Lexically guided perceptual retuning studies have demonstrated that listeners use their knowledge of phonemes in words to accommodate to artificially generated sounds that are halfway between two phonemes. However, it is unknown whether listeners accommodate in the same way when words are pronounced with an incorrect native phoneme (e.g., flower pronounced as “thlower”). Monolingual Australian-English listeners completed one of two exposure phases where the /f/ or /s/ in words was pronounced as /Θ/ (“th”), followed by a visual lexical decision task with cross-modal priming. If training is effective, identity-priming should be observed when a /Θ/-bearing auditory prime (e.g., thoil) precedes a trainingcongruent matched visual target (e.g., foil or soil). Priming was observed for participants in the /f/-training, but not the /s/-training condition. A second experiment with intact /f/- or /s/-primes confirmed stimulus validity by showing an identity priming pattern. We conclude that lexically-guided perceptual retuning may be possible across a category boundary, but the native phonological system and/or acoustic similarity may impose limits on which native phonemes can be substituted effectively.

Index Terms: speech perception, lexically guided retuning, cross-modal priming, spoken word recognition.