Elamur Research⌕
☰
home / papers / 2017 / Code-Switching as a Social Act: The Case of Arabic Wikipedia Talk Pages

Code-Switching as a Social Act: The Case of Arabic Wikipedia Talk Pages

Paper · 2017 · Michael Yoder, Shruti Rijhwani, Carolyn Rosé et al.

Summary

The paper asks whether switching away from Arabic in Arabic Wikipedia discussions carries social meaning, and whether that meaning can be measured. The authors use markedness theory (Myers-Scotton, 1998), which holds that deviating from a conversational norm signals something. From it they derive three competing hypotheses: switching is neutral, switching marks an outsider and hurts acceptance, or switching signals expertise and helps.

To test them, they built and publicly released a dataset that pairs editors' talk page contributions with their simultaneous article edits. Editor success is the share of an editor's word-level changes that survive one day after the discussion closes, following Priedhorsky et al. (2007). A linear regression with code-switching features predicts this score better than a unigram baseline. Switching correlates positively with success, which supports the expertise hypothesis. An annotated sample adds nuance: switching helps most for technical terms on articles unrelated to the Arabic-speaking world, and it tends to fail on Arabic-centred topics.

The work matters because it treats code-switching as a pragmatic, socially consequential choice, not merely a language-identification problem.

Key points

  • The dataset has 5259 editor–discussion-thread instances, each linking a talk page interaction to a measurable influence on the article; it is published on GitHub (michaelmilleryoder/wikipedia-codeswitching-data).
  • Success score: each edit is represented as a vector of per-word frequency changes (additions positive, deletions negative). The score is the fraction of that change still present in the revision one day after the discussion ends. Word reordering is ignored.
  • "Code-switching" is defined broadly to include script-switching, such as Arabizi (Arabic written in Latin script, usually dialectal), as well as French, Hebrew and other languages. In the annotated sample, about 94% of switches go to English.
  • A 100-instance sample was annotated for switch function and article topic, with categories drawn from DBpedia. Short non-Arabic fillers like "sorry" made up about 16% of the sample. Switching was more successful on technical and non-Arabic-topic pages.
  • Conclusions:
    • Hypothesis 3 (prestige or expertise) is best supported.
    • Hypothesis 1 (no social meaning) looks unlikely.
    • Hypothesis 2 (outsider penalty) may still hold for culture-related pages, such as those on Yazdanism or Egyptian ancestry.

Interesting ideas

  • I1. The authors replace a subjective label like "influence" with a behavioural one: how long an editor's word-level changes survive. This yields supervision at scale without human judgment.
  • I2. Theory is turned into several mutually exclusive hypotheses, each predicting a different sign of correlation, and the data is then used to pick among them. This is a cleaner design than a single "does X help" test.
  • I3. Writing Arabic in Latin script is treated as a meaningful choice in its own right, not as noise to normalize away.
  • I4. Markedness frames a choice that breaks the local norm, including apologizing for the language one writes in, as carrying stance or identity information.
  • I5. The effect of the same behaviour flips depending on the topic category of the page, which was taken from an ontology (DBpedia). Context, not the feature alone, decides the meaning.

For Mova

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

  • I1 (learn): a silver label or self-induced repair rule could be scored by how often later witness parsers or corrections leave its output untouched, giving a cheap durability signal for trust.
  • I2 (learn): evaluations could state in advance what positive, null and negative results would mean, pairing naturally with negative controls.
  • I3 (tokens): the tokenizer could tag foreign-language or off-script tokens as an explicit, explainable attribute instead of silently mapping them to unknown words.
  • I4 (frames): prag could record norm-breaking choices (register shifts, inserted foreign terms, meta-comments on language) as speech-act or implicature cues about the speaker's claimed expertise or status.
  • I5 (world): ontology categories of a text's topic, drawn from Wikidata or WordNet, could condition how the world model interprets a speaker's choices instead of applying one global reading.

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

🎓Carnegie Mellon UniversityPittsburgh · United States539 works · 179,533 citations

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksSocial Sciences · 1,772 worksCommunication · 97 works

🧭 Topics 2Research subjects and languages the work is about.Semitic languages · 2,908 worksMultilingual & cross-lingual NLP · 828 works

Cite

Cite as Michael Yoder, Shruti Rijhwani, Carolyn Rosé and Lori Levin. 2017. Code-Switching as a Social Act: The Case of Arabic Wikipedia Talk Pages. Proceedings of the Second Workshop on NLP and Computational Social Science.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/w17-2911

@inproceedings{yoder2017code,
  title = {{Code-Switching as a Social Act: The Case of Arabic Wikipedia Talk Pages}},
  author = {Michael Yoder, Shruti Rijhwani, Carolyn Rosé and Lori Levin},
  booktitle = {Proceedings of the Second Workshop on NLP and Computational Social Science},
  year = {2017},
  doi = {10.18653/v1/w17-2911},
  url = {https://doi.org/10.18653/v1/w17-2911}
}

