Theory, Assessment, Praxis: On the Body and Soul of CALL

Duke University

Many routes lead into Computer Assisted Language Learning (CALL). In one scenario, a computer literate parent may discover a delightful reward and punishment system with sirens and melodies by which to encourage a youngster to learn whatever the parent could get onto the computer, maybe even a foreign language: thus the foundation myth of the Transparent Language company (Michael Quinlan). In another scenario, a far-sighted language and literature professor (Leland R. Phelps) might have witnessed instructional software in another field, say, economics (Allen Kelley), and recognized its applicability to the language learning enterprise: thus the CALIS foundation myth. Yet others may have begun in the realm of testing and evaluation, first on paper and face-to-face but in due course also by computer adaptive means (Danielle Janczewski, and Pardee Lowe). And a whole other cadre of practitioners established their credentials in Linguistics and Philology before stumbling over a computer and applying their genius to CALL (e.g., Donald Becker, James Noblitt).

Each of these access routes, not to mention the countless possible variations, has in common with the others, explicitly, classroom practice, and implicitly, some methodology or theory of learning andsome means of determining in general where in the process the learner needs to be placed. Making explicit this triad--theory, assessment, and praxis--is the mission of this paper, where "praxis" is employed to mean the array of real and actual practices in the language classroom and laboratory, out in the field, and facing a computer (see fig. 1), to build a bridge between a or any theory of language learning (or theory of knowledge) and CALL (Computer Assisted Language Learning). The milestones en route would be (1) the implied methodology of a theory, (2) identification of the stage a learner may have reached, so that (3) appropriate activities may be invented in accord with the theory, method, and level. On the opposite shore, the computer is waiting to determine which of these activities lend themselves to electronic representation and which not, which can be replicated, which simulated, and which need to be ignored. Such questions as refer to content having been asked, those questions referring to form press forward: what characteristics of computer mediation best serve the content? How can feedback, adaptive branching, communications, and interactivity in their peculiarly computer based realizations best enhance the content suggested by the theory-methodology-level-activities continuum? What about multi-media? What about questions of "Human Factors"? e.g. legibility, font size, contrast, "animated text" (figure 2)?

It does not require the pioneering genius of a century ago to acquire or develop someone else's or one's own theory of (language) learning. In broad strokes, most theories can be encompassed in a field limited by Pavlov's dog, salivating to the sound of a bell, "structuralism," "generative grammar," the "natural method," "parallel distributed processing" (PDP), and "Cognitive Theory" (see fig. 3, adapted from Omaggio Hadley, p. 45).

The oldest and most influential of modern learning theories surely goes back to Pavlov and was codified for twentieth-century audiences by B.F. Skinner. The "behaviorist" mode begins with the empiricist axiom that all psychological data should be limited to that which is observable. On that basis, the thesis maintains that human and animal learning closely resemble one another. To begin with, the mind of the human infant is a tabula rasa, a blank slate. There pre-exists no programming for very much other than suckling, and certainly for nothing as complex and gratuitous as language. Then learning of language, as indeed of anything else, is accomplished exclusively by experience, specifically by sequences of stimulus and response. "Conditioning" is the automatic production of an appropriate or expected response upon any given stimulus. It is the consequence of extensive experience in which the associations between stimulus and response are cultivated and strengthened. Learning consists in the cultivation of these associations. Language is nothing more than an elaborate response mechanism developed through years of conditioning across critical periods in human development (fig. 5).

"Structuralism" looks at languages as closed, internally more or less consistent systems. Any component of such a system may or may not be "referential," that is, have something useful to reveal about the ambient universe. It will, however, have something useful to say about the other components, by similarity and difference from which each component becomes identifiable. For example, the plural of pronouns achieves its identity not from the relationship it may or may not have with multiple objects in the real world but rather from the difference it displays from the singular. "We" may refer to single or multiple entities in the world, depending on the speaker's custom or intention. It is "plural" because it is different from "I," not because of its referent. The implications of this thesis for language learning included masses of contrastive drills, by which the interrelated components, one hoped, would be learned and eventually woven into one magnificent tapestry.

"Generative grammar" (fig. 6) argues that human beings seem to have an innate predisposition toward the creation of language and that the system is remarkably finite, with most human languages looking, at least at a distance, very like one another. Learning takes place by the activation of this predisposition in specific linguistic environments, and the major relationships can be described in a series of abstractions that look very much like rules. A finite number of these "rules" ("deep structures") describes language in general and each actual language in particular.

