Talker specificity and the perception of grammatical variation

Abstract

Language processing has been shown to be affected by both recent exposure to linguistic structures (priming effects) and non-linguistic, indexical information (talker effects). Three experiments explored two hypotheses: that sociolinguistically variable subject–verb agreement constructions are susceptible to grammatical priming, and that comprehension of these constructions is susceptible to talker-specificity effects. Experiment 1 tested for grammatical priming effects on the interpretation of two subject–verb agreement constructions, there's+NPSG/PL and NPSG/PL+don't. In a forced choice task, participants chose which of two images represented the subject noun that was in a sentence; participants read an unambiguous prime sentence and then heard an ambiguous target sentence. Experiment 1 found a priming effect: nonstandard primes increased participants’ perception of nonstandard agreement in target sentences. Experiment 2 tested whether this priming could be modulated by non-linguistic cues by manipulating whether the prime and target sentence were spoken by the same talker or by different talkers, cued by voice and photograph. Experiment 2 did not find a talker specificity effect. In an alternative design using identical prime and target sentences, Experiment 3 tested for talker-specific lexical repetition of subject nouns across primes and targets separated by intervening trials. Experiment 3 found a talker effect: participants were better at matching a subject noun in the target block to the subject noun in the prime block when the talker was identical than when the talker differed. These results are discussed in terms of competing linguistic and non-linguistic cues during language processing, and the activation of sociolinguistic representations.