The main purpose of this work is to outline the historical development of CALL from the 1960s to nowadays, considering it in the light of major shifts in FLT and SLA theory, and practice perspectives, and deliberately avoiding any detailed technical description of hardware or software. In particular, I will focus on the most recent resources offered by computers for language teaching: Multimedia and the Internet, trying to discuss their potentialities as means to foster language learning.

Mainly because of time constraints, due to my many commitments at school, the handling of the above issues will not claim to be wide ranging and exhaustive, and not even brief, but clear; it will be simply an attempt to explore a field which today seems to have great potential for FLT, made by a teacher who is well aware of the need for further reading and, above all, more personal experience.

The first use of computers in language learning dates back to the 1960s, with the PLATO (Programmed Logic for Automatic Teaching Operations) project at the University of Illinois and the TICCIT (Time–Shared, Interactive, Computer-controlled Information Television) project at Brigham, followed by a number of others, all sharing the purpose of exploring how the computer could be used for foreign language instruction in higher education. These projects, which Chapelle (2001: 3) relates to "CSLA before the microcomputer", were carried out by researchers who used equipment and software which had been intended for other purposes falling within the broader boundaries of Computer Assisted Instruction (CAI).

These first experiences of CALL were supported by mainframe computers connected to terminals on a single campus or by telephone lines to terminals off-campus. The computer- based learning activities were developed through programming languages and were stored on a mainframe for students to access.

In the 1970s the general-purpose programming languages were used to produce CALL "courseware" (computer-based learning activities) which was not technologically sophisticated by today’s standards, but was interactive enough to satisfy the instructional design for this era of CALL. Despite obstacles such as cost, many teachers throughout the world understood the potentialities of computer-assisted language learning and started working on it. In the UK Last and Graham, first individually and then together, explored the construction of authoring software. In North America, the first CALL projects were started as part of larger Computer-assisted Instruction projects. One of the best known was the one initiated at Stanford University in the Institute for Mathematical Studies. Atkinson’s research on learning foreign language vocabulary (Atkinson, 1972) "was based on his mathematical learning theory rather than on then-current foreign language pedagogical practice" (Chapelle, 2001: 4). He found that learning could be optimised by a computer program which selected items for practice on the basis of learners’ past history of performance. In Canada, a CALL project was initiated thanks to the coordinated efforts of three Ontario universities, and resulted in a series of 62 lessons covering basic French Grammar, called CLEF (Computer-assisted Learning Exercises for French).

Many other CALL projects were undertaken in the 1970s. Chapelle (2001: 5) remarks that their pedagogical principles "tended to go beyond the behaviourist/audio-lingual paradigms of early teaching machines by providing learners with grammatical explanations and specific feedback about their responses.

In his paper "Computer-assisted language Learning" (1999), David Eastment defines the role of the computer in these early years of CALL as that of an Instructor and specifies that "computers – then as now – were clearly good at certain kinds of things: they could accept input, and give some sort of feedback; they could store a great deal of information; they could allow learners to work at their own pace; and unlike the teacher, they would never get tired or bored". Eastment, however, clearly states that one of the reasons why the large-scale projects of the 60s and the 70s were not the successes that had been hoped for, was that they were based on a behaviouristic principle, according to which learning a language consisted above all in practising form, rather than content.

In their paper "Theory and practice of network-based language teaching" (2000), Kern and Warschauer express the same view about the strict links between the structural-behaviouristic approach to FLT and the earliest CALL programs, consisting of grammar and vocabulary tutorials, drill and practice programs, and language testing instruments. They describe these first CALL activities as "the computer-as-tutor model" and point out that these programs "were designed to provide immediate positive or negative feedback to learners on the formal accuracy of their responses. This was consistent with the structuralist approach, which emphasized that repeated drilling of the same material was beneficial or even essential to learning. The authors do not consider these sorts of programs very motivating either for teachers or learners, since "they merely perpetuated existing instructional practices, albeit in a repackaged form" (2000: 9).

