With the ascendance of new information technologies, the significance of writing has, it seems, slipped from view, in spite of the fact that the conceptual and cognitive implications of the newer technologies is a matter of enthusiastic speculation rather that serious research. On the other hand, it is now reasonably well established that the invention of the first “information” technology, namely writing, has had a profound effect on the ways in which we think about language, the mind, and the world, effects which have taken millenia to unfold. “Effects” is perhaps too strong a term as it is less a matter of how technology affects people than a matter of the ways in which people in different cultures have used and applied the technology and the ways they have altered the technology to suit their purposes. In the West, some of these uses have involved institutional change; thus, to make use of a technology such as writing requires the development of monasteries, schools, and other institutions. Indeed, some of the cognitive effects we usually attribute to schooling are better thought of as consequences or implications of literacy.

Tracing the sociocultural influence of any technology is fraught with problems. First, many of the influences cited are likely to be too large and diffuse to be tested under experimental conditions in the laboratory. Second, the technology is likely to be, at most, an accessory to many other influencing factors rather than a singular cause. Third, insofar as the technology can be isolated as a factor of influence, the direction of the influence is often two way. The technology may cause changes in sociocultural states, but existing sociocultural states are also likely to result in the technology being used and evolved in unanticipated ways.

As the opening quotation suggests, a shift in the practice of literacy such as is suggested by the term, literacy online, has epochal significance: The death of one world (that of print) and its replacement by something else (the online world) signals not just a change in communication or technology but a change in civilization itself. Just what is this new world of literacy online that seems to await us, and how should we approach it—with open arms or poised for battle? Behind such questions lies a host of definitional difficulties. Surely the temptation is here for a kind of simplicity: “Online literacy” means reading and writing with a computer. But reading and writing what and how? In theory, we can read a dense, complex novel like Henry James' Golden Bowl—a work created in and designed for the privacy, solitude, and deep interiority of print culture—sitting in front of a computer terminal. Of central importance, however, is not the theoretical possibility of being able to read anything from a computer screen but whether or not readers and writers fully acclimated to the computer screen as the primary source of literacy would ever think to read, much less produce, such a work in the first place.

The inventions of writing and printing constitute the first two revolutions in information recording and transmission. The current development of electronic management of text, somewhat prefigured by Remington and the typewriter, consitutes a third revolution. This article deals with the current state of play in text production and consumption in the specific context of electronic text production and comsumption.

Work in computational linguistics began very soon after the development of the first computers (Booth, Brandwood and Cleave 1958), yet in the intervening four decades there has been a pervasive feeling that progress in computer understanding of natural language has not been commensurate with progress in other computer applications. Recently, a number of prominent researchers in natural language processing met to assess the state of the discipline and discuss future directions (Bates and Weischedel 1993). The consensus of this meeting was that increased attention to large amounts of lexical and domain knowledge was essential for significant progress, and current research efforts in the field reflect this point of view.

Computers have become much more widely used in translation since the early 1980s but in unexpected ways. From the beginnings of machine translation research in the 1950s until recently, it was expected that computers would be in direct competition with human translators for the same work. Instead, it has turned out that most translation done by computers fills latent needs that do not reduce the amount of work available to professional translators. Computers have also turned out to be useful as productivity tools for human translators who still perform the central translation task. Today, the relationship between computers and human translators is often seen as synergistic rather than competitive.

Over the past ten to fifteen years, the discipline of lexicography has changed almost beyond recognition. This change is due to the technological revolution which has computerized the lexicographers' working environment to a very high degree and which has permitted a veritable quantum leap in the amount and variety of resources that can be brought to bear on the lexicographical process. The most important of these resources are computerized corpora of real, mostly written, but now increasingly also spoken, running text. When the first entirely corpus-based dictionary—COBUILD1—came out in 1987, it was on the basis of a corpus of around 20 million words of connected text. Now all major British dictionary publishers use corpora of at least one hundred million words of text. Harrap/Chambers, Longman, and Oxford University Press have built the 100 million word British National Corpus (BNC), HarperCollins has the 200 million-plus word Cobuild Bank of English (BoE), and Cambridge University Press has compiled the 100 million word Cambridge Language Survey corpus (CLS).

On a basic level, there are two main areas of study within linguistics: language structure and language use. Language practioners as well as theoreticians must be concerned with both areas; that is, they need a full understanding of the structural resources available in a language as well as analyses of what speakers and writers actually do with those resources. Investigations of a representative text corpus—a principled collection of texts stored on computer—provide important insights into both of these domains and open new avenues of inquiry.

