Saturday, September 18, 2004

Learning Object Production and Implementation: UT Telecampus

Learning objects produced and implemented for multiple e-learning applications. Jennifer Rees and Michael Anderson of UT (University of Texas) Telecampus discuss an innovative, large-scale project that required the expansion and development of learning objects. Interest and discussion in learning objects has picked up considerable steam in the last year, Jennifer Rees points out, which has resulted in the circulation of multiple definitions of what constitutes a learning object. Jennifer and Michael participated in an online interview with Susan, which was first published in Xplanazine. Susan was thrilled with the thoughtful, detailed, and accessible descriptions of learning objects, their development and implementation, and was delighted to be able to share the interview here.

Most agree that the learning object is a small, digital, accessible, transferable packet of information. From there the definitions vary and may include something as simple as what we call an information object - such as a jpeg or gif - to a learning object as complex as a medical simulation. When the use of learning objects (LO) began in the arena of corporate training it was simpler to define. As it migrated to the arena of higher education's online offerings and was tied to multiple instructional design theories, the complexity of the definition grew.

Michael Anderson, Manager, Course Development and Technology ServicesAnderson leads a team of instructional designers and multi-media programmers. Course developers work with academic and corporate partners to create and oversee innovative online courses. Technology staff administer course hosting, manage technical support services, and design and deliver solutions including a 15-campus student information system.

Jennifer Rees, Manager, Communication ServicesRees leads a team of communication specialists to promote awareness and understanding of the UT TeleCampus and the online programs it supports. Services offered include website design and content management, surveys, market research, graphic design, advertising placement, media relations, publication design and organizational communications.
Overview of the TRACK projectOn February 16, 2004, the University of Texas TeleCampus launched a free (grant funded) web application named TRACK (TAKS Readiness and Core Knowledge). The site is designed to help prepare 11th graders for the TAKS (Texas Assessment of Knowledge and Skills) exit-level test administered throughout Texas beginning spring 2004. Students must pass the 11th grade exit-level TAKS to graduate.

The UT TeleCampus (UTTC) was charged with design and delivery of a dynamic, highly-interactive program that would provide students, teachers, and parents the ability to access informative webpages, sample diagnostic tests, and learning materials in the four subject areas tested under the 11th grade TAKS: algebra and geometry; biology and integrated physics and chemistry; English; and social studies. Ultimately, 230 diagnostic items were selected for use in the site along with more than 340 learning objects, comprising 400 hours of instruction.

The plan's important and noteworthy strategy to take a learning object approach in developing TRACK learning materials represents a departure from the standard approach to Web-based learning materials. Tactically, UTTC needed to distribute the learning object design and development workload to as many authors as possible in order to meet the ambitious schedule. Typically, a large organization might require two years to develop a comparable amount of material. The project was developed in six months and saw more than 50,000 registered high school student users in its first 90 days live on the Web.

--- A Discussion with the UT TeleCampus about Learning Objects -

(Rees)Interest and discussion in learning objects has picked up considerable steam in the last year, resulting in the circulation of multiple definitions of what constitutes a learning object. Most agree that the learning object is a small, digital, accessible, transferable packet of information. From there the definitions vary and may include something as simple as what we call an information object - such as a jpeg or gif - to a learning object as complex as a medical simulation. When the use of learning objects (LO) began in the arena of corporate training it was simpler to define. As it migrated to the arena of higher education's online offerings and was tied to multiple instructional design theories, the complexity of the definition grew.

(Anderson) The definition we've used for learning objects for a while is fairly simple: a digital resource that can be reused and supports learning. But that's just the definition.

More important are the characteristics we've attached: our LOs are:
o encapsulated and discrete: objects are self-contained and do not depend on anything external; they can accept and send data through an interface point, but everything an object can do is represented by its message interface.
o discoverable: objects contain data about themselves so that other objects know how to use and communicate with them. Metadata is the implementation of discoverability.o coherent: objects utilize internal navigation for a consistent UI presentation layer.
o responsive: objects provide a feedback loop to the learner (and perhaps others) for verification of the learning outcome.
o polymorphic: objects have the ability to change at the moment they are called; this ability is represented by accepting messages (such as userIDs) and sending messages (such as assessment results) to provide object individualization.

(Rees)So what this means to our learners, whether they are high school students or graduate students, is the interactive learning content they receive online will support learning and will arrive via a common user interface that doesn't cause users to spend more time on navigation than learning content. What it means to us here is we can efficiently (time and finances) build encapsulated parts that, through good instructional design theory, can be woven together in various (but not unlimited) ways to convey knowledge - a verifiable process from viable student feedback.

(Anderson)With the more than 100 graduate and undergraduate online courses we offer each semester, we typically build rather than encounter LOs. But whether we are creating or viewing LOs, 90 percent of UTTC online course content is built using LOs. Faculty develop self-contained lessons (or modules or topics or sessions--you pick the name) around a learning objective (perhaps several); that objective is taught via content pages (text and images, Flash, audio, JavaScript interactivity, video, Java--you pick the technology) and assessed (self-contained quizzes, external high-stakes tests, uploaded homework documents, mandatory discussion board posts, group design projects presented in a synchronous chat session--you pick the assessment type). More important than where we've used LOs is why we've used them: for the most part, by virtue of the design modularity, we've used LOs to prevent our content from being "locked up" inside an LMS.

