Date of Award


Document Type

Campus Access Thesis



First Advisor

Barbara Schulz


There are obvious differences between a second language (L2) and a native language (L1). When processing an L2, a person is slower and prone to more errors than when processing an L1. However, the reasons for these differences are not clearly understood. A possible explanation is a limitation in available working memory (WM) resources, which may result in shallow parsing behavior (Hopp, 2006).

Regardless, we still do not have a clear understanding of the resources available during L2 processing. L1 reading span as measured by the reading span task (Daneman and Carpenter, 1980) has been shown to be a good indicator of biological WM capacity (Conway et al., 2005). However, studies have shown that L2 reading span is generally lower than L1 reading span (Service et al., 2002). These results suggest that L2 reading span is affected by far more than just biological factors. Factors such as proficiency and processing speed may greatly influence the availability of WM resources. While previous work has investigated the relationship between L2 reading span and various predictor variables, the predictors have always been analyzed in isolation.

This study investigates how L2 reading span is affected by combined factors to give a more accurate depiction of WM resources under the strain of processing a second language. The reading span of German participants in their native language and English, as well as measures of language background, proficiency, and reading speed were compared. These comparisons were run through a multiple regression analysis to see how well combined factors can predict available L2 resources evidenced in L2 reading span. Results indicate that advanced non-natives did not possess statistically significant differences between their first and second reading span scores. These two scores correlated highly with each other, but the measures of language skill and background did not. Incorporating multiple predictors in the regression model added no significant predictability for L2 reading span. Nevertheless, given all of the participants' high English proficiency, it is possible that the group was too homogeneous to show effects of language. Furthermore, these results question the assumption that non-natives naturally have lower working memory capacities.