Changes in Content and Source Use

Only recently have researchers started to examine the way L2 learners integrate material from source texts in their own writing. Because most tests of L2 writing used to include only independent writing tasks, source use was not among the aspects that studies on L2 writing performance examined. However, with the growing use of integrated writing tasks in major L2 proficiency tests, which require students to listen to and/or read source texts and then use these sources in their writing, the number of studies that analyze L2 learners’ written texts in terms of content and source use is growing. This growing interest in source use in L2 writing is reflected in the growing number of studies examining how L2 learners use source texts in their academic writing in non-test settings as well. Some of these studies examine the process that L2 learners engage in when using source texts while writing, but the majority of these studies analyze L2 learners’ texts to identify how these learners select and integrate content and/or language from source texts into their own writing. As noted in Chapter 2, these studies often aim to identify which and how many points from the source texts L2 learners integrate in their texts, how they do this (e.g., quote, paraphrase, summary), and/or the quality (i.e., appropriateness and effectiveness) of use of material from sources. While there is a growing interest in source use in L2 writing, content or development is not usually examined in studies on L2 learners’ written texts. Because the two writing tasks in this study have different demands in terms of content and source use, in this study, we examined content (i.e., topic adherence) for the independent task, and source use for the integrated task.

Content for the Independent Task

We examined two indices of topic adherence for the independent task: semantic overlap (LSA cosine scores) and lexical overlap (synonym overlap for nouns and verbs), both were obtained using the computer program CRT/'. PERMANOVA, with occasion, ELP group, and student (within ELP) as predictors and the two indices of topic adherence as dependent variables, detected significant main effects for occasion (F(l, 134) = 4.87,p = .0092, rf = .09),ELP group (F(3,134) = 4.14, p = .0008, rf = .32), and student (within ELP) (F(78,134)= 1.98,p = .0002, rf = .76) for the full sample. For the longitudinal sample, there were significant main effects only for student (within ELP) (F(44, 95) = 1.98,p = .0013, rf = .66).

Semantic Overlap: Table 8.1 reports descriptive statistics for LSA cosine scores, a measure of semantic overlap between the independent task and essays, across test occasions, and ELP groups. For the full sample, ANOVA indicated significant effects for occasion (F(l, 134) = 6.80,p= .0108, rf = .12) and student (within ELP) (F(78,134) = 2.15,p = .0021, ff = .77). For the full sample,ANOVA indicated significant effects for student (within ELP) (F(44, 95) = 2.26,p = .0045, ff = .69). As Table 8.1 shows, for the full sample, participants obtained significantly higher LSA cosine scores (M = .61) after instruction than they did before instruction (Af = .59).

Lexical Overlap: Table 8.1 reports descriptive statistics for synonym overlap (nouns and verbs), a measure of lexical overlap between the independent task and essays, across test occasions, and ELP groups. For the full sample, ANOVA detected significant effects for ELP group (F(3,134)= 9.07,p = .0001, ff = .48) and student (within ELP) (F(78, 134) = 1.80,p = .0135, ff = .74).There were no significant main or interaction effects for the longitudinal sample. As Table 8.1 shows, for the full sample, participants with higher ELP had lower synonym overlap indices. However, only the difference between the very low group and each of the other ELP groups was significant. The group with the lowest ELP had significantly higher synonym overlap indices (M = .46) than did each of the other ELP groups.

Overall, the degree of semantic overlap between the essays and the independent task increased after instruction, suggesting that the independent essays written after instruction were more relevant to the independent task than were those written before instruction.There was no significant change in lexical overlap indices across test occasions, but the low ELP groups had significantly higher lexical overlap indices than did the other ELP groups, perhaps because they copied more words from the independent task into their independent essays than did the other groups.