Word-to-text integration in English as a second language reading comprehension

We assessed the relationship between word-to-text-integration (WTI) and reading comprehension in 7th grade students (n = 441) learning English as a second language (L2). The students performed a self-paced WTI reading task in Fall (T1) and Spring (T2), consisting of three text manipulation types (anaphora resolution, argument overlap, anomaly detection), divided in simple and complex passages. The passages contained proximate versus distant anaphora, explicit repetitions versus implicit inferences, and no anomalies versus anomalies. We first examined how WTI complexity was related to reading times on target, target plus one, and target plus two, controlling for word frequency, decoding fluency, gender, and age. Mixed-effects models showed shorter reading times on T2 than on T1 and for simple compared to complex passages, indicating improvement of L2 reading speed. Complexity affected WTI for our L2 learners, as was reflected by longer reading times on complex compared to simple argument overlap and anomaly detection passages. We then assessed whether reading comprehension could be predicted by WTI. Longer reading times on complex compared to simple argument overlap and anomaly detection passages predicted offline reading comprehension. These WTI-measures of complexity are thus indicators of WTI proficiency for novice L2 learners.