The Language of Content

NSF Phase One R&D Shows Results

Written by Ben Grimley | January 15, 2016

In January we issued our final report to the National Science Foundation for Phase I SBIR award IIP-1448520. This NSF grant supported development of many of the features you see today with Speak Agent's academic language and vocabulary learning platform, especially those directed at ELLs and bilingual or dual-language Spanish programs. Here’s a quick overview of what we achieved in Phase I: We built and beta-tested a authoring system that enables customization of interactive English and Spanish academic language learning content. This includes a mechanism to clone and fork various types of digital lessons, vocabulary sets, and activities; a publishing engine for game generation and delivery to students' computers and tablets; a set of basic game engines for independent academic vocabulary practice; an assignments tool; and embedded learning analytics data collection. This all runs on a highly scalable cloud infrastructure.

While there are many websites that provide open-license teaching materials, the problem with these sites is that they are simply file exchanges where teachers can swap or buy downloadable files (PDF, PowerPoint, Word, etc.) or else are just links to resources on various websites that are often outdated or hit 404 errors. These sites have little in the way of engaging game-based interactivity or linguistic capabilities. They typically offer self-study and self-quizzing tools, but do not support teacher-moderated learning (guided practice) or facilitate peer collaboration. Nor do they provide supports for ELLs. And, since they are cross-disciplinary and serve Pre-K all the way up to University levels, they lack a developmentally appropriate environment for learners and have little to no age-appropriate reading content or writing/speaking capabilities.

Speak Agent instead took a different approach in its NSF project: All content resides in the cloud and all authorized ESL/STEM coordinators, specialists, and teachers can modify it online at an atomic level without any proprietary desktop software, downloads or plug-ins. They can edit every learning object (games, activities, vocabulary sets, and lessons), as well as underlying linguistic data. These changes cascade in real-time. This means that a teacher who assigns a lesson and later makes changes to one of its constituent elements (e.g., spelling, visuals, audio, vocabulary definitions, etc.) has that change reflected immediately for all students in their learning environments. ESL/STEM specialists and teachers can reuse, revise and remix the work of peers, since all most of the underlying learning objects are open-license. They can reuse all of Speak Agent’s media resources and its lexical database contents (as of this writing over 8,600 English and Spanish words frequently studied by elementary students). Multiple authors may collaborate. Content dissemination is intelligent and automates tasks for time-constrained teachers. This reduces teacher burden.

Based on positive feasibility results, we are continuing to develop the Speak Agent academic language platform, thanks in part to funding from the Maryland Technology Development Corporation (TEDCO), as well as our own investment. The Speak Agent beta platform is now accessible to any U.S. educator who requests an invitation at speakagent.com.