DOCUMENTS AND MATERIAL

User needs
Learner Translator Corpus
Online courses
Collaborative Platform

User needs

The EU-sponsored LETRAC survey and the industry surveys by LISA recommend adapting translator curricula to the globalisation process as well as including language technology tools and language management in the skills base. Therefore, one of the project objectives of MeLLANGE was to improve new skills and competences of trainee and professional translators by providing access to continuing professional development (CPD) on an e-learning platform. With this objective in mind, the MeLLANGE team conducted a paper-based survey from April to May 2005 and a web-based survey from December 2005 to January 2006 covering internet search techniques, corpora and e-learning preferences. Over 1,000 people responded to the questionnaire. To see the results, click here.

Learner Translator Corpus

One of the objectives of the MeLLANGE project was to create a multilingual annotated aligned corpus of translations produced by trainee and professional translators. This Learner Translator Corpus (LTC) was enriched with stand-off linguistic annotations (part-of-speech tags and lemma information), as well as error annotations (using the error typology designed by the consortium). The purpose of the error typology is not to encode any quality evaluation but rather to identify and classify types of errors, since the corpus is intended to contribute to the formative process of training, rather than playing any role in assessment. Additional information is recorded with the files: the key properties of the source text and meta-data concerning the translator. These pieces of information, though not specific to the error typology, may correlate with different types of errors.

The partnership has selected four text types - legal, technical, administrative, and journalistic - and has published translation kits in up to 10 languages - ca, de, el, en, es, fi, fr, it, nl, and pt - the majority of which contain a translation brief, the source text, and a reference text with additional background information.

A translation collection mechanism was designed and localised in the project languages in order to capture metadata about the translation and its author alongside the translated text.

There has been an enthusiastic response to our initiative, and we have gathered 429 translations into the trainees' mother tongues - 63 into Catalan, 61 into English, 76 into French, 17 into German, 147 into Italian, 1 into Slovak, 61 into Spanish, and 3 into Romanian . Out of these, a total of 232 were annotated using the MeLLANGE translation error typology: 12 into Catalan, 54 into English, 48 into French, 13 into German, 56 into Italian and 49 into Spanish.

We have already conducted two multilingual, multi-site MeLLANGE annotation exercises involving a total of 59 translations in the following language pairs: de-ca, de-es, en-it, fr-en, fr-it, en-fr, and en-de. It gave the partnership the opportunity to share their experience regarding the annotation process, and consequently fine-tune it.

We are already able to obtain interesting and useful statistics based on the current data. One such example is the difference across the target languages of the project of the three most frequent errors overall: incorrect terminology and lexis (LA-TL-IN), distortion of the original content (TR-DI), and inconsistent terminology and lexis within the target text (LA-TL-IT).

LTC errors by language

By April 2007, we finalised the annotation of a total of 360 translations into the project languages: ca, de, en, es, fr, and it, which allowed us to produce more revealing statistics. A complex and flexible query mechanism was put in place , enabling translation trainers and trainees to analyse translation errors, as well as investigate translation choices which have received significantly less attention from the research world so far. Moreover, the data extracted from the LTC can be used to produce data-driven learning materials - i.e. materials developed around real texts produced by students, which better reflect the problems faced by trainees.

For further information on the LTC, please consult the LTC site at the University of Leeds.

Online Courses

E-Learning content was developed on Moodle, an open-source SCORM-compliant platform, which was chosen for its sound pedagogical background as well as the features it offers. E-Learning content, compliant with the LOM standard, was developed. The partnership has been worked on designing a general methodology for e-learning content development in translation and translation technology. At the same time, course outlines have been designed for a choice of topics, which are currently:

specialised translation
corpus use in translation
translation memory
machine translation
localisation
markup languages
project management
information management
terminology

The partners have focussed on creating user-friendly, interactive courses, thus fully exploiting the possibilities the platform offers, and on creating content appropriate to the user needs.

The developed courses were tested at a dissemination workshop organised in Vienna in May 2006, where participants from several Eastern European countries were introduced to the methodology developed by MeLLANGE. Participants to this workshop will continue to provide feedback as they test content developed by MeLLANGE at their home universities. The content was also presented at the Dissemination workshop on translation technologies and e-learning development in Paris, France, April 14, 2007. Online content was constantly improved according to the feedback received, and best practices for the creation of online content were developed.

If you wish to see the courses proper, please visit the MeLLANGE Moodle platform .
The MeLLANGE courses as well as their source files (either eXe or SCORM packages) may be downloaded from the MeLLANGE Plone platform.

Collaborative Platform

The collaborative platform used by MeLLANGE is Plone, an open-source content-management system with strong multilingual support. Maintained by Barcelona, one of the project partners, Plone for MeLLANGE is used as a collaborative platform for the internal development of the project, but also as a portal for collecting translations and other information from outside contributors.