CALL for English as a second language1 is an interdisciplinary area of inquiry which has been influenced primarily by educational technology (Reiser, 1987) but also by fields such as computational linguistics2 and recently by applied linguistics as well. These related fields contribute diverse epistemologies which shape CALL research questions and methods. The diversity in CALL research can also be explained in party be the current variety of approaches to CALL development and use. Through the 1970s and early 1980s, pedagogical objectives in CALL were focused primarily, although not exclusively, on improving specified areas of learner's grammatical knowledge through approaches borrowed from educational technology (Hart 1981, Hope, Taylor and Pusack 1984, Wyatt 1984). Today, in contrast, CALL is used for a variety of pedagogical objectives through many different types of software such as microworlds (Coleman 1985, Papert 1980), grammar checkers (Hull, Ball, Fox, Levin and McCutchen 1987), pronunciation feedback systems (Anderson-Hseih 1994, Pennington 1991), intelligent tutoring systems (Chanier, Pengelly, Twidale and Self 1992), concordancer programs (Johns 1986, Tribble and Jones 1990), word processing (Pennington 1993), and software for computer-mediated communication (Kaye 1992). These diverse approaches to CALL are predicated on different beliefs about teaching and learning (Higgins 1995, Kenning and Kenning 1990, Sanders and Kenner 1983, Stevens 1992). Rather than reviewing these “CALL philosophies,” this article will focus on the evolution of research traditions dedicated to the empirical study of CALL use for ESL. Accordingly, the term CALL research is employed to refer to empirical research on the use of CALL.

Research in computer-assisted language learning (CALL) has neither traditionally nor recently had a language or setting-specific emphasis, one that would have established a distinction between second language learning (SLL) as opposed to foreign language learning (FLL) applications in its attempts to answer principled CALL questions. For example the two anthologies of essays, studies, and reports on CALL theory and practice by Smith (1987; 1989), the papers on CALL research compiled by Dunkel (1991b), and the collection of international perspectives edited by Pennington and Stevens (1992) summarize work done in the 1980s and provide directions and guidelines for work to be done in the 1990s with implications for both SLL and FLL. As this continues to be the case in recent CALL research, an FLL emphasis in the chapter of non-English L2 instructional context has been maintained exclusively for the discussion of software; whenever the rationale for the design and implementation of FLL software was based on second language acquisition theory or findings from empirical ESL-oriented CALL investigations, the respective research source was examined and cited.

In recent years there have been many claims made for the importance of corpus linguistics to the fields of language description and language teaching. The strong case suggests that without a corpus (or corpora) there is no meaningful work to be done. The weak case is that there are additional descriptive and pedagogic perspectives facilitated by corpus-based work which improve our knowledge of the language and our ability to use it.

In Physical terms, the Net—a term used generically—is a global system of computer linked by optical cable, telephone connections, microwave, and satellites. It comprises a number of individual networks, from LANs (Local Area Networks in individual institutions) to the largest, the Internet. (See Appendix Glossary for terminology explanations.) In January 1993, the Matrix News estimated that the Net comprised 2,152,000 host computers and 18,150,000 users. Now, in mid-1995, these figures are 3,500,000 host computers and 55 million users world-wide. From its beginnings in the U.S. military establishment, it now reaches into education, research, business and commerce, government, and private use. France is connecting its entire education system to the Internet. And the U.S.A. is leading the way not only in numbers of connections to the Internet, but also in providing access for non-university education fields. The Net is arguably the most dramatic new communications technology of the latter part of the 20th century.

Advances in media technology have made it technically possible to meet pressing worldwide demands for language and literacy training by means of distance instruction. Knowledge gained from a substantial body of experience in distance instruction is available to guide educators in combining technology with effective pedagogical programs. Each medium and each associated technology has unique capabilities and limitations and all media have been applied to distance instruction in varying combinations and with varying degrees of success. Most recently, computers and electronic networking, with their capabilities and their limitations, are making an impact on language instruction (Susser 1993, Warschauer, Turbee and Roberts 1994). Lessons learned from past experience in distance instruction are applicable to new technologies. A common reason for failure has been allowing a fascination with a new media technology to obscure the need for sound pedagogical design; a second reason has been a failure to consider socioeconomic factors (Schramm, Nelson, and Betham 1981).

Technology continues to move forward with smaller and faster hardware, more efficient software, multimedia capabilities, and universal reach through advanced telecommunications. Evidence for the successful use of technology in education is overwhelming (Kulik 1994, Snow and Mandinach 1991) given that there are reasoned use of that technology—uses that reflect an understanding of what is to be taught and tested. With the above in mind, this paper reviews current and developing technology uses that are relevant to language assessment and discusses examples of recent liguistic applications from our own laboratory at Educational Testing Service. Below, we describe the processes of language test development and the functions that they serve from the perspective of a large testing organization. We encourage readers to think of technology broadly, to include not only hardware, software, and telecommunications, but other technologies as well, such as job analysis, Item Response Theory (IRT) scaling, standards-setting, and group assessments. Only in this way can the broad activities of language testing, and the relevance of new technologies, be assessed.