Deflected contrast

Eckman et al. point out some situations where some phonemic mismatches between L1 and L2 result in an intersection of two interlanguage substitutions, and that one of these substitutions is systematically blocked. The rendition of English interdentals /θ, ð/ by Portuguese speakers provides a good case for this. The typical substitution for the /θ/ target is [t] by the learners (e.g. thank realized as [tænk]). As we saw earlier, Portuguese also under-differentiates the English /t/ – /tʃ/ contrast. Since [t] and [tʃ] are the allophones of a single phoneme in Portuguese, learners pronounce the target tip and chip homophonously. While the learners realize the English target /t/ as [tʃ] before a high front vowel, they do not reveal the same tendency when the target word has /θ/ before a high front vowel. Thus, a word like think [θIŋk] is not expected to be rendered as [tʃɪŋk], but rather as [tɪŋk]. In other words learners distinguish the fate of two different [t] sounds. While the native allophonic rule converts the /t/ into [tʃ] before high front vowels, the [t] sound that is the substitute for the target /θ/ does not follow the same path. In this way, learners distinguish the three target language phonemes /θ/, /t/, and /tʃ/, and prevent the neutralization of any contrast. Eckman et al. state that their studies with Korean and Japanese speakers also confirm this tendency by maintaining the target contrasts.