The magnet effect on interdialectal speech perception: evidence from the palatalization of /t/ in Brazilian Portuguese

Abstract

The main objective of this dissertation is to investigate the magnet effect on listeners’ discrimination of the duration of /t/ frication burst, from the palatalization of /t/ before /i/ in Brazilian Portuguese. The palatalization is generally characterized by when an alveolar plosive is produced as an affricate (BISOL, 1986; HORA, 1997; BATTISTI; ROSA, 2012; AMORIM et al., 2019). The magnet effect refers to the influence of language exposure on listeners’ discrimination. It is characterized by a reduction in sensitivity to acoustic changes in the region near good exemplars (prototypes), and an increase near poor exemplars (non-prototypes) (KUHL et al., 1992; KUHL et al., 2007). For this dissertation, five experiments were conducted with two groups: natives and non-natives. Natives were the individuals from João Pessoa, Paraíba, Brazil, where the palatalization is the most frequent variant. The non-natives were the individuals who live in João Pessoa, but lived most of their life in a place where the affricate is the most frequent variant. The first experiment investigated the distribution of the duration of /t/ frication burst and provided helpful insights for the elaboration of stimuli from the other experiments. The second experiment was a discrimination test, which analyzed the effect of language exposure on listeners’ discrimination when comparing tokens that the burst differed in 5, 10, 15 and 20 ms. The third experiment was a pilot, an ABX task. It examined the influence of language exposure on listeners’ responses, but it failed because the stimuli differed only in 5 ms. The forth experiment is a goodness judgment task. It showed that the goodness varied with the stimuli. It contributed to validate the stimuli to be used in the following experiment. The fifth experiment was a ABX test, which was similar to the pilot, but using the stimuli from the goodness test. It confirmed the main hypothesis of the current study when demonstrated that listeners’ responses depend on language exposure. Concluding, the current dissertation is in line with prior studies on the Native Language Magnet model, which show that listeners’ sensitivity to acoustic changes is motivated by language experience. The stimuli that were good exemplars yielded a decrease in sensitivity to acoustic changes, whereas those that were poor exemplars yielded an increase, what indicates the magnet effect.