Distinguished Voices
in Arabic Linguistics

Arabic is best understood not as a single language target but as a structured space of variation across geolects, registers, genres, and culturally anchored meaning. This keynote argues that Arabic linguistics and Arabic NLP share a methodological bottleneck: we discuss “competence” while relying on resources and evaluations that often under-specify what they measure. This has become more visible with the rise of large language models, whose fluent outputs can mask systematic gaps in dialectal control, cultural inference, and context-appropriate choice. I propose treating corpora, tasks, and benchmarks as measurement instruments designed to make variation variables explicit, controllable, and comparable, so computational results can speak to linguistic hypotheses and linguistic theory can constrain computational claims. I present a unified program around three increasingly stringent notions of Arabic competence. (i) Variation competence: recognizing and preserving dialectal signals under controls that reduce spurious correlates such as topic, named entities, and geography-as-content. (ii) Cultural–pragmatic competence: interpreting conventionalized, non-compositional meaning and felicity conditions, illustrated through multidialectal proverb understanding and other culturally saturated constructions. (iii) Situated competence: choosing appropriate forms when meaning and appropriateness depend on grounded context, including visually and socially specified settings, interactional roles, and culturally salient cues. Across these layers, I show how community benchmarks and diagnostic protocols expose failure modes that matter to both fields. I conclude with a joint agenda for cumulative science: phenomenon-driven evaluations that encode linguistic variables by design, analyses that connect model behavior to Arabic generalizations, and inclusive data practices that treat Arabic diversity as the scientific object.

The linguistic landscape -the constellation of written signs in public space- has gradually evolved from being largely approached in descriptive terms into a robust cue of ethnographic typology (Blommaert & Maly, 2019), a strategic site for language planning and policy (Shohamy, 2015) and a dynamic space for pedagogical engagement (Cenoz and Gorter, 2021). This talk shows how the use of Franco (Arabizi) -writing Arabic in Latin letters- on the linguistic landscape creates a crossroads of these trajectories. Drawing on the Cairene linguistic landscape, I trace the migration of Franco/Arabizi from digital interactions between millennials into physical signage, as a parallel code alongside Modern Standard Arabic, dialectal Arabic, and English. Its physical visibility signals a horizontal shift in linguistic capital (Bourdieu, 1991), indexing a social group whose symbolic power is addressed via a newly designed (Bell 2001) hybrid textuality that challenges established classificatory frameworks. Arabizi also unsettles vertical hierarchies of the “levels of modern Arabic” as defined by Badawi (1975) and later scholarship (e.g., 2020), both theoretically and pedagogically. This talk offers critical implications for the broader field. For the theoretical linguist, it invites a re-evaluation of what constitutes a "written variety" and/or a level of contemporary Arabic in the 21st century. For practitioners and educators, it calls for a dialogue on the relevance of "standard" varieties in an increasingly Arabizi-mediated digital and public sphere. Ultimately, it seeks to answer: who owns the linguistic landscape? And equally important, which linguistic capital would gain more value over the near and distant future?