Evaluating Passage-Level Contributors to Text Complexity

• Main Author: Munir-McHill, Shaheen
• Other Authors: Good, III, Roland
• Language: en_US
• Published: University of Oregon 2013
• Subjects: Curriculum-based measures; Oral reading fluency; Readability; Text cohesion; Text complexity
• Online Access: http://hdl.handle.net/1794/13422

The complexity of text has a number of implications for educators in the areas of instruction and assessment. Text complexity is particularly important in formative assessments, which utilize repeated, alternate, equivalent forms to capture student growth towards a general outcome. A key assumption of such tools is that alternate forms of the assessment are of equal complexity. Consequently, there is a need to better understand what variables contribute to text complexity and how they impact student performance. This study was designed to evaluate features of text that are not typically included in readability estimates but may contribute to the text complexity: text cohesion and genre. Currently, text complexity of oral reading fluency measures is often quantified using readability estimates. It is hypothesized that a factor generally excluded from readability estimates, text cohesion&mdashthe extent to which the text functions as a cohesive, meaningful whole&mdashcontributes to text variability and variability in student performance. This research evaluated the role of a type of text cohesion (referential cohesion) in text complexity by manipulating the cohesion of passages otherwise assumed to be of equal difficulty. Genre was also considered, as research suggests that genre may impact complexity ratings of texts. Passages were strategically selecting to capture four conditions&mdash1) informational text/low cohesion, 2) informational text/high cohesion, 3) narrative text/low cohesion, and 4) narrative text/high cohesion. Data were collected on reading rate, accuracy, and passage-specific reading comprehension Results were analyzed using two-way, univariate ANOVA with dependent observations. Results indicate effects for each of the dependent variables included in the design. For rate and accuracy, results indicate significant interactions between genre and referential cohesion; scores were significantly higher for high cohesion narrative text than low cohesion narrative text and high cohesion informational text. There was a significant main effect of genre on comprehension, with students performing significantly better on the comprehension measure for narrative texts than informational texts. Altogether, these results indicate direct effects of genre and referential cohesion on student reading performance and provide evidence that text cohesion may be a meaningful component of text complexity. === 2015-10-10