Date of Award

5-2020

Degree Type

Report

Degree Name

Master of Science (MS)

Department

Communicative Disorders and Deaf Education

First Advisor

Sandra Gillam

Second Advisor

Tyson Barrett

Third Advisor

Ronald Gillam

Abstract

The purpose of this study examined the extent to which measures calculated using an automated, online discourse analysis program demonstrated predictive variance of a hand-scored metric of oral narrative proficiency with regards to macrostructure elements. An additional goal of this study was to explore the potential clinical utility of these automated measures by examining the extent to which they correlated with a hand-scored metric of macrostructure proficiency.

Pearson product correlation and multiple regression analyses were utilized to examine 240 oral narratives of children ages 6-9 that were analyzed using both an automated discourse analysis program and a hand-scored metric of narrative proficiency. The results showed that the automated Coh-Metrix measures of connectivity were predictive of the hand-coded macrostructure score. Coh-Metrix measures of narrativity and thematic relationships also accounted for additional variance to the macrostructure score. Certain Coh-Metrix measures of narrativity and connectivity correlated moderately with the hand-scored measure of narrative proficiency. These findings provide preliminary evidence that some aspects of the freely available Coh-Metrix automated scoring system, specifically measures of connectivity, narrativity, and thematic relationships predicted a macrostructure score based on human coding. Additionally, certain Coh-Metrix measures of narrativity and connectivity demonstrated moderate correlations with the narrative macrostructure proficiency score. However, the correlations did not demonstrate sufficient strength to provide an automated progress monitoring tool of narrative macrostructure proficiency.

Available for download on Wednesday, January 29, 2025

Share

COinS