S-assimilation in English and Italian: implications for foreign language learning and teaching

The alveolar fricative /S/ exhibits a strong crosslinguistic tendency to assimilate in voicing to the neighbouring sounds. In accordance with this trend, both English and Italian display a range of s-assimilation processes, from the most stable alternations rooted in the lexicon to the most variable connected speech phenomena. This study takes into consideration two obligatory processes, each of which is present in one language and absent in the other, causing some of the most persistent transfer phenomena in the pronunciation of Italian learners of English. The first one is the well-known phenomenon whereby word-final /S/ undergoes progressive assimilation of voice to a preceding segment whenever a morphological boundary intervenes, namely when it makes up the regular plural and the genitive noun affixes and the third person singular verb affix. No similar process applies in Italian, where an unassimilated word-final [s] appears very rarely and only in postvocalic position. On the contrary, in the Romance language /S/ assimilates regressively in voice to a following consonant, independently of the presence of a morphological boundary (e.g. [z]mania ‘frenzy’ vs. [s]trano ‘strange’). In the same environment, English /S/ is never voiced; rather, it triggers the (at least partial) devoicing of a following sonorant. Given the different distribution of these assimilation phenomena, Italian learners of English seem to encounter particular difficulty in adjusting to the target system. The question pertaining more specifically to the field of language learning and teaching is: should these typologically unmarked but to some extent language specific phenomena be explicitly taught when teaching English? Some (cf.. Jenkins 2000, Roach 2003) say ‘no’, based both on their purported naturalness, which would make them accessible to anyone getting the ‘general feel’ of the language, and on their scarce influence on intelligibility. To test, in a preliminary way, both hypotheses, I set up an experiment involving six Italian advanced learners and two native speakers of English as controls. In the first phase of the experiment, the Italian learners were made to produce the target sounds in different contexts, in order to see whether and to what extent assimilation would be performed correctly by them. The results showed considerable divergence from the L2 forms, even where these had been explicitly taught. In the second phase, the two native speakers of English were made to listen to the recordings, to see whether and to what extent the mispronounced forms were intelligible to them. Apart from a few minimal pairs produced in isolation, incorrect voicing of /S/ did not seem to constitute a particular obstacle to comprehension. For this reason, and to prevent the risk of transfer of training, I suggest that both types of assimilation should be taught only to advanced learners of English, as a metalinguistic reflection rather than as a rule to be applied.