Proceedings of the Fourth Workshop on NLP for Similar Languages, Varieties and Dialects (VarDial)

Abstract

PrefaceVarDial is a well-established series of workshops held annually and co-located with top-tier international NLP conferences. Previous editions of VarDial were VarDial'2014, which was colocated with COLING'2014, LT4VarDial'2015, which was held together with RANLP'2015, and finally VarDial'2016 co-located with COLING'2016. The great interest of the community has made possible the fourth edition of the Workshop on NLP for Similar Languages, Varieties and Dialects (VarDial'2017), co-located with EACL'2017 in Valencia, Spain.The VarDial series has attracted researchers working on a wide range of topics related to linguistic variation such as building and adapting language resources for language varieties and dialects, creating language technology and applications that make use of language closeness, and exploiting existing resources in a related language or a language variety.We believe that this is a very timely series of workshops, as research in language variation is much needed in today's multi-lingual world, where several closely-related languages, language varieties, and dialects are in daily use, not only as spoken colloquial language but also in written media, e.g., in SMS, chats, and social networks. Language resources for these varieties and dialects are sparse and extending them could be very labor-intensive. Yet, these efforts can often be reduced by making use of pre-existing resources and tools for related, resource-richer languages.As part of the workshop, we organized the first VarDial evaluation campaign with four shared tasks: Discriminating between Similar Languages (DSL), Arabic Dialect Identification (ADI), German Dialect Identification (GDI), and Cross-Lingual Parsing (CLP). The campaign received a very positive response from the community. A total of 28 teams subscribed to participate in the four shared tasks, 19 of them submitted official runs, and 15 of the latter also wrote system description papers, which appear in this volume along with a shared task report by the task organizers.We further received 14 regular VarDial workshop papers, and we selected nine of them to be presented at the workshop. The papers that appear in this volume reflect the wide range of interests related to language variation. We include papers applying NLP tools to perform dialect analysis, to study mutual intelligibility and diatopic variation in historical corpora, as well as core NLP tasks and applications such as dialect and similar language identification, adaptation of POS taggers, and machine translation between similar languages and dialects.We take the opportunity to thank the VarDial program committee and the additional reviewers for their thorough reviews. We further thank the VarDial Evaluation Campaign participants, as well as the participants with regular research papers, for the valuable feedback and discussions.