A Computational Perspective on the Romanian Dialects

A Computational Perspective on the Romanian Dialects Alina Maria Ciobanu, Liviu P. Dinu Faculty of Mathematics and Computer Science, University of Bucharest Center for Computational Linguistics, University of Bucharest alina.ciobanu@my.fmi.unibuc.ro, ldinu@fmi.unibuc.ro Abstract In this paper we conduct an initial study on the dialects of Romanian. We analyze the differences between Romanian and its dialects using the Swadesh list. We analyze the predictive power of the orthographic and phonetic features of the words, building a classification problem for dialect identification. Keywords: Romanian, dialects, language similarity. 1. Introduction and Related Work The rapid development of the online repositories has lead to a significant increase in the number of multilingual doc- uments, allowing users from all over the world to access information that has never been available before. This ac- celerated growth created the stringent need to overcome the language barrier by developing methods and tools for pro- cessing multilingual information. Nowadays, NLP tools for the official languages spoken in the European Union and for the most popular languages are constantly created and im- proved. However, there are many other language varieties and dialects that could benefit from such NLP tools. The ef- fort for building NLP tools for resource-poor language vari- eties and dialects can be reduced by adapting the tools from related languages for which more resources are available. The importance of adapting NLP tools from resource-rich to resource-poor closely related languages has been ac- knowledged by the research communities and has been ma- terialized through multiple events, such as the workshop on Language Technology for Closely Related Languages and Language Variants (Nakov et al., 2014) or the workshop on Applying NLP Tools to Similar Languages, Varieties and Dialects (Zampieri et al., 2014). A related problem occurs when researchers are interested in the cultural heritage of small communities, who developed their own techniques of communication and prefer using dialects instead of the official language of the region they live in. In some situations, these dialects are close enough to the standard language, but in other situations the differ- ence is consistent, so much so that some dialects have be- come languages of their own (for example Friulian, spoken in the North-East of Italy). These matters raise interesting research problems, since many such dialects are used only in speaking. Moreover, they often tend to be used only in very specific situations (such as speaking in the family), very rarely being taught in schools. Thus, many dialects are in danger of extinction, according to the UNESCO list of endangered languages (Moseley, 2010). In this paper, we conduct an initial study on the dialects of Romanian. This investigation has the purpose of providing a deeper understanding of the differences between dialects, which would aid the adaptation of existing NLP tools for related varieties. The aim of our investigation is to assess the orthographic and phonetic differences between the di- alects of Romanian. In this paper, we quantify only the orthographic and phonetic differences, but the morphology and the syntax are other important aspects which …

Content truncated to 500 words