Liberating the Learning Environment – Automating Multimodal Content
By Keith Bain, Saint Mary’s University
Like it or not, lecturing is still an integral part of the university learning experience. Even pure eLearning offerings contain elements of lecture. These are often disguised as articles posted in text (html) format or blogs, which may simply be edited and formatted transcriptions of a given lecture topic. In essence, whether one listens to a lecture or reads one, there remains a transmission of ideas, thoughts, and facts from those possessing these ideas, thoughts, and facts (i.e. the “instructor”) to those who don’t (i.e. the “learner”). One challenge, therefore, is to ensure that content is provided in an accessible format. Whether done live during a lecture or via eLearning platforms, this has proven challenging.
Accessibility is a topic rife with legal, cultural, and social complexities, but for our purposes, accessible content can be thought of as providing flexibility to meet different user needs, preferences, and situations. “Users” in this context refers to an increasingly diverse student body, which includes mature and foreign students, students with obvious disabilities, those with aptly named hidden disabilities, and distance learners.
At the individual level, all students have relative strengths and learning preferences. Auditory learners for example, may find it challenging to understand information presented visually and vice versa. To further complicate matters, instructors are usually unaware of these learning preferences. Sometimes the students themselves don’t have a firm grasp of how they learn best, although increasingly instructors are recommending online self-evaluators such as VARK or the Felder-Soloman diagnostic.
Even in conventional situations such as the lecture hall, not all learners have equal access to spoken information. Deaf or hard of hearing students may not be able to access speech without supports such as sign interpretation, stenography, or note taking. Many students may be taking classes taught in a non-native language and struggling to keep up. Note-taking for many is a particularly difficult challenge.
In the online world of eLearning, similar accessibility challenges abound. Students lacking high speed Internet connections will struggle to access high bandwidth multimedia. Visually impaired students may not be able to fully access Web pages due to accessibility design issues and compatibility with screen reader technology (software such as JAWS or Home Page Reader) that “reads” content to the user through text-to-speech synthesis. Content is typically provided in a “one size fits all” format, although if asked users would choose differing forms that accommodate personal preferences. These and other realities have prompted a movement to adopt universal design principles in the traditional lecture hall and online. As a former teacher, I know that this is sometimes easier said than done. That perfect (or so it seemed) learning activity or lecture is likely biased toward a particular learning modality and targeted to reach a particular learning group, despite best efforts to engage all students.
One approach that enables a more universally designed environment is to present and capture content using a new application of speech recognition (SR) technology. Since 1999, an international consortium of university and industry partners has been advancing a concept called Liberated Learning.
Liberated Learning courses use specially designed SR technology to automatically provide real-time captioning of speech so all learners simultaneously “see” and hear a presentation. The instructor’s speech is transcribed and displayed as text while he or she lectures. The software generated transcript is subsequently available as lecture notes for further access after class through an online portal. This transcribed text is also synchronized with the digital audio, as well as any other media used during class. This approach creates a more “universally designed” learning environment, not only in real time but via the online notes that allow for easy information search and retrieval.
The Liberated Learning approach also has eLearning implications. With the explosion of podcasting in higher education, more online content is available as streaming audio via RSS feeds or as simple MP3 downloads. However, search and retrieval of specific information from multimedia sources without a corresponding transcript is a difficult proposition. Instructors who use the Liberated Learning approach automatically create a rich set of learning objects from their everyday classes, which facilitates the easy creation of a blended learning environment.
Online text can be made more dynamic through SR applications being developed by the Consortium. This can be achieved in two ways: the instructor can create the content using SR software, which digitally records the speech and synchronizes it with the resulting text. Alternatively, the software can provide a text-to-speech computer synthesized output that is synchronized with the text. Either way, the user can choose how to access the content, which through this application of SR is now available in the following combinations: text only (html), audio only (i.e. a podcast), text and audio (an “enhanced” podcast), as well as any combination of media sources used as the content is created (i.e., slides, screen capture programs, video, etc.). The key to this approach is preparing the content in such a way that it is easily customizable by the end userto provide flexibility.
Members of the Liberated Learning Consortium are engaged in a variety of research and development projects focused on SR applications. One of the principle efforts is developing new SR software that works in both lecture and eLearning environments. When it was discovered that commercially available dictation products were poorly suited for the dynamic classroom environment, IBM Research, one of the founding members of the Consortium, developed a new technology called IBM ViaScribe. ViaScribe provided a number of fundamental enhancements that allowed it to be successfully used in the classroom and of course, an eLearning environment.
Other Consortium members study the technology’s impact on various stakeholders, with promising results to date. In one study, over 90% of students indicated they would recommend a Liberated Learning class to their peers. Various project teams have also implemented corporate and public applications in nontraditional learning settings, where access to information in a universally friendly and flexible way is an ongoing challenge.
Although the technology and novel applications present unique challenges, creating a cohesive, goal congruent, multidisciplinary team from 15 member institutions in nine countries across five languages may be one of the most significant accomplishments of the Liberated Learning story. As SR and a clearer understanding of its potential evolve, so must the Consortium model that has been instrumental in engendering the Liberated Learning concept. New discoveries championed through collaboration will pave the way, ensuring that a greater proportion of the learner universe will be reached.
Keith Bain is the International Manager of the Liberated Learning Consortium hosted by Saint Mary’s University in Halifax, Nova Scotia, Canada.