Discourse analysis is widely recognized as one of the most far-reaching, but also one of the least well defined, areas in linguistic. One reason for this is that discourse itself has has often been defined in two different ways: as a unit of language that is larger than the sentence, and as the use of language. Whereas the former definition focuses attention mostly on the linguistic regularities characterizing texts, the latter definition focuses attention mostly on the social and cultural functions underlying ways of speaking. Despite these two different foci, many who analyze discourse combine the two concerns, analyzing how the linguistic regularities found in ways of speaking are constrained not only by the structures and patterns inherent in the language, but also by the social and cultural meanings which frame the production and interpretation of messages.

In its early stages, “text linguistics” was a sufficiently compact and localized enterprise that describing it would not have seemed inordinately difficult. Today, however, text linguistics has attained such prominence that it formed the largest contingent at the last International Congress of Linguists in Berlin in 1987. This rise has been accompanied by a widening of scope, a profusion of models, theories, and terms, and a diversification of the phenomena it is intended to capture or designate. The very term “text linguistics” may seem too narrow, and wider alternatives have been proposed, such as “text studies,” “text science,” “textology,” and above all “discourse analysis.” Similarly, some scholars have proposed that text linguistics be incorporated into a broader domain, such as semiotics/semiology, ethnography/enthnomethodology, communications, cognitive science, and so forth.

The discourse analysis typical of language-oriented ethnographic research is distinct form the that which is more linguistically driven. From a linguistic perspective, ethnographic data may seem impressionistic, and the units of analysis (e.g., clarification, shaming, or visiting ritual) too ill-defined for rigorous or detailed investigation. The perspective taken in this review, however, is that ethnographic research has greatly expanded understanding of what can consitute both valid data and appropriate analytical units. Through focus on categories such as speech event or genre, a large number of previously unexamined discourse phenomena have emerged as appropriate and significant domains of inquiry.

The institutionalized, or conventionalized, properties of human language, somewhat neglected as objects of study in contermporary linguistics, have lately been looked upon with renewed interest. Tannen (1987), for example, synthesizing earlier work by Bakhtin (1981), Becker (1979), and Boliner (1976), has argued for a view of discourse as relatively prepatterned across linguistic domains. Similarly, the notion of “emergent grammer” (i.e., of grammer as “set of …recurrent partials, whose status is constantly being renegotiated in speech”; Hopper 1988:118) has been advanced in opposition to more widely-accepted models of grammatical knowledge. Recent research of this type, as well as a number of older studies, indicates a serious interest among certain linguists in formulating a theoretical basis for the study of conventionalized language.

Complex social activities—such as maintaining a legal system, fostering a literary system, or developing communally validated knowledge—rely on language as the medium through which these activities are accomplished. Law, literature, science, religion, politics, and even economics are socially constructed through discourse. Special language tools and uses have developed in conjuction with the rise of these activities. Thus, we may well say that the construction of legal language is part and parcel of the construction of the legal instructions that order social lives, and that the language acts using legal language were developed in coordination with the elaboration of roles, responsibilities, and relationships of legal actors.

The label Critical Discourse Analysis (CDA) is used by a significant number of scholars with a diverse set of concerns in a number of disciplines. It is well-exemplified by the editorial statement of the journal Discourse and Society, which defines its envisaged domain of enquiry as follows: “the reproduction of sexism and racism through discourse; the legitimation of power; the manufacture of consent; the role of politics, education and the media; the discursive reproduction of dominance relation between groups; the imbalances in international communication and information.” While some practitioners of Critical Discourse Analysis might want to amend this list here or there, the set of concerns sketched here well describes the field of CDA. The only comment I would make, a comment crucial for many practitioners of CDA, is to insist that these phenomena are to be found in the most unremarkable and everyday of texts—and not only in texts which declare their special status in some way. This scope, and the overtly political agenda, serves to set CDA off on the one hand from other kinds of discourse analysis, and from textlinguistics (as well as from pragmatics and sociolinguistics) on the other.