Chapelle (2001: 7) remarks how, despite its limitations, the courseware developed in the late 70s, was seminal for the future developments of CALL. She cites Robert Hart’s summary of accomplishments and directions for growth reached by PLATO IV: "eight years of intensive development have brought the PLATO IV grammar drill design to a high state of sophistication, so that further work in this direction will bring diminishing marginal returns. If we wish to make [CALL] a more powerful tool for language instruction, we really must begin to investigate qualitatively new design possibilities". (Hart, 1981:16) The following were the new design possibilities suggested by Hart:

(1981:20).

From the early 1980s microcomputers, which did not require to be connected to a mainframe computer, became easily available to language teachers. It was in this period that major development in CALT (Computer-assisted language testing), CASLR (Computer-assisted second language research) and CALL occurred both in Europe and America. The 1983 annual TESOL convention in North America included papers dealing with methodological issues in CALL. The following year, TESOL members established a CALL Interest Section. In 1985, the British Council sponsored a course on CALL at Lancaster University, followed by a gathering in 1986 which was the founding meeting for the EuroCALL professional organization, later funded by the European Commission.

Many CALL developments of this period were strongly influenced by Steven Krashen, who in his 1982 book, distinguished between unconscious "acquisition" and conscious "learning", maintaining that the former was the most effective. Krashen’s views on SLA (Second Language Acquisition) brought many to consider CALL as strictly related to acquisition rather than learning and to label as "learning oriented" the 1960s and 1970s projects. In 1984 Underwood made this point as follows:

In the same year Higgins and Johns expressed their hope "to be able to show…that the computer is quite flexible enough to serve a variety of learning theories" (1984:17).

Underwood (1984) proposed a "communicative CALL" which, through the use of techniques from artificial intelligence to recognize learners’ input and generate responses, could create an environment for acquisition. What was aimed at was "meaningful conversation" between computer and learning. Programs such as games and activities based on collaborative learning were considered as providing good contexts for acquisition.

Chapelle remarks how the strand of SLA research deriving from Krashen’s ideas about acquisition "failed to provide guidance for empirically based evaluation" (2001: 9). Evaluation of CALL activities was based instead on the extent to which it seemed "communicative" and on the type of tasks in which it involved the learners.

Text reconstruction activities, consisting of a variety of cloze exercises (words deleted on a fixed-ratio basis, words deleted on the basis of some criteria, "storyboards" (with all words deleted) complete with help options and scoring, with the end results being the completed text or the responses to comprehension questions, were considered sufficiently learner-controlled and communicative.

Eastment (1999) notes how, among the many programs of the early 1980s which could jumble words, sentences or paragraphs, create cloze tests, play "Hangman" or practise grammar, the many versions of the total cloze Storyboard (still in common use today), were the most successful, thanks mainly to the flexibility of the program. In fact, it can be used for dictation, for translation, for information gap activities, for transformation etc.; furthermore, it is authorable, since it allows teachers to create their own materials without any specialized programming skills. It takes only a few minutes to create a text and the only skill needed by the teacher is being able to type in the text and save it.

In the same paper Eastment identifies the limitation of text reconstruction activities in that their main aim is to get students to provide the right answer, the only one which matches the answer in the computer’s memory. This inability of text-reconstruction programs to allow different end results led many teachers in the late 80s to use the word processor in a different way, namely to allow learners to work on texts in order to improve them. These types of activities, consisting in improving a text which lacks the beginning, the ending, the adjectives, the adverbs, etc., do not require any single correct answer and can foster discussion of possible variations among students working in pairs or groups, resulting in a variety of end results. This more "creative" use of the word processor is certainly beneficial for the development of writing skills but, as Eastment remarks, can also play a role in the development of integrated skills; to this purpose, he suggests a typical model activity involving intensive reading of the original text, listening to additional input from the teacher, group discussion on content and proposed changes to the text as well as writing and editing.

