We welcome both undergraduate and graduate applicants who are interested in learning about and conducting research on phonology.
Phonology investigates the organization of sounds in the world's languages and the principles that give rise to those patterns: Why are certain sound patterns common and others rare or nonexistent? Which combinations of sounds can languages exhibit, in terms of both inventories and phonotactics? How do these segmental considerations interact with other structures like prosody and morphology? What is the possible range of variation across languages and dialects? And most fundamentally, what principles are responsible for these things?
My research probes the kinds of operations phonological theory must formalize. Noniterative and optional processes are central to this endeavor because they seem to require explicit mention of processes (as opposed to surface-form criteria), which OT eschews. I am also interested in prominence-based licensing phenomena, in which some element is permitted only if it has membership in some prominent position such as a stressed syllable. The range of phenomena of this sort imposes stringent demands on the constraint types (Positional Licensing and Positional Faithfulness) that produce them; my research aims to refine these constraints to reflect the typology of prominence-based systems and examine the behavior of these constraints in different frameworks: OT, Harmonic Grammar, Harmonic Serialism, etc.
We argue that while Harmonic Grammar allows Positional Licensing to do some of the work that Positional Faithfulness must do in OT, we cannot dispense with Positional Faithfulness altogether. Harmonic Grammar requires both Positional Licensing and Positional Faithfulness, just as OT does. (Here are a handout from our talk at the 2014 LSA Annual Meeting and posters from NELS 44 and Phonology 2013.)
This paper argues that to avoid certain pathologies, under Harmonic Grammar positional licensing must be a gradient, positive constraint (Here is a much shorter version from the proceedings of AMP 2015, and here is the corresponding poster. Here is a handout from a talk I gave on the subject at the 37. Jahrestagung der Deutschen Gesellschaft für Sprachwissenschaft (2015) [37th Annual Meeting of the German Society for Linguistics].)
When Positional Licensing causes a strong position like a stressed vowel to harmonize with a non-local trigger such as the word-final vowel, the vowels that appear between these positions show three crosslinguistic patterns: harmony, transparency, and opacity. I argue that while these three options receive unified treatments in OT and Harmonic Serialism, in Serial Harmonic Grammar a unified account is impossible: Positional Licensing constraints must be responsible for the harmony and transparency options, but we must invoke other constraints to produce opaque vowels.
This poster uses Harmonic Grammar to construct an alternative analysis of the vowel harmony dealt with in the 2017 mfm poster below. Vowel harmony in Tudanca Montañés (Romance; Spain) occasionally overshoots its stressed-vowel target, and I argue that this is best captured with a Positional Licensing formalism that rewards harmony not just in the licensor, but in other positions as well (as in the theory developed in a paper forthcoming in Phonology).
In most Positional Licensing systems, harmony stops at the licensor. But in Eastern Andalusian and Tudanca Montañés (Romance; Spain), harmony extends beyond the licensor under specific conditions. I argue that this provides evidence for the theory of Positional Licensing developed in a forthcoming paper in Phonology and against standard theories of Positional Licensing.
Vowel harmony in Tudanca Montañés (Romance; Spain) fosters a non-local derived environment effect: pretonic non-mid vowels undergo an independent centralization process only if post-tonic centralization harmony occurs. Standard approaches to derived environment effects do not capture the interaction, but using Dep-[F] to regulate featural faithfulness does. Dep-[F] blocks the pretonic process from introducing a centralization feature, but once centralization appears for post-tonic harmony, the pretonic domain can make use of that feature.
In other work I have argued that formulating positional licensing as a positive constraint in Harmonic Grammar has significant benefits. In this talk I ask whether the same goes for positional faithfulness. I argue that the answer is no: pathologies introduced by positional faithfulness persist under a positive formulation of that constraint type.
I argue in this paper that positional licensing constraints may target only "maximally" prominent positions (e.g. primary stress but not secondary stress), while positional faithfulness can target any prominent position. This explains crosslinguistic asymmetries in the kinds of processes the two constraint types give rise to. (Here is an earlier version that I presented at the 21st Manchester Phonology Meeting (2013)).
