Epsilon Learnng Systems
EPSILON LEARNING SYSTEMS - "Enabling People to Learn"

neurons banner
:: About Us ::
Learning and the Brain
The Learning Process
Focus On Learners
Learning Objects






Learning Objects

Learning objects are increasingly seen as key to a technology-based revolution in education and training — even to an emerging global knowledge economy. An international effort is underway to formulate standards that will enable their exchange, and the topic is popular in trade journals and at conferences. The vision encompasses the reuse and exchange of learning content among multiple educational settings, instructors, courses, and institutions.

Many educators and trainers have heard the hype and wish to get involved with producing, sharing, and using them. But most are confronted with a confusion of technical jargon and mixed messages that makes it difficult to know how or where to start:

  • There are disparate definitions of learning objects. They may or may not include non-digital resources, be based on an explicit learning objective, or have internal structure.
  • A pervasive conceptual confusion exists between learning objects as content for learning, as pointers to learning resources, or as metadata (data about data) about such resources. And is the metadata part of the object? Or does it reside in a separate database?
  • One must sort through much jargon, specifications and standards (recognizing the difference between them), and an alphabet soup of technical TLAs (Three-Letter Acronyms).
  • And unclear technology requirements don't help. Does one need databases? Learning object repositories? Flash animations? Video, audio, and other digital media? XML-structured content?
  • It seems such an overwhelming undertaking requiring a spectrum of in-depth technical expertise. It's certainly too much for individuals to accomplish on their own.

The obvious appeal of learning objects lies in two basic expectations about them:

  • The idea is attractive to educators because of an implicit promise that they somehow facilitate or incorporate the notion of learning. The vision is of easily accessible learning resources shared among instructors — or of quality objects developed by specialists who may license or sell specialty items.
  • The idea of learning objects also attracts the technically inclined as it appears to incorporate the powerful notion of object-oriented programming (OOP) and libraries of preprogrammed code. This embodies the dream of reusability based on the popular image of "lego blocks" that can be gathered and assembled as required.

Unfortunately, neither of these notions is yet supported in practice.

The Standard Definition

A primary source of confusion lies in the very definition of 'learning object.' The Learning Technology Standards Committee (LTSC) charged with formulating the Learning Object Metadata (LOM) standard for the IEEE (Institute of Electrical and Electronics Engineers) defines a learning object as:

any entity, digital or non-digital, that may be used for learning, education or training.

Although this definition fails to exclude anything (since any "entity" may be used for learning), it should be considered in the context of supporting an indexing scheme that simply describes such objects for the purpose of cataloguing and exchange. The LOM standard constitutes 58 data elements in nine primary categories — all of which are optional. The idea is to construct consistent databases of metadata that refer to anything (digital or otherwise) that could be used for learning. Publication of such catalogues would allow for identifying the "learning objects" themselves (which according to the definition may be Web pages, digital files, books, people, geographic locations, etc.). So having as wide as possible a definition allows for the broadest usage of the standard. It is left up to organizations interested in adopting the LOM standard to determine which of the 58 optional elements to use for their particular purposes.

A somewhat narrower definition of learning object has been proposed by David Wiley:

any digital resource that can be reused to support learning.

This definition limits learning objects to the digital domain and focuses on their reusability — thus eliminating geographic locations, people, and physical books from the category. Underlying this definition is a vision of a computer-enabled system for locating and retrieving digital materials to use and reuse in instructional situations.

It's important to recognize that reusability implies more than one-time access, so learning object technology implicitly entails some form of knowledge- or document-management, and therefore some kind of consistent indexing scheme. The LOM standard provides only limited guidance here as its focus is on technical interoperability, with elements like general.description, lifeCycle.version, technical.format, rights.cost, and classification.keyword. While there are eleven "educational" elements in the standard, they are extremely limited and include: interactivityType (active, expositive, or mixed), difficulty (very easy,..., very difficult), typicalLearningTime (e.g., 50 minutes), and semanticDensity (very low,..., very high). Learner profiling information is limited to two elements: context (school, higher education, training, or other) and typicalAgeRange (e.g., 9-12).

Clearly, the current LOM standard doesn't provide for the explicit description of how an object is to be used for learning. This again reflects the intended use of this indexing scheme for basic cataloguing and sharing: there is an implicit assumption that some knowledgeable agent (presumably a teacher or instructor) will provide the instructional framework for the given information or content. Since these objects include minimal metadata to index the learning, and so require human intervention and expertise to supply the context for learning, many consider it misleading to refer to them as "learning" objects.

Putting the Learning into Learning Objects

An alternative school of thought resolves this problem by keying each learning object to a learning objective. A learning objective (or "learning outcome") is an explicit statement of what the learner is expected to demonstrate after the learning has been completed. Clearly stated objectives form the foundation for systematic instructional design: independent of the content, they inherently capture the essence of how the learning is to be assessed as well as guide the design of the instructional strategies used to achieve those ends. Because they identify learning in a precise (and demonstrable) manner, learning objectives form a standard basis for specifying the learning enabled by a learning object. Thus, an instructional designer's definition:

A learning object is a collection of content that supports a learning objective with associated learning activities and assessment.

