Analysis of Reading Passages With the Reading Maturity Metric

• In Event: 70.021 - Multiple Perspectives on the Complexity of First-Grade Texts Over the Past 50 Years

Abstract

PERSPECTIVES: The Pearson Reading Maturity Metric (RMM) represents a new generation of automated text complexity measurement which incorporates modern approaches to natural language processing and artificial intelligence to assess broader and deeper levels of information in reading passages. The RMM uses computational linguistic variables to measure coherence, flow, word difficulty, the complexities of the grammar construction, sentence length, and semantic coherence in order to determine the overall difficulty and complexity of the reading passage. It also incorporates the computational language model Latent Semantic Analysis (LSA) to estimate how much language experience is required for a student to be able to process the meaning of each word, sentence, and paragraph in a text (e.g., Landauer, Kireyev, & Panaccione, 2011). The RMM analyzes passages and provides predicted text complexity scores aligned to student grade levels (e.g., K-14) as well to Common Core Grade bands.
METHODS: The RMM further provides information about words in the passage that may be more difficult for student comprehension in order to help text developers and teachers in choosing appropriate passages and instructional material around the passages. The measure has been validated across a range of standardized data sets and shown to provide 20-30% more accurate measures of text level than more traditional measures (see Nelson, Perfetti, Liben, & Liben, 2011).
DATA SOURCES & RESULTS: The first-grade passages were analyzed to derive overall RMM scores for each passage as well as measures of individual features that contribute the RMM scores. These features included measures of length, coherence, rate of introduction of new words, diversity of vocabulary, and usualness of the flow of words. The results indicate that generally, passage complexity, as measured by RMM, has increased from 1962 to 2013. Analysis of individual features show that passages from earlier decades tend to be more coherent, but less like standard written English. Passages from later decades introduce new words at much higher rate (30 words/text in 2013 vs. 4 words/text for 1962) and use words of greater difficulty.
SCIENTIFIC SIGNIFICANCE: The presentation will provide additional details of results and conclusions on the kinds of features that have changed in the complexity of the passages over the past 50 years.

Authors

• Peter W. Foltz, Pearson

• Mark Rosenstein, Pearson