Adapting real Web content for adult learners

with real needs


Thomas Cobb

Dept de linguistique et de didactique des langues

Université du Québec à Montréal


Michael Gallagher

Centre aux adultes Gédéon-Ouimet




For The Canadian Modern Language Review / La Revue canadienne des langues vivantes, Journals Division, University of Toronto Press Incorporated, 5201 Dufferin Street, North York, ON, Canada M3H 5T8

January 3, 2005




While computer assisted language learning (CALL) has been shown to succeed mainly with problem learners, few Canadian studies to date have detailed any substantial or sustained use of CALL in this context. Our learners are several cohorts of francophone adult learners in the east end of Montreal who have returned to school after some time in the workforce only to run up against a 'wall of English' - a mandatory English examination standing between them and their high school diplomas. Traditional self-paced language learning materials have not been successful with most of these learners. We have developed a Web-based system for multimedia materials development that provides them instead with a rich, varied, yet focused multimedia exposure to real English at the learners’ own pace and interest level. We report on learner situation and needs, several iterations of both system design and materials development, initial results, and the expansion of our approach beyond its institution of origin.







This report deals with a very specific application of Web based computer technology to the specific problems of specific learners, but it begins with a brief account of where this work fits into the general endeavour to make more and better use of learning technologies in second language acquisition.

We are currently experiencing a number of imbalances in the attempt to make the considerable language resources of the World Wide Web useful to real language learners. For example, (1) The amount of research in Web-based computer assisted language learning (CALL)resources far surpasses the amount of practical use these resources are being put to; (2) researchers’ interests tend in the direction of sophisticated technologies like speech recognition and natural language processing, while the potential of now commonplace technologies like informational websites remains largely unrealized for language learning purposes; (3) those few learners who do make use of Web-based materials and resources, and who tend to populate the available empirical studies, tend to hail from relatively privileged backgrounds and to be using the computer as an optional extra in their learning – while 20 years of research has shown it is the less privileged learners, and learners experiencing various kinds of difficulty in their learning, for whom computation can potentially make a difference; and (4) whether in research or classroom practice, the available Web-based learning materials tend to be employed on an ‘as is’ basis with no attempt, let alone an extensive attempt, to adapt these tools to the language learning abilities and purposes of particular learners.

In summary, what we have in CALL development today is a prospect where relatively abstruse research proceeds more or less independently, on one side of the picture, unconnected to, on the other side, a haphazard and unfocused use of a new, major, but unadapted learning resource, mainly by learners who do not particularly need it. We report on an ongoing attempt to redress all of these imbalances.

Our approach to using the Web for language learning involves bringing together theory, research, development, and at-risk adult learners. The first author is an experienced researcher who has conducted a number of theoretical (Cobb, 1997; 1999; in press) and empirical CALL-related studies (Cobb 1997; Cobb 1999; Cobb, Horst & Greaves, 2001; Gaskell & Cobb, 2004; Horst Cobb & Ionescu, in press), and in addition has engaged in several rounds of ESL internship supervisions in a variety of Centres aux adultes throughout the Montreal area. The second author is an instructor and instructional developer at one such center in a working class area of Montreal, who has more than 17 years experience working with large numbers of at-risk language learners facing high stakes examinations. This first author is also a developer of Internet-based Builder tools, which are basically ‘software for building software,’ and which teachers including the second author may use to adapt authentic Web materials for learning purposes. The rest of this report will deal with each of the matters in turn:

The report will conclude with an outline of future plans and a report on endeavors currently underway to expand this approach throughout the adult TESL milieu in other areas of Montreal. 

Theoretical perspective

Language learning bearings

While most current approaches to language learning give a large importance to language ‘input’ (Lee & Van Patten, 2003), and to the processing of input for meaning, with disadvantaged or at-risk learners input is pretty much all there is. The alternative to input as a source of learning is of course such standbys of the language classroom as rules, language comparisons, or other manifestations of declarative or metalinguistic knowledge, which disadvantaged learners have normally not had much success with in the past. A particularly strong form of input-driven learning is the approach known as ‘data-driven language learning,’ an approach the first author is associated with (as an illustration, running a website devoted to facilitating ‘data-driven language learning on the World Wide Web’). Two key components of input or data driven learning are motivation (finding motivating input) and enhancement (highlighting language patterns in input, or directing attention to different features of input). In sections below we will describe our attempts to build both these characteristics into our multimedia environment.

