David Temperley
Eastman School of Music

In person: LDC Conference Room (Room 135), 3600 Market Street, Suite 810

For the Zoom link, please email: pennmindcore@sas.upenn.edu

Repetition and Information Flow in Music and Language

The theory of Uniform Information Density makes two predictions regarding the use of repetition: (1) when a pattern is repeated, variable aspects of the pattern will be less probable in the second instance than in the first; (2) less probable patterns will have a higher tendency to be used repetitively. In this talk I present recent research testing these two predictions with regard to language and music. Regarding prediction (1): in syntactically-matched coordinate constructions (e.g. the big dog and the small cat), the second coordinate tends to have less frequent words than the first; in music, when a melodic pattern is immediately repeated in an altered form, the alterations tend to lower the schematic probability of the pattern (e.g. by increasing the interval size). Regarding prediction (2): in music, unusual melodic devices (such as escape tones and anticipations) tend to be used repetitively; in language, rare syntactic constructions show a higher tendency than common ones to be used repetitively (in coordinate constructions and elsewhere). Intriguingly, the syntactic constructions that are most often used repetitively tend to be associated with persuasive rather than informative discourse, implying an emotional commitment on the part of the speaker (such as the construction Det Adj, e.g., (the rich); this suggests a further connection with music.

David Temperley is a music theorist, cognitive scientist, and composer. Since 2000, he has been a professor of music theory at Eastman School of Music. Using computational modeling, Temperley has explored aspects of music cognition such as meter perception, key perception, harmonic analysis, and melodic expectation. He has also used corpus methods to explore questions of musical style and broader issues of music cognition. Temperley also has a strong secondary interest in language research: parsing, sentence production/comprehension, and corpus research.