Translation Crowdsourcing: Creating a Multilingual Corpus of Online Educational Content

Translation Crowdsourcing: Creating a Multilingual Corpus of Online Educational Content Vilelmini Sosoni1, Katia Lida Kermanidis2, Maria Stasimioti1, Thanasis Naskos2, Eirini Takoulidou2, Menno van Zaanen3, Sheila Castilho4, Panayota Georgakopoulou5, Valia Kordoni6, Markus Egg6 1Department of Foreign Languages, Translation and Interpreting, Ionian University, Corfu, Greece 2Department of Informatics, Ionian University, Corfu, Greece 3Department of Communication and Information Sciences,Tilburg University, The Netherlands 4ADAPT Centre, Dublin City University, Dublin, Ireland 5Deluxe Media Europe, Deluxe Entertainment Services Group, London, UK 6Department of English Studies, Humboldt-Universität zu Berlin, Germany vilelmini@hotmail.com; kerman@ionio.gr; stasimioti.maria@gmail.com; anaskos@ionio.gr; rinoulit@gmail.com; mvzaanen@uvt.nl; sheila.castilho@adaptcentre.ie; yota.georgakopoulou@bydeluxe.com; kordonie@anglistik.hu- berlin.de; markus.egg@anglistik.hu-berlin.de Abstract The present work describes a multilingual corpus of online content in the educational domain, i.e. Massive Open Online Course material, ranging from course forum text to subtitles of online video lectures, that has been developed via large-scale crowdsourcing. The English source text is manually translated into 11 European and BRIC languages using the CrowdFlower platform. During the process several challenges arose which mainly involved the in-domain text genre, the large text volume, the idiosyncrasies of each target language, the limitations of the crowdsourcing platform, as well as the quality assurance and workflow issues of the crowdsourcing process. The corpus constitutes a product of the EU-funded TraMOOC project and is utilised in the project in order to train, tune and test machine translation engines. Keywords: parallel corpus, MOOCs, online educational text, crowdsourcing 1. Introduction Massive Open Online Courses (MOOCs) have been growing rapidly in number of enrollees and participating universities, in impact, and in target groups. According to 20161 statistics, they involve around 60 million students worldwide. In their majority, MOOC users are non-native English language speakers, and the language barrier proves to be the main obstacle towards further expansion of the MOOC market, as most courses are offered in English only. The EU-funded TraMOOC (Translation for Massive Open Online Courses) project aims at providing machine translation solutions for the educational content available in MOOCs, thus enhancing access to online education. The content genre varies from video lecture subtitles, slides, assignments, and quiz text, to course discussion forum text (Kordoni et al., 2016). Given that the source language is English, in-domain trained and tested translation engines are built for 11 European and BRIC target languages (Bulgarian, Chinese, Croatian, Czech, Dutch, German, Greek, Italian, Polish. Portuguese, and Russian). In an attempt to achieve optimal translation output, the TraMOOC goal is to develop as much in-domain data as possible, especially for the language pairs that are not adequately supported by the required infrastructure. To this end, crowdsourcing is adopted (as an alternative to using professional translators) for collecting translations on a large scale for all language pairs involved. The present work describes the in-domain source data collected, the crowdsourcing experiment and the resulting multilingual parallel dataset, taking into consideration the 1 https://www.class-central.com/report/mooc-stats-2016/ challenges imposed by the text genre, the number of language pairs involved, the large data volume aimed at, the quality assurance of the experimental process, and the related crowdsourcing workflow issues. 2. Related Work The creation …

Content truncated to 500 words