Minimal Overlap in Language Control Across Production And Comprehension: Evidence from Read-Aloud Versus Eye-Tracking Tasks

Abstract

Bilinguals are remarkable at language control-switching between languages only when they want. However, language control in production can involve switch costs. That is, switching to another language takes longer than staying in the same language. Moreover, bilinguals sometimes produce language intrusion errors, mistakenly producing words in an unintended language (e.g., Spanish-English bilinguals saying "pero" instead of "but"). Switch costs are also found in comprehension. For example, reading times are longer when bilinguals read sentences with language switches compared to sentences with no language switches. Given that both production and comprehension involve switch costs, some language-control mechanisms might be shared across modalities. To test this, we compared language switch costs found in eye-movement measures during silent sentence reading (comprehension) and intrusion errors produced when reading aloud switched words in mixed-language paragraphs (production). Bilinguals who made more intrusion errors during the read-aloud task did not show different switch cost patterns in most measures in the silent-reading task, except on skipping rates. We suggest that language switching is mostly controlled by separate, modality-specific processes in production and comprehension, although some points of overlap might indicate the role of domain general control and how it can influence individual differences in bilingual language control.

Keywords: Bilingualism; Language control; Language production; eyetracking; read–aloud; silent–reading.

Figure 1.

Distribution of the number of bilinguals by the number of summed partial and complete language intrusion errors during reading aloud mixed–language paragraphs.

Figure 2

a. Relationship between target–word predictability and language switch depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task, as indicated by the single–fixation duration. b. Relationship between target–word predictability and language switch depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task, as indicated by the gaze duration. c. Relationship between target–word predictability and language switch depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task, as indicated by the go–past time. d. Relationship between target–word predictability and switching as indicated by the regression rate and depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task. e. Relationship between target–word predictability and switching as indicated by the skipping rate and depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task.

Figure 2

a. Relationship between target–word predictability and language switch depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task, as indicated by the single–fixation duration. b. Relationship between target–word predictability and language switch depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task, as indicated by the gaze duration. c. Relationship between target–word predictability and language switch depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task, as indicated by the go–past time. d. Relationship between target–word predictability and switching as indicated by the regression rate and depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task. e. Relationship between target–word predictability and switching as indicated by the skipping rate and depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task.

Figure 2

a. Relationship between target–word predictability and language switch depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task, as indicated by the single–fixation duration. b. Relationship between target–word predictability and language switch depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task, as indicated by the gaze duration. c. Relationship between target–word predictability and language switch depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task, as indicated by the go–past time. d. Relationship between target–word predictability and switching as indicated by the regression rate and depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task. e. Relationship between target–word predictability and switching as indicated by the skipping rate and depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task.

Figure 2

a. Relationship between target–word predictability and language switch depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task, as indicated by the single–fixation duration. b. Relationship between target–word predictability and language switch depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task, as indicated by the gaze duration. c. Relationship between target–word predictability and language switch depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task, as indicated by the go–past time. d. Relationship between target–word predictability and switching as indicated by the regression rate and depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task. e. Relationship between target–word predictability and switching as indicated by the skipping rate and depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task.

Figure 2

a. Relationship between target–word predictability and language switch depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task, as indicated by the single–fixation duration. b. Relationship between target–word predictability and language switch depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task, as indicated by the gaze duration. c. Relationship between target–word predictability and language switch depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task, as indicated by the go–past time. d. Relationship between target–word predictability and switching as indicated by the regression rate and depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task. e. Relationship between target–word predictability and switching as indicated by the skipping rate and depending on the number of language intrusion errors (<5 or >5) made during the read–aloud task.