The Interface between Readability and Automatic Text Simplification: Identifying Difficulties to Support Simple Writing

摘要

For nearly a century, readability formulas have focused on the complex task of outputing a single numerical value consisting in an estimate of the difficulty of a text for a given population of readers. Although this synthetic approach has virtues in certain contexts, its main limitation is that it analyses how dozens or even hundreds of linguistic characteristics of a text affect the reading process, but lets the user know about this process only through this single numerical value. Automatic text simplification (ATS), for its part, aims to identify complex features in a text (words, syntactic structures, numbers, etc.) and automatically simplify them. Despite being a finer-grained approach, due to the lack of theoretical and empirical data, ATS still struggles to identify all linguistic characteristics that should be simplified.