---
title: "Roles of Predictability and Acoustic Distance — Zhang 2026"
description: "Listeners tend to hear allophones as more alike than phonemes, even when the acoustic gap is similar."
authors: ["Shuhao Zhang", "Youngah Do"]
year: 2026
venue: "Proceedings of the Society for Computation in Linguistics 2026"
source: "https://aclanthology.org/2026.scil-main.45/"
doi: "10.18653/v1/2026.scil-main.45"
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/en/papers/zhang-2026-roles-predictability-acoustic-distance-sound/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/zhang-2026-roles-predictability-acoustic-distance-sound/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/8/) / Roles of Predictability and Acoustic Distance in Sound Discrimination via Contrastive Learning

# Roles of Predictability and Acoustic Distance in Sound Discrimination via Contrastive Learning

**Authors:** Shuhao Zhang, [Youngah Do](https://md.research.elamur.ai/free/en/authors/youngah-do/) · **Year:** 2026 · **Venue:** Proceedings of the Society for Computation in Linguistics 2026 · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.scil-main.45/) · [PDF](https://aclanthology.org/2026.scil-main.45.pdf) · [DOI 10.18653/v1/2026.scil-main.45](https://doi.org/10.18653/v1/2026.scil-main.45) · [OpenAlex](https://openalex.org/W7166801330)

**Citations:** 0 citations (openalex-snapshot · 2026-10-14)

## Summary
Listeners tend to hear allophones as more alike than phonemes, even when the acoustic gap is similar. One explanation is that allophones occur in predictable, complementary contexts. The authors test this account computationally. They ask three questions: whether predictability affects discrimination categorically or gradually, how it interacts with acoustic distance, and whether other contrasts in the sound system matter.

They built synthetic CV syllables from 16-dimensional acoustic feature vectors. Fricatives were varied in center of gravity (CoG), and vowel contexts were arranged to give five levels of predictability. A small encoder was trained with supervised contrastive loss (Khosla et al., 2020). How well the two sound categories separated in its latent space was measured with silhouette and distance distribution overlap (DDO) scores.

Only complete predictability clearly hurt discrimination, which suggests a largely categorical effect. Larger acoustic distances progressively offset this deficit. An extra contrast on the same acoustic dimension also improved separability. The study offers a controlled simulation of how context, acoustic cues and the wider phonological system together shape perceptual sensitivity.

## Key points
- **Stimuli:** Eleven synthetic fricatives formed ten pairs. Each pair combined an anchor with a mean CoG of 9000 Hz and a partner whose CoG decreased in 500 Hz steps, from 8000 down to 3500 Hz.
- **Predictability levels:** Predictability was set at 0%, 25%, 50%, 75% or 100% by distributing eight training vowels across the two fricatives. Testing used /A/ as the context.
- **Design and data:** Crossing the levels with the pairs gave 50 conditions, each with 8,000 training tokens. Vowel formants were based on Hillenbrand et al. (1994).
- **Model:** A two-layer ReLU encoder mapped inputs to a 4-dimensional embedding. It was trained for 200 epochs with temperature τ = 0.07.
- **Main findings:**
  - Only 100% predictability produced a significant drop in discrimination. Occasionally, 75% predictability also differed from lower levels.
  - A related additional contrast helped most when its own acoustic distance was small.
  - An unrelated contrast had no effect.

*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/en/shelves/linguistics/)

## Cite

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

```bibtex
@inproceedings{zhang2026roles,
  title = {{Roles of Predictability and Acoustic Distance in Sound Discrimination via Contrastive Learning}},
  author = {Zhang, Shuhao and Do, Youngah},
  booktitle = {Proceedings of the Society for Computation in Linguistics 2026},
  year = {2026},
  doi = {10.18653/v1/2026.scil-main.45},
  url = {https://doi.org/10.18653/v1/2026.scil-main.45}
}
```

- [← previous Rhetorical Questions in LLM Representations: A Linear Probing Study](https://md.research.elamur.ai/free/en/papers/yao-2026-rhetorical-questions-llm-representations-linear/)
- [next → Romanized Arabic Across Dialects: Views, Usage Patterns, and Linguistic Variation](https://md.research.elamur.ai/free/en/papers/keleg-2026-romanized-arabic-across-dialects-views/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/8/)

