---
title: "Potential of ASR for the study of L2 learner — Ayari 2024"
description: "The authors test whether automatic speech recognition (ASR) is suitable for working with oral corpora of learners of French as a second language."
authors: ["Sarra El Ayari", "Zhongjie Li"]
year: 2024
venue: "Proceedings of the 13th Workshop on Natural Language Processing for Computer Assisted Language Learning"
source: "https://aclanthology.org/2024.nlp4call-1.4/"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/papers/ayari-2024-potential-asr-study-l2-learner/"
---
[Elamur Research](https://md.research.elamur.ai/) · [this page in HTML](https://research.elamur.ai/free/papers/ayari-2024-potential-asr-study-l2-learner/)

[home](https://md.research.elamur.ai/) / [papers](https://md.research.elamur.ai/free/papers/) / [2024](https://md.research.elamur.ai/free/papers/2024/page/3/) / Potential of ASR for the study of L2 learner corpora

# Potential of ASR for the study of L2 learner corpora

**Authors:** Sarra El Ayari, Zhongjie Li · **Year:** 2024 · **Venue:** Proceedings of the 13th Workshop on Natural Language Processing for Computer Assisted Language Learning · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2024.nlp4call-1.4/) · [PDF](https://aclanthology.org/2024.nlp4call-1.4.pdf) · [OpenAlex](https://openalex.org/W4403431612)

**Citations:** 0 citations (openalex-snapshot · 2026-10-14) · 1 in our library, of the 140673 papers whose bibliographies we read

## Summary
The authors test whether automatic speech recognition (ASR) is suitable for working with oral corpora of learners of French as a second language. The speech of such learners (interlanguage) is non-standard at the levels of pronunciation, lexicon, morphology and syntax. For an ASR system, it is therefore essentially a low-resource language. The researchers processed semi-structured interviews from the longitudinal LANGSNAP corpus with Whisper. The manual transcripts in CHAT format and the automatic output were normalized (partly by hand) and aligned. They were then compared using WER and CER, and insertions, substitutions and deletions were also analyzed qualitatively. Comparable interviews with native speakers of French served as a point of comparison. It turned out that Whisper "hypercorrects": it fits the learners' non-standard forms to the standard language and thus erases precisely the phenomena that interest researchers of language acquisition. In addition, the system has failures of its own. At the same time, it produces a useful draft transcript, already aligned with the audio, which is easier for a human to correct than to transcribe from scratch. The paper shows the risks of handing transcription over to a machine in SLA research. It also makes the case for fine-tuning ASR on learner speech corpora.

## Key points
- **Data:** the LANGSNAP corpus (available via TalkBank). It contains 29 English-speaking students who studied French for 21 months, including 9 months spent abroad. 14 participants were analyzed at three time points: October 2011, May 2012 and October 2012.
- **Model and metrics:** Whisper (release 20231117), deployed on the French public servers Huma-Num and CINES. Quality was evaluated with WER and CER.
- **Three types of Whisper failures:** fabricated text with sudden switches into other languages; repeated output of a word (e.g., "oui") that is not in the recording; omitted stretches of speech.
- **Useful findings:** Whisper sometimes recognizes fragments that the human transcriber did not catch. This helps check and complete manual transcripts. The authors added import of Whisper output to their tool Sarramanka.
- **Limitations and plans:** the study uses only one ASR system. Next, the authors plan to fine-tune Whisper or wav2vec on learner speech corpora. They also want to carry out a similar study on the VILLA corpus: Polish after 14 hours of instruction, learners with five different L1s.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

Fields: [Linguistics & languages](https://md.research.elamur.ai/free/shelves/linguistics/) · [Education](https://md.research.elamur.ai/free/shelves/education/)

