Determiner Asymmetry in Arabic–English Code-Switching: Behavioral and Eye-Tracking Evidence

Aims and Objectives: This study investigates whether determiner selection in Arabic–English code-switching is predicted by the Matrix Language Frame (MLF) model, which posits that in the formation of code-switched sentences, the matrix language (ML) provides all functional morphemes. Specifically, the study examines whether the determiner choice is asymmetric across switching directions and whether this asymmetry reflects structural, morphological, or cognitive constraints beyond the MLF model. Method: Three experiments were conducted with early Arabic–English bilinguals. Experiments 1 and 2 used two-alternative forced-choice (2AFC) tasks to assess determiner preference when either Arabic or English served as the ML, and Experiment 3 employed eye-tracking to measure real-time processing. Data and Analysis: In Experiments 1 and 2, participants selected the more acceptable option between two code-switched sentences that included a mixed noun phrase (e.g., Arabic determiner + English noun vs. English determiner + Arabic noun). Reading-time measures from Experiment 3 were analyzed via mixed-effects models to compare non-switched and switched conditions. Findings: A consistent asymmetry emerged across all experiments: the Arabic determiner was preferred and processed more rapidly, regardless of the ML. This partially supported the MLF model only when Arabic served as the ML, revealing additional influence from morphological definiteness marking, language dominance, and processing economy. Parallel evidence from spoken data confirms this preference is modality-independent, suggesting that determiner asymmetry in Arabic–English code-switching is a fundamental feature of bilingual processing. Originality: This study provides novel, converging evidence from acceptability judgments and real-time eye-tracking that challenges the universality of MLF constraints in a typologically distinct language pair. It highlights the critical role of morphological properties and first language (L1) dominance over the structural constraints of the ML in functional element selection. Significance: These findings underscore the importance of integrating language-specific morphological features and L1 dominance into current models of bilingual sentence processing. The study establishes the Arabic determiner preference as a consistent, community-specific strategy, illustrating how typological differences can affect functional element selection in code-switching.