The pedagogical principle of teaching writing as a process, which had already been in fashion for some time before the advent of word processing, found its effective implementation only with the use of the word processor which, through desktop publishing software (DTP) and spell/grammar checkers, allow students to create satisfactory finished-looking products consisting of class newsletters, cookbooks, family and personal histories and so on. An important element of writing-as-process is also the "talk" occurring both before and after the writing itself. Furthermore, writing-as-process involves brainstorming, pre-writing, outlining and organizing, allowing students to find a topic, develop its schema, imagine an audience, take on a point of view.

Kern and Warschauer (2000) parallel the "cognitive" approach to CALL in the 1980s and the early 1990s with the shift from the structuralist-behaviourist to the cognitive/constructivist perspective in FLT, which they attribute in primis to Noam Chomsky’s ideas about language learning as the result of the individual’s innate cognitive structures rather than behavioural reinforcement. As a consequence, fostering the learner’s mental construction of a second language system came to be considered the key strategy for learning to take place, with the obvious corollary that errors are natural by-products of a creative learning process. This new perspective at first led to stress on the need to teach grammar rules, but later resulted in an emphasis on comprehensible input rather than on an explicit focus on grammar (Krashen, 1982). According to Kern and Warschauer, the "comprehensible input" advocated by Krashen did not refer to authentic social interaction, but rather to the necessity of giving individuals an opportunity to "mentally construct the grammar of the language from extensive natural data" (2000: 4).

In the 1980s this new perspective shifted the purpose of CALL activities from teaching the correct structure to the development of learners’ cognitive, problem-solving strategies through heuristic exercises and collaborative tasks organized in staged processes, such as idea generation, drafting and revising. The computer, then, provides tools and resources, but it is the learner who has to do something with them.

Another novel application of computers to FLT in the 1980s were the so-called "concordancing activities". Concordancers ( see graphic n° 2 in Appendix) are programs which can very quickly scan through millions of words of text, find all the examples of a given word, and display it in the contexts where the word occurs; the user can specify how much context is to appear and sort the word in various ways: for example, according to the word which follows or precedes it. This software, strictly derived from corpus linguistics, was proposed for classroom language activities aimed at empowering the learner to investigate questions of vocabulary use and grammatical collocation in the same way that teachers and lexicographers could do. Eastment (1999) observes that this kind of data-driven learning has never really taken off because "it is simply too academic".

Although the combination of graphics, video, text and audio in a single "box" had been available since the eighties thanks to the laser disc technology, it was only in the early nineties that multimedia software became increasingly available thanks to CD-ROM technology, which was much cheaper than previous technical resources.

Eastment, who defines the early 1990s CALL as the era of "computer as integrator"(1999), estimates that in 1997 there were over 300 pieces of ELT multimedia software on CD-ROM, apart from more than 10,000 non-ELT products, including reference works, games and other educational products.

The main appeal of multimedia for language education is certainly the fact that they allow information to be conveyed through spoken and written language, as well as through

images, video-pictures, graphics or photographs. Furthermore, it allows learners to control and interact with the different media. Paul Brett (2001) remarks how these features have led some to claim that language learning through multimedia is the best way to learn. He underlines the fact that so far there has been little research into the effects of multimedia environments on second language processing and gives a very interesting review of the literature on media and multimedia for language learning (in particular the use of video, computer feedback and tasks), followed by a review of the research into the effects of combinations of media, considered in the light of SLA theory. Finally, he presents his study on the effects of combinations of media in a multimedia environment on the recall of information and on language re-use.

The results of the review of the research into the use of video, computer feedback and comprehension tasks outside of a multimedia environment, demonstrate the positive effects of these individual media for L2 learning and comprehension. The results of the review of the research into different combinations of media show that positive effects have been found for certain combinations of media, especially subtitled video, and that multimedia applications appear to be positively received by learners. Moreover, multimedia software seems to have the potential to replicate some of the same conditions which are thought to promote language acquisition in natural learning contexts. In particular, from the studies reviewed, it seems that multimedia environments can fit in with the negotiated interaction model of SLA. According to this model, comprehensible input is essential for acquisition and it is the process of making non-comprehensible input comprehensible through interactional modifications (requests for clarifications, confirmation checks and comprehension checks) that promotes language acquisition.

