This talk examines the emergence of ethnic indices in the context of regional vowel shifts. My analysis draws on the local meanings of phonetic variables, showing that those meanings blur boundaries between traditional notions of ethnicity and region.

Data come from a subset of sociolinguistic interviews conducted in Spring 2008 with 86 English-dominant speakers from the Sunset District, a working/middle class neighborhood of San Francisco, California. The neighborhood has undergone rapid demographic change over the past few decades. What was once a predominantly Irish Catholic neighborhood is now approximately 43% White and 51% Asian, with 77% of those Asian Americans identifying as Chinese. In a city full of trendy neighborhoods, this "sleepy" residential district is seen by its residents as authentically San Francisco. For some people, the Sunset's dominant Asian population adds to this sense of local authenticity. Most residents, regardless of ethnicity, have multiple social ties to Asian Americans, and Asian identities are considered "cool" for some younger people. Through these links to San Francisco authenticity, I suggest that indexical resources of Asianness become available to the wider neighborhood community. At the same time, sound changes not indexing ethnicity are progressing across the population as a whole.

This talk focuses on three linguistic variables: the split movement of the (ow) vowel, the vocalization of coda-/l/, and the merger of the low back vowel classes. The fronting of (uw) and (ow) are well- recognized components of the Northern California vowel shift (NCVS), occurring in all phonological environments except preceding /l/ (Hinton et al. 1987; Luthin 1987). As a relatively recent innovation in California, fronting is more frequent among younger speakers in the Sunset District, both White and Asian American. At the same time, other young speakers, particularly Asian Americans, produce extremely backed variants of these vowels. Given the prevalence of back vowel fronting across California and across ethnicities, production of backed variants within the context of the NCVS appears to index an orientation to the Sunset District neighborhood and local meanings of ethnicity, rather than marking ethnicity as a context-free category (Eckert, to appear).

Secondly, some second-generation Asian Americans produce fully vocalized coda- or coda-cluster-/l/ (e.g., COLD approximates CODE). That Whites and higher-generation Asian Americans show more constrained /l/-vocalization suggests that this variable is a nativization of a second language feature, and a potential linguistic resource for Sunset District speakers that is more clearly indexical of ethnicity than is the movement of (ow). Finally, despite evidence for the regional prominence of the low back merger (Labov et al. 2006) and its early presence in San Francisco in particular (DeCamp 1953), my data supports the finding that the COT and CAUGHT classes are still distinct for some San Franciscans (Moonwomon 1987, 1992; Labov et al. 2006). Furthermore, the distinction is found for both Whites and Asian Americans, correlating significantly with age group rather than ethnicity.