Note that this kind of learning object has some internal structure: learning objective, content, activities, and assessment are separable components of the object — components variously called (reusable) information objects ("RIOs") or knowledge objects. In fact, according to this way of thinking the same terminology would apply to the objects referred to in the LOM standard.

The distinction between information and knowledge objects follows from some basic concepts of knowledge management:

  • First, data are meaningless "stuff" with no contextual relationship to anything else. In the digital domain, data are the raw "bits and bytes" that constitute the medium for messaging, documents, and software applications. Linguistic examples include a syllable or a word out of context.
  • Information is data with internal relationships or a structure with inherent meaning. For example, an e-mail message has a "header" for metadata (with various data fields like To:, From:, Subject:, etc.) and a "body" for content — and certain computer programs "understand" how to process the former while the sender and receiver (presumably) provide the context for and understand the meaning of the latter. Linguistic examples include a sentence or a paragraph.

    An information object is any chunk of information or content. In particular, any (reusable) digital content one can access directly on the Web to support learning would qualify. Other examples include a memo, any isolated digital file, a graph, a video, a book, an instructional exercise, or a test — the key point being that the content exists in isolation from any context.
  • Knowledge is information that has been semantically contextualized with consistent external relationships. A library is a source of knowledge because the books have been catalogued in a way we mutually understand. Consistent application of a standard classification scheme (like the Dewey Decimal or Library of Congress schema) provides the context of meaning and the framework for accessing the books and other items of information in the collection.

    A knowledge object, therefore, is an information object that has been meaningfully contextualized — generally through the application of standard indexing schema. For example, one might access digital knowledge objects through a database that catalogues information objects according to a standard metadata scheme (such as LOM).

What sets knowledge objects apart from information objects is that the latter exist in isolation while the former are accessed within the framework of a common context of meaning (or semantics). For example, a pile of books is a set of information objects while the same books catalogued in a library become a collection of knowledge objects. Similarly, your typical Web search results in a bunch of information objects (sorting out the relevance of which is left up to you), while a knowledge object would result from a database query that delivered precisely the item of information required without human intervention (e.g., a certain stock price pushed to a particular position in your personal Web portal). So the objects defined by the LOM standard (and by Wiley) are information objects, while a database of such objects indexed according to the LOM scheme would constitute knowledge objects.

From a learning designer's perspective, a learning object is a knowledge object that has been contextualized to the domain of learning — including a learning objective. Underlying this approach is the vision of computer-managed systems that dynamically assemble and serve appropriate objects to meet particular learning situations. Learning objects become stand-alone, modular entities, each incorporating its learning context (semantic relationships) in metadata included within the object itself. Hence, the alternate definition:

A learning object is a knowledge object that includes relevant learning metadata, objectives, content, activities, and assessment.

Of course, this presupposes a robust and clearly defined indexing scheme.

LOIS: A Learning Object Indexing Scheme

The LOM standard provides only minimal indexing of learning. Continuing activities by such groups as the IMS Global Learning Consortium are developing specifications that may later be ratified by such bodies as the International Organization for Standardization (ISO) or the IEEE as international standards. However, most current activities remain focused on technical interoperability and are ignoring the need to contextualize the learning itself.

A more explicit and complete scheme for encoding the learning in learning objects would include the following.

Learner profile information

Well designed instruction is personalized by targeting learner needs, and an object that successfully enables learning in one person may completely miss the mark with another. A learner profiling scheme is used to "tag" the object according to whether it requires a certain reading level, assumes a particular learning environment, employs specific motivational strategies, and so on; in particular, it should include an index of learning styles. Tagging learning objects according to learner needs enables the systematization of maximally effective learning.

Learning index

  • The key to good learning design is a framework of nested and clearly articulated learning outcomes (or demonstrable objectives). Established taxonomies can be used effectively to index cognitive, affective, and psychomotor objectives. Outcomes may also be cross-referenced to general skills and competencies essential for socioeconomic success (communication skills, problem solving, critical thinking, numeracy, etc.). Such an outcomes framework facilitates management of learning programs within an organization as well as providing a common language for inter-institutional articulation and learning object exchange.
  • A taxonomy of learning activities is used to index the kind of instructional activities included in a learning object. The LOM standard includes an educational.learningResourceType element, but its range of values is limited (exercise, simulation, questionnaire, diagram, figure, graph, index, slide, table, narrative text, exam, experiment, problem statement, self assessment, and/or lecture) and requires substantial elaboration to form a comprehensive scheme.
  • Any assessment mechanism associated with the learning object should also be specified. A comprehensive assessment taxonomy would accommodate not only the types of questions used (multiple choice, fill-in-the-blanks, problem, essay, case study, etc.), but also types of instruments (pretests, self-evaluation, assignment, discussion participation, take-home exam, etc.), assessment criteria and rubrics, metacognitive factors, and integration across multiple objectives. Such indexing enables the provision of alternate assessment strategies for the same content in different circumstances.

