PrevDistro (Preverb Distributions) is an open-source dataset containing 41.5 million corpus occurrences of 49 preverb-verb construction types. It consists of the following columns: 1 sid: ID 2 constype: construction type 3 subtype: construction subtype 4 prevpos: preverb position 5 prev: preverb 6 verb: verb lemma 7 intervening: intervening words (as lemmas) 8 actform: actual form (the same content as in column 10, but this column is lowercase) 9 left: left context 10 kwic: keyword in context 11 right: right context 12 docid: document ID from the Hungarian Gigaword Corpus 13 title: document title 14 style: document style (e.g. official, press, ...) 15 region: document region (e.g. Transylvania, Subcarpathia, ...) 16 year: year of publication (sometimes several years can be found in one document) The first row stands for the header. If a cell's value is unspecified, it is marked with underscore (_).

PrevDistro 1.0.0 (deprecated) can be found at https://science-data.hu/dataset.xhtml?persistentId=doi:10.5072/FK2/TRSD50 In PrevDistro 2.0.0, several new columns were added and the already existing data has undergone some fixes as well.