Integrating Selice Romani into the UD framework

In this talk, I discuss ongoing work aimed at creating a treebank for Selice Romani – a low-resource language without a written standard – within the Universal Dependencies framework. This variety of Romani is spoken in the Slovak village of Selice, where Hungarian is the prevalent language, leading to a high number of Hungarian loanwords and some grammatical borrowings. After outlining the sociolinguistic background of Selice Romani, I describe the methodological steps taken in the project: from processing and prosodically annotating raw spoken data, through morphological analysis with an FST tool, to the syntactic analysis of dependency relations.