'Ota witch?': modeling the relationship between linguistic variation and acquisition of branched attacks via the principle of tolerance

Abstract

This paper aims to revisit the acquisition of CCV (Consonant1+Consonant2+Vowel) branching onsets in Brazilian Portuguese. We investigate how the syllable structure is categorized throughout children’s phonological development and why this development occurs the way it does. To better describe the target adult speech (from São Paulo), a corpus study quantifies the CCV→CV variation (as in ‘outro’→[ʹo.tʊ]) (SP2010 Project, MENDES, 2013). To describe the child development, we conducted a speech production and comprehension test (mispronunciation detection task). As the theoretical framework, we adopted the Tolerance Principle (YANG, 2016), using it to model CCV-CV contrast construction. We assume that the contrast stablishment is based on the growth of the Lexicon’s phonological density (JUSCZYK, LUCE & LUCE, 1994). Our study finds that, when acquiring CCV, the child acquires a poorly dense and phonetically variable syllable, which is susceptible to sociolinguistic processes of CCV-CV contrast neutralization. We argue that such input characteristics lead to an incorrect generalization of CCV as optional, taking CV as an alternate form of CCV – both in unstressed contexts (which is found in adult speech) and in stressed contexts (which does not occur in adult speech). The productivity of this hypergeneralization is captured by the Tolerance Principle and stems from the high concentration of reducible CCVs in the child's initial vocabular. The hypergeneralization of CCV~CV variation is reflected into the mispronunciation detection task. The test points out to the recognition of CV→CCV (‘dente’→‘d[ɾ]ente’), but not of CCV→CV (‘prato’→’pato’, ‘preto’→’*peto’) by children who simplify CCV in their speech. The higher detection rate of CCV-CV neighbors (‘prato’→‘pato’ is better detected than ‘preto’→‘*peto’) points to the construction of contrast as a keypoint in phonological development. Therefore, we stand that CCV acquisition goes through a moment of incorrect neutralization of syllable contrast guided by sociolinguistic variation in the input.