Data-Driven Learning for Pronunciation: Perception and Production of Lexical Stress and Prominence in Academic English

Kevin Hirschi, Okim Kang

Research output: Contribution to journalArticlepeer-review

Abstract

Issues of intelligibility may arise amongst English learners when acquiring new words and phrases in North American academic settings, perhaps in part due to limited linguistic data available to the learner for understanding language use patterns. To this end, this paper examines the effects of Data-Driven Learning for Pronunciation (DDLfP) on lexical stress and prominence in the US academic context. 65 L2 English learners in North American universities completed a diagnostic and pretest with listening and speaking items before completing four online lessons and a posttest on academic words and formulas (i.e., multi-word sequences). Experimental group participants (n = 40) practiced using an audio corpus of highly proficient L2 speakers while comparison group participants (n = 25) were given teacher-created pronunciation materials. Logistic regression results indicated that the group who used the corpus significantly increased their recognition of prominence in academic formulas. In the spoken tasks, both groups improved in their lexical stress pronunciation, but only the DDLfP learners improved their production of prominence in academic formulas. Learners reported that they valued DDLfP efforts for pronunciation learning across contexts and speakers. Findings have implications for teachers of L2 pronunciation and support the use of corpora for language teaching and learning.

Original languageEnglish (US)
JournalTESOL Quarterly
DOIs
StateAccepted/In press - 2024

ASJC Scopus subject areas

  • Language and Linguistics
  • Education
  • Linguistics and Language

Fingerprint

Dive into the research topics of 'Data-Driven Learning for Pronunciation: Perception and Production of Lexical Stress and Prominence in Academic English'. Together they form a unique fingerprint.

Cite this