Student Achievement Indicators at Defense Language Institute Foreign Language Center
Abstract
The Defense Language Institute (DLI) trains most of the cryptologic language analysts (CLA) that perform translation and analysis of data to support the United States military and intelligence communities. Students take the Defense Language Proficiency Test (DLPT) when graduating, passing if they achieve a score of L2/R2 (2+ on the Listening portion, 2+ on the Reading). DLI has been ordered to improve its students scores upon graduation. It seeks an improved model to screen applicants for the potential to achieve the new, more difficult grading benchmark of 2+ Listening, 2+ Reading. Former NPS student Jonathan Bermudez-Mendez looked into predicting student test scores based on grades, prior language experience, Defense Language Aptitude Battery (DLAB) scores, whether a student was recycled from a different language program, language category, and whether the student attended an immersion program, using stepwise logistic regression. We show that random forests and neural networks, especially the former, can improve on existing predictive models. We also investigate some univariate relationships based on prior language experience and show that many aspects of prior language exposure are statistically significantly related to the event of a student passing at the new benchmark.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 2021
- Accession Number
- AD1150437
Entities
People
- Ian A. Brenner
Organizations
- Naval Postgraduate School