Syntagmatic Distractor Generation for Vocabulary Multiple Choice Questions

Abstract

Learning assessments that are currently conducted online pose new problems for educators as they demand extensive resources, take time, and require a deep understanding of the materials involved. Standardized English tests that have been used for a long time often use a multiple-choice format for convenience in scoring. The problem with creating multiple-choice questions for vocabulary tests is how to generate distractors that are plausible while not being acceptable as answers. This research focused on generating distractors for vocabulary questions that were syntagmatically related to the reading passage. To date, there has not yet been any research that focuses on the syntagmatic aspect of generating English vocabulary distractors for multiple-choice questions. This research generated distractor candidates from the LinguaTools word collocation dictionary that were filtered according to part of speech and syntactic relation, then ranked by the significance of co-occurring relevance to the context. We generated distractors for vocabulary questions which were then evaluated by educators in high schools and colleges who were experts in setting English tests. The results of the proposed method indicated that 80.83% of generated distractors qualified as vocabulary items for multiple-choice questions.