Assessing Readability Formulas: A Comparison of Readability Formula Performance on the Classification of Simplified Texts

EasyChair Preprint 3851

Abstract

This study compares the performance of five different traditional and new readability formulas in the task of classifying simple Wikipedia and authentic Wikipedia articles (N = 4,000). Results indicated that a new formula, the Crowdsourced Algorithm of Reading Comprehension (CAREC) performed the best. The traditional readability formula, Flesch-Kincaid Grade Level, also showed reliable performance. The results suggest the linguistic features used in newer readability formulas are capable of reliably representing the difficulty of a text.

Keyphrases: comprehension, readability, text difficulty

@booklet{EasyChair:3851,
  author    = {Joon Suh Choi and Scott A. Crossley},
  title     = {Assessing Readability Formulas: A Comparison of Readability Formula Performance on the Classification of Simplified Texts},
  howpublished = {EasyChair Preprint 3851},
  year      = {EasyChair, 2020}}