Although much research has reported the effectiveness of task repetition on oral performance (Bygate, 2018), few studies have investigated its effectiveness on writing performance (Manchón, 2014), especially in languages other than English. For instance, Nitta and Baba’s (2014) longitudinal study revealed that EFL undergraduates considerably progressed their syntactic complexity and lexical aspects, but not fluency, through repeating a timed writing task. In relation to the task repetition, however, whether and how L2 learners develop their grammatical accuracy and communicative adequacy (Pallotti, 2009) has remained unclear in the literature. Furthermore, in addition to the linguistic measurements and the qualitative assessments, scant research has attempted to investigate whether any significant changes are brought about in terms of learners’ perceptions through repeating language tasks. <br>Therefore, the current study has shed new light on the developmental changes in the writing performance of Japanese-as-a-foreign-language (JFL) learners. It investigated whether any remarkable changes are brought about in terms of overall complexity, complexity by subordination, accuracy, and fluency through repeating a weekly “15-Minute Writing Task” throughout one academic semester (16 weeks) and one academic year (32 weeks). The writing task topics were considered in terms of the Cognition Hypothesis (Robinson, 2001), which states that different cognitive demands of tasks will lead to different L2 output. Regarding this point, this study explored whether there were any significant differences between two task types: descriptive and argumentative essays. JFL learners who were enrolled in an intermediate-level course at an American university engaged in the two different types of timed writing tasks.<br>First, the one-semester investigation, based on the pre/posttest analysis, revealed different patterns between the two types of writing tasks. For the descriptive essays, despite the improvements in overall complexity, complexity by subordination, and fluency with a large effect size (r ≥ .6) (Plonsky & Oswald, 2014), no significant findings were confirmed for accuracy. In contrast, in the argumentative essays, the learners improved all the linguistic aspects but with a medium effect size (.4 ≤ r < .6).<br>Second, in the one-year investigation, the JFL learners significantly improved overall complexity, complexity by subordination, and fluency during the study period. The dynamic systems approach (Verspoor & van Dijk, 2011) also unraveled the developmental trajectories to show how different variables interacted in the two different types of writing tasks, respectively, throughout the measurement period. Although there were no statistically significant differences in grammatical accuracy measures, the process of L2 writing development showed fluctuations, demonstrating that the improvements in syntactic complexity seemed to have caused many grammatical errors temporarily. Lastly, the learners’ compositions, which were also assessed qualitatively by two native Japanese speakers in terms of readability, indicated significant improvements in communicative adequacy.<br>Finally, to investigate any changes in the learners’ beliefs toward Japanese writing before and after the task repetition, the JFL learners completed the Belief Questionnaire About Writing in Japanese (Ishibashi, 2009). In addition, to examine any changes in foreign language anxiety with a focus on Japanese writing, the learners were required to complete the second-language version of the Daly-Miller Writing Apprehension Test (Cheng, Horwitz, & Schallert, 1999). The study found that the extensive writing experience had a positive impact on the JFL learners’ confidence and willingness when writing in L2 Japanese.<br><br>