Multimedia development bearings

If the goal is to impact language learning for the largest number of learners, then the relevant task at present is not software development but software wiring. The quantity and quality of potential language learning opportunities and resources available on the World Wide Web is simply enormous. But it needs to be wired together before learners will get much use from it. There are two problems with the Web’s language learning opportunities and resources, as found. First, they are not focused on language learning, although they can be. Second, many of the opportunities are widely distributed, although it is possible to bring them together. In our system development section we will describe ways of bringing written texts, spoken texts, tasks, and resources together in a varied yet cohesive format.

Participants & setting

The participants in this longitudinal, observational study are successive cohorts of adult ESL learners from the Hochelaga-Maisonneuve  district of the east end of Montreal. This district comprises about 20 per cent of Montreal’s population and is almost exclusively francophone.  The participants are largely young (16 to 26 years of age), low income, poorly educated, and include a high proportion of single parents. There is little community support for the task of acquiring English as a second language, except of course at school where the Minstère de l’Éducation du Québec (MEQ) has made a basic level of English mandatory for high school completion.

On entering the Centre, students are placed in one of five levels from beginner to ‘bilingual.’ Those classified as beginners start working their way through roughly 100 hours per level of a combination of and teacher-provided and self-instructional ESL materials, for a total of 500 hours. The early emphasis in the program is on the skills of listening and speaking, since it cannot be assumed the learners have more than basic literacy in their first languages. The emphasis shifts toward literacy as the courses progress. The largest challenge is to prepare three  simple oral ‘show and tell’ presentations in each semester.

The learners’ motivation is to earn the high school leaving certificate that will offer them access to trade school or community college (CEGEP). And while many have concentration problems and learning disabilities, it is typically endurance rather than ability that is the greatest barrier to success. Many learners have difficulty getting through a whole two-hour class, asking permission to leave the room twice or more in this time, or through a whole term, many leaving their course twice of even more in a term for a brief jobine. The most motivated learners tend to be either retired people returning to school after a lifetime of work in low income employment, or else female single parents, yet it is often the latter who miss the most classes with sick children or other family or financial emergencies. It is not surprising, then, that by the time of the first oral exam in Week 3, class size has often dropped from 40 students to under 20.

The learning opportunities are divided into teacher fronted activities and self-paced, auto-instructional course books that the students work though on their own, requesting help as needed or practice tests when ready by adding their names to a list on the board. The adult learners complain that the materials are boring, childish, and irrelevant to their lives, which is hardly surprising since the content must be conveyed in extremely basic language. Still, the auto-instructional materials are effectively the mainstay of the instruction, since most of the classrooms in these centers cater to two or more levels, and this limits what can be done as a whole class. Sometimes adjacent levels can be combined for a whole class activity when a skill is effectively non-existent at either level, as is often the case with English listening ability. The second author reports conducting regular 20-minute, whole-class sessions listening to cassette recordings, which he describes as never producing a result in 17 years with test scores typically under 50% after any amount of classroom practice.

It is against this scenario of low preparation, low motivation, low success, high drop-out, and stalled lives that the two authors came together over an internship supervision and began to wonder whether computers might be of use to these learners. It should be added that the school had provided a PC lab for the ESL program so the question could be rephrased as how this could best be used to meet some of the challenges described above.

Computing and the adult ESL learners – the case in principle

     Providing adult content to learners with minimal language resources is a challenge, but one that can be met through several types of input enhancement that networked computing makes possible. 

It is a truism that linguistic messages can be conveyed using lesser or greater degrees of accompanying context (actual situations, still pictures, moving pictures, redundancy, repetition, cross mediation of writing and speech, and so on). Language and context lie in a compensatory relationship (Cummins, 1979), such that if language carries less of a given message then context must carry more of it if the message is to arrive at its destination. The modern high bandwidth World Wide Web allows a language course designer to create almost any blend of language and context imaginable. It contains spoken texts supported by written texts, written texts supported by spoken texts, either supported by video clips, or by resources such as learner dictionaries, or by motivation enhancers such as music or movie trailers, on all manner of topics, and in great number. In other words, the WWW offers a means to deliver adult content to linguistically weak learners by dividing this content between language and other types of media.

The problem with shifting the message from language to context is, of course, that this strategy risks reducing the learner’s need to process the linguistic input at all and hence make progress with acquiring the language. (If you can understand the movie from the pictures, why bother reading the sub-titles?) Therefore it is necessary, while offering the learner a way out of the language, to simultaneously redirect them back into it, normally through answering questions of performing other tasks related to the language. A problem here, of course, is that most of the Web’s most interesting texts whether spoken or written do not come with handy questions attached. This is an instructional design problem with an instructional design solution.  

Another huge advantage of choosing Web delivered materials for adult learners is that these can be accessed outside the classroom. It was mentioned above that the most motivated learners are female single parents who are often prevented from attending class by family responsibilities; for these learners, online materials allow them to work from home if they have access, or to catch up quickly with what they missed when they can attend.

Instructional system design

Initial concept

Several iterations of system design have now gone into making the potential of the Web work for these learners, this first of which was a networked ‘Builder’ program devised by the first author (for a screen shot of one of the Builders, see Figure 1a). This program builds resource-linked cloze passages from any text or blend of text and context that a teacher wishes his or her students to work with.  The basic idea is that the learner needs to do more with a text that just read it or listen to it in order to make much progress with language learning. The procedure is this: the teacher/course designer finds a text or a source of texts, either just written or else both written and spoken or of course sung (see several such sources found by the second author and shared in the References below). The text is copy-pasted into a window on the cloze builder Web page (which can be visited at, and if there is a corresponding sound file its Web address or URL is also pasted into another window on the page. The teacher next decides between two types of cloze passages (nth-word or rationale deletion), and between two modes of interaction for learners (paper and pencil for in-class individual or group work, or computer screen with automatic scoring). Finally, the teacher clicks Submit and the exercise is produced: text, sound file, language learning activity, and learner dictionary are all assembled on a new Web page, which the teacher can then Save for use on his or her own media (whether network or printer) for learners to access. Figure 1b shows a computer screen assembly or these items, ready for the learner to use, based on the Demo included on the builder page (based on the text and audio file of a 2002 speech by Jean Chretien accessed from the CBC’s online archives). The learner listens, straight or with pauses, reads, fills in missing words in any order, checks answers at any time, and at any point clicks up key word definitions in a linked bilingual or learners’ dictionary (such as the Cambridge Advanced Learner’s Dictionary shown in the screen print in Figure 1b). The teacher-designer then saves the file for upload to his own nor his institution’s Web server for students to access, and builds a portal page to hide or show selected pages at selected times (see the URL of the second author’s current portal below).

Figures 1a and 1b about here

Initial implementation

To say that not every school, even if it provides computers, necessarily offers a server to upload such media and task files to - would be an understatement. A great deal of time was spent in the present case finding suitable upload arrangements, which in itself could furnish the material for another full-length study, and in the end a server account for these learners was established at the Université du Québec à Montréal. 

Using these Builder tools and a university server, the second author set about building an impressively large library of authentic Web based reading and listening materials for his adult learners, as well as devising ways to make these materials the basis for the required oral presentations.

While learners responded well to these activities, it was nonetheless true that hard use of the system by real users suggested or rather necessitated a number of modifications almost immediately, and these will be discussed in briefly expanded point form below.

Design iterations

Removal of free answers. Some learners using the cloze exercises simply bypassed the reading or listening activity and completed the task using the ‘Check’ button, which told them the correct answer. The scoring routine was re-written to remove this possibility (see Figure 2 for the current state of error feedback).

Figure 2 about here

Addition of video as alternative to sound file. If we are looking for contextual enhancement of linguistic messages, then adding text to speech is one step in this direction but adding moving pictures is several steps. It was determined that the Web now offers several fairly reliable sources of text with corresponding ‘streaming’ video clips, some of them quite extensive, and hence a second cloze builder was developed to incorporate video files (or rather their URLs, since the file is not actually moved from where it normally lives).

Modification of video page structure. The video pages were initially modeled on the listening pages, but in fact there is a big difference between listening to a sound file while completing a text exercise and watching a video file, namely that as soon as you scroll down in the text you can no longer see the video. Thus a ‘frameset’ approach was adopted to the design of the page, such that the video stays put at the top of the page while the text below can be scrolled at will. In a similar way it is good if the Check button can always be accessed regardless of where in the text the learner is working, so this button was also installed in the top frame beside the video picture. Figure 3 shows this frameset arrangement, employed for an exercise  using a 1999 PBS (U.S. Public Broadcasting System) Frontline documentary about university entry SAT (strategic aptitude) testing in the United States). Eventually the full set of Builders were recoded to include the frameset option.

Expanded use of framesets. As mentioned above, it is necessary to both enrich context and ensure engagement with language, not just the enriched context. Filling the blanks in a cloze passage is admittedly a relatively minor engagement with a text, and hence the next phase in development was to incorporate into the frameset a further frame bearing comprehension questions into the single-page activity set. Learners were advised to read and listen, then work on the cloze, then answer the questions, or write their own questions.

From here it was a fairly small step to getting learners to develop their oral presentations based on their rather elaborate engagements with this interesting, grown-up Web-based content. To facilitate this transfer, one of the frames in the frameset was specifically adapted to providing an attractive, printable presentation outline -  once filled in.

How elaborate or the engagements? It eventually became clear that no amount of rich context, adult content, or the latest in learner dictionaries would be enough to keep every learner on task long enough to learn anything. Thus it was decided to develop a submission system beyond simple answer checking. A ‘Submit’ button was added to the web pages linking to a simple Perl script which sends the students’ names plus all the work they have accomplished on the page (gaps, answers, etc) to the top of a simple text file for the instructor to check, print out if desired, and empty at the end of the sequence. (Interested users can receive a copy of this script and instructions for use by email from the first author.)

     With the system more or less stabilized at this point, the second author set about amassing an extensive library of context rich multimedia input focused on language processing, use, and acquisition.


It should be mentioned at the outset of this section that it is almost impossible to undertake controlled research in the setting described above. Group instability, attrition, and low attendance in this milieu have reasons that go beyond the quality of learning materials. Our evidence at this point is observational and even anecdotal.

Positive outcomes

Increased time on task. In the first six months that the system was up and running, the learners demonstrated a strong proclivity to listen to the Web based materials for much longer periods of time than had been usual in the past, namely 40 minutes from a 150 minute class. Where previously there had been complaints about being forced to listen to cassettes played in front of the class, now learners would ask,  “When are we going to go to the computers?” They were clearly enjoying the process of listening far more than in the past. But just listening? 

Increased volume of writing. The large quantity of data submitted to the teacher via the Submit button showed that learners were answering many more comprehension questions than they had before they used this system.  While getting a few answers written down had previously seemed arduous, now students are asking “Did you get my answers to these questions?” and asked for copies of their work to be corrected by the teacher.

Development of computer skills. Since most of these students are learning English not for love of the language but as part of a broader strategy to improve their job prospects, it was a not inconsiderable side-benefit of the Web approach to ESL that many of the learners vastly improve their computer skills (some of the interactions with framesets, pause buttons, submit buttons, dictionary windows, etc., while basic are nonetheless not totally obvious).  Anyway it is a standard ploy of educators to piggyback target instruction to an ulterior motivation.

Collegial enthusiasm. One of the most positive outcomes was the enthusiasm this approach has received throughout the adult ESL teaching community. His topic gets its own section below.

Negative outcomes

Too much context? As mentioned above, there is a risk to offloading linguistic content onto an attractive supporting context (Why read the subtitles?), and vigilance must be maintained to keep some of the learners’ processing energy flowing through text not just context. It quickly became clear that music videos tended to replace rather than support the song lyrics provided, particularly with today’s high level of sound quality and the provision of earphones. Music and movie videos have now been reclassified as Friday afternoon treats!

Too much choice? As is true of the World Wide Web generally for many people, multiplication of choice can tend to restrict rather than expand some learners’ experience. It gradually became clear that fewer choices led to more language being processed, and the use of a portal Web page exposing some not all available options is de rigueur.

Too passive? It was the learners themselves who began to ask, ‘Why don’t we have discussions about some of these topics back in the classroom?’ Clearly the effort involved in getting this Web approach up and running, and the focus on oral presentations and listening tests, may have caused us to forget about some of the ‘traditional’ communicative activities of a language classroom. On the other hand it was a credit to the increased richness of the stimulation the learners were receiving from the Web that they were even interested in having discussions – generations of attempted discussions had fallen flat in previous years.

     Future developments

     In terms of system development, there is clearly more work to get the various balances in our approach just right, and this is under way on an ongoing basis, but we are already convinced that a properly handled Web component is extremely valuable for these at-risk adult learners. In terms of outcome data, we clearly need to set up a more controlled experiment to measure the actual effect this enriched instruction is having on learners where it counts – in getting them over the Wall of English on with their lives. In fact, the present report may well serve to clarify where we have all got to in this extended experiment, and where we are going next. It may also serve to inform school principals and administrators of what we are doing and why we need to have equipment, do controlled research, access test results, and so on.

The response of Montreal’s adult TESL milieu

One very satisfying outcome from this work has been the enthusiastic response from fellow instructors and potential instructor-developers throughout Montreal’s adult TESL milieu. Both authors but particularly the second have presented the work described above at a long series of conferences over 2003-2004, including SITSAT (or ‘sites satellite’ of the provincial Recit or réseau de personnes-ressources pour le développement des compétences des élèves par l'intégration des technologies) at several locations (Comission Scolaire Marguerite-Bourgeoys, Comission Scolaire de la Monteregie), SPEAQ (Societé pour l’encouragement de l’anglais langue seconde au Québec), and others. These demonstrations and workshops have been very well received and have led to the setting up of a number of working groups. There is clearly a felt need in the classroom for the approach outlined above – principles based, content flexible, and extensively adapted to particular learners.













      1. Paper

      Cobb, T. (1997). Is there any measurable learning from hands-on concordancing? System 25 (3), 301-315.

      Cobb, T. (1999). Applying constructivism: A test for the learner-as-scientist. Educational Technology Research & Development 47 (3), 15-33.

      Cobb, T. (1997). Cognitive efficiency: Toward a revised theory of media. Educational Technology Research & Development, 45 (4), 21-35.

      Cobb, T., Greaves, C., & Horst, M. (2001). Can the rate of lexical acquisition from reading be increased? An experiment in reading French with a suite of on-line resources. In P. Raymond & C. Cornaire, Regards sur la didactique des langues secondes. Montréal: Éditions logique.

      Cummins, J. (1979). Cognitive/academic language proficiency, linguistic interdependence, the optimal age question, & some other matters. Working Papers on Bilingualism, 18, 197-205.

      Gaskell, D., & Cobb, T. (2004). Can learners use concordance feedback for writing errors? System, 32(3), 301-319.

      Horst, M., Cobb, T., & Nicolae, I. (In press). Expanding Academic Vocabulary with a Collaborative On-line Database. Language Learning & Technology, Accepted July 2004; to appear Volume 9, Number 2, May 2005 .

      Lee, J., & VanPatten, B. (2003). Making communicative language teaching happen (2nd ed). New York: McGraw Hill.

2. Online

The Compleat lexical tutor for data-driven language learning on the Web. <>.

Michael’s ESL listening site. Consulted in January 2005.























Figure 1a: Initial builder - input










Figure 1b: Initial builder - output











Figure 2: Feedback, not freebies







Figure 3: Video cloze using frameset