References

27 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Penelope Gardner-Chloros. 2009. Cambridge University Press. find
  2. Celso Alvarez-Cáccamo. 1990. Rethinking conversational code-switching: Codes, speech varieties, and contextualization. In Annual Meeting of the Berkeley Linguistics Society. volume 16, pages 3–16. Code-switching. ↗
  3. H. Paul Grice. 1975. Logic and conversation. In Peter Cole and Jerry L. Morgan, editors, Syntax and Semantics: Speech Acts, Academic Press, New York, pages 41–58. ↗
  4. Peter Auer. 2013. Code-Switching in Conversation: Language, Interaction and Identity. Taylor & Francis. find
  5. John J. Gumperz. 1982. Discourse Strategies. Studies in Interactional Socio. Cambridge University Press. find
  6. Inma Munoa Barredo. 1997. Pragmatic functions of code-switching among basque-spanish bilinguals. Retrieved on October 26:2011. find
  7. Hyeju Jang, Yohan Jo, Qinlan Shen, Michael Miller, Seungwhan Moon, and Carolyn Rose. 2016. Metaphor Detection with Topic Transition, Emotion and Cognition in Context. Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) 1:216– 225. find
  8. Reem Bassiouney. 2006. Functions of code switching in Egypt: Evidence from monologues, volume 46. Brill. find
  9. Reem Bassiouney. 2009. Arabic Sociolinguistics. Edinburgh University Press. find
  10. Jens Lehmann, Robert Isele, Max Jakob, Anja Jentzsch, Dimitris Kontokostas, Pablo Mendes, Sebastian Hellmann, Mohamed Morsey, Patrick van Kleef, Sören Auer, and Chris Bizer. 2014. DBpedia - a large-scale, multilingual knowledge base extracted from wikipedia. Semantic Web Journal . ↗
  11. Rafiya Begum, Kalika Bali, Monojit Choudhury, Koustav Rudra, and Niloy Ganguly. 2016. Functions of Code-Switching in Tweets: An Annotation Scheme and Some Initial Experiments. In LREC. i, pages 1644–1650. find
  12. Jan-Petter Blom and John J. Gumperz. 1972. Social meaning in linguistic structures: code-switching in Northern Norway. Directions in Sociolinguistics: The Ethnography of Communication pages 407– 434. find
  13. Marco Lui and Timothy Baldwin. 2012. langid.py: An off-the-shelf language identification tool. Proceedings of the ACL 2012 System Demonstrations (July):25–30.
  14. Thamar Solorio, Elizabeth Blair, Suraj Maharjan, Steven Bethard, Mona Diab, Mahmoud Gohneim, Abdelati Hawwari, Fahad AlGhamdi, Julia Hirschberg, Alison Chang, and Pascale Fung. 2014. Overview for the First Shared Task on Language Identification in Code-Switched Data. Proceedings of The First Workshop on Computational Approaches to Code Switching, held in conjunction with EMNLP 2014. pages 62–72.
  15. Jeff MacSwan. 2000. The architecture of the bilingual language faculty: evidence from intrasentential code switching. Bilingualism: Language and Cognition 3(1):37–54. https://doi.org/10.1017/S1366728900000122. ↗
  16. Giovanni Molina, Rey-Villamizar, Thamar Solorio, Fahad AlGhamdi, Mahmoud Gohneim, Abdelati Hawwari, and Mona Diab. 2016. Overview for the Second Shared Task on Language Identification in Code-Switched Data. Proceedings of The Second Workshop on Computational Approaches to Code Switching, held in conjunction with EMNLP 2016. pages 62–72.
  17. Thamar Solorio and Yang Liu. 2008. Learning to predict code-switch points. EMNLP ’08 Proceedings of the Conference on Empirical Methods in Natural Language Processing pages 973–981. https://doi.org/10.16373/j.cnki.ahr.150049. ↗
  18. Carol Myers-Scotton. 1995. Social Motivations for Codeswitching: Evidence from Africa. Oxford studies in language contact. Clarendon Press. find
  19. Oren Tsur, Dan Calacci, and David Lazer. 2015. A Frame of Mind: Using Statistical Models for Detection of Framing and Agenda Setting Campaigns. ACL pages 1629–1638. find
  20. Carol Myers-Scotton. 1998. Codes and Consequences: Choosing Linguistic Varieties. Oxford University Press. find
  21. Dong Nguyen, A. Seza Doğruöz, Carolyn P. Rosé, and Franciska de Jong. 2016. Computational sociolinguistics: A survey. Computational Linguistics 42(3):537–593. https://doi.org/10.1016/j.jksus.2015.08.001. ↗
  22. Elinor Ochs. 1992. Indexing Gender. In Alessandro Duranti and Charles Goodwin, editors, Rethinking context: Language as an interactive phenomenon, Cambridge University Press, chapter 14, pages 335– 358. find
  23. Fabian Pedregosa, G. Varoquaux, A. Gramfort, V. Michel, B. Thirion, O. Grisel, M. Blondel, P. Prettenhofer, R. Weiss, V. Dubourg, J. Vanderplas, A. Passos, D. Cournapeau, M. Brucher, M. Perrot, and E. Duchesnay. 2011. Scikit-learn: Machine learning in Python. Journal of Machine Learning Research 12:2825–2830. ↗
  24. Mario Piergallini, Rouzbeh Shirvani, Gauri S Gautam, and Mohamed Chouikha. 2016. Word-Level Language Identification and Predicting Codeswitching Points in Swahili-English Language Data pages 21– 29. find
  25. Reid Priedhorsky, Jilin Chen, Shyong Tony K Lam, Katherine Panciera, Loren Terveen, and John Riedl. 2007. Creating, destroying, and restoring value in wikipedia. Proceedings of the 2007 international ACM conference on supporting group work - GROUP ’07 page 259. https://doi.org/10.1145/1316624.1316663. ↗
  26. Koustav Rudra, Shruti Rijhwani, Rafiya Begum, Kalika Bali, Monojit Choudhury, and Niloy Ganguly. 2016. Understanding Language Preference for Expression of Opinion and Sentiment: What do Hindi-English Speakers do on Twitter? pages 1131–1141. find
  27. Sabah Safi-Stagni. 1991. Agrammatism in Arabic. Perspectives on Arabic Linguistics. John Benjamins Publishing Company. ↗

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/yoder-2017-code-switching-as-social-act/