(Rees)That's an important point to consider. Since we launched in 1998 we have used three courseware platforms. With the first we had the content locked in each course. When we went to switch vendors it took the better part of the year to convert the courses to the new platform, during which time we were essentially at the mercy of the vendor. When our second LMS was purchased and we needed full-scale course conversion again, the content, which had been packaged in LOs, was converted in one month. Michael's team has built a course in our current LMS detailing our process.

(Anderson)The first conversion was literally one page at a time, copying and pasting the text into Dreamweaver; right-click and save the image; change the image SRC parameter to match the saved file. Once we had pulled all of the assets, we constructed self-contained LOs that were packaged as zipped files and uploaded. It did take a year. For the second conversion, we downloaded the dozen or so package files per course from Prometheus and uploaded those same dozen files to Blackboard. It took a month.

The best explanation of how the objects are built is located in the UTTC Capabilities course, as well as the TRACK project. We acquired hundreds of objects from outside sources for TRACK, but for UTTC courses, the objects are built inside the UT System. The process for identifying and selecting TRACK LOs was iterative and collaborative: teams of teachers and curricula experts surveyed Web and CD resources to identify materials that matched the TAKS objectives; the teams then analyzed, debated, and ultimately scored the materials in online discussion forums; when acceptable LOs were located, we negotiated with the owners for non-exclusive distribution rights. If no objects could be located, or if negotiations failed, LOs had to be produced by the TRACK team.

The priority was to cover every learning objective broadly and add depth as time permitted; the primary goal to launch TRACK with at least one LO for every objective was accomplished. Where LOs had to be built by the TRACK team, this primary goal demanded simultaneous subject development; as a practical result, we set up a Social Science team (90% created by TRACK), an English team (100% created by TRACK), and a Math/Science team. The latter group created relatively few new objects (20%) but integrated assets (information objects) such as video clips, PowerPoint slides, PDF documents, and test item banks into a customized QuickTime player and Flash self-assessment engine.

The TRACK project is an expansive test readiness program. One point of particular interest in TRACK is the use of LOs from multiple sources given a common look and feel. I think it's important to note that because LOs are independent pieces doesn't mean they have to look like it. But, the plan to take a learning object (LO) approach to the development of TRACK learning material represents an innovative departure (NLII, 2003) from the standard "all-encompassing textbook" approach to Web-based learning materials. The plan was predicated on both strategic and tactical goals.

Strategically, we needed to design the learning materials in such a way that they could be easily and constantly (perhaps on a monthly schedule) updated with new materials. Tactically, we needed to distribute the learning object design and development workload to as many authors as possible in order to achieve the ambitious schedule. Typically, a large organization such as the University of Texas Dana Center might require two years to develop an entire 4-subject online curriculum; a commercial developer such as Lightspan might require a $5 million investment. The TRACK project demanded the same breadth and quality of content with 25% of the time and 10% of the budget.

(Rees)Yes, Michael brought in two short-term contract Flash designers and teamed them with his designers. The teams produced an impressive volume of LOs in very short order and with extremely effective results based on pilot study and survey results. It was critically important that we used educators to help gather the initial "seed" content that Michael's team used to create the TRACK LOs.

(Anderson)Fortunately, the extremely granular nature of the TRACK learning content-based as it is on well-defined and specific TAKS (the state mandated test - Texas Assessment of Knowledge and Skills) objectives-lends itself to an LO approach. The acquisition of the learning content was developed by first establishing teams of experienced classroom teachers who surveyed and selected applicable Web resources, including commercial sources of digital content whose assets could be repurposed for Web delivery. The end result from the teacher teams was a series of maps, tables of content of the best the Web has to offer, keyed to the specific TAKS objectives.

(Rees)One of several formidable skills that Michael's group brought to the project was their diligence in creating an elegant, seamless user interface. Students and teachers using TRACK would never know that TRACK's content had been farmed from a multitude of sources because of the interface design.

(Anderson)That was a concern we addressed early in the design process. The potential drawback of the LO approach is cognitive dissonance as students move from a blue Flash animation to a green Java simulation-was addressed in two ways:
(1) a coherent and comprehensive interface menu and navigation was developed aftersurveying the initial collection of objects for the presentation layer of the learning objects;
and (2) the development by the teacher teams of bridge material which would serve to provide a consistent voice and tone and a common structure of:
(a) applicability and contextualization of the object to the TAKS objective;
(b) the object itself; and
(c) a self-assessment, summary, and transition back to the learning objective.

The result is a heterogeneous collection of world-class, multimedia-rich learning materials tied together by a strong narrative thread.

The learning objects are managed as zipped package files. The questions around naming protocols, directory structures, and functions are more applicable to information objects-smaller "chunks" that compose our learning objects. The distinction (for us) is important. For example, for some institutions, a photo is a LO; at UTTC, it is not. We made a decision NOT to catalogue objects at that fine a level (although we sometimes regret that decision when we are hunting for a specific image) because the expense (time) spent describing and then searching exceeded the expense (time) to recreate the information object. Because we manually assemble LOs (as opposed to using a machine to assemble them), we are not as dependent upon the extensive metadata generated in cataloging functions.

