The Impact of Using Automated Writing Feedback in ESL/EFL Classroom Contexts

It is undeniable that attempts to develop automated feedback systems that support and enhance language learning and assessment have increased in the last few years. The growing demand for using technology in the classroom and the promotions provided by automated- written-feedback program developers and designers, drive many educational institutions to acquire and use these tools for educational purposes (Chen & Cheng, 2008). It remains debatable, however, whether students’ use of these tools leads to improvement in their essay quality or writing outcomes. In this paper I investigate the affordances and shortcomings of automated writing evaluation (AWE) on students’ writing in ESL/EFL contexts. My discussion shows that AWE can improve the quality of writing and learning outcomes if it is integrated with and supported by human feedback. I provide recommendations for further research into improving AWE tools to give more effective and constructive feedback.

Artificial Intelligence, Teacher Tasks and Individualized Pedagogy

This paper investigates how educational technologies that use different combinations of artificial and human intelligence are incorporated into classroom instruction, and how they ultimately affect learning. We conducted a field experiment to study two technologies that allow teachers to outsource grading and feedback tasks on writing practices of high school seniors. The first technology is a fully automated evaluation system that provides instantaneous scores and feedback. The second one uses human graders as an additional resource to enhance grading and feedback quality in aspects in which the automated system arguably falls short. Both technologies significantly improved students' essay scores in a large college admission exam, and the addition of human graders did not improve effectiveness in spite of increasing perceived feedback quality. Both technologies also similarly helped teachers engage more frequently on personal discussions on essay quality with their students. Taken together, these results indicate that teachers' task composition shifted toward nonroutine activities and this helped circumvent some of the limitations of artificial intelligence. More generally, our results illustrate how the most recent wave of technological change may relocate labor to analytical and interactive tasks that still remain a challenge to automation.

Exploring General Versus Academic English Proficiency as Predictors of Adolescent EFL Essay Writing

Language learning is context-dependent and requires learners to employ different sets of language skills to fulfill various tasks. Yet standardized English as a foreign language assessments tend to conceptualize English proficiency as a unidimensional construct. In order to distinguish English proficiency as separate context-driven constructs, I adopted a register-based approach to investigate academic English proficiency (i.e., specific set of language skills that support academic literacy) and general English proficiency (i.e., wide range of language skills undifferentiated by context that are measured by traditional assessments) as separate predictors of overall essay quality. In the study, students completed a general English proficiency assessment and an academic language proficiency assessment, and essays were coded for academic writing features at the lexical, syntactic, and discourse levels. Beyond the contribution of academic writing features and general English proficiency, academic English proficiency emerged as a significant contributor to essay quality. Findings suggest that academic English proficiency scores more precisely identified a subset of academic language skills that is relevant to essay writing. The article concludes by discussing implications for strategic writing instruction that articulates the key expectations of academic writing used in and beyond school contexts.

Ethnic and Social Class Discrimination in Education: Experimental Evidence from Germany

Even though social class is at least as predictive of educational achievement as ethnicity in virtually all developed countries, experimental research on discrimination in education has overwhelmingly focused on the latter. We investigate both ethnic discrimination and social class discrimination by elementary school teachers in Germany. We conceptualize discrimination as causal effects of signals and use directed acyclic graphs (DAGs) to disentangle ethnic from social class discrimination. In our experiment, we asked randomly sampled elementary school teachers who teach immigrants to evaluate an essay written by a fourth-grader. Employing a 2x2x3 factorial design, we varied essay quality, child's gender, and ethnic and socioeconomic background using names as stimuli. We do not find evidence for discrimination in grading. However, our findings for teachers’ expectations of children's future performance suggest a discriminatory bias along the lines of both ethnicity and social class. The effect is conditional on essay quality---it only holds true for the better essay. We interpret our findings as evidence for models that highlight situational moderators such as the richness of information and ambiguity---e.g., statistical discrimination---but as evidence against simpler models of ingroup-favoritism or outgroup derogation, e.g., social identity theory or taste discrimination.

Persuasive Quick-Writing about Text: Intervention for Students with Learning Disabilities

Researchers examined the effects of self-regulated strategy development (SRSD) to teach students with learning disabilities (LD) to compose persuasive quick-writing about text. The study included a multiple-baseline design with multiple probes for eight students with LD in grades four and five. Researchers observed a functional relationship by systematically replicating the intervention across all student participants. Following SRSD instruction for paraphrasing text and persuasive quick-writing, students increased their persuasive writing outcomes. Improvements were also noted for essay quality and writing length. Implications for future integrated writing and reading interventions are provided.

INTERACTION IN ARGUMENTATIVE ESSAYS: THE CASE OF ENGAGEMENT

Given the importance of interaction in academic discourse, researchers have investigated the use of stance and engagement markers in written academic texts. However, few of these studies have analyzed engagement features in a genre such as the argumentative essay. Employing Hyland’s (2001) engagement framework, this paper examined the use of five engagement markers in a corpus of argumentative essays written by EFL learners across three levels of essay quality. All features of engagement were manually coded and normalized and a one-way ANOVA was subsequently run for analyzing the frequency counts. The findings showed a disassociation between the presence of engagement markers and the overall quality of EFL learners’ essays. The current study offers insight into the interactional nature of argumentative essays. We conclude with a discussion of some of the implications so that educators may provide EFL learners with more adequate support in order to write in a more persuasive way.