Cluster

Syntax-Prosody in Optimality Theory (SPOT)

Share

Print Print

About the Cluster

Syntax (from Greek σύνταξις, an abstract derivative of συντάσσω ‘put in order together, organize, arrange’), is the set of linguistic rules, principles, and processes that govern the structure of sentences, while prosody (from Greek προσῳδίᾱ ‘song accompanying words, tone or accent of a syllable’; Latin accentus) deals with syllables, rhythmic feet, and larger units of speech that contribute to intonation, tone, stress, and rhythm. The syntax-prosody interface is the study of how syntactic (grammatical) structures are mapped onto prosodic form in different languages.

The fundamental problem in syntax-prosody research is to precisely predict, for any kind of syntactic structure in a given language, what kind of prosodic parse it will be assigned by the language’s grammar. Languages are known to show wide variation in this respect, and the Linguistics Department at UC Santa Cruz has a long history of work on these issues dating back to the 1990’s. In order to approach the problem in a rigorous and precise way, it is necessary to consider the full set of prosodic structure candidates for any syntactic input, a task for which no formal tool has so far been developed. At the same time, automation is essential since the number of prosodic structure candidates increases exponentially, as length and syntactic complexity increase: Under widely shared representational assumptions, 3 words receive 48 different parses, 4 words 352, 5 words 2880, etc.

The project developed the pilot program SPOT (Syntax-Prosody in Optimality Theory), a JavaScript application that provides both exhaustive generation of the prosodic candidates, and automatic evaluation of the violation profiles of each candidate. Even though still in a preliminary form, the program has been successfully tested on well-known prosodic parsing data in Japanese and Basque. The pilot investigation has revealed that automating the generation and evaluation of candidates makes it possible to see clearly many hitherto unknown factors coming into play in the overall analysis, with interesting predictions that need to be tested in the future.

The computational program SPOT will be made publicly and freely accessible. Beyond its immediate goals, it has clear application possibilities in natural language processing, speech technology (recognition and production, translation), as well as in second language research and teaching (e.g., to explicate the differences in syntactic and prosodic structure between the first and the second language).

Principal Investigators

Ryan Bennett, Assistant Professor of Linguistics, rbennett@ucsc.edu
Junko Ito, Chair and Professor of Linguistics, ito@ucsc.edu
James McCloskey, Professor of Linguistics, mcclosk@ucsc.edu
Armin Mester, Research Professor of Linguistics, mester@ucsc.edu

UC Santa Cruz Faculty and Student Participants

Gorka Elordieta, University of the Basque Country
Boris Harizanov, Stanford University
Shinichiro Ishihara, Lund University
Maziar Toosarvandani, Linguistics, UC Santa Cruz
Lisa Selkirk, UMass, Amherst

Andrew Angeles (UCSC Linguistics Ph.D. student, 2nd year)
Jenny Bellik (UCSC Linguistics Ph.D. student, ABD)
Richard Bibbs (UCSC Linguistics Ph.D. student, 1st year)
Nick Kalivoda (UCSC Linguistics Ph.D. student, ABD)
Nick Van Handel (UCSC Linguistics Ph.D. student, 2nd year)

Events

November 18, 2017: SPOT (Syntax- Prosody in OT) Workshop