Voices of California

A major part of my work involved data collected during field work With the Stanford Voices of California project, which interviews English speakers across the state anually and represents one of the largest and most geographically diverse data sets of California English. I have conducted interviews at the three most recent field sites: Humboldt County, Redlands, and Amador County. My chief interest in this work is to understand how language is changing across the state, particularly with regards to measures not frequently studied in the existing literature.

My second qualifying paper used the data from the five field sites to challenge the claim that Californians have a merger in production between the LOT and THOUGHT vowels. Better known as the cot-caught merger, previous work has found significant overlap in formant frequencies among Californians, but work on other mergers across North America has found that vowel classes which appear to have merged along the first and second formants may still differ on a third dimension such as dynamism, length, or vowel quality. Using the Voices of California data, I replicated previous findings of convergence in first and second formants, developed a novel method to anaylze whether the formant trajectories of the vowels were converging in both formants simultaneously, and analyzed whether the vowel classes were diverging in length. The paper, in preparation to be submitted, presents evidence that while the formants of the vowels are converging over apparent time, they are simultaneously diverging in length suggesting that our previous claims of merger should be revisted with more scrutiny.

Alongside my colleagues, I also use the data to investigate how these changes are used by speakers to construct an identity and convey social meaning. At the 2020 Meeting of the American Dialect Society, I presented joint work investigating how the phonetic and phonological patterning of the TRAP vowel is used by Latinx Salinas speakers to index their orientation towards town or country ideologies (in many ways similar to Paul Reed’s “rootedness”). Among white Californians, TRAP shows a pronounced nasal split, where the vowel is raised before nasals and backed elsewhere. The backing pattern is seen amon Latinx Californians in our data, but the nasal split is less common. Our data show that among Latinx speakers, more “country” oriented speakers show a smaller difference between prenasal and non-nasal low-front vowels. This finding is the reverse of what has been foud among white speakers, and shows that depending on the particular social landscape a language is enbedded within, the same change can acquire different social patterns.

I have contributed to this project in other, often more technical, capacities. As of 2020, I am working with Lily Clifford to improve our corpus management infrastructure. As part of my update to FAVE, I am also planning to create an automated data pipeline for the alignment of the corpus and its metadata which I hope will be generalizable to other sociolinguistic corpora. As part of the interview, participants draw on a map to indicate what regions they see California divided into, and to digitize and analyze this data I have helped design and document a geographic information system (GIS) workflow. Much of this work comes out of ad hoc scripts I wrote in the course of working with the data, and so I view this cycle of development helpful not only for making my own work easier in the future, but improving the usability of the corpus for everyone.


Ende (ISO 639: kit) is a Pahoturi River language spoken in southwestern Papua New Guinea. To the field’s detriment, many of our sociolinguistic theories of language change are not robustly tested on data from understudied languages. Along with Kate Lindsey and other field workers in the area, I have started a number of projects using their data to investigate changes in Ende and its relationship to its sprachbund neighbors. Most recently I have been helping with a comparative analysis of the Pahoturi River languages to better understand their historical relationship to each other and hopefully their connection to other language families.

My main work has been on invetigating a pattern of nasal elision in an Ende copula which is socially patterned and potentially a change in progress. Like in California’s low-back vowels, it is not clear whether a distinction was lost or whether it has moved to a different dimension. Our work so far has been investigating the quality of the nasal consonant and whether the change represents a categorical loss or a phonetic reduction. Using acoustic measurments as articulatory proxies we find evidence that suggests the tongue-tip gesture in /n/ is not lost, but reduced. Further work will examine the velar gesture and the quality of the preceding vowel to determine whether the vowel is becoming nasalized due to the consonant weakening.