Perceptual similarity across multiple sociolinguistic variables

Although variationist studies of sociolinguistic performance have greatly enhanced our understanding of speakers’ use of linguistic variables and their correspondence with social and contextual factors, these works have largely overlooked the perceptual complement of the production behavior they capture. This paper reports the results of a study which uses multidimensional scaling (MDS) to model listeners’ similarity judgments of tokens of the word going produced by male and female talkers of Southern and non-Southern dialects of American English, with and without g-dropping and vowel quality differences in the suffix –ing. Listeners were found to weigh heavily the more fluid factors (g-dropping and vowel quality) and nearly ignore the more static characteristics (talker dialect and gender).