## Cite

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{ayari2024potential,
  title = {{Potential of ASR for the study of L2 learner corpora}},
  author = {Ayari, Sarra El and Li, Zhongjie},
  booktitle = {Proceedings of the 13th Workshop on Natural Language Processing for Computer Assisted Language Learning},
  year = {2024},
  doi = {10.3384/ecp211004},
  url = {https://doi.org/10.3384/ecp211004}
}
```

- [← previous Picking Up Where the Linguist Left Off: Mapping Morphology to Phonology through Learning the Residuals](https://md.research.elamur.ai/free/papers/aclbulk-2024-khalifa-picking-up-where-linguist-left/)
- [next → Pragmatic Competence Evaluation of Large Language Models for the Korean Language](https://md.research.elamur.ai/free/papers/park-2024-pragmatic-competence-evaluation-large-language/)
- [↑ back to papers of 2024](https://md.research.elamur.ai/free/papers/2024/page/3/)

## References

17 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. R. Ellis and G. Barkhuize. 2005. Analysing Learner Language. Oxford:Oxford University Press. [find](https://scholar.google.com/scholar?q=R%2E+Ellis+and+G%2E+Barkhuize%2E+2005%2E+Analysing+Learner+Language%2E+Oxford%3AOxford+University+Press%2E)
2. A. Baevski, H.Zhou, A.Mohamed, and M. Auli. 2020. wav2vec 2.0: A framework for self-supervised learning of speech representations. [find](https://scholar.google.com/scholar?q=A%2E+Baevski%2C+H%2EZhou%2C+A%2EMohamed%2C+and+M%2E+Auli%2E+2020%2E+wav2vec+2%2E0%3A+A+framework+for+self%2Dsupervised+learning+of+speech+representations%2E) [our card](https://md.research.elamur.ai/free/papers/baevski-2020-wav2vec-2-0-framework-self/)
3. Calbert Graham and Nathan Roll. 2023. Evaluating openai’s whisper asr: Performance analysis across diverse accents and speaker traits. Cambridge Open Engage. [find](https://scholar.google.com/scholar?q=Calbert+Graham+and+Nathan+Roll%2E+2023%2E+Evaluating+openai%E2%80%99s+whisper+asr%3A+Performance+analysis+across+diverse+accents+and+speaker+traits%2E+Cambridge+Open+Engage%2E)
4. N. Ballier, A. Meli, M. Amand, and J.-B. Yunès. 2023. Using whisper LLM for automatic phonetic diagnosis of L2 speech, a case study with French learners of English. In Proceedings of the 6th International Conference on Natural Language and Speech Processing (ICNLSP 2023), pages 282–292. Association… [find](https://scholar.google.com/scholar?q=N%2E+Ballier%2C+A%2E+Meli%2C+M%2E+Amand%2C+and+J%2E%2DB%2E+Yune%CC%80s%2E+2023%2E+Using+whisper+LLM+for+automatic+phonetic+diagnosis+of+L2+speech%2C+a+case+study+with+French+learners+of+English%2E+In+Proceedings+of+the+6th+Internat)
5. W. Klein. 1989. L’Acquisition de langue étrangère. Paris: Armand Colin. [find](https://scholar.google.com/scholar?q=W%2E+Klein%2E+1989%2E+L%E2%80%99Acquisition+de+langue+e%CC%81trange%CC%80re%2E+Paris%3A+Armand+Colin%2E)
6. W. Klein and C. Perdue. 1997. The Basic Variety (or: Couldn’t natural languages be much simpler?). Second Language Research, 13(4):301–347. [find](https://scholar.google.com/scholar?q=W%2E+Klein+and+C%2E+Perdue%2E+1997%2E+The+Basic+Variety+%28or%3A+Couldn%E2%80%99t+natural+languages+be+much+simpler%3F%29%2E+Second+Language+Research%2C+13%284%29%3A301%E2%80%93347%2E)
7. S. Benazzo and M. Watorek. 2021. Transcription de corpus oraux d’apprenants débutants en français l2 : quelques enjeux théoriques. In L. Spreafico, G. Bernini, A. Valentini J. Saturno (éds.) Superare l’evanescenza del parlato. Un vademecum per il trattamento digitale di dati linguistici, pages 1… [find](https://scholar.google.com/scholar?q=S%2E+Benazzo+and+M%2E+Watorek%2E+2021%2E+Transcription+de+corpus+oraux+d%E2%80%99apprenants+de%CC%81butants+en+franc%CC%A7ais+l2+%3A+quelques+enjeux+the%CC%81oriques%2E+In+L%2E+Spreafico%2C+G%2E+Bernini%2C+A%2E+Valentini+J%2E+Saturno+%28e%CC%81ds%2E%29+Super)