## References

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

1. Shannon L Barrios, Joselyn M Rodriguez, and Taylor Anne Barriuso. 2023. The acquisition of L2 allophonic variants: The role of phonological distribution and lexical cues. Second Language Research, 39(3):899–924. [find](https://scholar.google.com/scholar?q=Shannon+L+Barrios%2C+Joselyn+M+Rodriguez%2C+and+Taylor+Anne+Barriuso%2E+2023%2E+The+acquisition+of+L2+allophonic+variants%3A+The+role+of+phonological+distribution+and+lexical+cues%2E+Second+Language+Research%2C+39%28)
2. Amanda Boomershine, Kathleen Currie Hall, Elizabeth Hume, and Keith Johnson. 2008. The impact of allophony versus contrast on speech perception. In Peter Avery, B. Elan Dresher, and Keren Rice, editors, Contrast in Phonology, pages 145–172. Mouton de Gruyter. [find](https://scholar.google.com/scholar?q=Amanda+Boomershine%2C+Kathleen+Currie+Hall%2C+Elizabeth+Hume%2C+and+Keith+Johnson%2E+2008%2E+The+impact+of+allophony+versus+contrast+on+speech+perception%2E+In+Peter+Avery%2C+B%2E+Elan+Dresher%2C+and+Keren+Rice%2C+editor)
3. Ting Chen, Simon Kornblith, Mohammad Norouzi, and Geoffrey Hinton. 2020. A Simple Framework for Contrastive Learning of Visual Representations. arXiv preprint. ArXiv:2002.05709 [cs]. [find](https://scholar.google.com/scholar?q=Ting+Chen%2C+Simon+Kornblith%2C+Mohammad+Norouzi%2C+and+Geoffrey+Hinton%2E+2020%2E+A+Simple+Framework+for+Contrastive+Learning+of+Visual+Representations%2E+arXiv+preprint%2E+ArXiv%3A2002%2E05709+%5Bcs%5D%2E) [our card](https://md.research.elamur.ai/free/en/papers/chen-2020-simple-framework-contrastive-learning-visual/)
4. Matthew H. Davis, Ingrid S. Johnsrude, Alexis Hervais-Adelman, Karen Taylor, and Carolyn McGettigan. 2005. Lexical Information Drives Perceptual Learning of Distorted Speech: Evidence From the Comprehension of Noise-Vocoded Sentences. Journal of Experimental Psychology: General, 134(2):222–241. [find](https://scholar.google.com/scholar?q=Matthew+H%2E+Davis%2C+Ingrid+S%2E+Johnsrude%2C+Alexis+Hervais%2DAdelman%2C+Karen+Taylor%2C+and+Carolyn+McGettigan%2E+2005%2E+Lexical+Information+Drives+Perceptual+Learning+of+Distorted+Speech%3A+Evidence+From+the+Compreh)
5. Alain de Cheveigné. 1999. Formant bandwidth affects the identification of competing vowels. ICPhS, pages 2093–2096. [find](https://scholar.google.com/scholar?q=Alain+de+Cheveign%C3%A9%2E+1999%2E+Formant+bandwidth+affects+the+identification+of+competing+vowels%2E+ICPhS%2C+pages+2093%E2%80%932096%2E)
6. Qisheng Liao, Meiting Lai, and Preslav Nakov. 2023. MarsEclipse at SemEval-2023 Task 3: Multi-lingual and Multi-label Framing Detection with Contrastive Learning. In Proceedings of the The 17th International Workshop on Semantic Evaluation (SemEval-2023), pages 83–87, Toronto, Canada. Association fo… [find](https://scholar.google.com/scholar?q=Qisheng+Liao%2C+Meiting+Lai%2C+and+Preslav+Nakov%2E+2023%2E+MarsEclipse+at+SemEval%2D2023+Task+3%3A+Multi%2Dlingual+and+Multi%2Dlabel+Framing+Detection+with+Contrastive+Learning%2E+In+Proceedings+of+the+The+17th+Intern)
