perceived anxiety towards translation learning, their dependency on MT technology to complete translation tasks, and traditional teaching assessment methods being rendered ineffective. In response to these challenges, this article proposes that translation instructors pass on to students knowledge concerning MT to reduce their anxiety and turn stress into motivation and help them develop scientifically sound strategies for the use of MT technology to further reduce their anxiety while mitigating their excessive dependency on it. Furthermore, classroom teaching should be reformed by introducing “flipped learning” to ensure that these strategies are implemented under the supervision and guidance of the instructors. The teaching assessment should also be changed to regulate students’ use of the strategies outside the classroom, while at the same time eliminating the adverse effects brought about by MT technology on teaching assessment.

Predicting energy efficiency and greenhouse gases reduction potential under different tillage management and farm size scenarios for winter wheat production

Winter wheat is the most popular crop in Europe and at the same time a major contributor to environmental pollution, as its production process is energy-intensive and emits high levels of greenhouse gas (GHG). Therefore, tackling the pollution of agroecosystems requires development of cleaner production technologies by integrating sustainable agricultural production practices that reduce the use of machinery, diesel and chemicals. The aim of this study was to determine the main indicators of energy and carbon emissions in winter wheat production taking into account different scenarios based on tillage technologies and farm sizes. Three sizes of farms (<10 ha, 10–100 ha, and >100 ha) and three different tillage systems (conventional (CT), minimal (MT) and no tillage (NT)) were used in this study in order to build up different scenarios.The cost of fuel, time, energy and carbon emissions for the size of the farm, the area of wheat and the amount of crop harvested were analysed based on the developed scenarios. The lowest total cost of fuel and working time for the total area of winter wheat (621 032 ha) were needed under the scenario SC 3 – 65.37 million L and 3.05 million h or 105.26 L ha−1 and 4.90 h ha−1, respectively. The best energy efficiency ratio (3.84), energy productivity ratio (0.26 kg MJ−1) and specific energy ratio (3.80 MJ kg−1) for winter wheat production were found in large farms (>100 ha) under no tillage. A comparison of the tillage and the farm size scenarios showed that by applying the scenario SC 3, where CT is 30–70%, MT is 20–50%, and NT is 10–20%, the carbon emissions for the total area of winter wheat could be reduced by 12.6 kilotonnes carbon emission, compared to the scenario SC 1, where CT is 50–100%, MT is 0–40%, and NT 0–10%. In conclusion, the modelled scenarios results suggested that implementation of NT and MT technologies using instead of carbon emission more efficiently mitigate environment pollution, decreasing fuel and energy consumption and working time in winter wheat production.

Moses, time, and crisis translation

Abstract Training translators to react to sudden emergencies is a challenge. This article presents the results of a training experiment testing the speed of acquisition of the skills necessary to operate the open-source Moses statistical machine translation (SMT) system. A task-based approach was used with trainee translators who had no experience working with MT technology. The experiment is a feasibility study to ascertain whether training on Moses SMT could be considered for long-lasting crisis scenarios. The article reports its findings in four sections. The first section discusses the research context in which ‘crisis translation’ is defined; the second section illustrates the rationale of the experiment; the third section looks at the results of the training experiment; and the fourth at the trainees’ perceptions of their learning processes. The conclusion reflects on the viability of using Moses and on the next phases needed to refine the findings of this first experiment.

Machine Translation in the Hands of Trainee Translators – an Empirical Study

Abstract Automated translation (machine translation, MT) is systematically gaining popularity among professional translators, who claim that editing MT output requires less time and effort than translating from scratch. MT technology is also offered in leading translator’s workstations, e.g., SDL Trados Studio, memoQ, Déjà Vu and Wordfast. Therefore, the dilemma arises: should MT be introduced into formal translation training? In order to answer this question, first, it is necessary to understand how trainee translators actually use MT. This study is an attempt to obtain this knowledge. The methodology applied in this investigation is text analysis. During the experiment sessions the students were asked to translate a legal text using MT tools, which in practice meant the post-editing of the MT raw output. The post-edited versions of the text underwent analysis in order to answer the following research questions: - What are the most typical errors contained in both French and English MT output? - How critical are the students towards the text generated by MT? - How perceptive are the students during the post-editing task? - Are they able to detect and correct errors using their knowledge and skills? The results of this study suggest that the post-editing of the MT raw output is as demanding for translation students as traditional translation, however, it requires a different set of skills, such as critical thinking and perceptiveness. Therefore, a special kind of training related to the effective use of MT technology should be implemented during translation classes.