8. P. J. Lou and M. Johnson. 2020. End-to-end speech recognition and disfluency removal. In Findings of the Association for Computational Linguistics: EMNLP 2020, pages 2051–2061. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=P%2E+J%2E+Lou+and+M%2E+Johnson%2E+2020%2E+End%2Dto%2Dend+speech+recognition+and+disfluency+removal%2E+In+Findings+of+the+Association+for+Computational+Linguistics%3A+EMNLP+2020%2C+pages+2051%E2%80%932061%2E+Association+for+Computa)
9. B. MacWhinney. 2000. The CHILDES Project: Tools for Analyzing Talk. 3rd Edition. Mahwah, NJ: Lawrence Erlbaum Associates. [find](https://scholar.google.com/scholar?q=B%2E+MacWhinney%2E+2000%2E+The+CHILDES+Project%3A+Tools+for+Analyzing+Talk%2E+3rd+Edition%2E+Mahwah%2C+NJ%3A+Lawrence+Erlbaum+Associates%2E)
10. C. Blanche-Benveniste. 2000. Approches de la langue parlée en français. Paris: Ophrys. [find](https://scholar.google.com/scholar?q=C%2E+Blanche%2DBenveniste%2E+2000%2E+Approches+de+la+langue+parle%CC%81e+en+franc%CC%A7ais%2E+Paris%3A+Ophrys%2E)
11. R. Bley-Vroman. 1983. The comparative fallacy in interlanguage studies: The case of systematicity. Language Learning, (1):1–17. [find](https://scholar.google.com/scholar?q=R%2E+Bley%2DVroman%2E+1983%2E+The+comparative+fallacy+in+interlanguage+studies%3A+The+case+of+systematicity%2E+Language+Learning%2C+%281%29%3A1%E2%80%9317%2E)
12. C. Perdue. 1993. Adult Language Acquisition. Vol 1: Field Methods. Cambridge University Press. [find](https://scholar.google.com/scholar?q=C%2E+Perdue%2E+1993%2E+Adult+Language+Acquisition%2E+Vol+1%3A+Field+Methods%2E+Cambridge+University+Press%2E)
13. A. Radford, J. Xu Kim, Brockman T., McLeavey G., C., and I. Sutskever. 2023. Robust speech recognition via large-scale weak supervision. Proceedings of the 40th International Conference on Machine Learning. [find](https://scholar.google.com/scholar?q=A%2E+Radford%2C+J%2E+Xu+Kim%2C+Brockman+T%2E%2C+McLeavey+G%2E%2C+C%2E%2C+and+I%2E+Sutskever%2E+2023%2E+Robust+speech+recognition+via+large%2Dscale+weak+supervision%2E+Proceedings+of+the+40th+International+Conference+on+Machine+Lea) [our card](https://md.research.elamur.ai/free/papers/radford-2022-robust-speech-recognition-large-scale/)
14. L. Selinker. 1972. Interlanguage. International Review of Applied Linguistics, (10):209–231. [find](https://scholar.google.com/scholar?q=L%2E+Selinker%2E+1972%2E+Interlanguage%2E+International+Review+of+Applied+Linguistics%2C+%2810%29%3A209%E2%80%93231%2E)
15. Elise Tancoigne, Jean Philippe Corbellini, Gaëlle Deletraz, Laure Gayraud, Sandrine Ollinger, and Daniel Valero. 2022. Un mot pour un autre ? Analyse et comparaison de huit plateformes de transcription automatique. Bulletin of Sociological Methodology/Bulletin de Méthodologie Soci-ologique, 155(1)… [find](https://scholar.google.com/scholar?q=Elise+Tancoigne%2C+Jean+Philippe+Corbellini%2C+Gae%CC%88lle+Deletraz%2C+Laure+Gayraud%2C+Sandrine+Ollinger%2C+and+Daniel+Valero%2E+2022%2E+Un+mot+pour+un+autre+%3F+Analyse+et+comparaison+de+huit+plateformes+de+transcripti)
16. X. Wei, C. Cucchiarini, R. van Hout, and H. Strik. 2022. Automatic speech recognition and pronunciation error detection of dutch non-native speech: cumulating speech resources in a pluricentric language. Speech Communication, 144:1–9. [find](https://scholar.google.com/scholar?q=X%2E+Wei%2C+C%2E+Cucchiarini%2C+R%2E+van+Hout%2C+and+H%2E+Strik%2E+2022%2E+Automatic+speech+recognition+and+pronunciation+error+detection+of+dutch+non%2Dnative+speech%3A+cumulating+speech+resources+in+a+pluricentric+langua)
17. H. Yamasaki, J. Louradour, J. Hunter, and L. Pre-vot. 2023. Transcribing and aligning conversational speech: A hybrid pipeline applied to french conversations. In IEEE Automatic Speech Recognition and Understanding Workshop (ASRU), Taipei, Taiwan. [find](https://scholar.google.com/scholar?q=H%2E+Yamasaki%2C+J%2E+Louradour%2C+J%2E+Hunter%2C+and+L%2E+Pre%2Dvot%2E+2023%2E+Transcribing+and+aligning+conversational+speech%3A+A+hybrid+pipeline+applied+to+french+conversations%2E+In+IEEE+Automatic+Speech+Recognition+and)