Discourse Analysis in Language Learning
Discourse analysis is the study of language above the sentence: how sentences and utterances combine into texts, conversations and stories, and how their meaning depends on who is speaking, to whom, and in what situation. For most of the history of linguistics the sentence was the ceiling — grammar could tell you whether a sentence was well formed, but had nothing to say about why one sequence of perfectly grammatical sentences reads as a text while another reads as noise. Discourse analysis is the family of approaches that took that question seriously.
For language learning the stakes are concrete. A learner who can build a correct sentence may still be unable to build coherent talk: link one utterance to the next, take a turn at the right moment, open and close a conversation, refuse an invitation without sounding rude, or hold a paragraph together in writing. None of this is grammar in the textbook sense, and all of it is learnable — once it has been described. Describing it is precisely what discourse analysis has spent seventy years doing.
From sentence to discourse: Harris, Halliday and Hasan
The term was coined by the American structuralist Zellig Harris, whose paper “Discourse Analysis” appeared in the journal Language in 1952. Harris' aim was characteristically formal: to extend the distributional methods of structural linguistics beyond the sentence, finding patterns in how elements recur across a connected text, and to correlate language with its social situation. Little of his machinery survives in modern discourse analysis, but the move he made — treating connected text, not the isolated sentence, as the object of analysis — named the field and set its agenda.
The question Harris opened — what makes a text a text and not a random collection of sentences — received its classic answer in M. A. K. Halliday and Ruqaiya Hasan's Cohesion in English (1976). Their answer was cohesion: the visible grammatical and lexical ties that bind sentences together. They catalogued five kinds — reference (pronouns and demonstratives pointing back or forward: she, this, the same one), substitution (my laptop broke, so I bought a new one), ellipsis (omitting what can be recovered: “Coffee?” — “Please.”), conjunction (however, therefore, meanwhile) and lexical cohesion (repetition, synonyms and collocation weaving a text's vocabulary into chains).
Cohesion, however, is not the whole story, and the field soon sharpened a second concept against it: coherence — the underlying unity of sense that makes a text interpretable. The two can come apart in both directions. A passage can be stitched dense with pronouns and connectives yet add up to nothing; and an exchange can lack any visible tie yet be perfectly coherent — “That's the telephone.” / “I'm in the bath.” / “OK.” contains no cohesive link, yet every speaker of English reconstructs the household scene instantly. Cohesion lives in the text; coherence lives in the interpretation. A learner needs both: the ties as concrete, teachable items, and the sense of how a given language and culture expects ideas to be sequenced.
Conversation analysis: the hidden order of talk
The second great tributary of discourse analysis came not from linguistics but from sociology. Harvey Sacks, Emanuel Schegloff and Gail Jefferson began, in the 1960s, doing something then radical: recording ordinary conversations and transcribing them in obsessive detail — pauses, overlaps, in-breaths and all. What they found is that casual talk, which looks chaotic, is one of the most tightly organized systems humans operate.
Their most famous result is the model of turn-taking (“A Simplest Systematics for the Organization of Turn-Taking for Conversation”, 1974). Speakers build turns out of recognizable units, project where a unit will end, and apply an ordered set of rules at each possible completion point — which is why conversation flows with astonishingly few gaps and overlaps even though nobody has agreed in advance who speaks when. The system is universal in outline but locally calibrated: how long a silence is tolerable, whether overlap signals enthusiasm or rudeness, differs across languages — one reason fluent grammar does not protect a learner from constantly interrupting, or from being unable to get a word in.
A second discovery is the adjacency pair: paired utterance types in which the first part makes the second expectable — question and answer, greeting and greeting, invitation and acceptance or refusal. Crucially, the two possible second parts are not equal: acceptance comes fast and plain, while refusal — the dispreferred response — comes delayed, hedged and justified (“well, I'd love to, but…”). This machinery, invisible to native speakers, is a genuine learning target: a refusal delivered without the softening apparatus is grammatically flawless and socially jarring.
Even beginnings and endings turn out to be engineered. Schegloff and Sacks showed in “Opening Up Closings” (1973) that conversations are not simply stopped but jointly closed, through pre-closing moves (“well…”, “anyway…”) that give each party a last chance to raise a topic before the goodbyes. And when talk breaks down, speakers deploy an orderly system of repair — self-corrections, clarification requests, paraphrase — which for a learner is not an admission of failure but a core survival skill.
Genre, register and discourse competence in language teaching
Above the single conversation, discourse organizes into genres — recognizable, staged types of text and talk: the recipe, the complaint letter, the academic abstract, small talk at a bus stop. John Swales' analysis of research articles showed that genres have describable move structures that expert members of a community simply expect; a text can be grammatically perfect and generically wrong. Cutting across genre is register — the way language varies with situation: field (what is being talked about), tenor (the relationship between speakers) and mode (speech or writing). Choosing the wrong register — a chatty email to a ministry, a stiff formal reply to banter — is a discourse error, not a grammar error.
The bridge from all this research into the classroom was built by the communicative turn. When Michael Canale and Merrill Swain modelled communicative competence in 1980, refined by Canale in 1983, they made discourse competence — the ability to combine forms and meanings into unified spoken or written texts, cohesive in form and coherent in meaning — one of its four components, alongside grammatical, sociolinguistic and strategic competence. That decision, at the heart of communicative language teaching, turned discourse from a research topic into a curriculum goal: if building text and talk is a quarter of knowing a language, it has to be taught.