Context as Achilles’ heel of translation technologies

The tools of translation memories and machine translation can be viewed as (not) being able to draw on different aspects of context that are relevant to a particular translation project, such as bilingual text, portions of a text, versions of a text, related text, or extralinguistic context. The aspects of context that the tools can indeed draw on and how well they do so highlight their most important benefits, whereas the aspects of context that the tools may fail to draw on reveal their weaknesses. In cases where, depending on the context, a piece of text being translated may have more than one meaning or a translation of a piece of text may be rendered in more than one way, the contexts the tools are able to draw on are critical. An analysis of different kinds of context in different applications highlights the circumstances in which users of TM and MT tools may risk accepting semantically and lexically undesirable output, as well as when TM tool users may risk inputting contextually inappropriate translations. Further, context is an important consideration when integrating MT output into TM tools. Although TM and MT technologies differ in how they produce translation output, end-users of both tool types may face similar contextual challenges.

Adoption behaviour of resource conservation technologies in paddy cultivation in Punjab

The study revealed that the adoption of direct seeded rice and mechanical transplanting of paddy technologies were adopted by large farmers having, higher off farm income and extension contacts with the officials of PAU and State Agriculture Department than their counterparts non-adopters. In the case of DSR adopter farmers, the yield was lowest at 6481 and 2962 kg per ha for common paddy and basmati, respectively and highest for MT at 7259 and 4713 kg per ha, respectively; yield varied between 6760 kg to 7109 kg for CM for common paddy and 3705 to 4115 kg for basmati. The reason of adoption of these technologies was save time and save water ranked first and second in DSR and solve labour problem ranked first in MT technology. Reasons for non-adoption were more weeds in case of DSR and costly machines and lack of awareness in MT. Farmers are willing to adopt mechanical transplanters provided the machinery is available with cooperative societies or on rent. More technical knowledge is required for the adoption of DSR technology. Among the problems faced in paddy cultivation, labour problem and high wage scored first and second rank, respectively in common paddy and volatility of price and only private procurement in basmati cultivation.

Translation project adaptation for MT-enhanced computer assisted translation

The effective integration of MT technology into computer-assisted translation tools is a challenging topic both for academic research and the translation industry. In particular, professional translators consider the ability of MT systems to adapt to the feedback provided by them to be crucial. In this paper, we propose an adaptation scheme to tune a statistical MT system to a translation project using small amounts of post-edited texts, like those generated by a single user in even just one day of work. The same scheme can be applied on a larger scale in order to focus general purpose models towards the specific domain of interest. We assess our method on two domains, namely information technology and legal, and four translation directions, from English to French, Italian, Spanish and German. The main outcome is that our adaptation strategy can be very effective provided that the seed data used for adaptation is ‘close enough’ to the remaining text to be translated; otherwise, MT quality neither improves nor worsens, thus showing the robustness of our method.

Why MT systems are still not widely used

Prof. Somers started his topics by raising the question: “Where will MT be in the next 20 years?” As his observations indicate: after years of development of various MT technologies, the second generation MT systems are not much better than the first generation systems, and many commercialized systems are still of the first generation architecture in essence. It is thus interesting to ask “WHY.” Prof. Somers attributes the major problems to the architecture of the second generation MT, where the stratificational paradigm was adopted. The stratificational paradigm results in inappropriate intermediate representations and inappropriate transfer and generation processes that are tightly bound to the source text. Structure preserving translation is common for essentially all such systems. The designers also tend to assume that source text contains enough information to permit translation under such an architecture. As such, the system designers tend to enhance the performance at the source side, including incorporating better semantic theories, dealing with several sentences, accounting the contextual knowledge, and so on. We generally agree with the observation that stratificational analysis tends to introduce some undesirable effects as mentioned by Prof. Somers. However, we believe that these effects are not the necessary consequences of a stratificational MT architecture. Technical and non-technical problems both contribute to the problems in MT development. So we would like to review the problems from a more general point of view by raising the following questions: