---
title: "A study on native American English speech recognition by Indian"
description: "The paper asks how well Indian listeners understand American English speech and whether vocabulary rarity drives their… 22 references."
authors: ["Abhayjeet Singh", "Achuth Rao MV", "Rakesh Vaideeswaran", "Chiranjeevi Yarra", "Prasanta Kumar Ghosh"]
year: 2021
venue: "arXiv:2112.04151"
source: "https://arxiv.org/abs/2112.04151"
doi: "10.48550/arxiv.2112.04151"
arxiv: "2112.04151"
licence: "arXiv — non-exclusive license to distribute"
licence_url: "http://arxiv.org/licenses/nonexclusive-distrib/1.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/singh-2021-study-native-american-english-speech/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/singh-2021-study-native-american-english-speech/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2021](https://md.research.elamur.ai/free/en/papers/2021/page/7/) / A study on native American English speech recognition by Indian listeners with varying word familiarity level

# A study on native American English speech recognition by Indian listeners with varying word familiarity level

**Authors:** Abhayjeet Singh, Achuth Rao MV, Rakesh Vaideeswaran, [Chiranjeevi Yarra](https://md.research.elamur.ai/free/en/authors/chiranjeevi-yarra/) and [Prasanta Kumar Ghosh](https://md.research.elamur.ai/free/en/authors/prasanta-ghosh/) · **Year:** 2021 · **Venue:** arXiv:2112.04151 · **Licence:** [arXiv — non-exclusive license to distribute](http://arxiv.org/licenses/nonexclusive-distrib/1.0/)

**Links:** [Publisher's page](https://arxiv.org/abs/2112.04151) · [PDF](https://arxiv.org/pdf/2112.04151) · [DOI 10.48550/arxiv.2112.04151](https://doi.org/10.48550/arxiv.2112.04151) · [arXiv 2112.04151](https://arxiv.org/abs/2112.04151) · [OpenAlex](https://openalex.org/W4226124814)

**Citations:** 3 citations (OpenAlex, 2026-10-14)

## Summary
The paper asks how well Indian listeners understand American English speech and whether vocabulary rarity drives their difficulty. The authors took the 2342 TIMIT sentences and scored each one by adding up the rank of every word in the American National Corpus (ANC) frequency list. They sorted the sentences by this word familiarity score (WF-score), cut the list into thirds and drew 500 sentences labelled Easy, Medium and Hard. Through online Google forms, 500 Indian undergraduates of many native languages each heard 10 clips. For every clip they typed what they heard, then saw the true text and rated, on a 1–10 scale, how hard the sentence was and how hard the speaker was. Difficulty ratings and word error rate (WER) all rose as familiarity fell. Human recognition was about as accurate as an ASR system using an Indian-accent acoustic model, while an American acoustic model paired with a combined language model did best. Some American dialect regions were noticeably harder for these listeners. The results matter for Indian learners who depend on MOOCs taught by American speakers.

## Key points
- Design: 50 forms with 10 unique sentences each, mixed 5 Easy, 3 Medium and 2 Hard. The first 10 responses per form were kept, giving 5000 responses in total. The selected sentences contain 1867 distinct words (910, 556 and 401 by level).
- Mean sentence-difficulty ratings were 3.25, 4.00 and 4.70 for Easy, Medium and Hard. Mean speaker-difficulty ratings were 3.45, 4.16 and 5.01. Differences between levels were significant (p \< 0.01).
- The two self-ratings correlated at 0.90, and both tracked WER.
- Three ASR setups were compared:
  - ASR1: Indian-speaker acoustic model with a TIMIT-text language model. It performed about as well as the human listeners.
  - ASR2: American acoustic model with a LibriSpeech-text language model.
  - ASR3: American acoustic model with a language model built on LibriSpeech plus TIMIT text. It was the best of the three.
- Dialects and listeners:
  - Speakers from DR4 and DR5 were harder to follow, while Western and Army Brat speakers were easier.
  - Listener nativity also mattered; Hindi was the largest group at 27.19%.
  - Speech rate showed no clear trend.

