norma mendoza-denton labov lecture: used telstar (sp?) data (interviewed subjects in different locations); how could he have said what he did? did he assume that social factors might be at the conscious level? was he trying to separate regional from social? defs of style (after patrick '95): compare labov '72,'82; bell 84; csc95, m-d '02 testing referee design: oprah monologues (m-d, hay jannedy '03). addressee/audience. monophtongization of /ay/; 229 tokens from '96-97. coded for linguistic factors (preceding/following segment); ethnicity of person she is referring to; varbrul; more likely to monophth if person referred to (referee) is af/am; following segmental context also influences and frequency of word [fn/content?] variation is multiply determined. oprah not shifting between discrete speech styles. tidewater voices project people need ML, stats; stuck on varbrul niedzielski '99: detroiters choose different 'matching vowels' depending on whether stimuli labeled as uttered by canadian or michigander strand '99 on different perception of /s/ and /sh/ depending on prototypicality of face given gender; why? making social assumptions pisoni: native speakers can recognize their own language only from lip movements (eng vs. spanish)