Using Restricted Natural Language for Data Retrieval: A Plan for Field Evaluation

Abstract

One strategy that has been proposed for dealing with thegrowing backlog for development of applications is to givecasual users languages for interacting directly with databases.Yet, there is little agreement on the form suchlanguages should take. Should they be natural-like, conformingclosely to a user's native tongue or should they bestructured to take advantage of the characteristics offormal languages?This paper presents the rationale for and design of afield evaluation of natural language for data retrieval.The natural language system and application are describedalong with the research design of the project. The resultsof the first part of the study, a laboratory experimentto investigate whether users perform better with an artificial or natural language, suggest that after equalamounts of training no difference in subject performanceis found between languages using a paper and pencil test .The insights gained to date are summarized.