This Research to Practice Full Paper investigates second language learners’ writing output using an online next-word prediction writing tool after exposure to training and metacognitive prompts to improve their critical thinking. Engineering graduates’ writing skills are often deemed lacking by industry standards; this can be even more challenging for English as a foreign language (EFL) learners. This study employs a randomized control trial with university-level participants using an internally developed writing aid with next-word prediction, reverse translation support, and metacognitive prompts. EFL participants were given question prompts in the TOEFL iBT independent writing task style, and the outputs were assessed (machine and human) using several measures for writing quality. All participants were shown short explanatory videos for TOEFL writing advice and metacognition training. The treatment group, exposed to the next-word prediction writing aid and metacognitive prompts, performed better than the control group even though both received the same training and writing opportunities. This study indicates there is value in providing writing support and metacognitive thinking practice to improve writing skills and, ultimately, writing output quality. This study’s implications can be applied not only to EFL learners but also to engineering-related fields using English as a lingua franca.

As Virtual Reality (VR) is becoming more viable for creating immersive learning environments, a promising area of study is in monitoring and assessing learner’s engagement. Several related works in this area have used gaze tracking. Unfortunately, gaze tracking in VR is currently limited by hardware availability. This paper presents an approach for engagement monitoring that does not require the traditional costs of more sophisticated hardware setup. A simple combination of video processing techniques was used to show which contents capture the user’s attention, similar to the information current gaze trackers provide. However, success with this method was only evident with a sample video; refinements are still needed to cater to unpredictable changes in field of view. After fine-tuning the process, this approach may potentially benefit instructional VR content developers in determining which content grabs their learners’ attention effectively.

In recent years, massive open online courses (MOOCs) have been rapidly growing with learners from both English and non-English speaking countries. Nowadays, many MOOC developers provide transcripts in multiple languages to attract a broader range of learners. However, providing transcripts is ineffective for visually impaired learners and tends to divert learners' attention away from the overall content of the video lecture. Thus, aside from offering English transcripts, Japanese lecture videos in the Tokyo TechIntroduction to Electrical and Electronic Engineering MOOC were automatically dubbed into English with a computer-generated voice. To understand the effect of English-automatically dubbed videos on learners, the learners' interactions, video completion rate, and course completions were compared to the original course in Japanese. The results show that English automatically dubbed videos have an overall high learner engagement when compared to Japanese videos with English transcripts. Survey results on learners' satisfaction towards the English dubbed videos also showed higher positive responses. These results show the effectiveness of dubbing Japanese lectures into English using text-to-speech technology which has signification for MOOCs offered in various languages.

Multiple studies have hinted at the positive effects of educational games on learner engagement. However, some games, if they exist, may not be readily usable for some lessons even though the same topics are covered. In this workshop, the participants will be introduced to Twine, an open-source development environment for nonlinear stories that requires little to no coding. Tips and tricks to making engaging stories that target desired learning outcomes derived from existing research will be shared. The participants will have the opportunity to apply these insights through a group activity where they create their stories for a lesson. They will then individually create games for their stories and share their works to the rest of the group. An online community will be created after the workshop for the participants to continue exchanging ideas or creations afterwards.

Massive open online courses (MOOCs) provide a great opportunity to use multiple means of information representation through a mixture of various media such as text, graphics, and video, among others. However, most research on MOOCs focused on learning analytics and not much attention is given to content analysis. We gathered all text corpora and video transcripts of selected MOOCs using a web crawler and looked at word counts, clustered by distribution, and measured readability of the crawled data. Analyzing content distribution allows for a comparison of MOOCs regardless of topics, thus giving us an idea of what most course developers might think is ideal in terms of content distribution. This comparison along with readability analysis can be useful for course pre-run quality assessment and gauging content sufficiency.

This workshop proposes specifically soliciting contributions and presentations from initiatives, programs, and platforms around the world. While many of these may already be presented at the full conference, we are also interested in more casual experience reports, case studies, and background presentations from individuals more closely acquainted with how learning at scale initiatives-including MOOCs, for-credit degree programs, informal learning environments, government initiatives, and so on-have unique needs and opportunities based on their local context. We refer to this as Global Learning @ Scale. For the purposes of this workshop, we take two views of Global Learning @ Scale.