---
title: "Cross-Linguistic Analysis of Eye Movement Patterns: Insights"
description: "The authors use AraEyebility, which they describe as the first Arabic eye-tracking corpus built for NLP, to address this. 20 references."
authors: ["Ibtehal Baazeem", "Hend Al-Khalifa", "Abdulmalik AlSalman"]
year: 2026
venue: "Proceedings fo the Second International Workshop on Eye-Tracking Resources and Evaluation for Human-Aligned NLP"
source: "https://aclanthology.org/2026.gaze4nlp-1.2/"
doi: "10.63317/4t25sphn6p46"
licence: "CC BY-NC 4.0 (ELRA, LREC proceedings)"
licence_url: "https://creativecommons.org/licenses/by-nc/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/baazeem-2026-cross-linguistic-analysis-eye-movement/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/baazeem-2026-cross-linguistic-analysis-eye-movement/)

[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/28/) / Cross-Linguistic Analysis of Eye Movement Patterns: Insights from the First Arabic Eye-Tracking Corpus for NLP

# Cross-Linguistic Analysis of Eye Movement Patterns: Insights from the First Arabic Eye-Tracking Corpus for NLP

**Authors:** [Ibtehal Baazeem](https://md.research.elamur.ai/free/en/authors/ibtehal-baazeem/), [Hend Al-Khalifa](https://md.research.elamur.ai/free/en/authors/hend-al-khalifa/) and [Abdulmalik AlSalman](https://md.research.elamur.ai/free/en/authors/abdulmalik-alsalman/) · **Year:** 2026 · **Venue:** [Proceedings fo the Second International Workshop on Eye-Tracking Resources and Evaluation for Human-Aligned NLP](https://md.research.elamur.ai/free/en/venues/gaze4nlp/) ([fo the International Workshop on Eye-Tracking Resources and Evaluation for Human-Aligned NLP 2026](https://md.research.elamur.ai/free/en/venues/gaze4nlp/2026/)) · **Publisher:** [ELDA](https://md.research.elamur.ai/free/en/publishers/elda/) · **Licence:** [CC BY-NC 4.0 (ELRA, LREC proceedings)](https://creativecommons.org/licenses/by-nc/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.gaze4nlp-1.2/) · [PDF](https://aclanthology.org/2026.gaze4nlp-1.2.pdf) · [DOI 10.63317/4t25sphn6p46](https://doi.org/10.63317/4t25sphn6p46) · [OpenAlex](https://openalex.org/W7166524102)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary
Most eye-tracking corpora used in NLP cover left-to-right Latin-script languages. Little is known about gaze behaviour in a cursive, right-to-left, morphologically rich language like Arabic. The authors use AraEyebility, which they describe as the first Arabic eye-tracking corpus built for NLP, to address this. In it, 15 native speakers read 587 paragraphs of Modern Standard and Classical Arabic. The authors compare its gaze statistics with values published for established corpora, mainly English ones: Dundee, GECO, ZuCo 1.0 and Provo. The comparison focuses on measures that nearly every corpus reports: first, single and total fixation duration, and regression duration. Arabic readers show longer average fixations and regressions than English readers. The authors link this to higher processing load from cursive letter joining, dot-based letter distinctions, diacritics, root-and-pattern morphology and left-to-right numbers inside right-to-left text. The work matters because reading models and cognitively informed NLP methods tuned on English may not transfer to other writing systems. The corpus is released publicly.

## Key points
- AraEyebility contains 57,617 words in 587 paragraphs on 13 topics, from texts written between the 8th and 21st centuries. It mixes MSA and CA, and the texts were partially diacritized after consulting linguists.
- Participants were 15 native readers, 7 male and 8 female, aged 20–45, from varied Arabic-speaking regions.
- Eye movements were recorded with a Tobii X120 at 120 Hz. Sessions with gaze sample quality under 80% were repeated, and the final average quality was 93%.
- Table 1 lists the comparison corpora: Dundee (EN 56,212 / FR 52,173 words), GECO (EN 54,364 / NL 59,716), ZuCo 1.0 (21,629) and Provo (2,689 words, 84 readers). Only AraEyebility uses a cursive RTL script.
- The authors acknowledge three limitations:
  - The comparison is made at corpus level across heterogeneous studies, so no matched statistical testing is possible.
  - The reader pool is small, and readability classes are imbalanced.
  - Dialectal Arabic is not covered.

## Interesting ideas
- **Choosing a shared metric set.** To compare heterogeneous resources, the authors restrict themselves to the measures every source reports and openly treat the result as corpus-level evidence rather than a controlled test. This keeps conclusions honest when data come from different setups.
- **A hard quality gate during collection.** Recordings below a fixed sample-quality threshold are redone rather than kept, and the achieved average quality is reported alongside the data. This makes the quality of the whole dataset auditable.
- **Single-idea paragraphs as the unit of text.** Splitting each text into coherent paragraphs that each express one idea balances having enough context against being able to trace behaviour back to a specific stretch of text.
- **Partial annotation as a design choice.** Diacritics were added only partially, weighing their help with ambiguity against the visual clutter they create. More markup is not automatically better input.
- **Regressions as a sign of difficulty.** The paper reads longer regression durations as higher processing cost, which treats re-reading as evidence that later context was needed to settle earlier material.

## For Mova

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

- **I1** (learn): when scoring parsers across different treebanks or dialect conversions, compare only on metrics like LAS that are computed identically everywhere, and label such cross-treebank comparisons as indicative.
- **I2** (learn): drop or re-run silver sentences whose witness-parser agreement falls below a fixed threshold, and publish the mean agreement of the accepted silver set.
- **I3** (world): build and check the world of a text paragraph by paragraph, so every entity or state change can be tied to a small, explainable span.
- **I4** (tokens): attach extra morphological or lemma hints only where a token is genuinely ambiguous, rather than marking everything and adding noise for downstream steps.
- **I5** (tree): log where the parser revises an earlier attachment and use those points as an explainable difficulty signal when selecting hard sentences for repair rules.

*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

- [King Saud University](https://md.research.elamur.ai/free/en/universities/king-saud-university/) — University, Riyadh, Saudi Arabia · 38 works · 527 citations
- [King Abdulaziz City for Science and Technology](https://md.research.elamur.ai/free/en/labs/king-abdulaziz-city-for-science-and-technology/) — Research institute, Riyadh, Saudi Arabia · 8 works · 74 citations

📚 Shelves (7): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 17,450 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 17,296 works · [💭 Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/) · 1,649 works · [💭 Developmental and Educational Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/developmental-and-educational-psychology/) · 523 works · [🖱️ Human-Computer Interaction](https://md.research.elamur.ai/free/en/shelves/computer-science/human-computer-interaction/) · 160 works

🧭 Topics (2): [Semitic languages](https://md.research.elamur.ai/free/en/topics/semitic-languages/) · 3,066 works · [🧒 Psycholinguistics & language acquisition](https://md.research.elamur.ai/free/en/topics/psycholinguistics/) · 108 works

## Cite

Cite as: Ibtehal Baazeem, Hend Al-Khalifa and Abdulmalik AlSalman. 2026. Cross-Linguistic Analysis of Eye Movement Patterns: Insights from the First Arabic Eye-Tracking Corpus for NLP. Proceedings fo the Second International Workshop on Eye-Tracking Resources and Evaluation for Human-Aligned NLP.

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

```bibtex
@inproceedings{baazeem2026cross,
  title = {{Cross-Linguistic Analysis of Eye Movement Patterns: Insights from the First Arabic Eye-Tracking Corpus for NLP}},
  author = {Ibtehal Baazeem, Hend Al-Khalifa and Abdulmalik AlSalman},
  booktitle = {Proceedings fo the Second International Workshop on Eye-Tracking Resources and Evaluation for Human-Aligned NLP},
  year = {2026},
  doi = {10.63317/4t25sphn6p46},
  url = {https://doi.org/10.63317/4t25sphn6p46}
}
```

- [← previous Cross-lingual Lexical Semantic Change in Romance Languages](https://md.research.elamur.ai/free/en/papers/uban-2026-cross-lingual-lexical-semantic-change/)
- [next → Cross-Linguistic Situation Entity Segmentation for Discourse Analysis in Diachronic English and German Text](https://md.research.elamur.ai/free/en/papers/schmuck-2026-cross-linguistic-situation-entity-segmentation/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/28/)

