Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Producing code-switches: Adaptation of cognitive control in code-switching
Poster Session D, Saturday, September 13, 5:00 - 6:30 pm, Field House
Demi Zhang1, Emiliana Pulido1, Maria Josefina Estrada1, Souad Kheder2, Edith Kaan1; 1University of Florida, 2University of Algiers2
Background: The role of language context in modulating executive control during production remains underexplored. Prior work often used constrained tasks, e.g., picture naming [2], which lack sentence-level activation. To bridge this gap, our study employs a reading-aloud paradigm—a balance between naturalistic production and control—given evidence of shared phonological [3] and planning processes between reading and speech [4]. We investigate how dense code-switching during production influences executive control by using a conflict adaptation paradigm in which reading dense code-switching or unilingual Spanish sentences is mixed with a flanker task [5,6]. We specifically test two competing hypotheses. Hypothesis A predicts that dense code-switching allows for a free gateway between languages and requires less inhibition, leading to increased flanker interference relative to unilingual sentences [7]. Hypothesis B predicts that dense code-switching recruits inhibitory control, thereby reducing subsequent flanker conflict due to cross-task adaptation. Method: Early Spanish-English bilinguals (n = 31 after 2 were excluded because of proficiency) read aloud experimental sentences in two language contexts: unilingual Spanish and dense code-switching (Table 1). Each experimental sentence was followed immediately by a flanker task with congruent (<<<<<, >>>>>) or incongruent (<<><<, >><>>) arrow arrays. Reading responses were recorded and annotated by trained bilingual coders. Trials were excluded if the audio was unclear, the participant made self-corrections, or the sentence did not contain the intended code-switching pattern. Trials with flanker response times <100 ms or >2000 ms and incorrect flanker responses were also excluded. Flanker RTs from remaining trials were analyzed using a linear mixed-effects model. The final model included flanker congruency, language context, and their interaction as fixed effects, and a by-participant random intercept. Results: Of the 1488 recorded experimental trials, 1163 trials (78.1%) from 31 Spanish-English bilingual participants were retained after applying pre-registered exclusion criteria. A significant main effect of flanker congruency was observed: participants responded more slowly to incongruent (M = 728 ms) than congruent (M = 602 ms) trials, indicating successful replication of the standard flanker interference effect (b = 0.310, SE = 0.018, t = 16.80, p < .001). There was no main effect of language context (b = –0.015, SE = 0.018, t = 0.83, p = .406), nor a significant interaction between flanker condition and context (b = –0.038, SE = 0.037, t = –1.02, p = .307). Reading aloud sentences with dense code-switches did not significantly influence subsequent performance on the flanker task compared to reading unilingual Spanish sentences. Discussion: Although bilingual language production is theorized to recruit domain-general control [1], the current findings suggest that merely reading code-switched sentences aloud does not exert a measurable influence on flanker task performance over reading unilingual sentences, thus neither Hypothesis A nor Hypothesis B was born out. This null effect may reflect the controlled nature of the reading-aloud task, which lacks the planning demands of spontaneous speech. Alternatively, code-switching may have involved cognitive control, but this may not have carried over to the subsequent flanker trial due to the reading and Flanker tasks being very different in nature.
Topic Areas: Control, Selection, and Executive Processes, Language Production