The GoTriple platform as a tool to support scientific translation
Last February, the city of Bonn hosted the final conference of the TRIPLE project, funded under the European Horizon 2020 programme in order to develop a multilingual and multicultural discovery platform for the humanities and social sciences. Coordinated by the French National Centre for Scientific Research with the participation of 21 partners from 13 European countries, the TRIPLE project is coming to an end and the GoTriple discovery platform is now available online.
The scientific coordinator of the Translations and Open Science project, funded by the French National Fund for Open Science and coordinated by OPERAS, has tested the GoTriple platform to see how it can be of help to scientific translators and hence support multilingualism in scholarly communication. The purpose of this article is to introduce some translation-related use cases of GoTriple and to identify possible synergies between the TRIPLE and the Translations and Open Science projects.
GoTriple: why a multilingual and multicultural discovery platform?
The GoTriple platform is a single, centralised point of access to open and multilingual scientific resources in the humanities and social sciences. Its model was inspired by the Isidore search engine, developed by the French research infrastructure Huma-Num at the Centre national de la recherche scientifique (CNRS – National Centre for Scientific Research). The platform can be searched in several languages, including (but not limited to) eleven official languages: Croatian, English, French, German, Greek, Italian, Polish, Portuguese, Slovenian, Spanish, and Ukrainian. The multilingual results are aggregated from several national and disciplinary scientific repositories. GoTriple also offers a set of innovative tools and services, such as advanced filters and content recommendations based on searches, document annotation and search result visualisation features, as well as peer networking and partnering opportunities.
The platform is intended to serve several ambitious purposes: improving the visibility and the discoverability of publications in the humanities and social sciences within the scientific community and beyond, promoting interdisciplinary and large-scale scientific collaborations, supporting linguistic and cultural diversity in Europe, as well as increasing the economic and social impact of research in SSH. GoTriple can therefore be used across a broad – and perhaps unexpected – variety of use cases, including the scenario which is the object of this article: a support tool for scientific translation.
The GoTriple platform as a support tool for scientific translation: considerations and use cases
Although the platform was not specifically designed as a support tool for scientific translation, language specialists will not fail to notice the interesting features offered by GoTriple. From the creation of monolingual or multilingual corpora to the export of specialised vocabularies, a translator or researcher dealing with translation activities will find valuable resources in the platform without a doubt. This article only suggests a few ideas that will hopefully inspire a more comprehensive list of possible use cases.
1. Terminology extraction from disciplinary vocabularies
One of the goals of the TRIPLE project was to create and leverage multilingual controlled vocabularies in the humanities and social sciences, in order to properly index and annotate publications, and to ensure their discoverability through search and navigation on the GoTriple platform. These vocabularies can now be freely downloaded in CSV format (among others), and can therefore be easily cleaned, refined as needed and converted to the interoperable TBX format in order to be imported into computer-assisted translation tools.
Data can be downloaded from this link to the Semantics platform, which contains other potentially interesting open lexical resources. It should be noted that the TRIPLE multilingual vocabularies are the result of a technology-based, collaborative work: the concepts were automatically pre-translated, then proofread and validated by experts. There is certainly room for improvement, but 3,375 terms translated into several languages remain a significant starting point, in particular for those looking to train and specialise in scientific translation, especially in the humanities and social sciences.
2. Search for reference documents and corpus creation
GoTriple is a discovery platform for the humanities and social sciences, which makes it possible to carry out multilingual searches and thus find publications in a given language by using keywords in one of the supported languages. This means that by searching for a term in Spanish, it is possible to find publications containing the same term not only in the language of the search but also in other languages. While it may seem trivial to explain to a translation specialist how useful such a feature can be, it is interesting to suggest here a few use cases in order to introduce working methods and practices to non-specialists, students or disciplinary experts wishing to get involved in translation work, among others.
Let’s consider, for example, translating into French a publication written in English, about urban studies and more specifically on the subject of “urban commuting”. By searching for this expression in English, GoTriple returns 765 results that can be filtered by discipline, type of publication, dissemination platform or publisher, access policy, licence and, even more interesting, by language. Thanks to this filter, it is possible to prioritise in the search results the publications containing text in French, the target language of our translation use case. In this way, it is easy to identify any publication likely to contain specialised terminology in the target language and therefore to be added to the translator’s reference corpus: for example, a thesis defended at the Université Paris-Est, an article published in 1998 available today on Persée, or a publication archived in the Dépôt Institutionnel de l’Académie Louvain.
