This study aimed to investigate the effects of reading instruction, where we set a specific goal for reading, on the detection and resolution of inconsistent information in several dimensions during narrative reading. A total of 39 Japanese university students, comprising learners of English as a foreign language (EFL), read short English texts introducing consistent or inconsistent information on the protagonist or spatial dimensions. We asked participants to verbalize their thoughts while reading the texts (think-aloud) under the following two conditions: (a) control condition, where the readers read for comprehension, and (b) image instruction condition in which they read while picturing the situation of the narratives in their minds. To explain whether the readers could understand and maintain coherence in the texts, that is, detect and resolve inconsistent information about the protagonist and spatiality, we analyzed their think-aloud protocols. The results showed that image instruction was effective in helping readers maintain protagonist coherence regardless of their proficiency level, whereas it was not significantly effective for spatial dimensions. The pedagogical suggestions from these findings are discussed in detail.
