Predicting the asymmetric intelligibility between spoken Danish and Swedish using conditional entropy

Abstract

The languages Danish and Swedish are so similar to each other that they are partially mutually intelligible. Speakers of these languages can communicate with each other each using their own language. The intelligibility between the spoken languages is asymmetrical, however: Danes can understand spoken Swedish better than Swedes can understand spoken Danish. This asymmetry is absent in the written language. In this article, we explore whether this asymmetry can be explained by conditional entropy. Conditional entropy is a way of measuring the amount of regularity in the sound or grapheme correspondences between the two languages. This study is based on the assumption that a high amount of irregularity among the correspondences will impair intelligibility, and conversely, a high amount of regularity will aid intelligibility. The entropy measure can be asymmetrical: the entropy of Swedish from a Danish perspective is not necessarily the same as the entropy of Danish from a Swedish perspective. We calculated the entropies between the two languages for both written and spoken language and compared the results to intelligibility data gathered through a word translation task. In agreement with previous research, the word translation task showed a higher intelligibility of spoken Swedish for Danes than vice versa, and only a very small asymmetry for the written language. The results of the entropy calculations followed the same pattern. In the spoken language, there was a higher entropy of Danish for Swedes than vice versa, but in the written language, there was virtually no asymmetry. This leads us to conclude that entropy is a promising predictor for intelligibility and should be further explored in this context.