## References

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

1. AlJassmi, M. A., Hermena, E. W., and Paterson, k. B. (2021). Eye movements in Arabic reading. Experimental Arabic Linguistics, 10:85–108. [find](https://scholar.google.com/scholar?q=AlJassmi%2C+M%2E+A%2E%2C+Hermena%2C+E%2E+W%2E%2C+and+Paterson%2C+k%2E+B%2E+%282021%29%2E+Eye+movements+in+Arabic+reading%2E+Experimental+Arabic+Linguistics%2C+10%3A85%E2%80%93108%2E)
2. Al-Wabil, A. and Al-Sheaha, M. (2010). Towards an interactive screening program for developmental dyslexia: Eye movement analysis in reading Arabic texts. In Proceedings of the 12th International Conference on Computers Helping People with Special Needs, pages 25–32, Vienna, Austria, July 14–16. Springer. [find](https://scholar.google.com/scholar?q=Al%2DWabil%2C+A%2E+and+Al%2DSheaha%2C+M%2E+%282010%29%2E+Towards+an+interactive+screening+program+for+developmental+dyslexia%3A+Eye+movement+analysis+in+reading+Arabic+texts%2E+In+Proceedings+of+the+12th+International+Conf)
3. Baazeem, I., Al-Khalifa, H., and Al-Salman, A. (2025). AraEyebility: Eye-tracking data for Arabic text readability. Computation, 13(5):108. [find](https://scholar.google.com/scholar?q=Baazeem%2C+I%2E%2C+Al%2DKhalifa%2C+H%2E%2C+and+Al%2DSalman%2C+A%2E+%282025%29%2E+AraEyebility%3A+Eye%2Dtracking+data+for+Arabic+text+readability%2E+Computation%2C+13%285%29%3A108%2E)
4. Blanken, G., Dorn, M., and Sinn, H. (1997). Inversion errors in Arabic number reading: Is there a nonsemantic route? Brain and Cognition, 34(3):404–423. [find](https://scholar.google.com/scholar?q=Blanken%2C+G%2E%2C+Dorn%2C+M%2E%2C+and+Sinn%2C+H%2E+%281997%29%2E+Inversion+errors+in+Arabic+number+reading%3A+Is+there+a+nonsemantic+route%3F+Brain+and+Cognition%2C+34%283%29%3A404%E2%80%93423%2E)
5. Cop, U., Dirix, N., Drieghe, D., and Duyck, W. (2017). Presenting GECO: An eyetracking corpus of monolingual and bilingual sentence reading. Behavior Research Methods, 49(2):602–615. [find](https://scholar.google.com/scholar?q=Cop%2C+U%2E%2C+Dirix%2C+N%2E%2C+Drieghe%2C+D%2E%2C+and+Duyck%2C+W%2E+%282017%29%2E+Presenting+GECO%3A+An+eyetracking+corpus+of+monolingual+and+bilingual+sentence+reading%2E+Behavior+Research+Methods%2C+49%282%29%3A602%E2%80%93615%2E)
6. El-Haj, M., Kruschwitz, U., and Fox, C. (2015). Creating language resources for under-resourced languages: Methodologies, and experiments with Arabic. Language Resources and Evaluation, 49(3):549–580. [find](https://scholar.google.com/scholar?q=El%2DHaj%2C+M%2E%2C+Kruschwitz%2C+U%2E%2C+and+Fox%2C+C%2E+%282015%29%2E+Creating+language+resources+for+under%2Dresourced+languages%3A+Methodologies%2C+and+experiments+with+Arabic%2E+Language+Resources+and+Evaluation%2C+49%283%29%3A549%E2%80%93580%2E)
7. El-Haj, M. and Rayson, P. (2016). OSMAN—A novel Arabic readability metric. In Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC’16), pages 250–255, Portorož, Slovenia. May 23–28. European Language Resources Association (ELRA). [find](https://scholar.google.com/scholar?q=El%2DHaj%2C+M%2E+and+Rayson%2C+P%2E+%282016%29%2E+OSMAN%E2%80%94A+novel+Arabic+readability+metric%2E+In+Proceedings+of+the+Tenth+International+Conference+on+Language+Resources+and+Evaluation+%28LREC%E2%80%9916%29%2C+pages+250%E2%80%93255%2C+Portoro%C5%BE%2C)