This bibliographic essay will only consider work published after 1986 in the belief that Volume VII of the Annual Review of Applied Linguistics (ARAL VII, Language in Professional Contexts, 1987) adequately covers earlier material. Professions will be understood in a non-elitist sense which includes service and other occupations )(c.f., Coleman 1989) as well as the more prototypical categories of medicine, law, etc. This essay will also cover those areas of academic discourse that can be reasonably viewed as professional; for example, the publishing of research papers (Myers 1990), the giving of research presentations (Dubois 1987), or the providing of instruction (Briggs, et al. 1990).

Language testing is concerned with the measurement of language. Language is the trait, and how we go about measuring it is the method. Trait involves the what, and method the how of testing. This paper will focus on the interaction between the what and the how, where the perspective of the what is the domain of discourse analysis. The paper will also review studies and tests that illustrate the degree to which the domain of discourse analysis has contributed to the theory, research, and development of language testing.

Only relatively recently has discourse analysis begun to have an impact on how English grammer (i.e., the rules of morphology and syntax) is taught to non-native speakers of English. In fact, a majority of teachers of English to speakers of other languages still conceive of grammer, and thus teach grammer, as a sentence-level phenomenon (if and when they teach it). This state-of-affairs reflects a rather counterproductive view of grammer since, as Bolinger (1968; 1977) has long argued, there are relatively few rules of English grammer that are completely context-free.

Several excellent articles and books have been written, based on discourse analyses of spoken language, which can infrom teachers (and teacher educators) about features and conventions that are unique to spoken English (Brown, et al. 1984, Brown and Yule 1983, Bygate 1987, Melrose 1989, Tannen 1989) or about differences between authentic, naturalistic discourse and that fount in textbooks (Cathcart 1989, Scotton and Bernstein 1988). These approaches, clearly, have applications and implications for language instruction. However, it is the intent of this paper to discuss the ways in which discourse analysis itself can be useful for language learners as a tool for speaking and listening instruction. Many of the techniques and activities described in this paper bring students into the role of active participant in their own language learning processes, requiring them to collect and produce their own language “data,” to analyze these materials, and to become conscious of the skills involved with language production.

In the 1984 volume of the Annual review of applied linguistics, Grabe (1985) presented a comprehensive discussion of discourse analysis explaning its history, frameworks, models, taxonomies, and operationalizations. The approach of the present article complements Grabe's as a reivew with a more direct concern for instructional applications of discourse analysis in student reading and writing.

The terms discourse analysis and stylistic analysis mean different thing to different people. Most narrowly defined, discourse analysis has only to do with the structure of spoken discourse. Such a definition separates discourse analysis from literany stylistics and pragmatics—the study of how people understand language in context. At the other end of the spectrum, discourse analysis can be carried out on spoken and written texts, and can include matters like textual coherence and cohesion, and the inferencing of meaning by readers or listeners. In this case, it includes pragmatics and much of stylistics within its bounds. Similarly, stylistics can apply just to literary texts or not, and be restricted to the study of style or, on the other hand, include the study of meaning. For the purposes of this review, relatively wide definitions of both areas have been assumed in order to make what follows reasonably comprehensive. The main restriction assumed is that the works discussed will be relevant to the examination of literature in some way. The section on literature instruction will include matters relevant to both native and non-native learners of English, and will also make reference to the integration of literary and language study.

One of the stated objectives of this series, the Annual Review of Applied Linguistics (ARAL), from its earliest apperance, has been to strive toward a gradual approximation of a functional definition for applied linguistics. Despite the fact that applied linguistics has existed as a field for more than thirty years, despite the fact that there are a few publications available which attempt to define the field (Kaplan 1980, Grabe and Kaplana in press, Bright, et al. in press), and despite the fact that there are in the world a number of universities offering graduate study in applied linguistics and rewarding such study with credentials and degrees carrying the specific desgination applied linguistics, the field remains constrained by the absence of a clear definition of its parameters and by the absence of a clear theoretical framework.