Chair: CULP, MARY BETH
DAI-A 59/05, p. 1547, Nov 1998

Two major questions were explored in this study: (1) whether reading in a hypertext environment affects the reading comprehension of second language learners in an intermediate-level college German class, and (2) whether the frequency of errors L2 readers make differs depending upon whether they have read print or hypertext. The experiment took place at a mid-size southern university over a nine-week quarter. The researcher used a modified A-B-A single-subject experimental crossover design in which data for each subject were collected and analyzed on a collective and an individual basis. L2 reading comprehension of subjects randomly assigned to one of two treatments, print text or hypertext, was evaluated both quantitatively and qualitatively. Reading comprehension performance data were gathered
for analysis on four separate occasions. Probe questions were used to establish a quantitative measure of readers' comprehension. Free recalls and selected one-on-one interviews were used for the qualitative analysis, which was based on the factors in Bernhardt's constructivist model of L2 reading comprehension. Additional data were gathered using assorted standardized tests as well as questionnaires. Twelve of the 14 subjects in the study scored higher on the quantitative measure when they had read hypertext. When stories were analyzed individually, subjects who had read hypertext outperformed those who had read print on three of the four stories. On two of these stories, the difference was statistically significant. Subjects reading print outperformed those reading hypertext marginally on one story, but the difference was not statistically significant. When all scores were listed from highest to lowest and divided at the median, 68 percent of the upper-half scores were for those reading hypertext. Analysis of the numbers of errors for each factor in Bernhardt's model for those reading print and those reading hypertext found only the difference for word recognition to be statistically significant, with those reading print making more errors on this factor than those reading hypertext. A
particularly salient secondary finding in the study was that hypertext appears to enable L2 readers to read more quickly.