Delayed feedback disrupts optimal strategies during foreign speech sound learning

Abstract

The Competition between Verbal and Implicit Systems (COVIS) model posits that an explicit hypothesis-testing system competes with an implicit procedural-based system to mediate category learning. During early learning, the hypothesis-testing system is dominant, whereas in later learning, the procedural system dominates. In visual category learning, delayed feedback is known to impair the procedural but not the hypothesis-testing system. We tested the COVIS model in natural auditory category learning. Young adult native English speakers learned to categorize lexical tones in Mandarin syllables. Timing of feedback was either immediate (0 s) or delayed (500 or 1000 ms). Consistent with COVIS, delay feedback affected accuracy only in later learning. Further, modeling analysis revealed that participants were more likely to adopt procedural strategies during later learning, but this transition was disrupted by delayed feedback. These results will be discussed in the context of developing methods to optimize foreign speech sound learning.