The Swedish Association of Professional Translators reports positively about the ELRC workshop in Stockholm

2018-12-19

Original article written by Charlotte Sjögren, SFÖ (Swedish Association of Professional Translators)

Translation from Swedish by Simon Dahlberg, Språkrådet

 

ELRC on Data Hunt

The Automated Translation platform of the European Union needs to be fed with big quantities of data to perform satisfactorily, especially regarding terminology. This is why the European Commission has started ELRC, an extensive initiative aimed at identifying and collecting language resources that are relevant for national public services, administrations and governmental institutions.

The European Language Resource Coordination (ELRC) project also includes the organization of workshops in all EU countries to showcase the need for collecting language data for the Automated Translation platform eTranslation. The second Swedish workshop of this kind was held in Stockholm on May 23rd and arranged by the Språkrådet (Language Council of Sweden). The first workshop was held in 2016, see Facköversättaren 2016/2, page 6-7.

Rickard Domeij from Språkrådet (Language Council of Sweden) started the workshop at Europahuset with a short introduction on the ELRC project and the efforts to get public services and other organisations involved. The rest of the presentations before lunch were mostly dedicated to different Swedish public agencies that are using eTranslation today, among others Livsmedelsverket (The Swedish National Food Agency) and Tillväxtverket (The Swedish Agency for Economic and Regional Growth), that runs the website www.verksamt.se. The demand for e-services to be available in different EU languages and other immigrant languages is big and this concerns www.verksamt.se in particular. Other examples of areas where the need is big are consumer complaints and procurement basis.

 

Comparison with Google Translate

Naturally, several of the speakers brought up the benefit and the point of using the eTranslation service. It was particularly interesting to notice that in this respect, they seemed to compare eTranslation with using other free of charge services like Google Translate rather than with using human translators. Kristina Lagestrand Sjölin from Livsmedelsverket had a very positive experience with eTranslation. Nonetheless, she pointed out that the service should not replace human translators, but instead could be used for making an initial evaluation of a text to see what competence needs to be hired to have it translated. She also made clear that this service requires knowledge and good judgement from the users.

The first session concluded with a panel session with the recent speakers on the advantages and disadvantages of the eTranslation service. Terminologist Henrik Nilsson asked what consequences the panelists expect for the translation business and the translators (thank you Henrik!). The answer from the panelists was that there is still so much to be done that human translators will be needed, while there are actually some language pairs that we have no human translators for. As the representative from Livsmedelsverket mentioned earlier on, machine translation can help the contracting authority to evaluate a source text before ordering a human translation, hence enabling them to make an informed decision compliant with the requirements of the translation.

 

Collecting Language Data in Practice

The second session, covering the afternoon, focused on the practical work with collecting language data, meaning translation memories and multilingual documents. Among the speakers were Anne-Marie Colliander Lind who represented Språkföretagen with a presentation on the role of the translation companies in disseminating language data – the presentation that was maybe the most interesting to us translators. Anne-Marie started by explaining that nowadays, machine translation is able to translate more data in one day than all human translators in the world could do together in one year – a calculation based on statistics from Google, among others. 
This simple fact was important to note and illustrated how integrated MT already is in our reality; it also gives a good indication about the amount of text we are talking of when we require translation.

Like previous speakers, Anne-Marie highlighted the fact that machine translation increases the need for human translation. On top of this, a need for a new type of competence appears as machine translation becomes increasingly better – since the mistakes that are still made by MT systems really become harder to spot. She also expressed a degree of nervousness about the quality of the data that is fed into the eTranslation system and reminded that here, there is an integrity aspect, too: there are documents that should never be made public and someone has to be responsible for not letting them reach the system.

Furthermore, Anne-Marie brought up the undoubtedly interesting aspect of who will get paid if eTranslation becomes a natural part of the translation work of public authorities especially if this also includes feeding the data back into the system to improve it. In this case, this is something that has to be included as a standard in the procurement basis. In that case, the whole price model would need changes. These changes need to give more consideration to the reviewing step, for example.

 

Question to the Translators

The workshop was concluded by Rickard Domeij opening up for a dialogue on the collection of language data and participation in the ELRC project. He also asked the translators in the audience whether they see any benefit in using MT in their work, to which many of the translators answered yes, but with certain reservation (not for all kinds of jobs and not for all clients). However, it did not sound like eTranslation will be available for individual translators in the near future, due to a lacking server capacity, among other things. The subcontractors of some public agencies can get access, but it will be decided on a case-by-case basis.

 

Editor’s note:

eTranslation is only granted to public administration bodies. If you are unsure whether your institution is eligible to use eTranslation, please contact: help@cefat-tools-services.eu