Parallel text typology dataset

This repository contains data accompanying the following paper: Neural models can sometimes discover typological generalizations. Computational Linguistics (2023) 49 (4): 1003–1051. doi.org/10.1162/coli_a_00491 It contains the following information for 1295 different languages: language vector representations from a range of neural models automatically derived lists of affixes automatically derived lists of inflectional paradigms typological features derived from annotation projection, and statistics on dependency relations typological features derived from classifiers trained on language vectors and typological databases automatically derived word lists data needed for automatic evaluation of language representations (code in separate repository) Note that the multilingual word embeddings described in the paper are very large, and therefore distributed in a separate public repository.

The computations were enabled by resources provided by the Swedish National Infrastructure for Computing (SNIC) at C3SE partially funded by the Swedish Research Council through grant agreement no. 2018-05973. This work was funded in part by the Swedish Research Council through grant agreement no. 2019-04129.