## Interesting ideas
- Rate each sentence's difficulty by summing its words' frequency ranks from a large corpus list, then split the sorted pool into tiers; this gives a cheap, reproducible difficulty axis for any test set.
- Let the subject attempt the item first, then reveal the gold answer and collect a self-rated difficulty score; the strong link between these ratings and actual errors shows that self-assessment can serve as a calibration signal.
- Change one component at a time (acoustic model versus language model) and compare each system against human performance to locate where errors actually come from.
- Treat a mismatch in word count between the transcription and the reference as a diagnostic of its own, since unfamiliar vocabulary made listeners lose track of word boundaries.
- Keep fixed difficulty proportions in every evaluation batch, never repeat an item, and keep the mix of subgroups (here, dialect regions) in line with the source corpus.

## For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

- **I1** (learn): Build evaluation slices by lexical rarity so we can check whether parser or reader errors cluster on sentences full of rare words, rather than reading one averaged score.
- **I2** (answer): After a reader answers a school test or fable question, compare its own confidence with correctness against the gold key to calibrate when it should flag uncertainty.
- **I3** (tree): Run gparse and the perceptron parser with swapped tagger or tokenizer inputs to tell whether a drop in LAS comes from morphology or from attachment decisions.
- **I4** (tokens): Log disagreements in token count against gold UD tokenization, split by word rarity, to find where the en tokenizer mis-segments rare or unknown words.
- **I5** (learn): Stratify silver-data evaluation sets by genre and difficulty in proportion to the source treebanks, so comparisons with negative controls are not skewed by an unbalanced sample.

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

## Institutions

- [International Institute of Information Technology, Hyderabad](https://md.research.elamur.ai/free/en/universities/international-institute-of-information-technology-hyderabad/) — University, Hyderabad, India · 24 works · 1,719 citations
- [Indian Institute of Science Bangalore](https://md.research.elamur.ai/free/en/universities/indian-institute-of-science-bangalore/) — University, Bengaluru, India · 8 works · 1,140 citations

📚 Shelves (7): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 9,086 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 8,889 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 9,296 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 9,206 works · [💭 Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/) · 798 works · [💭 Experimental and Cognitive Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/experimental-and-cognitive-psychology/) · 463 works · [🔊 Signal Processing](https://md.research.elamur.ai/free/en/shelves/computer-science/signal-processing/) · 213 works

🧭 Topics (2): [English](https://md.research.elamur.ai/free/en/topics/english/) · 1,373 works · [🎙️ Speech & spoken language](https://md.research.elamur.ai/free/en/topics/speech/) · 233 works

## Cite

Cite as: Abhayjeet Singh, Achuth Rao MV, Rakesh Vaideeswaran, Chiranjeevi Yarra and Prasanta Kumar Ghosh. 2021. A study on native American English speech recognition by Indian listeners with varying word familiarity level. arXiv:2112.04151.

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

```bibtex
@misc{singh2021study,
  title = {{A study on native American English speech recognition by Indian listeners with varying word familiarity level}},
  author = {Abhayjeet Singh, Achuth Rao MV, Rakesh Vaideeswaran, Chiranjeevi Yarra and Prasanta Kumar Ghosh},
  year = {2021},
  doi = {10.48550/arxiv.2112.04151},
  eprint = {2112.04151},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2112.04151}
}
```

- [← previous A Falta de Pan, Buenas Son Tortas: The Efficacy of Predicted UPOS Tags for Low Resource UD Parsing](https://md.research.elamur.ai/free/en/papers/anderson-2021-falta-de-pan-buenas-son/)
- [next → Approximation derived from a scalar exclusive particle associating with covert focus: The case of Hebrew be-sax ha-kol](https://md.research.elamur.ai/free/en/papers/orenstein-2021-approximation-derived-scalar-exclusive-particle/)
- [↑ back to papers of 2021](https://md.research.elamur.ai/free/en/papers/2021/page/7/)

