Post by Panayiota (Pani) Kendeou & Kristen McMaster, University of Minnesota
Despite the persistent efforts of researchers, policy makers, and educators to improve reading performance of all children across grade levels, year after year the National Assessment of Educational Progress (NAEP) releases statistics that show a significant percentage of U.S. students performing below basic proficiency levels in reading. For example, the latest Nation’s Report Card indicates that approximately 31% of 4th graders read below a basic proficiency level (NAEP, 2015)—that is, they fail to make simple inferences and understand the overall meaning of texts. Students who experience such difficulties are likely to struggle throughout their education and employment.
In the context of reading comprehension, an inference is information that is retrieved from memory or generated during reading to fill in information that is not in a text. Reading researchers have examined the conditions under which inferences are generated, the nature and types of inferences readers generate, and the neural correlates of inference generation. Findings have revealed that inference making is one of the unique, significant predictors of reading comprehension, with some studies indicating a causal link from poor inference making to poor reading comprehension (Oakhill & Cain, 2012). We consider inference making the cornerstone of reading comprehension.
How do we develop the ability to make inferences? The development of inference skills begins well before formal reading instruction starts. For example, 2-year olds can generate causal inferences between sequential events; 4-year olds can generate causal inferences of the events they experience or hear; and 6-year olds can generate causal inferences during comprehension of aurally presented or televised stories. Thus, even very young children engage in inferential processes to comprehend the events they experience in their everyday lives. As children get older, they generate a greater number and wider variety of inferences during everyday experiences and through both listening and reading comprehension.
Our focus on inference skills is also motivated by strong evidence that the ability to draw inferences is a general skill—it is not specific to reading (Kendeou, 2015). Strong evidence from developmental studies that examined inference skills in very young children (even two-year-olds) supports this claim. For example, children make causal inferences from what they see and hear (e.g., pushed the vase and heard a sound – inference that the vase broke); spatial inferences pertaining to the locations of different objects, and emotional inferences from people’s facial expressions (e.g., give mom flowers, mom smiles – inference that mom is happy). There is also direct evidence for the generalization of inference skills in older children and adults. For example, in one of our studies we assessed inference skills using aural, televised, and written stories in 4-, 6-, and 8-year-olds, and found that children generated bridging and elaborative inferences across different media that significantly predicted reading comprehension independent of the types of media (Kendeou et al., 2008).
What are Multi-Tiered Systems of Support? Multi-tiered systems of support (MTSS) have been developed in an effort to provide high quality instruction and differentiated support for all students (Fuchs, Fuchs, & Compton, 2012). They consist of prevention-based models that minimize the risk of failure by responding quickly with evidence-based approaches to meet student needs.
Within these systems, the emphasis is on a hierarchy of support that is differentiated through data-based decision making. In this context, instruction increasingly varies in intensity, frequency, and individualization at three levels or tiers. Tier 1 instruction includes high quality core instruction for all students in the general education classroom. This tier is often termed the ‘prevention tier’ and typically addresses successfully the needs of approximately 80% of the students. Tier 2 instruction includes targeted small-group interventions for those students identified as at risk. This tier addresses the needs of approximately 15% of the students. Tier 3 instruction includes high intensity, individualized intervention and serves approximately 5% of the students. At all tiers, teachers need instructional tools and assessments that meet a wide range of student needs and are efficient to implement. Furthermore, the hierarchy suggests that high quality Tier 1 instruction will reduce the probability and need for Tier 2 targeted intervention, and similarly high quality targeted Tier 2 intervention will reduce the probability and need for Tier 3 individualized high intensity intervention. In our work, we situate inference making training within MTSS and focus specifically on Tiers 1 and 2.
ELCII is designed to support reading comprehension by developing inference making for all students in Kindergarten. To achieve this goal, ELCII does not rely on decoding skills. It is an intelligent tutoring system (ITS) with 24 modules, which engage students to:
TeLCI is designed to improve reading comprehension by developing inference making for students who experience comprehension difficulties in Grades 1-2. We identify struggling comprehenders as those students who score at or below the 25th percentile in both language and reading comprehension measures. TeLCI does not rely on decoding skills. It is an ITS with 24 modules, which engage students to:
ELCII and TeLCI build on major findings of previous cognitive, developmental, instructional, and assessment work conducted by our research team, highlighting that (a) early language comprehension skills that are developed in non-reading contexts (e.g., video comprehension) contribute significantly to later reading achievement, (b) language comprehension skills--and specifically inference skills-transfer across different media, (c) children’s inference skills can be improved using questioning that includes scaffolding and specific feedback (McMaster et al., 2012), and (d) technology provides a cost-effective, standardized, individualized delivery of instructional tools in classroom settings.
