The comparative impacts of portfolio-based assessment, self-assessment, and scaffolded peer assessment on reading comprehension, vocabulary learning, and grammatical accuracy: insights from working memory capacity

Abstract This research was carried out to comparatively study the impacts of portfolio-based assessment, self-assessment, and scaffolded peer assessment on reading comprehension, vocabulary learning, and grammatical accuracy of Afghan English as a foreign language learners. To accomplish this, 172 learners enrolled at a language institute, through an Oxford Quick Placement Test (OQPT), 120 lower-intermediate learners and 5 higher-intermediate learners were selected. These selected participants were assigned into four groups: portfolio group ( N = 30), self-assessment group ( N = 30), scaffolded peer assessment group ( N = 35), and control group ( N = 30). The five higher-intermediate learners were injected into the scaffolded peer assessment group to function as the mediators, hence more participants in the group. After selecting the participants, through a reading-span test developed by Shahnazari (2013), learners’ working memory (WM) span was determined. It was discovered that 16 subjects in the portfolio condition, 14 self-assessment learners, 18 participants in the peer assessment group, and 13 participants in the control condition had high WM, while the rest of the participants had low WM. Thereafter, through validated instructor-made tests, subjects’ reading comprehension, knowledge of targeted lexical items, and grammatical accuracy at baseline were determined. Then, a ten-session treatment began. After the treatment, a follow-up post-test was administered. The results of three two-way between-group MANOVA disclosed that all three experimental conditions outstripped the comparison group on the second occasion and that high WM learners outstripped low WM learners (with a large effect size on reading comprehension test (partial eta squared = .365), a moderate effect size on the same test among high vs. low WM learners (partial eta squared = .095), a large effect size on vocabulary post-test (partial eta squared = .465), a moderate effect size on the same test among high vs. low WM learners (partial eta squared = .083), a large effect size on grammar test (partial eta squared = .500), and a moderate effect size on the same test among high vs. low WM learners (partial eta squared = .072)). The results further revealed that subjects in the scaffolded peer assessment group outstripped subjects in other experimental conditions, but the difference was non-significant. Additionally, the difference between the portfolio assessment and self-assessment group was not statistically significant. The implications of the study are reported.