This paper argues that even though OT with Candidate Chains offers a simple analysis of Chamorro umlaut, a standard OT analysis is more insightful and superior on typological grounds.
This poster is an early version of a paper that appeared in LI in 2011.
Chamorro umlaut seems to target a word-initial stressed syllable. It does not target non-initial stress. I argue instead that word-initial stress is the trigger of umlaut; failure of umlaut with non-initial stress is due to the lack of a trigger, not the distance of the target. (A longer version of the analysis appears in my dissertation.)
This handout addresses the same issue as MALC paper immediately above, but it focuses on the argument that umlaut is a response to the weakness of pretonic syllables in Chamorro.
This paper explores different implementations of Noisy Harmonic Grammar and MaxEnt and assesses how well they model variable ATR harmony in Eastern Andalusian. The most successful implementation, "classic NHG," adds noise to constraint weights at the outset of the evaluation. This has implications for CON: unlike other versions of NHG, classic NHG cannot produce harmonically bounded outputs, so we can use its success to narrow down the analytical options, picking a constraint set under which no attested form in Eastern Andalusian is harmonically bounded. Here is a different version of this work that I presented at the 2018 Analyzing Typological Structure workshop at Stanford.
This paper argues that Partial Orders Theory, an account of optionality that involves
adopting variable constraint rankings, can account for local optionality, contrary
to claims in the literature. If the constraints involved in a variable ranking refer
to specific positions (as many broadly accepted phonological constraints do), Partial
Orders can produce outputs in which an optional process has applied at a proper subset
of the available loci, an outcome previously argued to be impossible in this theory.
OTSoft/OT Help/OTWorkplace files:
English: marketability_repetitive.txt marketability_workbook.xls;
Pima (capitalization indicates a reduplicated stem): apricotpie.txt glassdishcloth.txt tamarack.txt wagonknife.txt
We present results from a corpus study of the optionality of French schwa to answer two main questions: Does schwa's optionality exist in the form of intra-speaker variation (i.e. do individual speakers produce multiple surface realizations for a single word)? And do different speakers show different rates of schwa omission? We answer "yes" to both questions. This has implications for theories of optionality: in particular, not all such theories allow the frequency at which a variable process is applied to vary across speakers. (Here is the poster we presented at Phonology 2013.)
This paper develops a new way of accounting for optional process application in OT that covers a wider range of phenomena than standard approaches.
This poster is an early version of the "Variation through Markedness Suppression" paper.
Building on Zwicky (1976), we investigate patterns of consonantal mismatches in imperfect rhymes, in which the rhyming words are not a perfect match (e.g. long/gone). We collected imperfect rhymes from popular music and investigated the mismatched consonants from the point of view of feature theory. We find that in general, consonant pairs that differ on few features are more common than those differing on more features. This suggests that featural measures of similarity provide a plausible model of speakers' judgments about consonantal similarity as those judgments are reflected in imperfect rhymes. But not all features are equal: some mismatch more than others, and some combinations of featural differences are more common than others. Above is the handout; here are the corresponding slides.
This paper examines the various ways in which tones interact with reduplication and the restrictions on reduplicant size in Adhola.
I argue that noniterativity in phonology always emerges from the interaction of independent factors (adjacency of trigger and target, uniqueness of target, etc.) and never from a requirement that a process may occur just once.
This is a condensed version of the analysis from my dissertation.
Myers (1999) presents evidence that noniterative tone spread in Chichewa is just peak delay, whereby a high pitch target is not reached until after the phonologically high-toned syllable. This paper formalizes peak delay in an OT analysis to account for tone spread in Chichewa and tone shift in Kikuyu. It is suggested that peak delay may be behind other reported cases of noniterativity in tone.
This paper argues for the position that the syllable is the universal tone-bearing unit, even in the face of evidence that the mora is the TBU in at least some languages.
This paper argues for an elaboration of syllable structure to account for coda cluster restrictions in Misantla Totonac.
This paper reconciles long-distance movement in Chamorro with Minimalism's Phase Impenetrability Condition.
This is a longer version of the paper immediately above.