My research focuses on phonetic detail within phonology and how it fits into representations, establishing phonetic correlates of contrasts in production and testing how they influence perception and how they can shift based on experimental manipulations.

My second focus is on sound change. Modern perception and production data, in combination with traditional reconstruction and philology, can help establish pathways of diachronic developments.

Phonetics and Recording Methods in a Pandemic

Because of restrictions on travel and in-person research during the Covid-19 pandemic, many researchers have started making remote recordings, using a range of methods.  It is important to know how these methods might affect acoustic measurements, particularly for underdocumented languages, but also in any phonetic research.  Variation might arise either because of different recording devices or because of the software used to make recordings.  In a collaborative project, I led the Yale fieldwork group in investigating how varied recording methods impact phonetic measurements.  Different recording devices and software programs have a substantial impact on many of the measurements that we tested, including measurements of duration, frequency, and aperiodic noise.

Phonetics in Phonology

Even when a contrast is described as hinging on a particular characteristic, there are often many acoustic characteristics that correlate with that contrast, which listeners attend to differently depending on their native language. Awareness of cues is likely to be reflected in two ways: Use of it in identifying the relevant contrast, and compensation for the expected effect when identifying the cue itself. The different behaviors of these effects suggest that there are two types of cue representation: Fully phonologized cues, reflected in both effects, and awareness of a gradient probabilistic relationship, which produces much weaker compensation. One line of my research is on coda voicing, which has many phonetic correlates, most of which are clear consequences of producing voicing. A smaller number of these characteristics influence listeners' perception of voicing.

Phonetic details in production are influenced by a range of conditioning factors, both phonological and higher level. Part of speech and predictability in context can produce significant differences between homophone mates, but these details are largely absent when the same words are produced in isolation, suggesting that they are not part of the representation, but are the result of factors like predictability in context.  Higher frequency words are accessed more easily and are correspondingly produced more rapidly; I show that repeated recent exposure to a word results in a lower VOT category boundary, regardless of if exposure items had shortened or lengthened VOT.  Recent exposure to homophones can increase listeners' accuracy in discriminating between homophone mates, suggesting that training can draw listeners' attention to acoustic differences caused by factors like lexical frequency and emotional valence.  On the other hand, I find no evidence that speakers can be trained in arbitrary new acoustic targets for particular words; in a shadowing task with stimuli manipulated in opposing directions in different words, the manipulation produced no post-task effect. 

Another line of my research looks at convergence, which provides a look into how input influences the representation and thus subsequent outputs. Elicited shifts are highly shaped by the phonology.  Acoustic shifts are generalized at the featural level: exposing listeners to a raised or lowered F1 or F2 in one vowel produces corresponding shifts in vowels that match in height or frontness, respectively. Shifting one word also influences other words with the same sounds, such that opposing shifts across words end up having no effect. Apparent word-specific convergence may result from interactions of lexical frequency and repetition effects; even in a reading task with no exposure to other speakers, speakers' productions become more similar to each other with more repetitions, particularly for low frequency words. 


Apparent word-specific convergence is particularly common when convergence is measured with the difference-in-difference method, which also can create the appearance of individual differences. However, there is limited evidence for individual differences in convergence; when compared across different characteristics beaing measured, some interlocutors are more likely to elicit convergence than others, but there are not some speakers who are more influenced by their interlocutors than others. Any speaker-specific tendencies in convergence are highly specific to the characteristic being measured, which suggests that variation in convergence is not driven by broad cognitive differences but by differences in cue weighting or attention.

Evidence for Sound Changes

While experimental data alone is not sufficient to provide conclusive evidence for pathways of sound change, patterns in perception and production can clarify the relative likelihood of developments and illustrate how they occur. For example, there are many parallels between misperception of sounds and sequences in laboratory experiments and in known historical changes, which can provide a model for sound changes that are caused by misperception.

One of the main patterns that I am investigating is voicing-conditioned vowel duration; some correlates of coda voicing provide a likely pathway for developing voicing-conditioned vowel duration.  I examine how vowels are influenced by coda voicing and how those characteristics influence perceived vowel duration. When the codas are removed, vowels that had been produced with voiced codas are perceived as longer than vowels produced with voiceless codas.  Perception tests isolating correlates of voicing confirm that spectral tilt and intensity contour can produce these effects on perceived duration.


In addition to broad connections between diachronic phonology and synchronic patterns of errors and variability in perception and production, I also seek synchronic parallels to test specific reconstructions.  This is particularly useful for changes which lack clearly reconstructed parallels that would provide a foundation for a typology, such as the developments of the Proto-Indo-European "laryngeals" or the sound change proposed in Winter's law, with lengthening only before voiced unaspirated stops.

I also have some more traditional reconstruction projects, particularly combining comparative reconstruction and written records to reconstruct sound changes and synchronic patterns in ancient languages of the Near East.

Language Documentation

Currently I am part of a group studying A'ingae, a South American language isolate; we have produced the first instrumentally based overview of its phonological system and continue to elucidate both areally motivated characteristics of its phonology and unique characteristics, such as narrative use of syllable-level falsetto.

As part of my collaboration with the A'ingae fieldwork project, we have been working on reconstructing phonological changes. Although A'ingae has no known linguistic relatives, synchronic allomorphy and limited distributions of sounds suggest several sound changes that substantially restructured the system, including development of contrastive prenasalized stops and development of vowel nasality.