Poster #51 - Social inference ability predicts reading comprehension in multilingual adolescents but not in children

• In Event: Poster Session 15

Abstract

Reading comprehension, relying on decoding and listening comprehension (Hoover & Gough, 1990), also requires readers to make inferences. Inference-making involves background knowledge, and the ability to follow writers’ motivations to convey information beyond what is explicitly written (e.g., Guajardo & Cartwright, 2016). Both children (e.g., Cain et al., 2004) and adolescents (e.g., Barth et al., 2015) rely on inference-making to succeed in reading comprehension. Daugaard et al. (2017) further showed that inference-making mediates the relationship between vocabulary and reading comprehension among monolingual children. Similar results were also observed in multilingual children and adolescents when reading in their second language (e.g., Hu & Nassaji, 2013; Li et al., 2021). However, readers’ multilingual language experience and background have not been considered in the relationship between inference-making and reading comprehension. In this study, we examine the relationship between inference-making and reading comprehension among children and adolescents with diverse linguistic experiences. Informed by the literature, we hypothesized inferencing would predict reading comprehension for multilingual children and adolescents and these relationships might be modulated by multilingual experiences.

One hundred and thirty-four multilingual children (age 6-10, n = 69, 38 females) and adolescents (age 12-15, n = 64, 36 females) participated. All participants were exposed to English and additional language(s) daily. Parents reported that they spoke English and additional language(s) at home with instructional languages being English and French. Children and adolescents were given online assessments: the Oral Passage Understanding Scale (OPUS; Carrow-Woolfolk & Klein, 2017), the Letter-Word Identification, the Picture Vocabulary, and the Passage Comprehension subtests from the Woodcock Johnson-IV (Schrank & Wendling, 2018). The Inference Subscale from the OPUS was extracted as a measure of social inferencing. Parents responded to the Language and Social Background Questionnaire (adapted from Anderson et al., 2018). We extracted the onset age of English acquisition (AoA) and parental reports of English exposure at home as their multilingual experiences.

Partial correlations controlling for age showed significant positive relationships among all English assessments within each age group (Table 1). Using stepwise regressions, the full models account for significant variances in children (F(2, 62) = 86.24, p < .001, R2 = 85%) and adolescents (F(1, 56) = 5.38, p = .024, R2 = 57%). For both children and adolescents, neither English AoA nor English exposure predicted reading comprehension after accounting for age and gender (Step 2, Table 2). For both age groups, English vocabulary and Decoding added significant variance predicting reading comprehension (Step 3, Table 2). Inference-making significantly predicted reading comprehension in adolescents (binference = .42, SE = .18, p = .02), but not in children (binference =.15, SE = .14, p =.27), after accounting all the other variables (Step 4, Table 2). Our results indicate that inference-making is a significant predictor of reading comprehension in multilingual adolescents, consistent with the literature (Barth et al., 2015). However, vocabulary and decoding accounted for significant variance in reading comprehension among multilingual children. Future considerations should be made to consider the interaction between social inferencing, multilingualism, and component skills of reading in children and adolescents.

Authors

• Justin Feng, McGill University
  Presenting Author

• Julie Oh, McGill University

• Katie Hartwick, McGill University

• Lauren Schellenberg, McGill University

• Gigi Luk, McGill University