On the identification of selected Danish intonation contours

Abstract

10 Copenhagen and 4 non-Copenhagen speakers identified 15 human utterances, differing only in their fundamental frequency course, as being either declarative, non-final, or interrogative (forced choice). Responses are very clearly correlated with Fo: the most steeply falling intonation contours are identified as being declarative, the least falling (i.e. "flat") ones as being interrogative, and contours in the middle of the continuum as being non-final. Copenhagen speakers clearly perceive three categories, whereas non-Copenhagen speakers seemingly operate with only two. Several, mutually interdependent, parameters in the Fo course may account for the results, the two most powerful ones, however, being the levels of the last stressed and the succeeding unstressed syllable in the utterance. In a subsequent experiment, 7 Copenhagen speakers identified the same utterances as being either declarative or non-declarative. The majority of the (formerly) non-final sentences were now labelled non-declarative, rather than being split into partly declarative, partly non-declarative categories. When a subset of the same utterances were mutilated, identification criteria changed, and identification deteriorated almost progressively with the number of syllables being cut away from the end of the utterance (but not seriously - until only the first stress group remained), whereas syllables cut away from the beginning hardly affected identification at all.