Teaching translation and interpreting in the age of neural machine translation

When DeepL and Google Translate launched their neural machine translation service in late 2016, it marked the beginning of a new era.

Neural machine translation (NMT) now significantly outperforms statistical machine translation (SMT) which itself had earlier displaced rule-based machine translation (RBMT).

Raw NMT output increasingly manages to meet the end user’s expectations in terms of translation quality, even though it depends on a number of variables and still relies on human post-edition for a more polished result. While both statistical and neural machine translation are corpus-based, the latter is nonetheless a new breed of corpus-based machine translation as it exploits text corpora through deep learning algorithms. The meaning of each word is encoded by the neural network into a real-value vector – or word embedding (Forcada 2017) - resembling a semantic representation (Koehn 2020: 108). Hence, NMT tends to generate paraphrases (Neubig, Morishita & Nakamura 2015). While the resulting translation product is typically more fluent, it sometimes lacks terminological accuracy (Forcada 2017). Diverse studies have also shown that omissions and additions of content are more prominent in neural machine translations than in statistical machine translations (Castilho et al. 2017), a phenomenon which has been dubbed as “neurobabble” (Hasler 2018).

In recent years, NMT has become the mainstream translation method and a central theme in translation studies and translation pedagogy.

It has even made inroads in the domain of literary translation (Hansen 2021) and interpreting (Defrancq & Fantinuoli 2021), where there is a growing interest in exploring the potential of MT, artificial intelligence and computer tools more generally.

The ISTI-Cooremans School of Translation and Interpreting is moving to the ULB Solbosch campus and wishes to mark the event by organising a conference on International Translation Day, in honour of St. Jerome, that will bring together researchers, trainers and professionals.

Translation and interpreting pedagogy is at the cross-roads of research and practice.

It is important to keep track of the latest developments in the domain of NMT research and its practical applications, to prepare students for the current and future job market. In doing so, it is essential to rise above widely held popular beliefs (which have no sound scientific basis and are often supported by supposed examples of ‘good’ and ‘bad’ translations) about the pros and cons of artificial intelligence in general and neural machine translation in particular. Translator and interpreter training must indeed be research-based and stimulate critical thinking about the state-of-the-art technological advancements and their impact on the occupational status, visibility, rights and income of professional translators.

This conference wishes to provide a platform for researchers, trainers, students and professional translators and interpreters in the form of plenaries, presentations and round table discussions. 

We welcome contributions that explore the following topics:

MT
and
the translation profession​

• _{Recent developments}
• _{The impact of MT on professional translation (copy rights, fees and expected skill set)}
• _{Quality assessment, post-edition efforts and error typology}
• _{The use of NMT for different types of translation such as localisation and audio-visual adaptation}

Corpus-based MT
 

• _{Developments in corpus-driven machine translation and interpreting}
• _{The impact of machine translation on quality, bias and copy rights}
• _{Machine translation and controlled languages}
• _{Machine translation and translation memories}
• _{Corpora of low-resource languages}

Translation
and
interpreting pedagogy

• _{Integrating NMT, post-edition and revision into translator training: methods, tools, timing, practical use, training MT systems and keeping up with the latest developments}
• _{MT/PE learning platforms}  
• _{Quality assessment and indicators of post-editing effort}

We invite researchers, professional translators and interpreters, translator and interpreter trainers to share their experience and ideas in the form of a 20-minute presentation followed by a discussion. 
 

Proposals

• Please submit anonymized abstracts of approximately 500 words, including relevant references (not included in the word count) in English or French, in .pdf, .docx, .rtf ou.odt, format
• Add 5 topic keywords in English or French
• Send to pascaline.merten@ulb.be
• Deadline for submission : 28 March 2022 

Notification of acceptance

• 2 May 2022

Deadline for submission of final extended abstract

• 2000 words
• 15 August 2022

Schedule

 

Scientific comitee

Pascaline MERTEN Présidente	Université libre de Bruxelles
Dragos CIOBANU	Université de Vienne
Guillaume DENEUFBOURG	Université de Mons
Damien HANSEN	Université de Liège
Koens KERREMANS	Vrije Universiteit Brussel
Nancy MATIS	Université libre de Bruxelles
Marianne STARLANDER	Université de Genève
Alina SECARA	Université de Vienne