Predictors of success in writing in English as a foreign language: reading, revision behavior, apprehension, and writing

College Student Journal, Dec, 2002 by Sy-Ying Lee, Stephen Krashen

The combination of the four predictors was able to account for a substantial 40% of the variability in grades.

The results of the second multiple regression analysis are presented in table 4. Results for the four predictors included in the first multiple regression are nearly identical to the results presented in table 3. Focus on organization in revision is also a good predictor of grades, falling just short of the .05 level of significance, and the overall r2 improved to .44.

We are unwilling to discard those predictors that did not reach statistical significance in the regression analysis, namely focus on grammar and focus on content during revision. Bivariate correlations between focus on content and grade and between focus on grammar and grade were statistically significant. When we ran subsequent multiple regressions using only focus on grammar without focus on content of organization, it was a significant predictor (p < .05). Entering only focus on content and not focus on grammar resulted in focus on content coming much closer to statistical significance (p =. 16). In addition, at this stage in our research with so many additional factors yet unknown, and with relationships among factors unclear, we are more concerned with making type II errors than type I errors: It is, we feel, premature to reject these factors as possible predictors of grades in composition class. More sophisticated analyses with more sensitive measures could result in stronger results.

Discussion

Methodological issues

Future studies of the success of students in writing classes at this level should consider the following methodological issues.

First, there was no control for previous knowledge of English. Fowler and Kroll (1980) found that measures of reading and scores on standardized tests of English were significant predictors of grades in a college composition course for native speakers of English, and in fact overwhelmed the effect of writing apprehension, which was not a significant predictor when other factors were considered.

One could argue that this variable is controlled to at least some extent in this study, because all participants had successfully completed freshman English, and chose to take the composition course, suggesting that they were among the more proficient and motivated English students and had lower writing apprehension than their peers (Daly and Miller, 1975b, discussed below). Nevertheless, including a measure of previous English competence would give us a clearer indication of improvement attributable specifically to the experience in the elective writing class.

Second, we acknowledge that our measures were crude. We asked only for yes/no binary responses with respect to reading and writing behavior, as well as with respect to revision strategies, and assumed that grades were a valid measure of writing competence. Our measures produced clear results, but more finely-tuned measures should be employed.

Relationships among predictor variables

Earlier research found a modest relationship between free reading and writing apprehension. This relationship was confirmed here. The correlation between reading and writing apprehension fell just short of significance, but was nearly identical to the size of the correlation reported by Lee and Krashen (1997). Lee and Krashen hypothesized that those who read more have more competence in the written language, which lowers apprehension.