Whether created before or during the translation process, corpora are widely used among translators, as shown by the use case study conducted as part of the Translations and Open Science project. In order to study in-domain terminology and phraseology, translators look for specialised texts, collect them and create corpora, which will be used as a reference throughout the entire translation process. What is the input of GoTriple compared to other conventional tools, such as Linguee or a standard search engine? The platform only harvests and indexes scientific content from specialised providers and aggregators (Isidore, DOAJ, OpenAIRE, BASE, EconStor, OpenEdition, CESSDA, Clarin, Biblioteka Nauki, ZRC SAZU, EKT, University of Coimbra), hence it filters the huge amount of data and information available on the Web. In addition, the harvested resources are open, which constitutes a valuable asset for translators who do not have an academic affiliation and therefore do not have access to paywall content (this issue has been raised several times during the interviews and workshops of the Translations and Open Science project). Finally, GoTriple offers features to easily retrieve publications by the same author, on the same subject, or containing citations of the consulted document, thus allowing translators to find other potentially interesting content to add to the reference corpus. The “knowledge map” also makes it possible to display a visual overview of the main subjects and the 100 publications which are most relevant to the current search.
With research being increasingly internationalised, the GoTriple platform also offers the opportunity to restore a truly multilingual translation model, overcoming the artificial bilingualism that often imposes English as the pivot language and which is, in reality, a legacy of language engineering. As Philippe Lacour of TraduXio points out, translators generally have a working knowledge of several languages and they use them all, not only the source and the target languages of the text they are working on. Thanks to the GoTriple platform, it is possible to find similar or equivalent terms in several languages, which is particularly interesting in the translation of the humanities and social sciences, where the role of English is less central than in engineering, physical, medical or natural sciences. If we search for our initial English expression “urban commuting”, for example, we can observe in GoTriple’s search results the Portuguese term movimento pendular, which can help the translator looking for the French expression mobilité pendulaire dans l’aire urbaine. In addition, the keywords retrieved from GoTriple’s controlled vocabularies and associated with the publication in question can be added to personal or shared glossaries.
3. Collaboration and interdisciplinarity
Another ambition of the GoTriple platform is to establish a collaborative and interdisciplinary dynamic within the scientific community and with society at large, in order to highlight the value of SSH research as well as its impact in economic and social terms. Such a dynamic could benefit not only research activities per se but also related activities, like scientific translation. Networking and collaboration opportunities, as well as professional valorisation are indeed urgently needed, according to the stakeholders involved in the scientific translation processes. GoTriple could help meet these needs through its functionalities, such as the creation of user accounts and profiles, the attribution of publications to authors (and why not the attribution of translations to translators?), the possibility to easily connect and communicate with disciplinary experts (and why not linguistic experts?), the document annotation system, etc. In such a scenario, a publisher or an author wishing to have a publication translated or revised could get access to a directory of specialised profiles, with their expertise and project history visible in just one click. Similarly, a translator in need of advice when translating a specialised text could easily work with disciplinary experts, thanks to the mailbox and annotation features, plus – maybe in the future? – the integration with collaborative and open-source computer-assisted translation environments. Far from adopting a logic of platformisation that neglects the human component, GoTriple offers the opportunity to optimise translation processes and to share linguistic and disciplinary knowledge.
Avenues for improvement of the GoTriple platform for a use in scientific translation
GoTriple was designed from the outset to address a variety of needs, and its use in scientific translation, as unexpected as it may seem to non-specialists, is clearly a possible scenario. However, some improvements would be welcome in order to make the platform an even more relevant tool for scientific translators.
With regard to search features, the first limitation observed is that the platform only harvests and indexes the metadata of the publications, therefore the keywords, titles and abstracts, but not the full texts, which significantly narrows the search results. Moreover, in the case of expressions made up of several words, the engine only searches for the single words, instead of the whole expression. In spite of these two limitations, interesting resources can still be found, but an improvement in this regard could undoubtedly benefit scientific translators.
One aspect that should be fixed as soon as possible is the mention of automatically translated texts. Some titles and abstracts were automatically translated into English in order to improve the discoverability of the associated publications, but the mention indicating the automatic nature of the translation should be made clearly visible. Currently, this information is rather hidden, since it is only available in the JSON export (in the publication page, the user has to click on Export, then select the JSON format and check whether the value of the “translated” attribute for a given language is “true”: in this case, the text is a machine translation output). According to the latest information, this point should be fixed soon.
Finally, it would be useful to add the “Language Expert” category in the available user profiles, so that the platform can truly become a place to enhance collaboration and visibility for the scientific translation experts and stakeholders involved.
The GoTriple platform is still young and there is obviously room for improvement. However, it has all it takes to eventually become an essential tool for scientific translators and to help overcome the compartmentalisation of people, skills and resources involved in the translation processes, which are at the heart of an open and multilingual science. Hoping to see new integrated features and publications soon, we wish a bright future to the GoTriple platform!
NOTE: This article is a translation of La plateforme GoTriple: un outil de support à la traduction scientifique, published on the Traductions et science ouverte research blog. https://tradso.hypotheses.org/215
The translation was made by the author of the article with the support of translation technology (interactive use of machine translation within a computer-assisted translation environment).