Quantifying sonority contour: A case study from American English

Previous studies have argued that the most reliable phonetic correlate of sonority is intensity (e.g., Parker 2002, 2008, Jany et al. 2007). However, those studies have only considered intensity of a single segment. This paper investigates the phonetic correlate of sonority contour in consonant clusters. 10 native speakers of American English (5 male, 5 female) read 33 monosyllabic English words that begin with a bi- or tri-consonantal cluster (e.g., play, stray) embedded in a frame sentence (“Father saw ‘____’ again,” used in Parker 2008). Measured first were (i) an average RMS and (ii) sound level minima of each consonant in the cluster C1C2, and the sonority contour was quantified by subtracting the intensity value of C1 from the intensity value of C2. Also, (iii) actual intensity slopes in the transition between the two consonants were measured. Results show that the intensity contours calculated based on (i) and (ii) do not always correspond to the intensity slopes (iii), while both of them are in general correlated with the sonority contour. It will also be suggested that it is intensity slopes (iii) that play a crucial role in consonant cluster perception and in phonological phenomena involving consonant clusters.