Accession Number : ADA318951
Title : Analysis of Russian and Spanish Subskill Testing at the Defense Language Institute.
Descriptive Note : Master's thesis,
Corporate Author : NAVAL POSTGRADUATE SCHOOL MONTEREY CA
Personal Author(s) : Lavinder, Carlton L., III
PDF Url : ADA318951
Report Date : SEP 1996
Pagination or Media Count : 66
Abstract : The Defense Language institute is responsible for training military and government service personnel requiring a foreign language skill. Ten Subskill tests have been developed to evaluate the graduating students' language abilities and to determine if they have met the sponsor's Final Learning Objectives. The Subskill tests in some languages have been in place long enough that they can now be studied. This thesis examines these Subskill tests for both Russian and Spanish to determine if the tests have been developed and implemented in a manner to efficiently and consistently discriminate between students of different abilities. Three different issues are treated. The ANOVA is used identify Subskill tests with significant rater effects and the magnitude of those effects when they are present. Item Response Theory is used to examine the Subskill tests at the question level in order to identify questions that poorly discriminate between students of different abilities. In addition, the ability range that students are tested over is examined. Finally, methods using principle components and multiple regression are used to determine which tests, if any, can be eliminated with an acceptable loss of information about the students.
Descriptors : *LANGUAGE TRANSLATION, *CONDITIONING(LEARNING), *TEST CONSTRUCTION(PSYCHOLOGY), MILITARY REQUIREMENTS, SKILLS, STUDENTS, PERFORMANCE(HUMAN), ANALYSIS OF VARIANCE, THESES, NONPARAMETRIC STATISTICS, RUSSIAN LANGUAGE, NAVAL RESEARCH, SPEECH RECOGNITION, MILITARY TRAINING, COURSES(EDUCATION), LINEAR REGRESSION ANALYSIS, MENTAL ABILITY, APTITUDE TESTS, SPANISH LANGUAGE, ACHIEVEMENT TESTS, APTITUDES, TRANSFER OF TRAINING.
Subject Categories : Linguistics
Distribution Statement : APPROVED FOR PUBLIC RELEASE