Each learning module in ELCII and TeLCI is embedded in a cloud-based software application that is individualized, fully automated and interactive. It is individualized with computer adaptive algorithms already established in FAST™, which facilitate the use of sophisticated branching in scaffolding and feedback. The interactivity is supported by an agent, which is the ‘face and voice’ of the instructions, questions, and feedback in each learning module. The agent was designed to function as a fictional, more knowledgeable peer whose mission is to help each child learn how to make inferences. The software application also has a built-in database and automated reports for teacher use.
Our team is engaged in ongoing research to test usability, feasibility, and promise for implementation of these ITS by school personnel. We have already developed TeLCI and tested its usability and feasibility. We followed a bottom-up, development approach during which we solicited input from teachers and parents at different points of time concerning software functionality, content, and appropriateness (cultural, age). Preliminary results suggest high levels of acceptance and feasibility for implementation in authentic school settings, and we are now working on revising the tool so we can examine its promise in an efficacy study. ELCII development is also starting soon. You can track this work here.
An important question that we will also address is whether the effectiveness of questioning may differs depending on when questions are asked. Is it more beneficial to prompt the students during (i.e., online) or after (i.e., offline) the comprehension task (watching the video)? The rationale for an online questioning approach is the focus on the cognitive processes that operate during comprehension, because it is during these moment-by-moment processes that comprehension succeeds or fails. Indeed, recent research has indicated that online questioning shows promise for improving reading comprehension of 4th grade struggling readers (McMaster et al., 2012), but has yet to be explored with younger readers. The rationale for an offline questioning approach is that it may help young readers make inferences, without overly taxing attention and working memory due to the interruptions.
ELCII and TeLCI are only a few examples of educational technologies that have the potential to enhance educational practice. In recent years, the scientific community has made significant progress in this direction, as there is evidence that educational technology (e.g., games, intelligent tutoring systems) improves a variety of student learning outcomes. With continued support through federal funding, technological advances will further solutions to reading comprehension problems. Given the importance of reading comprehension to academic achievement and lifelong success—and particularly to closing achievement gaps—our efforts must continue to both prevent and ameliorate reading comprehension difficulties.
The research reported herein was funded by grant numbers R324A160064 and R305A170242 from the U.S. Department of Education to the University of Minnesota. The opinions are those of the authors and do not represent the policies of the U.S. Department of Education.
This post is part of a special series curated by APA Division 15 President Bonnie J.F. Meyer. The series, centered around her presidential theme of "Welcoming and Advancing Research in Educational Psychology: Impacting Learners, Teachers, and Schools," is designed to spread the dissemination and impact of meaningful educational psychology research. Those interested can learn more about this theme in Division 15's 2016 Summer Newsletter.
Fuchs, D., Fuchs, L. S., & Compton, D. L. (2012). Smart RTI: A next-generation approach to multilevel prevention. Exceptional Children, 78, 263-279.
Kendeou, P. (2015). A general inference skill. In E. J. O’Brien, A. E. Cook, & R. F. Lorch (Eds.), Inferences during reading (pp. 160-181). Cambridge, MA: Cambridge University Press.
Kendeou, P., Bohn-Gettler, C., White, M., & van den Broek, P. (2008). Children’s inference generation across different media. Journal of Research in Reading, 31, 259-272.
McMaster, K. L., van den Broek, P., Espin, C. A., White, M. J., Rapp, D. N., Kendeou, P., . . . Carlson, S. (2012). Making the right connections: Differential effects of reading intervention for subgroups of comprehenders. Learning and Individual Differences, 22(1), 100-111.
Oakhill, J., & Cain, K. (2012). The precursors of reading ability in young readers: evidence from a four-year longitudinal study. Scientific Studies of Reading, 16(2), 91-121.