PDP would argue that human beings learn by another process, by which some connections in the brain become cultivated and enhanced though masses of positive feedback, and others atrophy from lack of use. The connections become, thereupon, the location of knowledge rather than any cell or cells. Rabbits are nowhere to be found in the brain, but soft, fuzzy, twitchy nosed, long eared, pink and white, quadruped rodents are somewhere in the brain. Upon learning or perception, "rabbit" is disassembled and its components stored in the brain amidst many difference connections in many different locations, redundantly, and reassembled from the scattered connections only at the instant of production.

The so-called "Natural" method which is associated nowadays with Krashen and Terrell (see fig. 7) takes on the whole tradition of classroom practice of the previous decades, chiefly the ancestor of us all, the "grammar-translation" method (modern languages as Latin), and the "audio-lingual" method. It distinguishes radically between "learning," which is conscious, deliberate, and reflective, and "acquisition," which is subconscious, spontaneous, and automatic. Most of these insights were gained from first language acquisition by children, and the application to adults has proven controversial. Regardless of that fact, their influence on the modern classroom is, to judge by the newest generation of textbooks, enormous and beneficial. The idea is wholly repudiated that exercises or activities should be organized around grammatical problems. Grammar will be "acquired" naturally in a more or less predictable order, as long as "real" or "authentic" situations provide "comprehensible input." It might be legitimate to provide a poem with a great many modal auxiliaries in it, if the poem is "authentic," but not if it is constructed for the sake of teaching modal auxiliaries. This basic premise of the "natural" method has continued to produce theory and experiment of the most fertile kinds, most recently John Oller, Jr.'s outstanding work on "True Narrative Representations" (1996).

Developing a theory of one's own, or appropriating an existing one, is much assisted by comparison and contrast. "Grammar/translation" or otherwise "rote" adult language learning models, prefer the learning or memorization of the paradigm with a goal of "correctness." "Correctness" becomes as important as any communicative skill which would have the goal of "fluency."

By contrast, in a behaviorist model, rewards and punishments (excitatory and inhibitory stimuli), e.g. smiles and comprehension, frowns and incomprehension, would instruct the child that "die," "das," and "den" are appropriate responses to the stimuli "kennt" but "der" and "dem" are not. In a connectionist model, the learner would be building what is called "an internal representation" of the language, or, here, of a particular language problem. That representation is contained in the connections. These connections, correcting for experience over time, would suggest to the learner, which sequences take place at an acceptable level of probability and which do not. In this schema (fig. 8: MacWhinney 1989), the thick lines indicate that the connections between "kennt" and "die," "das," and "den" are strong enough to indicate allowable sequences. The thin dashed lines between "kennt" and "der" and "dem" would indicate unallowable sequences. Which of the allowable choices is most probably correct would be determined by the next set of connections describing the next state of the sentence. Note that in this model the future (what gender and number the subsequent noun may be) influences the past (which of the allowable morphemes has the highest probability of fitting). Not all but some connectionist and statistical paradigms (e.g., "ergotic" HMMs [Hidden Markov Models]) understand and describe this persistent and perverse defiance of the laws of classical physics on the part of day-to-day language.

Those methodologies which seem to have persisted both into present classroom practice and the research agenda of large new interdisciplinary projects are those which can be classified under "cognitive." They are neither new nor uniform, indeed, they may find themselves at opposite ends of the "learning-teaching-acquisition" spectrum. Whereas the Cognitive "Anti-Method" (fig. 9) took a radical position in favor of adult language learning as childhood language acquisition, the Cognitive-Code Method (fig. 10) made a basic assumption: "that meaningful learning was essential to language acquisition, and that conscious knowledge of grammar was important. This viewpoint contrasted strongly with both audiolingualism and the cognitive anti-method" (Omaggio Hadley, p. 101).

This wide array of possibilities for the development of a theory of learning could be complemented by numerous other approaches. However, the spectrum displayed here is likely to contain between its extremities most of the possible variations. Actually, theories posited as oppositional, when investigated a bit more closely, often turn out to have a great deal more in common than at first meets the eye. The oppositions are largely rhetorical, and when they are made to prove themselves in the classroom, in the laboratory, on computer, or in the field, they will often result in remarkably similar activities. The crucial element of theory in pedagogy is not which theory is "right" and which "wrong" or even which one adopts or develops but rather just having one in place. Although it is easily possible to proceed without a theory, having one to invoke permits the classroom teacher (or "platform" trainer, as they are called in government) to test practice against result in a meaningful way. Furthermore, explaining to adults why a certain practice is taking place or a certain activity is being encouraged very much enhances the cooperative nature of the enterprise and includes the learner in the very act of instruction. This act of inclusion opens the way to any number of experiments, which learners will be willing to perform because they know the reason for it. It is, indeed, possible, to demand and get the most remarkable and unexpected behavior from adult learners, if they have been persuaded that the performance will assist in achieving the learning goal. Some such instrument as a "Discussion Guide" (here somewhat skewed in favor of "natural" or "communicative" methodology: figure 11) should be developed to assist teachers and trainers (and learners as well) to reflect about what is happening when they study a language.

Some methods presume a kind of global learning. The operative metaphor here is that of an anthropologist thrown into an unknown culture with an unknown language. The dataset eventually to be learned will appear to the subject virtually in its entirety at every instance of learning. All data will therefore be contextual and interconnected. The adult learner will, to be sure, first acquire those codes which appear most frequently in day-to-day activity. They will probably have to do first with personal survival (nutrition and shelter), and then, in due course thereafter, with successful social, economic, and biological interaction (status, production, and reproduction). Frequency of usage and a general conformity to the requirements of survival may provide "global" theories with the appearance of a "natural order of acquisition." Such theories hold that apart from frequency and survival such theories hold that a linguistic or language-oriented "natural order of acquisition" simply does not exist. The sequence of acquisition will be unique in each case. The unpredictable and unrepeatable unique combination of this particular subject in so and such particular place at so and such particular time will determine the sequence in which the features of the language will be mastered.

By contrast, those theories which hold that adult language acquisition differs fundamentally from childhood language acquisition incline early to abandon the strategy of global and simultaneous presentation of the entirety of the language system. These theories will have to conjecture some or other "natural order of acquisition" (see fig. 12: Nunan 1992).

There is good reason for this. History shows that most adults thus experienced (acquiring language without sequencing and reflective rule-learning) sacrifice correctness for fluency and will "fossilize" at a socially stigmatized argot ("GI German," Spanglish"), if not caught in time. "Fossilization" means that the speakers will never be able to abandon their argot for a generally acceptable form of the language (known as "terminal two's" to users of the Interagency Language Roundtable [ILR] Proficiency Guidelines).

The anthropologist metaphor and those like them rest on the conviction that adult language acquisition is somehow fundamentally similar to childhood language acquisition. The competing theories of adult language acquisition establish the requirement of some rational and reflective knowledge about language. When that requirement has been established, then the order in which this knowledge needs to be presented and learned becomes critical. In general, it conforms to the language learning of the ancient tradition (in the West, Latin and Greek). Nouns will be studied before verbs; singulars before plurals; present tenses before past and future; regular before irregular. This resemblance to an old tradition is perhaps the greatest public relations problem faced by the "natural order of acquisition." Nonetheless, some order has got to be conjectured and employed as the basis for formal instruction. The consensus of the local community of language professionals will produce generally similar "orders of acquisition" from time to time and permit a fair amount of variation. For these systems, distinct from (and opposed to) global "immersion" experiences, linguistic sequencing (as opposed to sheer "frequency" sequencing) is more or less inevitable. The results will be evident in the fashion prevailing in the commercial textbooks of any given period.

Given the present style of content-based or "communicative" instruction, an other-than formalistic "order of acquisition" suggests itself. Most modern textbooks take advantage of such a thesis, whether they express themselves overtly on the subject or not. One such order (see fig. 13) might begin with "greeting," on the assumption that the very first thing one "acquires" upon exposure to a new linguistic culture is how to say "hello." Thereupon one acquires "thank you." With those two and vigorous gesticulation it is possible to satisfy the great majority of survival needs. To acquire anything to eat or drink past what can be pointed to requires minimal lexical "acquisition," that is, the names of those beverages or comestibles one desires or requires with any regularity. Fairly early in the process one needs to learn how to ask how much things cost, although the actual acquisition of the foreign number system is probably the very last element of the language actually internalized, even after the ability to understand and transmit jokes

Depending upon the traveler's hierarchy of worries, habitation assumes greater or lesser importance; whichever it may be, it is a problem that has to be addressed fairly early in any sojourn in the homeland of another language. The lexical field may become much broader than one might initially imagine as necessary ("Do you have a single room for three nights?"), since the critical information may encompass terms for kind of bed, room facing street or back, breakfast included or not, cable, telephone, toilet down the hall. Asking for and giving directions will typically rise as a problem relatively early in a sojourn and require familiarity with a variety of directional terms: right, left, straight ahead, North, South, words for street, block, proximity and distance, destination. Transportation is adjacent to and coheres with the field of "directions," treating as it does all of human and vehicular locomotion.

Time in all its many interventions in human activity depends in part on passage through space: When does the train leave? How long does it take to get there? Does it run on holidays? As transportation treats space dynamically, environment describes space as a state or condition. The vocabulary field will include weather, lay of the land, natural or urban landscape. On the road, attire follows from weather more immediately than it does from fashion. Personal identification, though often the first thing taught in the language classroom after greeting, probably does not appear on the agenda until well after the first phase of survival skills since these address needs generally expressing themselves in public with anonymous interlocutors. By the time one has to reveal one's name or ask an interlocutor to do likewise, one is beginning a new and somewhat more advanced phase of linguistic and social activity. At this point those activities having more to do with real or virtual, long-term or ad hoc communities take over. Discovery of family ties presupposes exchanges of identity. Expression of opinion likewise has limited application in anonymous contexts. Learning to practice and recognize the local rules of deference can make an enormous difference in the success of a sojourn; as failure to do so almost always results in horrible misunderstandings.

Understanding the content and forms of job-related activity occupies most people's waking days, except when they are on vacation. Communities seize time and carve it away from the working day by way of family, communal, and larger-scale celebrations. Individuals accomplish similar ends by voluntary association with sport, avocation, or entertainment, each with its special vocabularies. In some countries, politics is recreation, but in most it is taken very seriously and has such a local flavor that visitors are almost certain to misinterpret analogues to their previous experience. It is very late in the process of familiarization that a visitor begins to understand the passions and peculiarities of another polity. The same may be said of religion, even of religions which may be nominally the same across borders. By contrast, the distinctive cultures of the world's elites have a fundamentally verbal (linguistic, rhetorical) and historical core in common. Presenting oneself as adorned with this culture requires a very high degree of linguistic competence, considerable experience with the target culture's literary artifacts, and the ability to speak knowledgeably about these and the other monuments of the target culture, combined with access to a relatively small body of intercultural reference. Emphasis may shift from one culture or generation to the other--the one more scientific, the other more poetic or musical--but the common treasury remains remarkably stable and remarkably small. Finally, the last skill a learner can acquire (except for the number system) is understanding and appreciating the humor of the target culture.

The topics enumerated above, except for frank discussions of ideology (politics and religion), inform most recent modern language textbooks, though the order differs, too often randomly, from one to the other. This enumeration is, furthermore, not intended in any way to be prescriptive. Numerous other sequences and categories are possible and even desirable, especially under the rubric "culture" with a lower case "c," government, media, education, for example. "Culture" with a capital "K," that is, history, art, poetry, drama, music, their history, and the propensity to talk or write about all of the above, also deserves fuller attention. Familiarity with or, even better, mastery of High Culture, empowers learners, whereas ignorance of it does the opposite.

Having surveyed an array of possible theories, admitted that language is acquired over time, and that therefore, some sequence of learning has to be proposed, the conscientious teacher cum courseware developer confronts the next step: At which stage in the process is the targeted student situated? How does the instructor coordinate appropriate activities to the level reached by or to be reached by the student? The answer is "assessment," and that word is far more easily written than made real.

There have, over the years, been many attempts at establishing a generally acceptable metric for communicating within the language teaching profession just where learners find themselves in the language acquisition process. The oldest and most successful are those which go back to the U.S. government's Foreign Service levels which eventually came to rest in the InterAgency Language Roundtable (ILR) Language Skill Level Descriptions (1983).

These remain palpable in the ancestry of all subsequent proposals for assessment, but themselves are authoritative only inside the U.S. federal government and among its contractors. The story of why that should be the case is, at best, untidy. Although some of the successors contribute interesting, sometimes new, perspectives, none has ever reached the ILR Skill Level Descriptions for sheer utility of measurement. They are far from perfect. Occasionally one hears well informed repudiation of one whole skill category, viz., "Listening," which has been called "spurious," because it appears to be based on reading as a reception skill and be not at all cognizant of the psychoacoustic research dealing with second-language acquisition. The accusation may be accurate but unfair, since that research had either nor yet taken place or had not yet been "received" by Second Language Acquisition Research and Practice as a discipline. Even so, the ILR Skill Level Descriptions, even for listening, remain the most discriminating of all attempts at establishing a common metric.

The earliest widespread successor to the ILR Skill Level Descriptions can be found in the "Proficiency Guidelines" published by the American Council on the Teaching of Foreign Languages in 1986. Indeed, there was actual personnel overlap in the commissions forming both. There is a clear bias in the ACTFL Guidelines in favor of "self-esteem" and with it, a quiet reluctance to express what the learner cannot do at a certain level. This very seriously impacts the utility of the "Guidelines" for assessment purposes. A part of this is motivated by how depressing it would be to admit that after three or four semesters of college foreign language (with perhaps as little as thirty-nine contact hours per semester), that the best one could hope for might be a "Zero Plus" or "Level One: Survival Proficiency," given the realities of college foreign language instruction. It would be somewhat more consoling to call the level "Novice High" or "Intermediate Mid" than ILR "0+" or ILR "Level 1" (see figure 14).

For some notion of the similarities and differences, consider ILR Speaking Level 1 Elementary Proficiency and ACTFL Speaking Intermediate Low and Mid (see figure 15). The significant features in the ILR Levels are: minimum courtesy requirements, conversation on familiar topics, can satisfy predictable, simple, personal and accommodation needs, able to formulate questions. Comparable features in the ACTFL Guidelines are: ability to combine and recombine learned elements, close basic communication tasks, ask and answer questions, can introduce self, order a meal, ask directions, make purchases, can generally be understood by sympathetic interlocutors. ILR employs "cohesion" as a basic concept and maintains that level ones cannot produce continuous discourse, except with rehearsed materials. The ACTFL Guidelines like the idea of "closure" and maintain that an "Intermediate Speaker" can bring closure in a simple way to basic communicative tasks.

At least three sets of descriptions are circulating in addition to ILR and ACTFL. The source of one is the Council of Europe (Council of Europe, 1996). The second circulates with the diplomas awarded by the EuroCentres. (See fig. 16 for a comparison.) The third is said to originate in Australia and stresses "competencies" rather than proficiency (Bottomley 1994). Since the number of categories under each skill varies from seventeen to thirty-two, it is clear that the "competencies" scheme is not meant for diagnosis nor to provide an easily remembered metric to discuss learner performance or placement. Rather, the scheme seems meant to provide an agenda of exercises proposed to explore a multiplicity of related activities of similar or slightly incremental difficulty with a target of gradually progressive capability to perform such activities in new but similar circumstances.

The organization is radically un-hierarchical. If any visual simile suggests itself it may be general horizontal exploration in many different directions on the same plane. Study of the document suggests a certain leaning toward lists for the practice and testing of reception (listening and reading) "competencies." These seem, especially, to be best practiced by highly structured activities, organized with linear, sequential logic, characterized by the overt presence of choices which contain the desired information and which provide the expectation of recognition.

The concreteness of the suggestions for explicit practice and testing activities for the "Listening Competencies" designations here seem to let that scheme escape the criticism applicable to the other proficiency guidelines when it comes to "Listening." However, after reflection, the sympathetic but thoughtful critic will conclude that even these "Listening Competency" activities derive to some extent from reading and speaking activities and could well profit from consideration of acoustic and psycho-acoustic variables specific to listening as a skill (e.g., accent, interference, cross-talk, background conversation, competition for attention, feedback), and, as well, visual corollaries (e.g. lip reading, gesture, body language).

Just as the Australian "competencies" scheme is actually an agenda of activities, the most recent "standards" documents, American, Standards for Foreign Language Learning: Preparing for the 21st Century (Phillips 1996) and the European "External Context of Use" (Council of Europe 1996), concentrate on content, to be more specific, on culture, at the expense of skill and do not even pretend to provide a metric but rather a set of desirable familiarities on the part of the language learner. They can be very useful for supplying content to the language learning process but chiefly in conjunction with some metric to inform the teacher/trainer where to guide the learner to explore next (see fig. 17).

A way of applying proficiency standards immediately in one's own classroom is to develop instruments for testing oral performance. One such might look like figure 18. There are numerous advantages to using instruments of this kind. Students incline to place great trust in them. Both the percentage grade and the letter grade (or number grade in many European countries) seem to provide a kind of objectivity normally missing in oral examinations. Furthermore the weighting (in figure 18 to prefer communication) can give students advanced notice of what is going to be considered most important and permit specialized preparation. Finally, scale descriptions supply achievable objectives for performance improvement in the next exercise. The scales can be adapted to collaborative or group performance (skits or dialogs) by identifying the "team" instead of or as well as the individual student.

These forms and this manner of making discriminations eventually goes back to the historic "proficiency" phase of professional language training (as distinguished from the present "communicative" fashion). The great advantages to these forms is that they can be adapted in any number of ways to any number of circumstances by teacher and learners alike. This kind of testing and assessing, furthermore, takes advantage of the decades-long experience of the "Oral Proficiency Interview" and places that experience at the disposal of the classroom teacher. The first one or two experiments may end up as "learning experiences." But just a little practice and personal tailoring to a teacher's preferences makes these instruments extremely useful for pedagogy, while gradually preparing the teacher to think in terms of achievable steps of improvement.

The OPI ("Oral Proficiency Interview") and its relatives have become an extremely controversial political issue because money and authority are at stake. Workshops to learn OPI testing can be very expensive for the learner and very profitable for the administering agency. At least one populous state refuses to reimburse teachers for training in the OPI and has established its own standards by way of protest. Institutions may be very reluctant to give up to others the right to evaluate or to issue credentials. The reasons may be both self-protection and (justified) skepticism toward external agencies. The notion that external agencies are somehow, miraculously, freer of self-interest and an own agenda than local assessment mechanisms is, of course, naive.

Once local performance requirements are established, rigorous testing and fair evaluation must be made to follow, locally inspired. No one can assess adherence to local performance requirements and understand the limitations of local capacities as accurately as responsible locals. This is not to say that the insights gained by decades of experience in professional, full-time training should be ignored, they just need to be adapted to local circumstances.

The same applies to the instrument in figure 18. Even while leaving text and letter grading intact, changing the order in which they appear (for example, lowest evaluations on top, highest evaluation on bottom) can produce startlingly different results. Experiment is called for, and plenty of practice. Both teachers and students will eventually grow comfortable with this kind of evaluation, once the descriptors are adjusted to the actual content of the event being evaluated and once the tester has enough experience to be confident in any given evaluation

.

There is more at stake than simply the advantage of being able to conduct some localized version of the "OPI." If the teacher (or trainer or courseware developer) has adopted any theory that calls for an "order of learning," then an assessment tool has got to be created and applied for the teacher to know where to begin! For the sake of the present exploration, consider the possibility of revising the "Oral Examination" instrument specifically for level assessment. The idea would be to provide a provisional assignment at × level so that tasks could be presented at level × + .5 (plus) or 1 (next level).

Of the countless possibilities of combination offered in figure 1, let us combine out of Connectionist, "Natural," and Cognitive theory an eclectic hypothesis. That hypothesis would say: learners acquire language on the basis of what they know already (cognitive "knowledge of the world"), by being exposed to and required to produce correct utterances (connectionist reinforcement), and by being challenged by materials just beyond where they find themselves at the moment ("Natural" method: Input + 1). A "provisional skill level assessor" (see fig. 19) would have two purposes. One would be to make practical the theory of assessment implied in the instrument. If "zero plus" to "level one" or "novice to intermediate low" were defined in the tester's mind (after studying the various descriptions) as "memorized proficiency," of single word or short phrase production and recognition, usually having to do with survival needs; inability to produce or fully grasp sentence length utterances except if memorized; ability only to terminate an exchange with a memorized formula, not to bring it to graceful conclusion. . . 

It might be well to look to the possibility of building on the survival skill knowledge of the learner with utterances that extend attention span to full sentence length. This might be done by providing first a reading passage of some length, perhaps twenty lines when done, dealing with survival needs, where stichomythy (alternating one line dialogue) shapes the conversation:

a: Good Morning! How are you?

b: I'm just fine. How are you doing?

a: I'm doing great. What's for breakfast?

b: Oh, they have a serve-yourself breakfast buffet here.

a: Is it included in the price for the room?

b: I don't think so, but you can charge it to your room.

a: What do they offer in the buffet?

b: Oh, the usual, orange juice and coffee, bread, rolls, muffins, and croissants, scrambled eggs, ham, bacon, and sausages.

a: Stop! You're making me hungry.

b: Are you trying to lose weight?

a: I never stop trying.

b: Why don't we jog around the park after breakfast?

a: That sounds like a good idea. Let's do it.

b: Let's meet in the lobby at 9:30, okay?

a: Okay!

Such a dialogue would meet the specifications of dealing with survival needs (plus a bit of fitness), going past one-word and short phrase utterances, while building on those to sentence length utterances. The longest sentence would not necessarily be the most complex, achieving its length by stacking related vocabulary. And finally, since the dialogue tells a tiny story, with a beginning, middle and an end (closure and not termination) it might meet John Oller's stipulations for context as "True Narrative Representations" (1996). For maximum effect on computer, the dialogue should probably be presented orally as well as in text, with the text pouring out letter by letter, at the pace of the audio, both (ideally) user controllable. A glossary providing meanings (or translations) for "new" words should probably be available in one frame by means of "hot spots," with the explanations pronounced as well as written. Cultural notes should probably be provided explaining the main conventions prevailing in the hotel customs in the country of the target language, likewise conventions for actual foods served and consumed, and customs of courtesy surrounding the meal. These could be presented aurally and in text with accompanying video in the target language, or the first language of the learner, depending on the teacher/trainer's preferences or the choice of the user. Note that both as far as structure and content are concerned, the dialogue tries to remain within the first column of the "Order of Acquisition" sequences suggested in figures 12 1nd 13.

After presentation of the dialogue, learners may be faced with true/false alternatives, multiple choice questions, open form answers checking for content retention, and finally, question and free-form answer alternation recreating the dialogue and challenging the learner to produce full sentence utterances. Ideally each such exercise would carry two designations: one for the level of the learner (here "zero plus" to "level one," novice high to intermediate low), and one for the level of the task ("level one" to "level one plus," intermediate high). Obviously a high degree of clarity needs to be dominating the thought of the courseware designer as to what the levels mean and what kind of activities are appropriate to what level. Conceptualization along the lines suggested here, from theory and methodology by way of assessment to actual activities (in the classroom, in the field, and on computer) would be one way of achieving that kind of clarity (see fig. 1).

All the original instances appearing in this document are, like the "Provisional Level Assessor," notmeant to be prescriptive, only possible examples of an attempt at fulfilling a perceived need. All of them contain the open invitation to be abbreviated, expanded, reordered, revised, reassembled, or rejected. They are meant only to provide concrete instances of the abstractions that come about as attempts are made at conceptualization of the problem: where does Computer Assisted Language Learning fit in the process of language acquisition? The possibilities of computer expression at the end point of this process are countless, and new ones are being discovered with every attempt at innovative courseware. One researcher (Piqué 1997: pp. 104-5) proposes at least nine kinds of user interaction with computer, coming over from traditional print and other hard copy classroom precedents and an additional two specific to the graphical user interface:

1) True/False

2) Multiple choice

3) Matching

4) Chaining

5) Tables

6) (De-)Scrambling

7) Dictation

8) Fill-in the blanks (incl. Cloze)

9) Open response and computer specific:

1) "hypertext"

2) Point-and-click, True/false, and multiple choice should be self-evident or familiar from normal, traditional testing experience, although both kinds of activities, especially multiple/choice, have grown highly sophisticated in the professional testing market. The casual user of these formats would do well to review the literature (e.g. Clark 1972, Bachman 1990, Haladyna 1994). "Matching" is familiar from "matching columns" in print-based tests, but could be expanded to match text, picture, and sound in many combinations. "Chaining" will be familiar to trainers of reading skills as the linking of items by similarity, either grammatical (all descriptive adjectives) or semantic (all words for body parts). "Tables" permit presentation of countless kinds of formatted information: television programs, railway schedules, recipes, help-wanted ads, any kind of indexed information, whether indexed by time, place, number, quality, function, genre, or any other imaginable category. Ordering or (De-) Scrambling is commonly used to encourage readers to acquire the rules of cohesion. Dictation with computers is virtually identical to "live" or recorded dictation. Fill-in-the-blanks including Cloze-style exercises permit the exploitation of context to suggest answers. Open response is perhaps the most difficult for computers to handle, but emulations of various kinds are possible even without resort to "Intelligent Tutoring Systems." The use of "hypertext" as a way of linking information is not entirely without print and manuscript precedent, if one considers the text and commentary links of medieval religious writing in the West. Of course "hypertext" is or can be far more dynamic, offering sound and sight in the links in addition to text. "Point and click" is specific to computers and is very suitable for navigating the technical highways and byways of computer technology. It is much less satisfactory as a general input device, where it inclines to resemble the infrared remote and the concomitant "couch potato" posture. Many programs stressing this form of input have the gall to call themselves "interactive," whereas, in fact, "interpassive" would be a far more appropriate designation.

