The past decade has witnessed rapid growth of user-generated content, such as blogs, forums and social media. This type of content offers an important source of information to diverse fields, such as social sciences, economics and computer science, both for research and business. But when dealing with user-generated content it is necessary to come to grips with the language of computer-mediated communication which is, due to social and technical characteristics, often very different from the standard, characterized by colloquialisms and borrowings, dialect-specific phonetic orthography and syntax, specific abbreviations, fast uptake of new vocabulary, etc.
This was achieved in the scope of the Slovene basic research project JANES, which compiled a large and representative corpus that covered a large portion of publicly available user-generated text in Slovene, in particular tweets, blogs, forums posts, news comments and Wikipedia talk pages (Fišer, Ljubešić and Erjavec 2018). The corpus is linguistically annotated with standardized spelling, lemma, part-of- speech, and names and is freely available via the two CLARIN.SI concordancers to make it useful for theoretical and applied linguistic research. The project further produced a series of manually annotated datasets, which were used to develop methods for automatic processing of non-standard Slovene texts. Finally, the project developed a dictionary of non-standard Slovene, available through a web portal. The dictionary should be useful for teachers, students, linguists, lexicographers and the general public. All the developed resources have been made openly available for download under the Creative Commons license through the CLARIN.SI repository for research and development in computational linguistics and other automatic data processing fields.
Apart from hosting the developed resources and tools, CLARIN.SI has also contributed to several user involvement events which presented the results of the project to different user groups: 2 summer camps on Slovene Netspeak for high school students from 20 high schools all over Slovenia, 1 summer school on Internet linguistics for university students of Slovene linguistics from Slovenia and abroad and 4 workshops on the resources, tools and methods for analyzing non-standard language for researchers and university lecturers.
Particularly notable was the JANES Express seminar series for fellow researchers in corpus and computational linguistics which have been organized in Ljubljana (Slovenia), Zagreb (Croatia) and Belgrade (Serbia). It was organized in collaboration with the Regional Linguistic Data Initiative. The seminar series presented the guidelines for manual annotation of training corpora of non-standard language varieties and the annotation platform WebAnno. As a result, three comparable gold standard corpora for tagging, lemmatization and normalization of non-standard Slovene, Croatian and Serbian have been developed and are available from the CLARIN.SI repository. Apart from testing and adapting the methodology originally developed for Slovene to two additional closely related languages, the JANES Express seminar series is also a best-practice example of knowledge transfer in the region.
More information about all these events as well as teaching materials are available on the project website.
Blog post written by Darja Fišer, edited by Tomaž Erjavec.