Metrics

Phonological theory and metrics are fundamentally interconnected. My research explores the connection between meter and phonology using statistics, computational methods and Maxent modeling. A central methodological goal of my recent work has been to test different ways of incorporating ordinary-language baselines into Maxent models of metrical data, which helps disentangle true metrical preferences from the effects of ordinary-language phonology and lexical statistics.

I am particularly interested in oral folk poetry, mainly from the perspective of linguistic theory. Oral traditions offer a unique window into how humans organize language under strict (and sometimes conflicting) formal and cognitive constraints. At the same time, oral composition tells us a great deal about language itself: oral verse is a rule-governed system, governed by a mental device that allows poets to generate and evaluate the well-formedness of verse forms. Often illiterate and lacking formal education in metrics, folk poets acquire and transmit the metrical grammar across generations, much like native speakers acquire and transmit language.

For my dissertation, I compiled an annotated and searchable corpus of tens of thousands of lines of South Slavic oral epic verse. I analyze this corpus quantitatively and through computational modeling. An important aspect of the meter is word order: the factors that shape it and the ways in which it diverges from ordinary language. Word order variation in BCMS oral epic poetry is the empirical focus of my dissertation. The meter appears to allow word orders that would be impermissible in everyday speech. This flexibility emerges from the interplay between linguistic (primarily phonological) constraints and the real-time cognitive pressures of live performance, such as memory load and planning. I provide a formal analysis of these competing factors in a probabilistic weighted-constraint framework.

I am also conducting a cross-linguistic survey of rhyme strictness effects. The central question of the project is how the strictness of correspondence between sounds in rhyming domains varies as a function of linguistic factors such as domain size (mono- vs. di- vs. tri-syllabic rhyme), the featural makeup of a given language and distance between the rhyming domains. This project aims to identify cross-linguistic universals of rhyme and explore their cognitive basis.