The implication of SLA theory for multimedia environments is that they should be designed in such a way as to allow learners to retrieve meaning from L2 input, which can be augmented through online dictionaries, subtitles or comprehension tasks with corrective feedback, in order to help learners to make non-comprehensible L2 input comprehensible. Furthermore, multimedia can allow the learner to control and determine their use of such features, thus personalizing the negotiation of meaning. To sum up, the interactional modifications initiated by the learner on the basis of input from the computer should produce similar psycholinguistic effects to those in the oral face-to-face linguistic exchanges in which they were first investigated (Chapelle, 1997: 27).

Finally, Brett presents his study aiming at investigating the effects of different combinations of media in a multimedia environment on the recall of information and on language re-use. To do this, he exposed four randomly formed groups of Business English academic students to classes which made use of a multimedia environment constituted of four different combinations of media: 1) video, tasks and subtitles; 2) video and tasks; 3) video and subtitles; 4) video only. The different combinations of media were made possible by the software used (see software interface in Appendix).

The results of Brett’s study are quite surprising: the learners with access to the most media recalled the least number of ideas and reused the fewest language items. The author tries to explains this by examining the nature of the combinations of media and concludes that when the same information is conveyed simultaneously through two different processing channels, as in the case of the video + subtitles environment, the learner’s processing load is reduced; conversely, different information conveyed simultaneously through both processing channels (visual and acoustic), as in the case of the video + tasks + subtitles, causes over-stimulation and cognitive overload, which may be an obstacle for learning and memory.

The results of Brett’s study are important for CALL multimedia software design, since they involve the need to take into account not only principles of FLT pedagogy, but also the processing demands created by different combinations of media. In "Technology in the Classroom: Practice and Promise", Hanson-Smith (1997) criticises the lack of pedagogical substance behind most multimedia software. She highlights the potentialities of multimedia as a means to fit different learning preferences or styles and foresees great improvements coming from the next wave of technology: "speaking" word processors, speech recognition devices, CD burners, DVD etc., but she complains that in the present state of the art, the media aspects often drive the content, rather than the way around. She also expresses her belief that the main function of many multimedia products is to entertain rather than to support cognition and memory, and she attributes the current success of multimedia to the fact that it keeps the student busy and makes him/her feel safe: s/he does not have to perform real communicative acts with other people and is very often involved in old-fashioned drilling in which s/he may do very well without any real improvement in his/her communication skills. Hence the need for further pedagogical research into multi-sensory learning and the need for more appropriate pedagogical models for the use of multimedia. One thing is clear: the initial hypothesis that one day computers, through multimedia software, could replace teachers, is certainly not going to be realised.

Kern and Warschauer (2000) relate the era of CALL multimedia to the cognitive/ constructivist view of language learning and identify its main pedagogical feature in the fact that multimedia environments, and in particular "microworlds", shift agency from the computer to the learner. As an example of microworld environment, they cite a multimedia videodisk program, A la rencontre de Philippe di Furstenberg et al. (1993). In this game, intended for intermediate and advanced French learners, full motion video, sound, graphics and text are incorporated, allowing learners to "walk around" and explore simulated environments by following street signs or floor plans. So as to facilitate the understanding of spoken French, the program provides optional tools such as transcriptions of dialogues, a glossary and a video album including samples of many language functions. Learners have the opportunity to create their own customised video albums.

The authors suggest that these kinds of multimedia microwords represent a significant advance over earlier tutorial and drill programs and correspond to the computer-as-pupil construction metaphor posited by Crook (1994). However, they remark that the computer activities based on this metaphor, like the one based on the computer-as-tutor metaphor, distance the teacher from what the students are doing individually and autonomously, and can therefore compromise the collaborative nature of classroom learning. In addition, they maintain that, although multimedia microworlds can provide an effective illusion of communicative interaction, the learner nevertheless acts in a principally consultative mode within a closed system and does not engage in genuine negotiation of meaning.

