My research interests lie in the structure of linguistic variation, and what it can tell us about the structure of language. In my work, I examine quantitative patterns of inter- and intra-speaker variation to determine what an observed distribution of alternating forms indicates about how those forms were generated. In doing this, I work toward the goal of developing a sociolinguistically-informed model of the grammar of individual speakers.

My dissertation concerns variation above the phonology. This encompasses variation in morphemes and lexical items (such as doublet phenomena) as well as the variable application of morphophonological and morphosyntactic processes. Such supra-phonological variables are complex and interesting because they implicate many levels of a derivation, meaning that the locus of surface variation is not always clear. As a result, they raise a number of questions about where in the grammar probabilistic behavior must be localized, and whether variables at different levels of the grammar display different patterns of linguistic and/or social conditioning.

In the dissertation, I use quantitative data gathered from large-scale corpora to inform theoretical models of variable linguistic phenomena. Specifically, I carry out a detailed case study of the linguistic and extra-linguistic constraints on variable English auxiliary contraction in corpora of telephone conversations and sociolinguistic interviews. Based on the results of this study, I argue that the observed patterning of variants has its source in both probabilistic alternations in the grammar and extra-grammatical performance effects. Consequently, the goal of this work is to develop an integrated model of variation that takes into account these two sources of variable surface data.

proposal abstract | proposal document | slides from proposal defense+presentation notes

As part of my thesis work, I've been carrying out a corpus study of the variable contraction of English auxiliaries (e.g. he has ∼ he's been there all day), looking at internal and external factors conditioning the observed surface variation. My database currently consists of over 14,000 tokens from the Switchboard and Philadelphia Neighborhood corpora.

By using natural-speech data, I am able to carefully consider the various phonological shapes that auxiliaries are found to surface in and their distribution. As a result, I draw novel conclusions concerning the grammatical mechanisms behind this alternation. Specifically, I propose that surface auxiliary realization is best accounted for by two stages of processes, with variation in the morphology followed by variable phonological processes.

I have presented on this work at NELS 41, NWAV 39, the LSA 2011 annual meeting, and WCCFL 29. Papers will appear in the NELS and WCCFL proceedings, and a longer manuscript is currently in preparation.

My data have revealed that contraction is conditioned by the length of an auxiliary's subject. I have further explored this effect of subject length in collaborations with Anton Karl Ingason and Constantine Lignos.

Anton and I presented on this work at MFM 19. Constantine and I will present on this work at GURT 2012. Additionally, I presented on the grammatical implications of the subject length effect at the LSA 2012 annual meeting.

Josef Fruehwald and I have been examining the rate of speaker-level divergence from overall community norms for contraction and other variable phenomena, using Markov Chain Monte Carlo Estimation.

We presented on this work at NWAV 40, and a manuscript is currently in preparation.

This work sprang out of a course in dialect geography taught by Bill Labov. Using linguistic atlases of southern France, I examined the weakening and loss of plural /s/ and its effects on the system of number marking in Modern Occitan. I found that weakening /s/ in Occitan goes through a stage of vocalization to [j], and that, in regions where /s/-loss is most advanced, vocalization of /s/ on the definite article concomitantly raises the preceding vowel. This raised vowel preserves plural meaning despite the weakening /s/.

I presented on this work at NWAV 38. A paper appeared in the NWAV 38 proceedings, and a manuscript is currently in preparation for the Journal of Dialect Geography.

This work examines a process of diphthongization that affects long vowels in Montreal French. In addition to providing acoustic measurements of this process, we have looked at the capacity of individuals to change their degree of diphthongization in the direction of community trends as they age.

We presented on this work at NWAV 37, the LSA 2009 annual meeting, and CVC III. A paper appeared in the NWAV 37 proceedings.

This work began with my undergraduate honors thesis, in which I examined the presence of excrescent velar nasals in Midi French. These velar nasals surface after what are nasalized vowels in Standard French. We investigated why nasal vowels would develop into vowel+velar nasal sequences, pursuing the hypothesis that velar nasals are acoustically and visually similar to nasalized vowels, and that the change could have been based in listener misperception.

We presented on this work at the LSA 2008 annual meeting. A paper appeared in the UC Berkeley Phonology Lab Annual Report.