## References

22 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. A. Warzybok, T. Brand, K. C. Wagener, and B. Kollmeier, “How much does language proficiency by non-native listeners influence speech audiometric tests in noise?” International journal of audiology, vol. 54, no. sup2, pp. 88–99, 2015. [↗](https://doi.org/10.3109/14992027.2015.1063715)
2. M. G. Lecumberri and M. Cooke, “Effect of masker type on native and non-native consonant perception in noise,” The Journal of the Acoustical Society of America, vol. 119, no. 4, pp. 2445–2454, 2006. [↗](https://doi.org/10.1121/1.2180210)
3. M. Cooke, M. Garcia Lecumberri, and J. Barker, “The foreign language cocktail party problem: Energetic and informational masking effects in non-native speech perception,” The Journal of the Acoustical Society of America, vol. 123, no. 1, pp. 414–427, 2008. [↗](https://doi.org/10.1121/1.2804952)
4. G. Borghini and V. Hazan, “Listening effort during sentence processing is increased for non-native listeners: A pupillometry study,” Frontiers in neuroscience, vol. 12, p. 152, 2018. [↗](https://doi.org/10.3389/fnins.2018.00152)
5. D. Meador, J. E. Flege, and I. R. MacKay, “Factors affecting the recognition of words in a second language,” Bilingualism: Language and Cognition, vol. 3, no. 1, pp. 55–67, 2000. [↗](https://doi.org/10.1017/s1366728900000134)
6. O. Scharenborg, J. M. Coumans, and R. van Hout, “The effect of background noise on the word activation process in nonnative spoken-word recognition.” Journal of Experimental Psychology: Learning, Memory, and Cognition, vol. 44, no. 2, p. 233, 2018. [↗](https://doi.org/10.1037/xlm0000441)
7. O. Scharenborg and M. van Os, “Why listening in background noise is harder in a non-native language than in a native language: A review,” Speech Communication, vol. 108, pp. 53–64, 2019. [↗](https://doi.org/10.1016/j.specom.2019.03.001)
8. A. R. Bradlow and J. A. Alexander, “Semantic and phonetic enhancements for speech-in-noise recognition by native and non-native listeners,” The Journal of the Acoustical Society of America, vol. 121, no. 4, pp. 2339– 2349, 2007. [↗](https://doi.org/10.1121/1.2642103)
9. A. R. Bradlow and T. Bent, “The clear speech effect for non-native listeners,” The Journal of the Acoustical Society of America, vol. 112, no. 1, pp. 272–284, 2002. [↗](https://doi.org/10.1121/1.1487837)
10. L. H. Mayo, M. Florentine, and S. Buus, “Age of second-language acquisition and perception of speech in noise,” Journal of speech, language, and hearing research, vol. 40, no. 3, pp. 686–693, 1997. [↗](https://doi.org/10.1044/jslhr.4003.686)
11. S. Brouwer, K. J. Van Engen, L. Calandruccio, and A. R. Bradlow, “Linguistic contributions to speech-on-speech masking for native and non-native listeners: Language familiarity and semantic content,” The Journal of the Acoustical Society of America, vol. 131, no. 2, pp. 1449– 1464, 2012. [↗](https://doi.org/10.1121/1.3675943)
12. A. K. Nábělek, “Identification of vowels in quiet, noise, and reverberation: Relationships with age and hearing loss,” The Journal of the Acoustical Society of America, vol. 84, no. 2, pp. 476–484, 1988. [↗](https://doi.org/10.1121/1.396880)
13. A. K. Nábělek and A. M. Donahue, “Perception of consonants in reverberation by native and non-native listeners,” The Journal of the Acoustical Society of America, vol. 75, no. 2, pp. 632–634, 1984. [↗](https://doi.org/10.1121/1.390495)
14. J. Aydelott, D. Baer-Henney, M. Trzaskowski, R. Leech, and F. Dick, “Sentence comprehension in competing speech: Dichotic sentence-word priming reveals hemispheric differences in auditory semantic processing,” Language and Cognitive Processes, vol. 27, no. 7-8, pp. 1108–1144, 2012. [↗](https://doi.org/10.1080/01690965.2011.589735)
