Learning to translate: in the age of AI, do we still need to do theme and version exercises?

Share:

“If translation tools work miracles, why practice transposing texts from one language to another?” students ask themselves. Far from being trivial, this learning allows us to become truly aware of the subtleties and functioning of the language we are learning.


“DeepL works miracles, so why should I learn to translate?” asks a student in a translation course .

Thematic courses (translating a text from French into the language studied) and version courses (translating a text from a language studied into French) occupy an important place in French university education , in language degrees up to the second cycle competitive examinations.

Already in high school, students discover the subtleties inherent in moving from one language to another and, as part of teaching ancient languages ​​and cultures , they regularly practice translation.

In a general context where neural translation , powered by artificial intelligence, increasingly rivals human translation, the interest of this learning can give rise to questioning.

Course objectives to be clarified

Without addressing the historical debates that mark the evolution of translation as a pedagogy and of translation studies as a discipline, it should be noted that numerous studies have highlighted the “strange status of translation at university, often based on a poor understanding of its interest”, in the words of academic Fayza El Qasem .

Although teachers set specific learning objectives for theme and version courses, many students still struggle to understand their purpose. It is no secret that the phrase “Be careful, do not use an automatic translator!” has long been part of the instructions given to students, under the pretext that the quality of the translation is deplorable. Only dictionaries were allowed.

Today, can teachers of theme and version courses still avoid automatic translators despite their obvious improvement  ?

It should be remembered that the skills targeted in the theme and translation courses range from understanding a text and the functioning of languages ​​through linguistic analyses (grammar, vocabulary, translation processes ) to the translation of a paragraph, as a means of assessing language skills. The extract to be translated is generally from the literary or journalistic register and only rarely opens up to so-called pragmatic translation (everyday, professional texts).

In recent years, there has been no shortage of criticism. Recent studies have emphasized the growing interest in integrating technological tools into the teaching of translation .

Beyond translation, understanding and analyzing linguistic subtleties

The theme and version courses , however, are real linguistic laboratories. They involve in-depth analysis of a source text by inviting learners to dissect the linguistic and extralinguistic structures that translation software still struggles to grasp. In theme or version , it is not simply a matter of translating isolated segments, but of grasping their overall meaning, identifying figures of speech or even tone, etc.

Each level of analysis allows for an “acceptable” translation, of course, but above all it promotes a fine manipulation of the language, transposable to other contexts. This is the case for the translation of syntactic ambiguities or wordplay, humor or even neologisms .

The work of linguist Natalie Kübler and her colleagues in specialist languages ​​further demonstrates “the limits of these [machine translation] systems, particularly in the processing of complex noun phrases, both at the level of the phrase itself (possible variations in the juxtaposition of constituents, identification of coordinated constituents, etc.) and at the level of the text (instability of translation choices, adequate identification of the specialist field, etc.)”.

Machine translation, as efficient as it is, remains imperfect despite its progress. While it is effective for simple and literal translations, it often struggles to capture essential contextual nuances. This is why the translation of idiomatic expressions (for example “the carrots are cooked”, “the turkeys of the farce”), instructions for use or certain advertisements sometimes produces renderings far removed from the original meaning, to the point of producing false meanings.

Theme and version courses can be an opportunity to raise awareness and support students towards a reasoned use of machine translation. It is also a space to practice identifying and correcting the pitfalls mentioned above, while strengthening the understanding of the linguistic systems of the languages ​​studied. In the long term, this analytical capacity is of fundamental importance in their future professional context. Communicators, journalists, translators or language teachers, these students will often be required to navigate between various sources of information, sometimes tainted with deepfakes to justify the possible failures that machine translators generate.

Strengthening intercultural understanding

In addition to the reinforcement of linguistic elements that can be supported by theme and version courses , taking into account cultural specificities constitutes a learning element in its own right, particularly because translation is, among other things, a means of mediation between two cultures .

Moreover, when the Canadian translation expert Jean Delisle speaks of the cultural dimension of translation, he uses the metaphor of the “hydra with a hundred thousand heads” to emphasize its multiple and dynamic nature.

The ability to detect and understand cultural differences thus helps to prevent misunderstandings that can so easily arise in a foreign language and that are sometimes already present in the source language. The humorous excerpt from Juste Leblanc with the confusions between the adverb “juste” and the first name “Juste”, in the film Le Dîner de cons, illustrates this.

Finally, by moving away from the sometimes artificial nature of the practices adopted in the teaching of the theme and the version and by taking into account the socio-economic evolution of society combined with the use of artificial intelligence, these courses could (re)gain the initial interest of the language learning path. Updated, they would be able to offer students a clearer understanding of the requirements of language-related professions, professions that today require specific human skills, complementary but distinct from those of machines.

The question that arises today is no longer “Why teach the theme and the version in the age of AI?”, but rather “How?” “Letting the relationship with AI systems exist, where everywhere we often only talk about their uses, also means leaving room for the indeterminate dimension of their inhuman intelligence, as we can read in the book by Apolline Guillot, Miguel Benasayag and Gilles Dowek entitled Is AI a chance  ?

Author Bio: Anissa Hamza-Jamann is a Lecturer in Language Sciences at the University of Lorraine

Tags: