Word order typology through multilingual word alignment

Abstract

With massively parallel corpora of hun-dreds or thousands of translations of the same text, it is possible to automatically perform typological studies of language structure using very large language sam-ples. We investigate the domain of word order using multilingual word alignment and high-precision annotation transfer in a corpus with 1144 translations in 986 lan-guages of the New Testament. Results are encouraging, with 86% to 96% agreement between our method and the manually cre-ated WALS database for a range of differ-ent word order features. Beyond reproduc-ing the categorical data in WALS and ex-tending it to hundreds of other languages, we also provide quantitative data for the relative frequencies of different word or-ders, and show the usefulness of this for language comparison. Our method has applications for basic research in linguis-tic typology, as well as for NLP tasks like transfer learning for dependency pars-ing, which has been shown to benefit from word order information.