The era of computers that are capable of evaluating the appropriateness of the learners’ language, diagnosing their difficulties and intelligently choosing among a range of communicative response options remains a distant dream, though some multimedia products are already beginning to display some degree of responsiveness to the learner.

In the early 1990s, a new input to CALL activities came from networked computers, which allowed a shift from a computer-learner interaction model to a learner-learner model. The Local area Network (LAN) technology for computer labs permitted the creation of CALL and CASLA activities based on collaborative work among students. However, it was only thanks to the Internet that instructional connectivity was expanded beyond the confined networks of LANs in computer labs. The Internet allowed students to search for information in archive sites spread around the world and to communicate with language learners from all countries. From the learners’ perspective, interesting opportunities for autonomous language learning and self-assessment became widely available.

Eastment (1999) identifies the turning points of the Internet growth in the creation of the World Wide Web (www) and the development of browsers which could be used by anyone, with minimal training.

The Internet turned CALL from a fringe interest to common practice, since it offered teachers and learners three main features: a delivery mechanism, a means of communication, and the collection of resources. As a delivery mechanism, it allows better implementation of all the achievements of earlier CALL: language manipulation exercises, word games, concordance activities, spread over hundreds of different websites. It also allows the delivery of multimedia material with texts, graphics, audio and, provided you have a fast connection, video. As a collection of resources, the Internet offers a huge amount of material which is of interest to language learners and teachers. On the Web there are a lot of ELT and FLT resources, which come in all shapes and sizes, ranging from very small sites run by individuals, to very large ones, made up of hundreds of pages and run by teams of teachers and by publishers. Many of these sites lack technical quality and educational weight, but some of them offer very interesting resources both for students and teacher education, such as varied language exercises, online dictionaries with recorded pronunciation of words, specialized journals and magazines, book reviews etc. Many authentic non-FLT sites are also exploitable for language teaching: museums, newspapers, schools and universities, club and societies etc. Thus, the multimedia world of the Internet, the "Giant CD-ROM", as someone has defined the World Wide Web, allows the global expansion of the four classroom walls, both for communication and for the research of information.

Many authors have proposed models of language learning activities based on the exploitation of this huge treasure of authentic material. Two of the most recent "practical guides" for classroom activities are worth citing: Windeatt, Hardisty and Eastment (2000) and Dudeney (2000). In the former, the activities are structured according to a pre-computer/computer/post-computer model, while in the latter they are articulated in four phases: preparation, online, offline, follow up. In both books there is an indication of the level of language proficiency of the activities (elementary, lower intermediate, intermediate, etc.).

Internet-based activities usually involve different basic language skills and get the students to cooperate and interact in pairs and in groups. I remember one in which students are given the electronic addresses of some museums which they are asked to "visit" and are asked to "steal" a painting they like by copying and pasting it into a Word document. Then they are asked to type in the document why they like the picture, and finally they are invited to move around, read each other’s documents, ask and answer about their classmates’ art preferences, their reasons for choosing a particular picture etc.

Of course, Internet activities have to be well planned by the teacher, who has to manage both technological and pedagogical aspects: pre-teaching preparation for language points, worksheets, homework assignment, clear instructions, demonstrations and above all, awareness of underlying FLT principles, are essential, as well as the readiness to be flexible and cope with the unexpected.

The third important feature that Eastment (1999) attributes to the Internet, namely the possibility of using it as a means of communication, is perhaps the one which is the subject of most investigation and theoretical speculation by CALL teachers and researchers. There are four possibilities to communicate through the Internet: Email, IRC (International Relay Chat), Web phone, Video-conferencing. Of these, Email and IRC surely offer the best opportunities for communication, while Web telephony and video-conferencing are at the moment technically less reliable and difficult to manage with large classes. Email, in particular, allows international pen pal exchanges and international joint curriculum collaborative projects; students (and teachers) can take part in discussion lists and even in the construction of websites in collaboration with people in other parts of the world. The Internet allows for collaborative tasks on a global scale, and this has led many authors to start research into their potentialities both for collaborative, task-based language learning and intercultural exploration.

