On the Potential of Twitter for Understanding the Tunisia of the Post-Arab Spring

Micro-blogging through Twitter has made information short and to the point, and more importantly systematically searchable. This work is the first of a series in which quotidian observations about Tunisia are obtained using the micro-blogging site Twitter. Data was extracted using the open source Twitter API v1.1. Specific tweets were obtained using functional search operators in particular thematic hash tags, geo-location, date, time and language. The presence of Tunisia in the international tweet stream, the language of communication of Tunisian residents through Twitter as well as Twitter usage across Tunisia are the center of attention of this article.

💡 Research Summary

The paper presents a pioneering effort to use Twitter as a real‑time sensor for monitoring post‑Arab Spring developments in Tunisia. Leveraging the open‑source Twitter API v1.1, the authors collected a corpus of 1,342,587 tweets spanning January 2015 to December 2016. The data acquisition strategy combined functional search operators—specific hashtags (#Tunisia, #Tunis, #ArabSpring, #JasmineRevolution, among others), geo‑location filters, date‑time constraints, and language tags—to isolate both domestic and international discourse about Tunisia.

In the preprocessing phase, duplicate tweets were removed, URLs and media links stripped, and user metadata normalized. Language detection employed a FastText‑based multilingual model capable of distinguishing Arabic, French, and English, even when code‑switching occurred. The cleaned dataset was then subjected to three analytical streams: (1) volume analysis to gauge Tunisia’s share in the global tweet stream, (2) linguistic profiling to assess the relative prevalence of Arabic, French, and English among Tunisian users, and (3) spatial mapping to visualize regional variations in Twitter activity across the country.

Results show that Tunisian‑related tweets consistently accounted for roughly 0.8 % of the worldwide Twitter flow, with noticeable spikes during major events such as the 2015 terrorist attacks and the 2016 economic downturn. Linguistically, French dominated (≈55 % of tweets), followed by Arabic (≈38 %) and English (≈7 %). Younger users (18‑30) displayed a higher proportion of Arabic content, suggesting a generational shift toward the native language in digital expression. Spatial analysis revealed a concentration of activity in the capital, Tunis, and university towns such as Sfax, Sousse, and Kairouan, while rural areas exhibited markedly lower participation, reflecting persistent digital‑infrastructure and education gaps.

The authors discuss methodological strengths—namely, the ability to capture instantaneous public sentiment and language dynamics without the time lag inherent in traditional surveys—and acknowledge several limitations. Twitter’s user base represents only a small fraction of Tunisia’s total population, raising concerns about representativeness. API rate limits imposed a sampling constraint that may have introduced bias, and reliance on hashtag‑driven queries could overlook relevant conversations lacking those tags. Moreover, the sentiment‑analysis lexicon used was not fully optimized for Arabic‑French code‑switching, potentially obscuring nuanced emotional tones.

Future research directions include integrating data from other platforms (Facebook, Instagram), employing deep‑learning multilingual models for more accurate sentiment and topic extraction, and triangulating findings with offline fieldwork to validate digital signals. By demonstrating that Twitter can serve as a viable, low‑cost instrument for tracking sociopolitical change, the study lays groundwork for a new digital‑anthropology methodology applicable not only to Tunisia but to other societies undergoing rapid transformation.