The term discourse is applied to language considerations beyond the boundaries of isolated sentences, although a discourse in its simplest form may be manifested as a single utterance in context, such as “Children at play.” Discourse studies emerge from a variety of disciplines, with major contributions from linguistics and psychology. Linguists have been motivated primarily by the desire to explain phenomena that cannot be accounted for at the word and sentence levels, such as reference or given/new information, while psychologists have emphasized strategic processes and the role of cognitive factors, such as memory, in the production and comprehension of discourse. This entry focuses on seminal linguistic research areas that have had a strong influence on the field.

A construct that defines the nature of discourse study, regardless of discipline or perspective, is coherence. A discourse is coherent when it “hangs together,” or makes sense. This notion of coherence pervades the approaches of a variety of discourse analysts. With some, it is used as a technical term in its own right. With others, it forms an underlying attribute of other constructs. Despite differences in terminology or focus, discourse approaches are similar in their analysis of an organization that supersedes any single sentence or utterance. Thus, the main goal of discourse analysis is to differentiate discourse from random sequences of sentences or utterances.

Discourse models generally assume that discourse coherence is realized through the integration of a variety of resources: the information contained in the text, shared knowledge, and the relevant features of the situation in which the text is embedded. Thus, linguistic form and meaning alone are insufficient for discourse comprehension and production. For that reason, discourse paradigms represent a dramatic shift from more traditional linguistic pursuits that focus exclusively on linguistic forms in isolation.

The knowledge structures that contribute to the formation of coherence are thought to be varied, conventional, and differentiated from each other by the kind of information they contain. For instance, story schema (Mandler, 1984) or superstructure (van Dijk and Kintsch, 1983) represent knowledge of the way in which events unfold in a story (narrative), and how a story typically begins and ends. Script knowledge (Schank and Abelson, 1977) specifies the sequence of steps in common everyday routines, such as going to a restaurant or making a sandwich. Knowledge structures also include knowledge of common patterns of conversational exchange, such as question-answer sequences (Schegloff, 1980).

Discourse analyses differ primarily in the degree to which they focus on the relative contributions of text, shared knowledge, and context. One influential model (van Dijk and Kintsch, 1983) is quite detailed in its account of the transformation of semantic content into cognitive information content. It represents the semantics of a text as a set of propositions. These propositions display coherence through inference at a local level, as microstructures, and at a global level, as macrostructures that represent the topic or gist of the text. The micro-and macrostructures constitute the text base, which is integrated with shared knowledge. The product of that integration is a representation of the events depicted in the text, conceptualized as the situation model. Thus, this is a model of discourse as a process involving transformation of information.

Linguists have also contributed greatly to our understanding of the various means by which discourse is coherently organized above the level of the sentence, and how the surface features of a text contribute to these higher levels of organization (Halliday, 1985). The research has centered largely on narrative, in part because of its universality as a genre and the fundamental nature of the temporal organization on which it is based. The extensive research on narrative has historical roots in Propp's (1928/1968) analysis of folk tales (from the field of rhetoric) and Bartlett's (1932) study of narrative remembering (from the field of psychology).

In linguistics, an early and influential framework for analyzing narrative structure was provided by Labov and Waletzky (1967) and extended in Labov's later work (1972). The model is organized around the role of sentential grammar in discourse-level structure. In this framework, the verbal sequence of clauses is matched to the sequence of events that occurred, as the means by which past experience is recapitulated in the narrative. The overall structure of the narrative progresses from orientation to complicating action to resolution. An additional component of the overall structure is evaluation, which is expressed through a wide range of lexico-grammatical devices. Thus, this work established a fundamental approach to analyzing how an event sequence is realized in linguistic form. Contemporary adaptations of this seminal approach to narrative analysis are numerous (Bamberg, 1997).

Work on textual cohesion (Halliday and Hasan, 1976) provides another view of the relationship between discourse organization and its component linguistic units. Halliday and Hasan define cohesion as “the set of possibilities that exist in the language for making text hang together [as a larger unit]” (p. 18). Cohesive ties, which are surface features of text, provide the relevant semantic relation between pieces of the text. These ties can be lexical or grammatical and include devices such as reference, conjunction, and ellipsis. Thus, cohesion is similar to coherence in that it is a relational concept; text is cohesive when it is coherent with respect to itself. The notion of cohesion has been widely applied, especially in the area of reference.

Linguistic accounts of discourse genre are another means of analyzing how discourse can be coherent. Although narrative has been the most extensively studied genre, Longacre (1996) provides a typology of various discourse genres (e.g., procedural, expository, narrative) in terms of both underlying knowledge structure and their linguistic realization. In his framework, discourse is classified by the nature of the relationship between the events and doings in the discourse, the nature of reference to agents in the discourse, and whether the events happened in the past or are not yet realized. As is typical of linguistic approaches, Longacre further specifies the surface linguistic characteristics of each discourse type, such as the types of tense and aspect markings on the verb, the typical forms of personal reference, and the nature of the linkage between sentences for each discourse genre. For instance, he specifies how the setting of narrative usually contains stative verbs, while the complicating action contains action verbs. He further analyzes how the peak or climax of the story can be marked through devices such as shifts in tense or changes in length and syntactic complexity (Longacre, 1981).

An especially detailed account of how thinking is transformed into language is found in cognitive-linguistic work using narratives (Chafe, 1980). Chafe's framework addresses ways in which the flow of thought is matched to units of language during the process of verbalization. Not only does he address lexicosyntactic contributions to discourse formation, he also considers the way in which syntax and intonation interact in discourse production.

Several of the discourse theorists explore the commonality and variability of narrative features across languages and cultures. These pursuits reflect an interest in universality that is pervasive in the field of linguistics in general. Contributions in this area include studies of cross-linguistic differences in expression of the basic features of narrative, such as verbs or the marking of reference (Longacre, 1996); ethnic linguistic devices used in the various narrative components, such as evaluation (Labov, 1972); variations in linguistic expression both within individuals and across individuals from different cultures (Chafe, 1980); and cultural presuppositions reflected in the content of narratives (Polanyi, 1989).

Conversational discourse represents a discourse type very different from the other discourse genres. Research on conversational discourse emphasizes the role of context and social interaction (e.g., Schiffrin, 1994). A basic unit of analysis for this genre is the speech act, a construct derived from early work in the philosophy of language (Austin, 1965; Searle, 1969). Speech acts are utterances defined by their pragmatic functions, such as making statements, asking questions, making promises, and giving orders. The sequence of speech acts can display a coherence that extends beyond any one speech act. On this basis, van Dijk (1981) proposes his notion of a macro-speech act, which consists of sequences of speech acts that function socially as one unit. The use of these speech acts during turn-taking in conversation is also rule-governed (Sacks, Schegloff, and Jefferson, 1974; Schegloff, 1982). Another important pragmatic framework that guides the overall coherence of information exchange is that of a cooperative principle (Grice, 1975). It subsumes the maxims of quantity, quality, relation, and manner, which guide the amount, clarity, and relevance of information required in conversations between interlocutors. Finally, some of the most recent and socially relevant work in conversational discourse extends the notion of context by focusing strongly on context in a variety of social settings. In this research, context is specified broadly to include participants and their relative roles in particular societal settings in a given culture (e.g., Tannen, 1994).

The advantage of discourse analysis lies in its potential to address both linguistic and cognitive factors underlying a range of normal and disordered communication performance.

See also discourse impairments.