7. Andrew Martin, Sharon Peperkamp, and Emmanuel Dupoux. 2013. Learning Phonemes With a Proto-Lexicon. Cognitive Science, 37(1):103–124. [find](https://scholar.google.com/scholar?q=Andrew+Martin%2C+Sharon+Peperkamp%2C+and+Emmanuel+Dupoux%2E+2013%2E+Learning+Phonemes+With+a+Proto%2DLexicon%2E+Cognitive+Science%2C+37%281%29%3A103%E2%80%93124%2E)
8. Sara Finley and William Badecker. 2009. Artificial language learning and feature-based generalization. Journal of Memory and Language, 61(3):423–437. [find](https://scholar.google.com/scholar?q=Sara+Finley+and+William+Badecker%2E+2009%2E+Artificial+language+learning+and+feature%2Dbased+generalization%2E+Journal+of+Memory+and+Language%2C+61%283%29%3A423%E2%80%93437%2E)
9. Abdellah Fourtassi and Emmanuel Dupoux. 2014. A Rudimentary Lexicon and Semantics Help Bootstrap Phoneme Acquisition. In Proceedings of the Eighteenth Conference on Computational Natural Language Learning, pages 191–200, Ann Arbor, Michigan. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Abdellah+Fourtassi+and+Emmanuel+Dupoux%2E+2014%2E+A+Rudimentary+Lexicon+and+Semantics+Help+Bootstrap+Phoneme+Acquisition%2E+In+Proceedings+of+the+Eighteenth+Conference+on+Computational+Natural+Language+Lear)
10. Jessica Maye, Daniel J. Weiss, and Richard N. Aslin. 2008. Statistical phonetic learning in infants: facilitation and feature generalization. Developmental Science, 11(1):122–134. [find](https://scholar.google.com/scholar?q=Jessica+Maye%2C+Daniel+J%2E+Weiss%2C+and+Richard+N%2E+Aslin%2E+2008%2E+Statistical+phonetic+learning+in+infants%3A+facilitation+and+feature+generalization%2E+Developmental+Science%2C+11%281%29%3A122%E2%80%93134%2E)
11. Bob McMurray and Allard Jongman. 2011. What information is necessary for speech categorization? Harnessing variability in the speech signal by integrating cues computed relative to expectations. Psychological Review, 118(2):219–246. [find](https://scholar.google.com/scholar?q=Bob+McMurray+and+Allard+Jongman%2E+2011%2E+What+information+is+necessary+for+speech+categorization%3F+Harnessing+variability+in+the+speech+signal+by+integrating+cues+computed+relative+to+expectations%2E+Psych)
12. Stella Frank, Naomi H. Feldman, and Sharon Goldwater. 2014. Weak semantic context helps phonetic learning in a model of infant language acquisition. In Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1073–1083, Baltimore, Marylan… [find](https://scholar.google.com/scholar?q=Stella+Frank%2C+Naomi+H%2E+Feldman%2C+and+Sharon+Goldwater%2E+2014%2E+Weak+semantic+context+helps+phonetic+learning+in+a+model+of+infant+language+acquisition%2E+In+Proceedings+of+the+52nd+Annual+Meeting+of+the+As)
13. Masaki Noguchi and Carla L. Hudson Kam. 2018. The Emergence of the Allophonic Perception of Unfamiliar Speech Sounds: The Effects of Contextual Distribution and Phonetic Naturalness. Language Learning, 68(1):147–176. [find](https://scholar.google.com/scholar?q=Masaki+Noguchi+and+Carla+L%2E+Hudson+Kam%2E+2018%2E+The+Emergence+of+the+Allophonic+Perception+of+Unfamiliar+Speech+Sounds%3A+The+Effects+of+Contextual+Distribution+and+Phonetic+Naturalness%2E+Language+Learning)
