Task differences do not impede overall learning when adapting to a novel sound

Abstract

Lexically guided perceptual learning (LGPL) and second-language learning (L2) research seeks to understand how listeners adapt to novel speech sounds. In LGPL, listeners shift their perception of a native phonetic contrast in response to hearing an ambiguous token embedded in an unambiguous lexical context. In L2 tasks, often involve categorizing non-native sounds, usually with explicit feedback. In general, L2 learning is seen as more effortful and individually variable than LGPL. However, paradigms differ in terms of stimuli (native vs. non-native phonetic contrasts) as well as tasks (lexically-guided implicit feedback vs. explicit feedback). To test whether L2-type vs. LGPL-type tasks yield differences in learning, participants were trained using an L2-style task to shift their native boundary along an /s/ to /sh/ continuum. With feedback, participants categorized the midpoint token as one novel object, and an endpoint token (counterbalanced across subjects) as another novel object. After training, a boundary shift comparable to previous studies using the LGPL task was found, suggesting that explicit and lexically-guided feedback produce similar magnitude of learning. Despite this, participants in the L2-style task experienced lower accuracy and more variability in perception of the continuum endpoints, suggesting that task differences partially explain less consistency across individuals in L2 research.