Use of language-specific speech cues in highly proficient second-language listening

Language-specificity in listening to speech occurs at all processing levels and even between structurally close languages (e.g., English, Dutch). Transitional cues to fricative place of articulation are used in English for identifying /f/ (which resembles theta) but not /s/, whereas in Dutch (without theta) they are used for neither. In spoken-word recognition, suprasegmental cues are used in Dutch, but not in English (with more segmental reduction); Dutch L2 listeners even outperform native L1 listeners in detecting origin of differently stressed English syllables (e.g., car- from CARton versus carTOON). Here, longterm residents in Australia with Dutch as L1 but predominantly using English completed each of these tasks. In the phonetic task, with cross-spliced nonsense words, these listeners performed just as Dutch listeners in the Netherlands, showing insensitivity to transitional cues for both /f/ and /s/. In the lexical task, with word fragments (e.g., car-), they however did not behave as L1 Dutch and outperform Australian English listeners, but instead resembled the latter, by ignoring suprasegmental stress cues. A (lexical) listening strategy available in L1 can apparently be abandoned if it delivers little payoff in L2, but acquiring for L2 listening a (phonetic) strategy not used in L1 seems less feasible