8. Hermena, E. W., Drieghe, D., Hellmuth, S., and Liversedge, S. P. (2015). Processing of Arabic diacritical marks: Phonological–syntactic disambiguation of homographic verbs and visual crowding effects. Journal of Experimental Psychology: Human Perception and Performance, 41(2):494. [find](https://scholar.google.com/scholar?q=Hermena%2C+E%2E+W%2E%2C+Drieghe%2C+D%2E%2C+Hellmuth%2C+S%2E%2C+and+Liversedge%2C+S%2E+P%2E+%282015%29%2E+Processing+of+Arabic+diacritical+marks%3A+Phonological%E2%80%93syntactic+disambiguation+of+homographic+verbs+and+visual+crowding+effects%2E)
9. Hollenstein, N., Barrett, M., and Björnsdóttir, M. (2022). The Copenhagen corpus of eye tracking recordings from natural reading of Danish texts. In Proceedings of the Thirteenth Language Resources and Evaluation Conference, pages 1712–1720, Marseille, France, June 20–25. European Language Resources Association (ELRA). [find](https://scholar.google.com/scholar?q=Hollenstein%2C+N%2E%2C+Barrett%2C+M%2E%2C+and+Bj%C3%B6rnsd%C3%B3ttir%2C+M%2E+%282022%29%2E+The+Copenhagen+corpus+of+eye+tracking+recordings+from+natural+reading+of+Danish+texts%2E+In+Proceedings+of+the+Thirteenth+Language+Resources+an)
10. Hollenstein, N., Rotsztejn, J., Troendle, M., Pedroni, A., Zhang, C., and Langer, N. (2018). ZuCo, a simultaneous EEG and eye-tracking resource for natural sentence reading. Scientific Data, 5(1):180291. [find](https://scholar.google.com/scholar?q=Hollenstein%2C+N%2E%2C+Rotsztejn%2C+J%2E%2C+Troendle%2C+M%2E%2C+Pedroni%2C+A%2E%2C+Zhang%2C+C%2E%2C+and+Langer%2C+N%2E+%282018%29%2E+ZuCo%2C+a+simultaneous+EEG+and+eye%2Dtracking+resource+for+natural+sentence+reading%2E+Scientific+Data%2C+5%281%29%3A1802)
11. Hollenstein, N., Troendle, M., Zhang, C., and Langer, N. (2020). ZuCo 2.0: A dataset of physiological recordings during natural reading and annotation. In Proceedings of the Twelfth Language Resources and Evaluation Conference, pages 138–146, Marseille, France. May 11–16. European Language Resources Association (ELRA). [find](https://scholar.google.com/scholar?q=Hollenstein%2C+N%2E%2C+Troendle%2C+M%2E%2C+Zhang%2C+C%2E%2C+and+Langer%2C+N%2E+%282020%29%2E+ZuCo+2%2E0%3A+A+dataset+of+physiological+recordings+during+natural+reading+and+annotation%2E+In+Proceedings+of+the+Twelfth+Language+Resources)
12. Just, M. A. and Carpenter, P. A. (1980). A theory of reading: From eye fixations to comprehension. Psychological Review, 87(4):329–354. [find](https://scholar.google.com/scholar?q=Just%2C+M%2E+A%2E+and+Carpenter%2C+P%2E+A%2E+%281980%29%2E+A+theory+of+reading%3A+From+eye+fixations+to+comprehension%2E+Psychological+Review%2C+87%284%29%3A329%E2%80%93354%2E)
13. Kennedy, A., Hill, R., and Pynte, J. (2003). The Dundee corpus. In Proceedings of the 12th European Conference on Eye Movements, Dundee, UK. [find](https://scholar.google.com/scholar?q=Kennedy%2C+A%2E%2C+Hill%2C+R%2E%2C+and+Pynte%2C+J%2E+%282003%29%2E+The+Dundee+corpus%2E+In+Proceedings+of+the+12th+European+Conference+on+Eye+Movements%2C+Dundee%2C+UK%2E)
14. Khateb, A., Taha, H. Y., Elias, I., and Ibrahim, R. (2013). The effect of the internal orthographic connectivity of written Arabic words on the process of the visual recognition: A comparison between skilled and dyslexic readers. Writing Systems Research, 5(2):214–233. [find](https://scholar.google.com/scholar?q=Khateb%2C+A%2E%2C+Taha%2C+H%2E+Y%2E%2C+Elias%2C+I%2E%2C+and+Ibrahim%2C+R%2E+%282013%29%2E+The+effect+of+the+internal+orthographic+connectivity+of+written+Arabic+words+on+the+process+of+the+visual+recognition%3A+A+comparison+between+)
