A Morphophonological Account of Spanish/English Word-Internal Codeswitching

The current study examines whether a codeswitched word (a word with morphological elements from two languages) can also contain phonological elements from two languages (MacSwan & Colina, 2014). For instance, consider the Spanish/English codeswitched word hangear. Even though this word is comprised of the English verb ‘to hang’ and the Spanish infinitival affix -ear, it is thought to not display corresponding phonology (i.e. English in the root and Spanish in the affixes). Instead, review of codeswitched words across the literature reveals a consistent pattern: codeswitched words evince the phonology of the language of the affixes. However, there are no studies that provide acoustic evidence to provide support or opposition for this pattern. In this dissertation, I aim to provide such evidence via two experimental paradigms: a production task and an aural acceptability judgment task. Both experiments investigated codeswitched verbs comprised of English roots and Spanish affixes. The English root in these mixed words contained one of three English phonemes /z/, /θ/ and /ɪ/ that do not exist in the Spanish of the participants. In the production task, the maintenance or substitution of these phonemes was used as indication of English or Spanish phonology, respectively. The results indicate that the bilinguals produced codeswitched words containing /z/ and /ɪ/ in the English root with Spanish phonology. For the codeswitched words with /θ/, some bilinguals maintained /θ/ in the English root while others did not. In the second experiment, participants were asked to rate codeswitched words with three different phonological instantiations (Spanish, English, mixed) on a 1-7 Likert scale. The results demonstrate that the participants provided the highest ratings to the codeswitched words with Spanish phonology and support the findings from the production study. The combined results of the production task and the acceptability judgment task provide evidence to support the prediction that codeswitched words evince the phonology of the language of the affixes. While there was considerable variability in the production study between the three phonemes, the majority of the codeswitched words comprised of English roots and Spanish affixes demonstrated Spanish phonology. These trends align with previous CS accounts (e.g. MacSwan & Colina, 2014) that posit a ban on phonological codeswitched words (albeit one that is subject to restrictions of applicability conditioned by factors such as phonological dissimilarity) and suggest that a bilingual’s two phonological systems do not interact at the word level.