Michael gave a colloquium talk at UPenn on March 25, titled “Two dimensional parsing and the iambic-trochaic law”.
Abstract:
The ‘Iambic-Trochaic-Law’ of rhythmic perception holds that alternating long and short sounds are perceived as sequences of binary groups with final prominence; alternating soft and loud sounds as sequences of binary groups with initial prominence. This talk reports on experiments that illustrate how the ITL emerges from the way listeners parse the signal along two in principle orthogonal perceptual dimensions, grouping and prominence. Evidence from production experiments shows that intensity and duration correlate when cueing prominence (syllables carrying word stress or focal stress are loud and long) and anti-correlate when cueing phrasing (word-final and phrase-final syllables are soft and lengthened, word- and phrase-initial syllables are loud). Listeners exploit this cue relation when deciding what aspects of the signal to attribute to each dimension. Syllables that are excessively long are perceived as final and prominent (leading to the perception of iambs), syllables that are excessively loud as initial and prominent (leading to the perception of trochees), but these two cases (which the ITL is based on) are only a small part of the more general pattern, to which the notions of iamb and trochee are not central. The decisions about grouping and prominence are orthogonal in principle, but they compete for explaining overlapping cues, and they mutually constrain each other.  This perspective on prosodic parsing raises new questions about why exactly we often even perceive a rhythm when listening to sequences of acoustically identical tones or syllables (a phenomenon called ‘subjective rhythm’), as well as about rhythmic differences between languages.