Variabilidade fonética na produção de formas de plural do inglês como segunda língua

Abstract

The aim of this thesis is to investigate the phonetic variability attested during the production of plural forms in English as a Second Language by Brazilian speakers. Production of such forms is evaluated from two aspects: (1) alternation of word-final [Cs±] ~ [Cis±] sequences (e.g. cakes [keɪks] ~ *[keɪkis]) and (2) sibilant voicing (e.g. jobs [dʒɔbz] ~ *[dʒɔps]). In order to assess whether an ongoing L1 sound change influences L2 productions, we also investigated emerging [Cs±] sequences in Brazilian Portuguese (BP), e.g. crepes [kɾɛpis] ~ [kɾɛps] and cheques [ʃɛkis] ~ [ʃɛks]. This study is mainly grounded on the assumptions of Exemplars Theory (JOHNSON, 1997; PIERREHUMBERT, 2001; BYBEE, 2001, 2008, 2016) and Complex Dynamic Systems Theory (VERSPOOR; DE BOT; LOWIE, 2011). Studies on the effects of orthography on L2 speech are also considered (DELATORRE, 2006; SILVEIRA, 2007; COLANTONI et al., 2015). Methodologically, this work is based on the premises of Laboratory Phonology (PIERREHUMBERT; BECKMAN; LADD, 2000). Data collection involved speech recordings of twenty Brazilian speakers of L2 English. Reading and picture counting tests were adtopted for the investigations of both languages. Statistical analyses were performed using Pearson's chi-square (χ²), Wilcoxon and Kruskal-Wallis tests. Overall, results indicate that [Cs±] sequences are robust in both BP and L2 English, pointing to the influence of empirical L1 data on the production of L2 sound patterns, as predicted by Exemplars Theory. In terms of orthography, results indicate that English [Cs±] sequences are favored by the orthographic pattern <Ces> regardless of the presence of written input, suggesting that speakers have access to pre-activated spelling patterns in their cognition. Additionally, it was found that word-final devoicing constitutes an ongoing sound change in both BP and L2 English. We also found that L2 learners tend to borrow two phonological patterns from their L1: sometimes they voice sibilants due to a following word-initial vowel (e.g. bags are [bӕɡzɑːr]) and sometimes they do not voice sibilants due to word-final devoicing (e.g. bags are [ bӕksɑːr]), leading to the production of both target and non-target forms. In terms of Complex Dynamic Systems Theory, we suggest that there is a competition between the following phonological environment and the phenomenon of word-final devoicing. Results are also in line with the assumptions of Exemplars Theory, which predicts that fine phonetic detail - be it associated with stable, allophonic or emerging sound patterns - is used to build L2 phonological representations.