Studies and Design Specifications for Computerized Measurement of Textual Comprehensibility.

摘要

A previous report (AD-A001 537) defined a series of 14 novel measures for determining the comprehensibility of English text on the basis of current psycholinguistic and Structure-of-Intellect oriented concepts. That report not only suggested the potential usefulness of the measures, but also conjectured the feasibility of automating the calculation of these measures. The present report takes the next logical steps in implementing these measures for computer application. First, these measures are analytically defined and described. Then, selected measures are subjected to 'laboratory' experimental investigation using Air Force Manuals, Career Development Course materials, and USAF Technical Orders as sample texts. Results of these experiments are presented. An automatic calculation method is then developed for each of the 13 selected measures. The structure of the programming specifications is modular and is intended to calculate the measures for variable size blocks of texts. Flow charts and summary descriptions of the program attributes are also presented, together with explanations of run request syntax, sample measures calculations, and output formats. This report then constitutes a complete definition of the program suitable for future implementation on an automatic data processing system.