Perception of tonal rises and falls for accentuation and phrasing in Swedish

In previous experiments with Dutch, French and Swedish listeners, it was shown that the location in the syllable of the onset of a rising or falling pitch movement is critical for the perception of accentuation. As the onset of the pitch movement is shifted through the syllable, there is a point at which the percept of accentuation shifts from one syllable to the next. This point is termed the accentuation boundary. It has also been proposed that in certain positions, the percept of accentuation conflicts with the percept of phrasing. An experiment with Swedish listeners was carried out using the same stimuli as used for the accentuation st udy, but now the task was to determine the phrasing of the syllables. The results indicate that perceptual phrase boundaries can be determined in the same way as accentuation boundaries. Differences in the locations of the boundaries can be interpreted in terms of strengths of tonal cues for accentuation and phrasing. were consistent with the finding for Dutch that the cue which induces the percept of accentuation is located at the onset of the pitch movement. However, the results for the French listeners differed in two ways from those of the Dutch and Swedish listeners. Firstly, the location of the accentuation boundaries for the accent-lending rises differed. For the Dutch and Swedish listeners, the boundary for the rises was located near the end of the vowel of Sn while for the French listeners, the boundary was located earlier, near the middle of the vowel of S n. Secondly, for the accent-lending falls, the AB could be determined quite accurately for the Dutch and Swedish listeners. The result was very vague for the French subjects, however, indicating that they had difficulty perceiving the falls as accent-lending. The accentuation boundaries for the three languages are plotted in Figure 1 for the /mamamamama/ stimuli with the /m/ of normal duration and in Figure 2 for the /.a.a.a.a.a/ stimuli with the intervocalic silent interval of normal duration. Similar results were obtained for stimuli with doubled and halved consonant and silence durations (2). The language differences found in the above study were interpreted in terms of temporal-alignment categories of accent- lending rises and falls (12). A model of perceived accentuation was proposed in which several different categories of accentuation are represented as the falling or rising movement is advanced through the test syllables. The perceived category would depend upon whether a pitch jump or the onset of the pitch movement (rise or fall) is perceived and whether or not the particular category is represented in the language in question. The percept of phrasing can also influence the perceived category. An attempt has been made to describe correspondences between these perceptually-based categories and phonological categories presented in theories of intonation for the three languages. For rises, a jump category was proposed and formalized as L+H* in autosegmental terminology (3). This jump category corresponds to rise 1 in the Dutch description of intonation as presented by 't Hart, Collier and Cohen (15) and to a focal accent realization of an accent I word in Swedish as presented by Bruce (4). The perceived tonal change is a jump up in tonal levels from low to high into the stressed vowel. Also for rises, a rise-onset category was proposed and formalized as L*+H corresponding to rise 3 in Dutch and to a focal accent realization of an accent II word in Swedish, especially in dialects of southern Sweden (5). Here the perceived tonal change is a rise on the stressed vowel with the onset of the rise