These conventions apply to Internet based activities as well, but far from exhaust the new territories opened by the ease of use of the World Wide Web. Combinations of E-mail with websites with organizing homepages have only begun to be explored (e.g. Oliver 1996). Solid theoretical foundations are also being constructed by some who recognize how totally revolutionary the Internet has become in respect to instructional technologies, who, indeed, argue that the theoretical promise of Computer Assisted Language Learning has never been technologically realizable until the World Wide Web (Zhao 1996 and Salaberry 1996).

Whether on the Web or locally, the choices as to where to employ which of these conventions should depend on the rest of the bridge: theory-methodology-assessment-activities. When the decisions about approach, content, and level have been made, appropriate activities should suggest themselves automatically. If they do not, then study of the various proposals for social, cultural, contextual content, should be of help, as long as one has evolved a methodological bag of tricks appropriate to one's theory or theories, and has considered the level of the learners and of the instruction. These elements constantly interact and inform one another. The final content is the body of Computer Assisted Language Learning. Theory-methodology-assessment is the soul.

Bibliography

Bottomley, Y. , J. Dalton, and C. Corbel (1994). From Proficiency to Competencies. Sydney: Macquarie University Press.

Clark, John L. D. (1972. Foreign Language Testing: Theory and Practice. Philadelphia: Center for Curriculum Development.

Council of Europe (1996). Education Committee. Common European Framework of reference for language teaching and learning: Draft 1 of a Framework proposal--Language learning for European citizenship. Strasbourg: Council of Europe.

Haladyna, Thomas M. (1994). Developing and Validating Multiple-Choice Test Items. Hillsdale, NJ: Lawrence Erlbaum Publications, 1994

MacWhinney, Brian , Jared Leinbach, Roman Taraban, and Janet McDonald (1989). "Language Learning: Cues or Rules?" Journal of Memory and Language, 28 (1989), 255-277.

Nunan, David (1992). Research Methods and Language Learning. Cambridge: Cambridge University Press, p. 138.

Oliver Walter and Terri Nelson (1996) "An E-mail Whodunit to develop Writing Competence in Intermediate Language Classes Offered at a Distance." "Un Meutre a Cine (Un homocidio en Toluca)," Proceedings of the Computer Assisted Language Instruction Consortium 1996 Annual Symposium "Distance Learning," Frank L. Borchardt, Clair L. Bradin, and Eleanor T. Johnson, eds. Durham, NC: CALICO, 1996, pp. 195-201.

Oller, Jr., John (1996). "Toward a Theory of Technologically Assisted Language Learning/Instruction," CALICO Journal, 13, 4 (Summer), 19-43.

Omaggio Hadley, Alice (1993). Teaching Language in Context. 2nd edition. Boston: Heinle & Heinle, 1993.

Piqué, Ramon (1997). Tipologia d'exercicis per a l'adquisicio d'una segona llengua I directrius per al disseny d'interfícies d'usari per a l'aprenentatge de llengües assistit per ordinador. Diss. Universitat Autònoma de Barcelona. 1997.

Phillips, June K. and Christine Brown, eds. (1996). Standards for Foreign Language Learning: Preparing for the 21st Century. Yonkers, NY: National Standards in Foreign Language Education Project.

Salaberry, Rafael (1996): Rafael Salaberry, "A Theoretical Foundation for the Development of Pedagogical Tasks in Computer Mediated Communication," CALICO Journal, 14,1(1996), 5-34.

Zhao, Yong (1996). "Language Learning on the World Wide Web: Toward a Framework of Network Based CALL," CALICO Journal, 14, 1 (1996), 37-51.

Return to Table of Contents