My dissertation [bib] argued that certain phonological patterns could best be modelled if it is assumed that phonological grammars can make reference to representations that are scalar but are not necessarily grounded in phonetics. The idea of phonological scales (n-ary features goes back to Trubetzkoy, if not earlier, but scalar features were typically assumed to follow a phonetic dimension (like vowel height). My proposal was a radical one that argued for a phonology that is grounded in phonetics only as a consequence of history.

This does not mean that phonetics is not important to phonology, only that the effect of phonetics on phonology is a diachronic one. As a result, languages often acquire phonological patterns which—while they once had a phonological basis—are phonetically nonsensical. An example of this is tonally conditioned vowel raising in Shuijinping Mang [bib].


I have a long-standing interest in compounding, especially in languages other than English. I once argued [bib] that the four-syllable constructions in Hmong called elaborate expressions (many languages of Southeast Asia have similar constructions) were actually coordinate compounds. Later, Mary Lou Vercellotti and I evaluated the the Scalise and Bisetto (2009) framework for classifying compounds by applying it to American Sign Language [bib].

Language documentation

I have been fortunate to do field work on a number of different languages, mostly from South and Southeast Asia. In the Tangkhulic group in the Tibeto-Burman family, I have worked on Ukhrul Tangkhul, Kachai, Huishu, and Tuson. Recently, my lab published a speech dataset for Tusom. My lab is currently producing other documentation on this language.

Among the members of the Tangkhul ethnicity (concentrated in Ukhrul District, northeast Manipur State, India) are speakers of non-Tangkhulic Tibeto-Burman languages. One of these is spoken in the area around Sorbung village and is called “Sorbung.” With a student, I produced a phonetic and phonological sketch of Sorbung [bib] with my student Jennifer Keogh. Hopefully, more work will be done on these langauges soon.

My longest linguistic association, however, has been with Hmong. I have produced numerous manuscripts on Hmong, over the years, and have compiled a digital corpus from postings to the soc.culture.hmong (SCH) Usenet group called the SCH Corpus [bib].

Historical linguistics

Being able to reconstruct the linguistic past provides an important window into other aspects of human history, but to me historical linguistics means more than just reconstructing forms and meanings at the maximum possible time depth. It means producing data that can be used to understand how language changes. That, it turn, provides our most powerful tool for understand how language works synchronically.

During graduate school, I worked on the Sino-Tibetan Etymological Dictionary and Thesaurus (STEDT) project, doing a lot of computational work with comparative lexical databases. I also did field work on Tangkhulic languages (a group within Tibeto-Burman) and, based on this field work, produced a preliminary reconstruction [bib]. With my student James Miller, I produced an updated reconstruction of the Proto-Tangkhulic Rhymes [bib], adding data from Tusom. We never produced an updated reconstruction of the onsets. Now that I have improved data from Tusom, and access to better computational tools, I hope to publish a new reconstruction of the whole group.

While I was reconstruction Proto-Tangkhulic, I noticed a peculiar sound change in the language Huishu—Proto-Tangkhulic *-u and *-i became Huishu -uk and -ik (a /k/ was inserted after high vowels). I knew that something like this had happened in another (distantly related) Tibeto-Burman language called Lhao Vo (Maru). Further investigation revealed that versions of this counter-intuitive sound change (which adds codas to open syllables) was also attested in Grassfields Bantu (in Cameroon) and in a few Austronesian languages. I published a paper [bib] arguing that there was a unified aerodynamic and perceptual basis for this class of changes.