Kern and Warschauer (2000) relate the Internet-driven development of CALL to the socio-cognitive perspective of language learning derived from the theories of the American sociolinguist Dell Hymes and the British linguist Michael Halliday. For both language is not a private affair in the head of the individual, but a socially constructed phenomenon. Hymes coined the term "communicative competence" in response to Chomsky’s "linguistic competence" and insisted on the social appropriateness of language use, which goes beyond the rules of grammar. Halliday, by positing the ideational, interpersonal and textual functions of the language, drew attention to the fact that language teaching had, up to then, only dealt with the first of these, namely the use of referential language to express content, neglecting the second (the use of language to maintain social relations), and the third (to create situationally relevant discourse). During the 1980s, the principle of teaching language as communication became the idea of the day and the attention of language teachers was directed towards the teaching of sociolinguistic competence, discourse competence and strategic competence, in addition to linguistic competence. Instruction became more learner-centred and was viewed not just in terms of individuals’ cognitive structures, but also in terms of the social structures of learners’ discourse and activity (Crook, 1994). Cognitive and social dimensions were thought to "overlap in a dialectical, co-constitutive relationship" (Nystrand, Greene and Wiemelt, 1993: 300).

This perspective on learning led to the consideration of language instruction as "helping students enter into the kinds of authentic social discourse situations and discourse communities that they would later encounter outside the classroom" (Kern and Warschauer, 2000: 5). Some believed that this objective could be reached through various types of task-based learning (e.g. Breen, 1987; Candlin, 1987; Prabhu, 1987; Long and Crookes, 1992), while others stressed the role of content-based learning, in which students learn language and content simultaneously (e.g. Snow, 1991; Flowerdew, 1993), In this approach, reading and writing are also regarded as processes belonging to particular sociocultural contexts.

The impact of the sociocognitive approach to language learning on CALL resulted in a shift of dynamic from learner interaction with computer to interaction with other learners via the computer. Computer networking allowed for interactive human communication in addition to access to information and data.

Going back to Email and IRC, it is worth pointing out that they allow individuals all around the world to communicate in synchronous or asynchronous mode; messages can be exchanged and simultaneous (typed) conversation can be effected, in addition to all sorts of written documents or multimedia files, thus facilitating also collaborative reading and writing. All these features can allow language teachers to use technology in order to integrate authentic and creative communication into their syllabuses.

Kern and Warschauer (2000) strongly suggest that today the new communication technologies (NCT) not only support new teaching/learning paradigms, but even contribute to shaping them. The new forms of communication are so widespread that language teachers are almost compelled to expose their students to them in the classroom. Of course, this is particularly true in ELT, since a very great deal of online communication is conducted in that language.

Computer-mediated-communication has been investigated by many authors; some have concentrated on its benefits for the development of reading and writing skills (e.g. Kern, 1995); others have investigated its influence on the development of speaking (Negretti, 1999). Very few studies, however, have investigated the quality of the language produced in CMC. Chun (1994) found that in CMC students use a wide variety of discourse functions; Sotillo (2000) found the same great variety of functions in synchronous CMC interaction; Kern (1995) compared the students’ oral and CMC interaction and found that online interactions had a wider range of morphosyntactic features and discourse functions; Warschauer (1996) also compared oral and CMC interaction and concluded that electronic language was more formal as well as syntactically and lexically more complex; Blake (2000) investigated negotiation of meaning and proved that it does take place in CMC interactions.

Pellettieri (2000), in a very interesting study on synchronous CMC interaction, demonstrated that, in the same way as in oral face-to-face interaction, "chatting" on the Internet (to perform cooperative tasks) can foster negotiation over all aspects of the discourse, which in turn pushes the learner to form focused linguistic modifications.

The latter aspect, in particular, seems to be essential for grammatical development in a second language (Schmidt, 1990; Spada and Lightbown,1993; Gass and Varonis, 1994; Long, 1996). Grammatical development here is intended as the knowledge of the features and rules of the language, including the lexicon, the syntax and the semantics (Canale and Swain, 1980).