(Rees)The metadata tagging Michael has mentioned is a growing topic of concern in the world of LOs. Designed to be efficient and financially viable, a repository of LOs is useless if it's inaccessible by vague or inadequate catalogs. To be a true LO, the object must be discoverable, yet it's equally inefficient if the cataloging itself is so time-consuming that the cost to do so exceeds the value of the content itself, which can happen if you aren't setting some parameters. Michael is intimate with the work of the Dublin Core Metadata Initiative for this subject, but chose a different naming convention based on our specific needs.

Since the UT System wasn't planning to individually distribute the LOs external to the System, but did need to access, edit and redistribute the objects, the naming protocol for TRACK was based on the TAKS objective and the accompanying Texas Essential Knowledge (TEKS) expectations. The compilation of these objectives comprises the questions/topics of the TAKS test. LOs are named thusly for quick, efficient reference: Subject Area/TAKS Objective/TEKS Objective/LO Identifier

While we're on the subject of metadata it would be good time to express concerns that we have here at the UT TeleCampus with regard to discussion of metadata being an 'end all' solution where machines simply compile LOs from repositories and create online courses. We disagree with that model for higher education. Using Wiley's (David A. Wiley, II, The Edumetrics Institute, Utah State) metaphor of metadata being like the content and nutritional value labels on cans, we'd like to take that metaphor a step further. Just because you could see a list of contents on a series of cans on a given grocery store shelf does mean pouring them together would make a gourmet meal. Good instructional design and good designers have to be involved in bringing the content together appropriately.

(Anderson)Exactly. Think of the difference too in the way a teenager might prepare a meal with little knowledge of recipes and then think of Julia Childs. This goes beyond tags and content into having deep understanding of how the pieces fit together.

(Rees)Yes, and again as Wiley points, the references in the last few years to LOs being like LEGO blocks is overly simplistic (and maybe even dangerous) for the same reason. We agree with his suggestion the metaphor move to atoms, as LOs are like atoms in structure and scope. They are small objects, but not the smallest, and they can be assembled to create larger objects; however, not every atom can be bonded with every other atom effectively. Just because a LO exists on a given topic does not make it the appropriate choice to package with other LOs in the goal of supporting a specific learning outcome.

One of things I've been thinking about is referential metadata. By adding qualitative statements to the metadata we can know more than say, this LO is a ___ file size, 32 second html doc based on ___ instructional theory covering the 1812 war. We can know, this LO is a 32 K html doc (etc) covering the 1812 war that was effective in raising scores for females in 10 - 12th grade. There's nothing in the object-oriented model that says it can't pass and receive messages. Of course, this is problematic from a variety of privacy points, but it's interesting to explore.
Finally, LOs are intimately related to SCORM, a LO specification that is becoming a de facto standard. For UTTC, LOs must be SCORM-compliant in the use of variable names.

(Rees)The UT TeleCampus partners with more than 100 faculty in a given full semester. It's both impractical and impossible for Michael's team to create the courses completely in-house. His department and our Student and Faculty Services staff work with faculty from course approval through course delivery providing a myriad of training and support services.

(Anderson)Faculty will be the managers of LOs in constructing a course, but (most) faculty will rely on course development teams to build LOs. The message interface aspect of LOs has our attention--and seems to hold the most promise.

Even the "infinite" variety of LOs promised by massive online global repositories comes down to this: some really cool computer-based interaction that eventually gets boring. However, if you concede that LOs include a message component, we can enable LOs to "talk" with other objects: tests can be posted to grade books; RSS feeds can be pulled into pages and pushed into blogs; student interactions can be tracked and guided; teams can explore and learn and solve complex problems together in an immersive, communication-rich online community. In what seems at times a silent digital wilderness, voices can be heard.

(Rees)To see what LOs look like in the real world, feel free to access TRACK. By going to the TRACK website at you can follow the links to the login page. Create an "other" account (please do not create a student account - we want the diagnostics available exclusively to the students using them in Texas) and explore subject areas.

(Anderson)And bear in mind that not all content is appropriate for LOs. The top five considerations:* Is this content valuable enough to build as a LO?* Is the size of the content appropriate for a LO?* Does the content have clear learning objectives?* Can mastery of the content be assessed via feedback?* Do the concepts build in complexity over the LO?

There are many, many people joining the discussion about LOs. A few leaders that come to mind include: Wayne Hodgins for the term and the analogy. David Wiley for the ontology and OSS. David Merrill for making sure "something happens." Ruth Clark for the cognitive theory. Ellen Wagner for application in performance support. Stephen Downes for insisting on RSS. Andrew Gibson for exploring the pedagogy. Chuck Barritt for the RIO/RLO model. Grady Booch for OO-software. And that's just the tip of the iceberg.

The UT TeleCampus is the centralized support center for online education among the University of Texas System's 15 academic and medical research facilities. To learn more:

Blog Archive