CAPT software shows promising results for segmental pronunciation, while prosodic features and fluency still require further research and development¶

Claim · All claims
Evidence · 2 studies · 1 quant-synthesis r2, 1 review r2 · q2–q3 · i2 medium · n=15 studies (38 effect sizes)
Currency · newest study 2024 · 1 of 2 entries checked against their sources (automated) · not reviewed by a person

Subclaims¶

q2 i? Most CAPT software shows promising results and very positive impacts on the pronunciation of segmental sounds among various types of learners, but prosodic features and fluency remain underdeveloped areas. → Ali 2016 q3 i2 Across 15 ASR studies with ESL/EFL learners, ASR had a large pooled effect on segmental pronunciation (g = 0.82) and a small one on suprasegmental pronunciation (g = 0.37) whose confidence interval crosses zero. → Ngo et al. 2024

Evidence¶

Ali 2016¶

Ali, S. (2016). Towards the development of a comprehensive pedagogical framework for pronunciation training based on adapted automatic speech recognition systems. In S. Papadima-Sophocleous, L. Bradley & S. Thouësny (Eds), CALL communities and culture – short papers from EUROCALL 2016 (pp. 7-13). Research-publishing.net. https://doi.org/10.14705/rpnet.2016.eurocall2016.530

q2 · i? · review · r2

This is the article's own narrative-review synthesis of existing CAPT research and tools, stating that "most CAPT softwares show promising results" for segmental sounds while prosody and fluency "still seem to require further research and development." No effect sizes are printed.

"On the whole and to our knowledge, most CAPT softwares show promising results and very positive impacts on the pronunciation of segmental sounds among various types of learners. Prosodic features and fluency generally speaking are areas of pronunciation training that still seem to require further research and development."

Ngo et al. 2024¶

Ngo, T. T.-N., Chen, H. H.-J., & Lai, K. K.-W. (2024). The effectiveness of automatic speech recognition in ESL/EFL pronunciation: A meta-analysis. ReCALL, 36(1), 4–21. doi:10.1017/S0958344023000113

q3 · meta-analysis (three-level, random effects) · i2 · medium effect, g=0.69 · n=15 studies (38 effect sizes) · quant-synthesis · r2

A meta-analysis of 15 studies (2008–2021, 38 effect sizes) comparing ESL/EFL learners who practised pronunciation with automatic speech recognition (ASR) against a non-ASR condition. The pooled effect was g = 0.69 (95% CI 0.31 to 1.08), with substantial heterogeneity (Q = 227.70; I² 56.56% between studies). The authors did not test for publication bias, and name the small number of primary studies as the main limitation. Read from the publisher's open-access full text; the article was published online in 2023. By outcome, the effect on segmental pronunciation was g = 0.82 (10 studies, 95% CI 0.24 to 1.41) and on suprasegmental pronunciation g = 0.37 (5 studies, 95% CI −0.22 to 0.97); studies measuring both gave g = 0.70. The synthesis covers ASR only, one kind of CAPT.

"The pooled effect size was medium (g = 0.69), and the confidence interval was not across zero (95% CI = [0.31, 1.08])."

Studies bearing on this claim¶

Which areas of pronunciation (segmentals, prosody, fluency) does computer-assisted or ASR-based practice improve, for whom, and with how much practice? This table gathers the studies behind this claim and the 6 narrower claims that bear on it: 4 distinct studies (1 review, 1 quant-synthesis, 1 causal, 1 design), q2–q3, 2 of 4 with an effect size coded. One row per study, codes as each entry records them; no effect is averaged and no verdict is drawn. Built by scripts/synthesize_claim.py from eval/synthesis/; a row that looks wrong is fixed on the claim it comes from.

Discussion¶

Ngo et al. (2024) support this claim for ASR-based training, with a large pooled effect on segmentals and a small, uncertain one on suprasegmentals. They also note that an earlier meta-analysis of CAPT that included multimedia tools rather than ASR found the reverse pattern, so the segmental-over-prosody pattern may be a property of ASR rather than of CAPT in general.

Related Claims¶

• Whether individual or peer practice works better with ASR-based pronunciation training is unsettled: one study found individual work best, a meta-analysis found peer practice gave larger effects — related
• The Fluspeak ASR-based pronunciation software gave good results with beginners focusing on phoneme production but poor results overall for advanced learners seeking fluency — a narrower finding that bears on this claim
• Students responded positively to ASR pronunciation training and the Rainbow passage, but were mixed on technical aspects of recording and voice typing — related
• Students most frequently attribute fluency and pronunciation gains to games — related
• Guided ASR practice did not produce statistically significant improvement in any specific pronunciation error type, with some errors unimproved or regressed — reports the opposite
• Explicit pronunciation instruction improves controlled production of targeted sounds and prosody, with smaller and less certain gains in spontaneous speech — related
• Resource scarcity and large class sizes limit individualized pronunciation feedback in Bangladeshi classrooms — related
• No surveyed presentation-coaching system covers all five taxonomy dimensions simultaneously — related
• Retrieved knowledge-graph subgraphs suggest dyslexic learners perceive existing AI tools as useful but still limited — related
• No existing corpus simultaneously provides L2 speech, long-form presentation structure, accent labels, and prosodic annotations — related
• Profy practice shows larger observed pronunciation-intelligibility improvement than elicited imitation, with non-overlapping pre/post confidence intervals — related
• ASR transcription serves as a diagnostic tool identifying individual learners' pronunciation errors, with function words most commonly mispronounced — a narrower finding that bears on this claim