15. “American national corpus, second release, frequency data, last accessed:29/03/2021.” [Online]. Available: [https://www.anc.org/data/anc-second-release/frequency-data/](https://www.anc.org/data/anc-second-release/frequency-data/) [↗](https://www.anc.org/data/anc-second-release/frequency-data/)
16. L. D. Consortium et al., “The darpa timit acoustic-phonetic continuous speech corpus,” NIST Speech CD, pp. 1–1, 1990. [find](https://scholar.google.com/scholar?q=L%2E+D%2E+Consortium+et+al%2E%2C+%E2%80%9CThe+darpa+timit+acoustic%2Dphonetic+continuous+speech+corpus%2C%E2%80%9D+NIST+Speech+CD%2C+pp%2E+1%E2%80%931%2C+1990%2E)
17. C. Spiccia, A. Augello, G. Pilato, and G. Vassallo, “Semantic word error rate for sentence similarity,” in 2016 IEEE Tenth International Conference on Semantic Computing (ICSC). IEEE, 2016, pp. 266–269. [find](https://scholar.google.com/scholar?q=C%2E+Spiccia%2C+A%2E+Augello%2C+G%2E+Pilato%2C+and+G%2E+Vassallo%2C+%E2%80%9CSemantic+word+error+rate+for+sentence+similarity%2C%E2%80%9D+in+2016+IEEE+Tenth+International+Conference+on+Semantic+Computing+%28ICSC%29%2E+IEEE%2C+2016%2C+pp%2E+266%E2%80%9326)
18. D. Povey, A. Ghoshal, G. Boulianne, L. Burget, O. Glembek, N. Goel, M. Hannemann, P. Motlicek, Y. Qian, P. Schwarz et al., “The kaldi speech recognition toolkit,” in IEEE 2011 workshop on automatic speech recognition and understanding, no. CONF. IEEE Signal Processing Society, 2011. [find](https://scholar.google.com/scholar?q=D%2E+Povey%2C+A%2E+Ghoshal%2C+G%2E+Boulianne%2C+L%2E+Burget%2C+O%2E+Glembek%2C+N%2E+Goel%2C+M%2E+Hannemann%2C+P%2E+Motlicek%2C+Y%2E+Qian%2C+P%2E+Schwarz+et+al%2E%2C+%E2%80%9CThe+kaldi+speech+recognition+toolkit%2C%E2%80%9D+in+IEEE+2011+workshop+on+automatic+sp)
19. V. I. Levenshtein, “Binary codes capable of correcting deletions, insertions, and reversals,” in Soviet physics doklady, vol. 10, no. 8. Soviet Union, 1966, pp. 707– 710. [find](https://scholar.google.com/scholar?q=V%2E+I%2E+Levenshtein%2C+%E2%80%9CBinary+codes+capable+of+correcting+deletions%2C+insertions%2C+and+reversals%2C%E2%80%9D+in+Soviet+physics+doklady%2C+vol%2E+10%2C+no%2E+8%2E+Soviet+Union%2C+1966%2C+pp%2E+707%E2%80%93+710%2E)
20. C. Yarra, R. Aggarwal, A. Rajpal, and P. K. Ghosh, “Indic timit and indic english lexicon: A speech database of indian speakers using timit stimuli and a lexicon from their mispronunciations,” in 2019 22nd Conference of the Oriental COCOSDA International Committee for the Co-ordination and Standardisation of Speech Databases and Assessment Techniques (O-COCOSDA), 2019, pp. 1–6. [find](https://scholar.google.com/scholar?q=C%2E+Yarra%2C+R%2E+Aggarwal%2C+A%2E+Rajpal%2C+and+P%2E+K%2E+Ghosh%2C+%E2%80%9CIndic+timit+and+indic+english+lexicon%3A+A+speech+database+of+indian+speakers+using+timit+stimuli+and+a+lexicon+from+their+mispronunciations%2C%E2%80%9D+in+2019)
21. V. Panayotov, G. Chen, D. Povey, and S. Khudanpur, “Librispeech: An asr corpus based on public domain audio books,” in 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2015, pp. 5206–5210. [find](https://scholar.google.com/scholar?q=V%2E+Panayotov%2C+G%2E+Chen%2C+D%2E+Povey%2C+and+S%2E+Khudanpur%2C+%E2%80%9CLibrispeech%3A+An+asr+corpus+based+on+public+domain+audio+books%2C%E2%80%9D+in+2015+IEEE+International+Conference+on+Acoustics%2C+Speech+and+Signal+Processing+%28IC)
22. “language-check, last accessed:02/04/2021.” [Online]. Available: [https://github.com/myint/language-check](https://github.com/myint/language-check) [↗](https://github.com/myint/language-check)

*Updated 24 October 2026*