Letting the genie out of the lamp: Using natural language processing tools to predict math performance

Abstract

This study examines links between natural language processing and its application in math education. Specifically, the study examines language production and math success in an on-line, blended learning math program. Unlike previous studies that have relied on correlational analyses between linguistic knowledge tests and standardized math tests or compared math success between proficient and non-proficient speakers of English, this study examines the linguistic features of students’ language production while e-mailing a virtual pedagogical agent. In addition, the study examines a number of non-linguistic features such as grade and objective met within the program. The findings indicate that linguistic features related to the use of standardized language use explain around 8% of math success. These linguistic features outperform non-linguistic features.