Volume 82, Issue 2

Whereas some authors claim that the distribution of prenuclear accents in English largely follows from rhythmic and other non-informational considerations, other authors report a small but meaningful effect of prenuclear accents on the interpretation of sentences. In this paper we report on an experiment where native English speakers were asked to repeat stimulus sentences with one of three different accentual patterns on a word in sentence-initial prenuclear position: unaccented, with a high pitch accent on the syllable with primary stress or with a high accent on an earlier syllable with secondary stress. Participants were moderately successful in reproducing the intonational patterns. The early high accent pattern was reproduced particularly well. An automatic classification algorithm nevertheless produced four clusters of contours, instead of the three patterns present in the stimuli. Two distinct contours were used to signal the presence of a high tone before the syllable with primary stress. We conclude that the early high accent pattern is a strong attractor in imitations, but it was implemented with F0 trajectories that would be analyzed as phonologically different, suggesting an equivalence class of prenuclear contours. We also note a preference for rhythmic anchoring in the prenuclear position.

There is now a large literature probing syllable affiliation of consonant sequences through phonetic measurements. These studies often use one of two diagnostic measures: (1) temporally stable intervals using relative standard deviation, and (2) compensatory shortening effects. In this study, we argue that both measures are difficult to infer from without precise theoretically predicted expectations and additional controls. We studied eleven native speakers of North-Central Peninsular Spanish who pronounced disyllabic real/nonce Spanish words with varying consonant sequences. On the face of it, our temporal stability and compensatory shortening results challenge the standard analysis of syllabic affiliation in Spanish phonology, potentially supporting a complex onset analysis for /sl/ and /sm/. However, in post hoc analyses we observed shortening effects outside the target syllable due to consonant sequences, indicating evidence for poly-constituent shortening. Therefore, compensatory shortening effects within a syllable cannot automatically be assumed to be due to syllable structure. Our results and simulations suggest that, despite superficial evidence of a c-centre alignment, the clusters are more consistent with a right-edge alignment once poly-constituent shortening and domain-initial lengthening are taken into consideration.