Mutual intelligibility

of closely related languages




Research projects

Ongoing projects


Matrix representation for sound matchings

Levenshtein distance is capable of measuring the distance between strings of sounds in words and large word corpora. This measure does not tell us about the nature of the sound matchings and dismatchings, nevertheless. We therefore plan to take a closer look at which sounds do match which other sounds in language pairs. In this way we will also be able to see which sounds correspond to a high degree with one another - e.g. to test the hypothesis that consonants are more stable across the languages than vowels, and if specific sounds are of high reliability between language pairs, whereas other sounds are less confident to match with the same (or a similar) sound. The project consists mainly on scripting. Word lists from two languages, which have been automatically aligned with the Levenshtein-method aleven, are given as an input. The script calculates the sound correspondencies for each of the sounds from both languages. The output is a matrix showing the correspondencies of all sounds found in the corpus, with ferquencies of all correspondencies. This matrix can be used as input for further visualizations etc.

Last update: 19.10.2010

Finished projects