Cross-linguistic similarity in L2 speech learning: Evidence from the acquisition of Russian stop contrasts by Mandarin speakers

Abstract

This study investigated the role of cross-linguistic similarity in the acquisition of Russian initial stop contrasts by Chinese learners, addressing two specific research questions: (1) How similar are Russian voiced stops to Mandarin stops for Chinese learners? (2) How can the speech learning model (SLM) be applied to account for the acquisition of Russian initial stop contrasts by Chinese learners? Regarding the first question, a hypothesis was proposed that Russian voiced stops could be regarded as dissimilar ‘new’ sounds by Chinese learners, as judged by three commonly adopted methods: IPA comparison, acoustic difference, and feature redeployment. The results of the perceptual assimilation task, however, disconfirmed this hypothesis, as Russian voiced stops were perceived as being highly similar to the sounds of Mandarin voiceless unaspirated stops. According to SLM, perceptually similar second language (L2) sounds are difficult to acquire; hence, a corresponding hypothesis was advanced in relation to the second question; namely, that the acquisition of Russian initial stop contrasts could be challenging for Chinese learners. This hypothesis was supported by the results of the perceptual discrimination and the production tests. These findings corroborated SLM regarding the difficulty in acquiring L2 sounds that are perceptually similar to their first language (L1) counterparts, while posing challenges to the hypothesis that Russian voiced stops could be dissimilar ‘new’ sounds for Chinese learners, as predicted by the three methods. The results indicated that perceived cross-linguistic similarity plays a key role in L2 phonetic acquisition. However, the degree of perceived similarity cannot always be accurately deduced solely through the three methods, at least not in the case of stops.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 34 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.