KantanMT.com was used in the course ‘Machine Translation and Post-editing‘, which was taught for the first time in the ‘Degree in Modern Languages Applied to Translation’ in UAH. English and Spanish were used as the main languages used during this course.
We caught up with Professor Cristina Toledo Báez, and in this post she describes her experience of using KantanMT during the course.
Professor Cristina Toledo Báez, University of Alcalá
KantanMT: Can you tell us how you used KantanMT during the course, and what were your key takeaways from using the platform?
‘Machine Translation and Post-editing‘ aims to provide a theoretical and practical approach to Machine Translation (MT) and Post-editing (PE). Thanks to this course, the students have acquired an overview of the history and evolution of MT as well as the key issues of using different types of MT and evaluation metrics. In addition, trainee translators were informed about the process of pre-editing, post-editing and controlled languages.
One of the specific competences developed during the course was to use different MT systems and compare their results. KantanMT was highly convenient in this regard because students could compare MT systems of different types (Google Translate, Systran, Bing, SDL BeGlobal, among others) and identify the strengths and weaknesses of each.
Students carried out two specific activities with KantanMT:
This activity consisted of three sub activities:
- Translating a text without training data.
- Translating the same text with parallel texts as training data.
- Translating the same text with a bilingual glossary and other xlsx documents (xlsx and brands.xlsx) as training data.
After each sub activity, students were asked to evaluate both the adequacy and fluency of each segment of each target text. They could easily see the improvement of the results thanks to the parallel texts and the bilingual glossary, and other xlsx documents.
For this activity, students were asked to compare the translation of the same text (a highly specialized text) without TM and then with TM, in order to see the difference between the two processes.
This activity consisted of the following four:
- Creating students’ own TMs: After having searched parallel texts of the source text in English, they manually translated the texts into Spanish. The target texts were the content of the TM.
- Translating the source text without training data, and selecting the English-Spanish IT stock engine.
- Translating the source text with the previously created TM as training data.
- Comparing the results of the second and third steps.
Students found this activity very interesting since they could see a real improvement in the target texts. They were a bit disappointed with the results in the second step, but they realised that the use of a tailored and suitable Translation Memory plays a crucial role as regards the effectiveness of a MT system.
These activities helped me as lecturer to introduce a positive outlook towards MT. Translation students, as linguists, are very picky about the results of MT and, while they are interested in MT, they are usually reluctant to use MT on a regular basis.
However, thanks to the course, they learned that MT, especially KantanMT, might be of great help in the translation process. The translation process needs to be managed effectively and the engines should be trained expertly, in order to get the best results from Machine Translation.
About Professor Cristina Toledo Báez
Cristina Toledo Báez is a Lecturer at the Department of Modern Philology in the University of Alcalá de Henares (Spain). She has a B.A. in Translation and Interpreting from the University of Malaga (2004) and earned her PhD in Translation and Interpreting with ‘Doctor Europeus’ mention in 2009.
She has been awarded several predoctoral and postdoctoral research grants, including ones from Dickinson College (USA), Centre Privé de Langues (France) and University of Wolverhampton (United Kingdom). She has also served as invited lecturer at several Master’s Programmes (University of Córdoba and University of Alcalá), and has taught internationally – (Erasmus Mundus) – in a Master’s Programme on Natural Language Processing (University of Wolverhampton).
Her research fields cover translation technologies, Computational Linguistics, technical and legal translation and translation assessment. Her research results have been disseminated in a number of international academic conferences and publications.
About the University of Alcalá
The University of Alcalá is a UNESCO world heritage site and one of the longest standing European universities which dates back to May 20, 1293. Today the University of Alcalá is a modern institution which covers all fields of knowledge, from humanities to engineering, and from social science to experimental and biomedical science. An estimated 28,000 students are currently pursuing their degrees at the University of Alcalá.
Note from the KantanMT Team
For more information on the KantanMT University Partner Program, download our brochure (PDF) or if you used our platform during your studies and looking for a job, please contact us, at firstname.lastname@example.org.