Furthermore, Pellettieri’s study indicates that in CMC simultaneous interaction, learners provide and are provided with corrective feedback, which results in the incorporation of target language into subsequent turns. The study also demonstrates the importance of the language task for the quantity and type of negotiation produced: the best tasks for synchronous CMC interaction seem to be goal-oriented, with a minimum of possible outcomes and designed in such a way as to allow all participants to request and obtain information from one another for successful completion.

In his paper " The death of cyberspace and the rebirth of CALL" (2001), Mark Warschauer expresses some ideas about the future of CALL with which I thoroughly agree. He maintains that, differently from what many people think, "information technology is transforming our societies and our lives and even, eventually, our minds, rather than creating alternate worlds". Cyberspace, then, is not a fantasyland to send our students into, but rather the real world around us, in which online communication is on the increase and is becoming a very important medium of 21st century life. Therefore, it is in this perspective that CALL must be reborn in our new century.

Upcoming developments in ICT, such as wireless communication, permanent connections, computing and online devices different from personal computers, broadbands etc., will not determine the way we teach, as it is true that technology in itself does not determine human behaviour in general, but they will create possibilities for new forms of education. New technological developments will affect the entire area of FLT, but in particular, they will have an impact on ELT, contributing to shape new contexts, new literacies, new genres, new identities and new pedagogies.

As regards new contexts, we are witnessing a major expansion in the number of second-language English speakers and this fact is leading to a change in the relationship between native and non-native speakers. It is highly probable that in the quite near future millions of people throughout the world will use English every day to communicate globally and access international media. Thus, if some years ago it was fashionable to say that the computer was just a tool for learning English, now many would say that they study English so as to be able to use computers, get information and communicate on the Internet. This fact has important implications for ELT and underlines the significance of the new literacies required by ICT. Online reading, for example, is different from traditional reading, being associated to such skills as being able to research, evaluate, archive and save information, which is becoming indispensable both for language students and teachers.

Online research and "reading", differently from library research, involve continuous critical decisions, such as whether to scroll down a page or follow an internal link or "quit" the research and start again. Also, writing skills will be more and more affected by the new contexts of communication: student writing of the future will probably be identified with the ability to use both traditional and multimedia resources; the traditional "essay" is no longer going to be the dominant form of student writing; many students of the future will need to carry out some form of collaborative long-distance inquiry and problem-solving as part of their jobs and community activities. Teaching the writing skills necessary for these kinds of tasks will therefore become necessary.

Many of us, as teachers, have students who appear unmotivated and not very proficient in traditional language classes, but who display much greater interest and linguistic skilfulness when they communicate on the Net. This means that learners who use new media develop new literacies and new personal identities.

All these aspects lead me to agree completely with Warschauer’s view of CALL in the 21st century as based on a sociocognitive view of language learning. Our role will be to help students enter new communities through electronic interaction, as well as to acquire new literacies. Of course, communication via the Internet must not be regarded as an end in itself; in order to make it authentic, students must be involved in real-life tasks and real-life problems. From this point of view, international collaborative email or IRC projects can play an important role, since they can allow the pursuit of what Warschauer considers the third great objective of CALL and language learning in general after accuracy and fluency: agency. Agency here means " the satisfying power to take meaningful action and see the results of our decisions and choices (Murray, 1997).

Undoubtedly, CALL can help a great deal to make the students exercise their agency and study languages with the purpose of using them and having a real impact on the world.

This would be the best way to take advantage of the power of modern information and communication technologies.

Someone said that computers will not replace teachers, but teachers who can use computers will replace teachers who cannot. In my view, computers should be used in language teaching in the most principled way as possible and this can be done only by teachers who become aware of the main pedagogical issues connected with CALL. This is what I have tried to do in this paper.

Graphic n° 1: Storyboard. As students type in words, the text starts to build up on the screen, until eventually the entire passage is reconstructed.