The Effect of Code Switching (Somali–Arabic) on Narrative Coherence in Contemporary Somali Short Stories: A Study of Mogadishu Based Literary Works

Background Somali–Arabic code-switching in Mogadishu short stories increased after the 1991 state collapse, diaspora returns, and Islamic revival. Treating switching as a literary strategy rather than a deficiency, this study established the first robust and reliable empirical baseline for Somali–Arabic literary code-switching and tested whether switching frequency affects narrative coherence. Methods Four hundred Somali short stories published between 2015 and 2025 were randomly selected from Mogadishu-based literary archives. Non-nativized Arabic insertions were coded per thousand words. Five bilingual raters evaluated narrative coherence using a four-dimensional rubric covering referential, temporal, causal, and thematic coherence. Polynomial regression, matched-pair t-tests, and moderation analysis were conducted to examine frequency–coherence relationships, compare high-frequency and low-frequency stories, and identify moderating effects of switch functions and quotation-bound status. Results Mean code-switching frequency was 14.8 per thousand words, ranging from 2 to 45. Mean narrative coherence was 67.3 out of 100. Polynomial regression revealed a significant inverted U-shaped relationship between frequency and coherence, with optimal coherence observed at 13 to 14 switches per thousand words and a sharp decline beyond 22 switches. High-frequency stories exceeding 22 switches scored substantially lower (54.2) than low-frequency stories below 8 switches (68.4), representing a mean difference of 14.2 points with a large effect size. Quotation-bound switches and switches serving foregrounding or characterization functions were associated with significantly higher coherence scores compared to unmarked lexical insertions. Conclusion Moderate and functional code-switching sustains narrative coherence effectively, whereas excessive switching disrupts it considerably. These findings strongly support the proposed Frequency-Adjusted Coherence Model. Authors and editors of multilingual Somali narratives may benefit from targeting 12 to 18 Arabic switches per thousand words and avoiding more than 22 switches. This study provides empirical benchmarks for evaluating code-switched literary works in typologically distant language pairs and offers practical and clear guidance for creative writing pedagogy and editorial review processes.