Michael McCarthy's Discourse Analysis for Language Teachers (1991) translated the research into pedagogy, and discourse analysis has since reshaped teaching in two practical ways. It supplied the strongest argument for authentic materials: invented textbook dialogues are typically cohesion-poor and machinery-free — no hesitations, no repair, no discourse markers, everyone speaking in complete sentences — so learners fed only on them never meet the talk they will actually hear. And it turned analysis on the classroom itself: Sinclair and Coulthard's study of lessons (1975) found a rigid three-part exchange — teacher initiates, pupil responds, teacher gives feedback — that exists almost nowhere outside school, which helps explain the classic puzzle of the learner who is fluent in classroom ritual and mute in the street.
What the learner takes from it: markers, connectives, structure
Discourse analysis earns its place in an encyclopedia of language learning because its findings convert unusually well into learnable material. Three conversions matter most.
First, discourse markers as vocabulary. Words and chunks like well, anyway, actually, by the way, I mean, you know do no grammatical work and carry little dictionary meaning, which is why coursebooks long ignored them — yet they are the traffic signals of conversation: they open turns, change topics, soften disagreement, signal that a story is ending. Speech without them can be perfectly correct and still sound translated. Each marker is a small, concrete item with a describable function: exactly the kind of thing a learner can collect and practice deliberately.
Second, connectives and cohesion for writing. Halliday and Hasan's catalogue reads like a syllabus: reference chains that keep pronouns unambiguous, conjunctions that make the logic of a paragraph explicit (however, therefore, on the other hand), lexical variation that carries a topic forward without numbing repetition. These are the most trainable ingredients of good writing in any language.
Third, the structure of spoken exchanges. Openings, pre-closings and closings; the softening apparatus of dispreferred responses; fillers that hold a turn while you plan (let me think, how should I put it); repair formulas that rescue a breakdown (sorry, what I meant was…). Most of these are fixed or semi-fixed expressions — stored and retrieved whole, exactly as described in the entry on formulaic sequences — which means they can be learned as ready-made units rather than assembled from grammar in real time.
What this means for language learning
The central lesson of discourse analysis for a learner is that the unit of learning should be bigger than the word — and often bigger than the sentence. Knowing a language means commanding the ties that bind sentences, the formulas that open, steer and close conversations, and the expectations of genre and register; none of that is visible in a vocabulary list of isolated nouns. The practical consequence: learn whole utterances that do discourse work — a hedged refusal, a topic change, a clarification request, a paragraph-opening connective in context. Items like these are ideally suited to active recall and spaced repetition, which is why the Taalhammer method has learners practice full sentences rather than words: the sentence is where cohesion, register and conversational function live. For the pedagogy this research helped create, see the entry on communicative language teaching; for the memory science of learning such chunks whole, the entry on formulaic sequences.
Frequently asked questions
What is the difference between cohesion and coherence?
Cohesion is the set of visible ties on the surface of a text — pronouns, connectives, ellipsis, repeated and related vocabulary — catalogued by Halliday and Hasan in 1976. Coherence is the underlying unity of sense that makes a text interpretable. They are independent: a passage can be full of connectives yet incoherent, and a two-line dialogue with no visible tie can be perfectly coherent because listeners supply the connecting scene themselves.
Is discourse analysis the same as conversation analysis?
No — conversation analysis is one branch of the broader field. Discourse analysis covers everything above the sentence, written and spoken: cohesion in texts, genre structure, register, classroom discourse, political rhetoric. Conversation analysis, founded by the sociologists Sacks, Schegloff and Jefferson, studies specifically the mechanics of naturally occurring talk — turn-taking, adjacency pairs, openings, closings and repair — from recorded conversations.
Why do discourse markers matter for language learners?
Because they are the machinery that organizes real speech — opening turns, changing topics, softening refusals, signalling endings — and their absence makes even grammatically perfect speech sound foreign and abrupt. They are also unusually learnable: each marker is a short fixed expression with a describable function, so it can be collected and practiced like vocabulary, ideally inside whole example sentences.
Sources
- Zellig S. Harris, “Discourse Analysis”, Language 28(1), 1952.
- M. A. K. Halliday, Ruqaiya Hasan, Cohesion in English, Longman, 1976.
- Harvey Sacks, Emanuel A. Schegloff, Gail Jefferson, “A Simplest Systematics for the Organization of Turn-Taking for Conversation”, Language 50(4), 1974.
- Emanuel A. Schegloff, Harvey Sacks, “Opening Up Closings”, Semiotica 8(4), 1973.
- Michael Canale, “From Communicative Competence to Communicative Language Pedagogy”, in J. C. Richards, R. W. Schmidt (eds.), Language and Communication, Longman, 1983.
- J. McH. Sinclair, R. M. Coulthard, Towards an Analysis of Discourse, Oxford University Press, 1975.
- Michael McCarthy, Discourse Analysis for Language Teachers, Cambridge University Press, 1991.
- John M. Swales, Genre Analysis: English in Academic and Research Settings, Cambridge University Press, 1990.