The present study used an artificial grammar learning paradigm to explore the prediction that exposure to anti-harmony might help learners infer that a neutral vowel in a vowel harmony language is transparent. Participants were exposed to a back/round harmony language with a neutral vowel [a]. This neutral vowel either always selected a back vowel suffix, always selected a front vowel suffix, or selected both front-and back vowel suffixes, in adherence to anti-harmony. Results indicated that exposure to a back/round harmony with the neutral vowel selecting either back vowel suffixes, or both front and back vowel suffixes, could induce a bias towards transparent vowels. Assuming that participants inferred that the centralized [a] paired with [o] harmonically, then the predictions that exposure to anti-harmony could induce a bias towards a transparent vowel interpretation were borne out. However, the bias towards a transparent vowel was not significantly different between the anti-harmony conditions and the harmony condition, suggesting that this effect should be replicated with other neutral vowels.