14. Tianyu Gao, Xingcheng Yao, and Danqi Chen. 2022. SimCSE: Simple Contrastive Learning of Sentence Embeddings. arXiv preprint. ArXiv:2104.08821 [cs]. [find](https://scholar.google.com/scholar?q=Tianyu+Gao%2C+Xingcheng+Yao%2C+and+Danqi+Chen%2E+2022%2E+SimCSE%3A+Simple+Contrastive+Learning+of+Sentence+Embeddings%2E+arXiv+preprint%2E+ArXiv%3A2104%2E08821+%5Bcs%5D%2E) [our card](https://md.research.elamur.ai/free/en/papers/gao-2021-simcse-simple-contrastive-learning-sentence/)
15. Judith E. Pegg and Janet F. Werker. 1997. Adult and infant perception of two English phones. The Journal of the Acoustical Society of America, 102(6):3742– 3753. [find](https://scholar.google.com/scholar?q=Judith+E%2E+Pegg+and+Janet+F%2E+Werker%2E+1997%2E+Adult+and+infant+perception+of+two+English+phones%2E+The+Journal+of+the+Acoustical+Society+of+America%2C+102%286%29%3A3742%E2%80%93+3753%2E)
16. James Hillenbrand, Laura A. Getty, Kimberlee Wheeler, and Michael J. Clark. 1994. Acoustic characteristics of American English vowels. The Journal of the Acoustical Society of America, 95(5_Supplement):2875–2875. [find](https://scholar.google.com/scholar?q=James+Hillenbrand%2C+Laura+A%2E+Getty%2C+Kimberlee+Wheeler%2C+and+Michael+J%2E+Clark%2E+1994%2E+Acoustic+characteristics+of+American+English+vowels%2E+The+Journal+of+the+Acoustical+Society+of+America%2C+95%285%5FSupplement)
17. Sharon Peperkamp, Rozenn Le Calvez, Jean-Pierre Nadal, and Emmanuel Dupoux. 2006. The acquisition of allophonic rules: Statistical learning with linguistic constraints. Cognition, 101(3):B31–B41. [find](https://scholar.google.com/scholar?q=Sharon+Peperkamp%2C+Rozenn+Le+Calvez%2C+Jean%2DPierre+Nadal%2C+and+Emmanuel+Dupoux%2E+2006%2E+The+acquisition+of+allophonic+rules%3A+Statistical+learning+with+linguistic+constraints%2E+Cognition%2C+101%283%29%3AB31%E2%80%93B41%2E)
18. Allard Jongman, Ratree Wayland, and Serena Wong. 2000. Acoustic characteristics of English fricatives. The Journal of the Acoustical Society of America, 108(3):1252–1263. [find](https://scholar.google.com/scholar?q=Allard+Jongman%2C+Ratree+Wayland%2C+and+Serena+Wong%2E+2000%2E+Acoustic+characteristics+of+English+fricatives%2E+The+Journal+of+the+Acoustical+Society+of+America%2C+108%283%29%3A1252%E2%80%931263%2E)
19. Prannay Khosla, Piotr Teterwak, Chen Wang, Aaron Sarna, Yonglong Tian, Phillip Isola, Aaron Maschinot, Ce Liu, and Dilip Krishnan. 2020. Supervised Contrastive Learning. arXiv preprint. Version Number: 5. [find](https://scholar.google.com/scholar?q=Prannay+Khosla%2C+Piotr+Teterwak%2C+Chen+Wang%2C+Aaron+Sarna%2C+Yonglong+Tian%2C+Phillip+Isola%2C+Aaron+Maschinot%2C+Ce+Liu%2C+and+Dilip+Krishnan%2E+2020%2E+Supervised+Contrastive+Learning%2E+arXiv+preprint%2E+Version+Number)
20. Patricia K. Kuhl. 2000. A new view of language acquisition. Proceedings of the National Academy of Sciences, 97(22):11850–11857. [find](https://scholar.google.com/scholar?q=Patricia+K%2E+Kuhl%2E+2000%2E+A+new+view+of+language+acquisition%2E+Proceedings+of+the+National+Academy+of+Sciences%2C+97%2822%29%3A11850%E2%80%9311857%2E)
