differs substantially in acoustic characteristics â some have significantly more centralised systems, which might affect boundary location (cf. Sadowsky 2012 for ...
Regional variation in loanword adaptation: phonological or acoustic repair? Fernanda Barrientos (U. of Konstanz) and Deepthi Gopal (U. of Cambridge) According to Calabrese (2009), two competing accounts of loanword adaptation exist: (a) naive reanalysis, where repairs are driven solely by the action of listeners' L1 perceptual systems on the foreign phonetic input, with no reference to external phonological systems (Silverman 1992); and (b) phonological adaptation, where the final state of adapted loans references L2 phonological information: the representations derived by bilinguals from their L2 knowledge are then coerced into the L1 system, which might result in a less acoustically-faithful solution (LaCharité & Paradis 2005). Given a donor-recipient language pair, Heffernan (2007) claims that the probability that adaptation may be caused by (a) rather than (b) corresponds to the population-level incidence of bilingualism. We consider here the adaptation of English words with /ʌ/ by speakers of different varieties of Spanish, all of which share the same 5-vowel system. This vowel triggers a situation wherein more than just one outcome is plausible: both /a/ and /o/ have been attested as possible candidates, and the overall winner varies significantly across Spanish-speaking regions. If we assume that all Spanish speakers use the first repair strategy (i.e. naive reanalysis), then all of them will strictly map /ʌ/ onto the acoustically-closest segment, unless each dialect has different perceptual spaces. In this regard, differences in perceptual boundaries among languages with acoustically-similar five-vowel inventories have been established (Boersma & Chládková 2011); furthermore, the Spanish 5-vowel system in fact differs substantially in acoustic characteristics – some have significantly more centralised systems, which might affect boundary location (cf. Sadowsky 2012 for overview and comparison between a peripheral system eg. Tijuana, Mexico and the centralised system of Concepción, Chile). If this phonetic variation causes perceptual boundaries to vary across different varieties of Spanish, then one hypothesis is that this variation drives differences in the adaptation of loans containing vowels such as /ʌ/. The second option is that the differences in adaptation result from different strategies: while some communities use an acoustically-driven strategy, others use phonological repair. We carried out an online survey which asked subjects from different Spanish-speaking countries about adaptations of English words with /ʌ/. The survey showed pictures of these words and asked subjects to choose one of three possible adaptations. The options were transliterations into Spanish, e.g. panc, ponc, punc, plus an “I don’t use this word” option (results in Fig. 1). A second online survey/experiment was carried out, this time with acoustic stimuli from two 7-step continua built with the Praat Klatt synthesizer (Boersma & Weenink 2017); one with Chilean mean values for /a/ and /o/ (Sadowsky, 2012) and the other one with Mexican values (Grijalva et al 2013). Listeners from Chile and Mexico were asked to categorise the stimuli into the Spanish categories /a/, /e/, and /o/. The results (Fig. 2) show that Chilean speakers have a perceptual space that accepts centralised acoustic values as instances of /a/, which could be explained by its centralised vowel system. However, the Mexican continuum yields the same perceptual pattern in both groups. This suggests that while Chilean speakers choose an acoustic solution (thus mapping /ʌ/ onto /a/), Mexican speakers prefer a phonologically-based approach (i.e. /ʌ/ mapped onto /o/). This hypothesis can be further sustained by the higher levels of bilingualism in Mexico and Puerto Rico (both with similar adaptation patterns) than in Argentina and Chile (also with similar patterns), thus confirming Heffernan’s claim. The possible origins that we suggest for variations in adaptation strategy thus conspire in the Chilean case – both the acoustically-compressed vowel space and the low rates of bilingual competence favour their variant mapping.
Figure 1. Output proportion for each loanword by country
Figure 2. Proportion of L1-like perception along Chilean and Mexican /a-o/ continua. Boersma, P. & Chládková, K. 2011. (2011). Asymmetries between speech perception and production reveal phonological structure. Proceedings of ICPhS XVII 2011, 328-331. Calabrese, A. 2009. Perception, production and acoustic inputs in loanword phonology. In Calabrese, A & Wetzels, W. (Eds.) Loan Phonology. Amsterdam: John Benjamins. 59 - 114. Heffernan, K. 2007. The role of phonemic contrast in the formation of Sino-Japanese. Journal of East Asian Linguistics 16: 61-86. Grijalva, C., Piccinini, P. E., & Arvaniti, A. (2013). The vowel spaces of Southern Californian English and Mexican Spanish as produced by monolinguals and bilinguals. In Proceedings of Meetings on Acoustics ICA2013 (Vol. 19, No. 1, p. 060088). ASA. LaCharité, D. & Paradis, C. 2005. Category preservation and proximity versus phonetic approximation in loanword adaptation. Linguistic Inquiry 36(2), 223-258. Sadowsky, S. 2012. Sociolinguistic stratification and phonetic description of the vowel allophones of Chilean Spanish. PhD dissertation. Univ. de Concepción. Concepción, Chile.
