Skip Navigation
Funding Opportunities | Search Funded Research Grants and Contracts

IES Grant

Title: Explicit Comprehension Instruction in an Automated Reading Tutor that Listens
Center: NCER Year: 2007
Principal Investigator: Mostow, Jack Awardee: Carnegie Mellon University
Program: Cognition and Student Learning      [Program Details]
Award Period: 4 years Award Amount: $1,999,543
Type: Development and Innovation Award Number: R305B070458
Description:

Purpose: Many students struggle to understand what they read even after they have achieved proficiency in basic reading skills (e.g., decoding, word recognition). Explicit instruction in reading comprehension, especially in the early grades, has been widely neglected in classroom practice, with relatively little research conducted that develops and evaluates instructional approaches for teaching reading comprehension in the primary grades. The purpose of this project is to develop and evaluate an automated tutorial intervention to help children in the first through third grades learn and use reading comprehension strategies. This research focuses on using explicit reading comprehension strategies to improve children's comprehension of both narrative and informational text. Although the researchers intend the intervention to be used with typically developing readers, the researchers are designing the intervention to be appropriate for readers with disabilities, and expect the automated tutor to have the greatest value for students with reading difficulties.

Project Activities: The initial activity is the development of the automated Reading Tutor for students in the first through third grades. The researchers are adapting an existing computer tutor to deliver new content for the new automated Reading Tutor. The existing platform displays stories on a computer screen, uses speech recognition technology to listen to children read aloud, and responds with spoken and graphical assistance. The automated Reading Tutor will take children through a five-phase instructional sequence for each reading comprehension strategy. First, the Reading Tutor provides students with an explicit description of a strategy (instruction is spoken by the Reading Tutor). Next, the Reading Tutor models how to use the strategy by displaying a written text and providing auditory instruction on what to do. In the third and fourth phases, the Reading Tutor provides the student with opportunities to practice using the strategy displaying a text, interrupting at specific points to prompt the student to use the strategy, and monitoring the student's responses. The final phase is independent use of the strategy by the student with the Reading Tutor fading out the comprehension strategy prompts.

An initial evaluation of the Reading Tutor will be conducted in elementary schools in a large urban city. To test whether the instruction helps scaffold comprehension, the researchers will conduct a within-subject randomized experiment comparing students' comprehension of text with and without support from the Reading Tutor.

Products: The outcomes of this study will include a fully developed automated Reading Tutor for teaching reading comprehension strategies to children in the first through third grades, and an initial evaluation of the effect of using the automated Reading Tutor on children's comprehension.

Structured Abstract

Purpose: The purpose of this project is to develop an automated tutorial intervention to improve reading comprehension of students in the first through third grades and conduct an initial evaluation of alternative variants of the automated tutor.

Setting: Research will be conducted in elementary schools in a large urban city in Pennsylvania.

Population: Participants include several hundred students in first through third grades, predominantly urban, low-income, and African-American. The sampling scheme will be systematic, including all students allowed by their school to participate, and excluding only students with disabilities that prevent them from operating the software.

Intervention: The automated Reading Tutor will take children through a five-phase instructional sequence for each reading comprehension strategy. First, the Reading Tutor provides students with an explicit description of a strategy (instruction is spoken by the Reading Tutor). Next, the Reading Tutor models how to use the strategy by displaying a written text and providing auditory instruction on what to do. In the third and fourth phases, the Reading Tutor provides the student with opportunities to practice using the strategy displaying a text, interrupting at specific points to prompt the student to use the strategy, and monitoring the student's responses. The final phase is independent use of the strategy by the student with the Reading Tutor fading out comprehension strategy prompts.

Research Design and Methods: In the first year of the project, the research team plans to complete a series of usability tests to determine characteristics of the automated tutor that are functioning as intended, and those which are not. As the project moves forward, the researchers will conduct multiple random assignment experiments - some using within-subject designs, some using between-subject designs -- to test variations in the structure and presentation of the automated Reading Tutor. The experimental findings will be used to support the design and development of the Reading Tutor. To provide a preliminary evaluation of the efficacy of including comprehension instruction in the automated Reading Tutor, students will be randomly assigned to receive the Reading Tutor with or without comprehension strategy instruction.

Control Condition: Students in the control conditions will use the Reading Tutor in its original form, without the strategy instruction turned on.

Key Measures: Predictor and outcome measures external to the automated tutor will include oral reading fluency, the Gates-MacGinitie Reading Comprehension Test, and finer-grained indicators of the comprehension strategies. Measures computed from data logged by the automated tutor will include performance on cloze questions and retelling tasks.

Data Analytic Strategy: Analysis of within-subject experiments will use logistic regression to predict performance on each comprehension question, using exposure to various tutorial prompts as predictor variables. Analysis of between-subjects experiments will use a repeated-measures analysis of covariance to compare students' growth in the Reading Tutor with embedded comprehension instruction versus without comprehension instruction.

Related IES Projects: Developing Vocabulary in an Automated Reading Tutor (R305A080157) andAccelerating Fluency Development in an Automated Reading Tutor (R305A080628)

Publications

Book chapter

Mostow, J., Beck, J.E., Cuneo, A., Gouvea, E., Heiner, C., and Juarez, O. (2010). Lessons From Project LISTEN's Session Browser. In C. Romero, S. Ventura, S.R. Viola, M. Pechenizkiy, and R.S.J.D. Baker (Eds.), Handbook of Educational Data Mining (pp. 389–416). New York: CRC Press, Taylor and Francis Group.

Proceeding

Chen, W. (2009). Understanding Mental States in Natural Language. In Proceedings of the 8th International Conference on Computational Semantics (pp. 61–72). Tilburg, Netherlands: Tilburg University.


Back