Subject cataloguing

It is of course important to catalogue content according to the subject under consideration. Various standard classification schema are available (Library of Congress, Dewey Decimal, Sears, etc.), the key being that the choice is explicitly declared and consistently applied. In particular, without an automated thesaurus, keywords may be meaningless unless applied using a consistent subject heading scheme — so that everyone knows, for instance, to use the word 'house' instead of 'abode,' 'domicile,' 'dwelling,' 'home,' or 'residence.'

Administrative metadata

Finally, the LOM standard provides some guidance for indexing basic metadata (such as language, version information, contributors, cost, etc.). A core subset of these elements has been identified and carefully defined in the CanCore Profile. Depending on organizational needs, other administrative elements may be required — for instance, more detailed technical metadata may be used to derive some usage requirements (such as bandwidth, browser plug-ins, etc.)

Structured Learning Objects

While appropriate metadata provides the semantics required to incorporate the context of learning into learning objects, the syntax required to make them object-oriented is provided by structured metalanguages such as SGML or XML — technology that has already created a revolution in publishing and continues to transform the Internet.

A metalanguage is a language used to define or create other languages. The Standard Generalized Markup Language (SGML, ISO 8879:1986) is the mother of all computer-processed "markup" languages, including HTML (HyperText Markup Language) and XML (eXtensible Markup Language), a simplified, "daughter" metalanguage optimized for the Web. Both SGML and XML are essentially object-oriented in nature, being based on hierarchical (treelike) structures of containers (elements or nodes) that possess properties (or attributes) and behaviors (methods, procedures, functions, or processing instructions). (This is formalized for XML in the W3C's Document Object Model, DOM.)

Key to this technology is the specification of a structure and some clear distinctions between that structure, the content of a document, and its presentation or format. This facilitates the creation, management, and publication of content that is reusable and media-independent, with single-source outputs to multiple formats.

Content is input to a pre-defined structure, and then output to various formats: HTML, print, or database.

Separation of structure, content, and presentation
in structured metalanguages

Either SGML or XML may be used to create a Learning Object Markup Language (LOML). This entails analysis and careful definition of the required informational structure. In particular, the learning designer's definition provides a first-level internal structure for a learning object comprising metadata, an objective or goal, informational subject matter, activities or practice exercises, and some form of assessment.

Each of these first-level components in turn has its own inherent structures and substructures. The metadata component would have the structure of a LOIS, and other components may contain elements that might be reused in other contexts. For example, a question used once for an exam may later be reused as the basis for a discussion or as an exercise. All such structures are formalized in a DTD (Document Type Definition) or an XML Schema that defines the LOML.

Once the language is defined, various software tools may be employed to create and manage a repository, to input the content, and to transform the materials for publication in the desired formats: to a Learning Management System, for a Web site, for print brochures, or to mobile devices.

The Challenge of Contextualization

Learning object technology is still in its infancy. While it might at first sight appear straightforward to "aggregate" learning resources developed in various places for various purposes, several challenges remain.

  • Different objects may use a different person (first, second, third) in sentence construction, or a distinctive voice, or a particular style of grammar usage.
  • There may be cross-references to other content that do not necessarily transfer with the object.
  • They may include or refer to information or procedures specific to a particular organization.

Such inconsistencies range from mildly jarring to intensely frustrating for learners. Some of these challenges may be resolved by adopting consistent house styles, by imposing strict rules for the separation of content, or through system architectures that maintain specific information in a separate layer. On the other hand, objects developed for multiple uses are in danger of being so generic that the needs of learners are not being met. As usual, it ends up involving a balancing act.

A Range of Solutions

Depending on your organizational needs, a range of solutions is available for implementing learning objects:

  • The simplest approach is to catalogue existing learning resources manually — even on paper. This is a basic knowledge management solution that would include a pointer or locator for each learning resource. It's a good way to try out indexing of information and testing out particular schema.
  • Once you've settled on your indexing schema, the metadata, together with the pointers to existing objects, may be stored in a simple relational database. The objects may well exist anywhere.
  • While maintaining a separate database of metadata, you can develop basic learning objects themselves as Web pages or as Binary Large OBjects (BLOBs — e.g., PDF documents or media files) in a Learning Management System (LMS). This is a common solution.
  • It might be preferable, however, to develop and manage a variety of content in a separate system — for instance, on a centralized server. You may use your server's filing system to manage version control, etc.
  • The storage and retrieval of objects can also be systematically managed through a repository, or Learning Content Management System (LCMS), with software-based document management, version control, and metadata capabilities.
  • The next level of sophistication is to create structured learning objects using SGML or XML solutions. This will involve formalizing your document and data structures, but it will allow you to manage the content separately from the presentation and so output the same content to multiple formats — print, HTML, different LMSs, etc.
  • While the metadata could still be maintained in a separate (relational) database, it is advantageous to include it directly into the structure of the learning objects themselves. This will allow for more options and control over output formats.
  • Structured learning objects may be managed in a fully object-oriented repository, with staged output to "flat files" that can be served from your LMS of choice.
  • Alternatively, structured learning objects may be developed and managed in an object-oriented repository and dynamically assembled at the time of output to meet precisely the needs of the user or learner.




Copyright © Epsilon Learning Systems