Automatic and Semi-Automatic Test Generation for Introductory Linguistics Courses Using Natural Language Processing Resources and Text Corpora

Abstract

This paper describes a collection of Natural Language
Processing (NLP) modules which automatically generate
exercises for introductory courses on structural linguistics and
English grammar at a Canadian University. While there is a growing demand for electronic exercises, online testing tools, and self contained linguistics and grammar courses, the exercises and tests offered on companion websites for popular textbooks consist largely of multiple choice type questions. The modules create exercises to practice and test part-of-speech identification, morphological analysis of complex words, and the analysis of sentences into phrase structure trees. They are part of an infrastructure capable of delivering instructional material, exercises for for self assessment, and online testing tools for courses which either use blended instruction or are taught exclusively online. Modules which are work in progress will be briefly discussed in the final section of this paper.