Preference for Distinct Variants in Learning Sound Correspondences During Dialect Acquisition.

The article investigates the role of sound similarity in learning sound correspondences (SCs)—systematic phonological relations between cognate sounds—in second dialect acquisition. Using an artificial language learning experiment with native Mandarin speakers learning “artificial dialects” modeled on southern Chinese varieties, the study tested whether learners prefer SCs with similar or distinct phonological variants. Results showed no effect of variant similarity in simple one-to-one SC mappings, but a preference for more distinct variants emerged in complex mapping structures (two-to-one and one-to-two), contingent on the distinct variant surpassing a threshold of phonological distance (at least three featural differences). These findings provide empirical support for the salience hypothesis, suggesting that learners favor salient, sufficiently distinct variants when resolving uncertainty in complex SC patterns, while overall sensitivity to subtle variant differences remains low due to high cross-dialect similarity.