15. Leal, S. E., Duran, M. S., and Aluísio, S. (2018). A nontrivial sentence corpus for the task of sentence readability assessment in Portuguese. In Proceedings of the 27th International Conference on Computational Linguistics, pages 401–413, Santa Fe, New Mexico, USA, August 20–26. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Leal%2C+S%2E+E%2E%2C+Duran%2C+M%2E+S%2E%2C+and+Alu%C3%ADsio%2C+S%2E+%282018%29%2E+A+nontrivial+sentence+corpus+for+the+task+of+sentence+readability+assessment+in+Portuguese%2E+In+Proceedings+of+the+27th+International+Conference+on+Co)
16. Leal, S. E., Lukasova, K., Carthery-Goulart, M. T., and Aluísio, S. M. (2022). RastrOS project: Natural language processing contributions to the development of an eye-tracking corpus with predictability norms for Brazilian Portuguese. Language Resources and Evaluation, 56(4):1333–1372. [find](https://scholar.google.com/scholar?q=Leal%2C+S%2E+E%2E%2C+Lukasova%2C+K%2E%2C+Carthery%2DGoulart%2C+M%2E+T%2E%2C+and+Alu%C3%ADsio%2C+S%2E+M%2E+%282022%29%2E+RastrOS+project%3A+Natural+language+processing+contributions+to+the+development+of+an+eye%2Dtracking+corpus+with+predictabili)
17. Luke, S. G. and Christianson, K. (2017). The Provo Corpus: A large eye-tracking corpus with predictability norms. Behavior Research Methods, 50(2):826–833. [find](https://scholar.google.com/scholar?q=Luke%2C+S%2E+G%2E+and+Christianson%2C+K%2E+%282017%29%2E+The+Provo+Corpus%3A+A+large+eye%2Dtracking+corpus+with+predictability+norms%2E+Behavior+Research+Methods%2C+50%282%29%3A826%E2%80%93833%2E)
18. Paterson, K. B., Almabruk, A. A. A., McGowan, V. A., White, S. J., and Jordan, T. R. (2015). Effects of word length on eye movement control: The evidence from Arabic. Psychonomic Bulletin & Review, 22(5):1443– 1450. [find](https://scholar.google.com/scholar?q=Paterson%2C+K%2E+B%2E%2C+Almabruk%2C+A%2E+A%2E+A%2E%2C+McGowan%2C+V%2E+A%2E%2C+White%2C+S%2E+J%2E%2C+and+Jordan%2C+T%2E+R%2E+%282015%29%2E+Effects+of+word+length+on+eye+movement+control%3A+The+evidence+from+Arabic%2E+Psychonomic+Bulletin+%26+Review%2C+22)
19. Rayner, K. (1998). Eye movements in reading and information processing: 20 years of research. Psychological Bulletin, 124(3):372–422. [find](https://scholar.google.com/scholar?q=Rayner%2C+K%2E+%281998%29%2E+Eye+movements+in+reading+and+information+processing%3A+20+years+of+research%2E+Psychological+Bulletin%2C+124%283%29%3A372%E2%80%93422%2E)
20. Zhang, G., Yao, P., Ma, G., Wang, J., Zhou, J., Huang, L., Xu, P., Chen, L., Chen, S., Gu, J., Wei, W., Cheng, X., Hua, H., Liu, P., Lou, Y., Shen, W., Bao, Y., Liu, J., Lin, N., and Li, X. (2022). The database of eye-movement measures on words in Chinese reading. Scientific Data, 9(1):411. [find](https://scholar.google.com/scholar?q=Zhang%2C+G%2E%2C+Yao%2C+P%2E%2C+Ma%2C+G%2E%2C+Wang%2C+J%2E%2C+Zhou%2C+J%2E%2C+Huang%2C+L%2E%2C+Xu%2C+P%2E%2C+Chen%2C+L%2E%2C+Chen%2C+S%2E%2C+Gu%2C+J%2E%2C+Wei%2C+W%2E%2C+Cheng%2C+X%2E%2C+Hua%2C+H%2E%2C+Liu%2C+P%2E%2C+Lou%2C+Y%2E%2C+Shen%2C+W%2E%2C+Bao%2C+Y%2E%2C+Liu%2C+J%2E%2C+Lin%2C+N%2E%2C+and+Li%2C+X%2E+%282022%29%2E+Th)

*Updated 2 November 2026*