Date of Award

2018

Document Type

Open Access Dissertation

Department

Linguistics

Sub-Department

College of Arts and Sciences

First Advisor

Nina Moreno

Abstract

Reading in a second language (L2) is a critical aspect of language acquisition, yet gaps remain in the literature regarding the extent to which textual factors impact reading difficulty. There is consensus that complex vocabulary and grammar affect L2 comprehension (Koda, 2005), and this is evidenced through the numerous traditional readability formulas, such as Flesch-Kincaid (FKGL). However, critics argue that discourse-level features, such as cohesion, also impact reading difficulty and must be included in difficulty analyses (Carrell, 1987).

One aspect of cohesion is content word overlap, or the number of content words repeated in a text; this measure is included in Coh-Metrix (Graesser, McNamara, Louwerse, & Cai, 2004), a recent readability tool. Content word overlap (CWO) is thought to facilitate reading by making explicit text connections to relieve the processing burden (Kintsch & Van Dijk, 1978). This claim is supported in the L1 literature (Britton & Gulgoz, 1991), but is relatively unexplored in L2 contexts.

This study utilized two experiments to capture effects of CWO in L2 reading comprehension. Experiment one looked at L2 reading comprehension scores (N=1131 scores on 18 tests), and compared the results to predictions of text difficulty made by two divergent readability tools. Results indicated that Coh-Metrix, which measures CWO, was more reliable in predicting comprehension than FKGL, which considered lexical/grammatical complexity alone.

Experiment two utilized eye-tracking measurements of L1 (n=31) and L2 readers (n=63) in passages with varying amounts of CWO of a target word. Results indicated that all nonnative speakers were susceptible to longer target word reading times when no overlap was present (p <.05). Additionally, lower proficiency L2 learners benefitted from more overlap than higher proficiency learners (p <.05) in target reading time and integrating the target into the surrounding text.

The results of both experiments support the role that content word overlap affects second language reading in both overall comprehension and localized processing. Thus, when considering models of L2 reading, it is necessary to include cohesion in addition to lexical and grammatical complexity. These findings have implications for pedagogues, and assessment developers to ensure that reading difficulty is accurately measured.

Included in

Linguistics Commons

Share

COinS