Studies have shown that perceived macro-social categories like location of origin, age and class can influence listeners’ perceptions of linguistic variables. Other work in sociolinguistics has demonstrated that variables can index multiple social meanings, often associable with personae that are more specific and complex than macro-social categories. This paper brings together these lines of inquiry, testing how persona-based social information influences linguistic expectations in a vowel categorization task. The experiment examines multiple social meanings of one sociolinguistic variable: the backing of the TRAP vowel. By virtue of its patterning in California, TRAP-backing has social meanings related to macro-social Californian location of origin, as well as to Californian social types like the Valley Girl. The feature has separately been associated with professional, formal personae. In a vowel categorization task, listeners categorized continua of ambiguous auditory words as either containing a TRAP vowel (e.g., SACK) or a LOT vowel (e.g., SOCK). Prior to the task, listeners were either told that the speaker was from California (macro-social information), the speaker had ‘been described as’ a Valley Girl, or a Business Professional (persona-based information), or they were not given any speaker information (Baseline). Listeners in both the macro-social and persona-based information conditions were more likely to respond to a given token as TRAP than listeners in the Baseline condition, indicating an expectation of TRAP-backing created by all three social meanings of the feature. The effect was strongest in the Business Professional condition overall, but when the token was particularly backed, the Business Professional effect disappeared, while a strong effect emerged in the Valley Girl condition. These findings demonstrate that persona-based information about a speaker can lead listeners to expect an associated linguistic feature as strongly, if not more strongly, than macro-social information. Crucially, the strength of the effect depends upon the phonetic manifestation of the variable, among other aspects of the speaker voice, listener background, and situational context.