MultiWord Verbs of Estonian: a Database and a Corpus

The paper describes two interrelated language resources: a database of 13,000 Estonian multi-word verbs (MWV) and a 300,000 word corpus with annotated MWVs. Both resources have been manually post-edited, and are meant to be used by a wide audience, from corpus linguists to language engineers. The paper gives a short overview of the types of MWVs in Estonian, followed by a description of some grammatical features – word order and inflection – of Estonian and their manifestation in the MWVs. The database is a table that has 13,000 rows and 11 columns and contains information about the source (dictionary or corpus) of the MWV, its linguistic category, frequency in the text corpus, and morphological description. The text corpus contains the morphological analysis of the source text and the annotated MWVs. The layout of the corpus is essentially a table, a row standing for a running word and the columns filled by annotations. The corpus contains 8,200 instances of tagged MWVs and 34,100 simplex main verbs, meaning that roughly every fifth predicate is represented by a MWV. The number of different types of the MWVs in the corpus is 3,500. 1