21. Sang-Im Lee. 2011. Spectral analysis of Mandarin Chinese sibilant fricatives. In Proceedings of the 17th international congress of phonetic sciences, pages 1178–1181, Hong Kong. [find](https://scholar.google.com/scholar?q=Sang%2DIm+Lee%2E+2011%2E+Spectral+analysis+of+Mandarin+Chinese+sibilant+fricatives%2E+In+Proceedings+of+the+17th+international+congress+of+phonetic+sciences%2C+pages+1178%E2%80%931181%2C+Hong+Kong%2E)
22. Shanpeng Li and Wentao Gu. 2015. Acoustic analysis of Mandarin affricates. In Interspeech 2015, pages 1680–1684. ISCA. [find](https://scholar.google.com/scholar?q=Shanpeng+Li+and+Wentao+Gu%2E+2015%2E+Acoustic+analysis+of+Mandarin+affricates%2E+In+Interspeech+2015%2C+pages+1680%E2%80%931684%2E+ISCA%2E)
23. D Norris. 2003. Perceptual learning in speech. Cognitive Psychology, 47(2):204–238. [find](https://scholar.google.com/scholar?q=D+Norris%2E+2003%2E+Perceptual+learning+in+speech%2E+Cognitive+Psychology%2C+47%282%29%3A204%E2%80%93238%2E)
24. Sharon Peperkamp, Michèle Pettinato, and Emmanuel Dupoux. 2003. Allophonic Variation and the Acquisition of Phoneme Categories. In Proceedings of the 27th Annual Boston University Conference on Language Development, volume 2, pages 650–661, Sommerville, MA. Cascadilla Press. [find](https://scholar.google.com/scholar?q=Sharon+Peperkamp%2C+Mich%C3%A8le+Pettinato%2C+and+Emmanuel+Dupoux%2E+2003%2E+Allophonic+Variation+and+the+Acquisition+of+Phoneme+Categories%2E+In+Proceedings+of+the+27th+Annual+Boston+University+Conference+on+Langua)
25. A. Seidl, A. Cristià, A. Bernard, and K. H. Onishi. 2009. Allophonic and Phonemic Contrasts in Infants’ Learning of Sound Patterns. Language Learning and Development, 5(3):191–202. [find](https://scholar.google.com/scholar?q=A%2E+Seidl%2C+A%2E+Cristi%C3%A0%2C+A%2E+Bernard%2C+and+K%2E+H%2E+Onishi%2E+2009%2E+Allophonic+and+Phonemic+Contrasts+in+Infants%E2%80%99+Learning+of+Sound+Patterns%2E+Language+Learning+and+Development%2C+5%283%29%3A191%E2%80%93202%2E)
26. Amanda Seidl and Alejandrina Cristia. 2012. Infants’ Learning of Phonological Status. Frontiers in Psychology, 3. [find](https://scholar.google.com/scholar?q=Amanda+Seidl+and+Alejandrina+Cristia%2E+2012%2E+Infants%E2%80%99+Learning+of+Phonological+Status%2E+Frontiers+in+Psychology%2C+3%2E)
27. Katrin Skoruppa, Anna Lambrechts, and Sharon Peperkamp. 2011. The role of phonetic distance in the acquisition of phonological alternations. In Proceedings of the 39th Meeting of the North-East Linguistics Society (NELS). [find](https://scholar.google.com/scholar?q=Katrin+Skoruppa%2C+Anna+Lambrechts%2C+and+Sharon+Peperkamp%2E+2011%2E+The+role+of+phonetic+distance+in+the+acquisition+of+phonological+alternations%2E+In+Proceedings+of+the+39th+Meeting+of+the+North%2DEast+Lingui)
28. D.H. Whalen, Catherine T. Best, and Julia R. Irwin. 1997. Lexical effects in the perception and production of American English /p/ allophones. Journal of Phonetics, 25(4):501–528. [find](https://scholar.google.com/scholar?q=D%2EH%2E+Whalen%2C+Catherine+T%2E+Best%2C+and+Julia+R%2E+Irwin%2E+1997%2E+Lexical+effects+in+the+perception+and+production+of+American+English+%2Fp%2F+allophones%2E+Journal+of+Phonetics%2C+25%284%29%3A501%E2%80%93528%2E)