{
  "id": "speech-act",
  "title": "Speech Act",
  "lang": "en",
  "source_lang": "en",
  "summary": "A speech act is an action performed through communication, such as asserting, requesting, promising, apologizing, or declaring.",
  "category": "language",
  "subcategory": "",
  "type": "concept",
  "aliases": [
    "Communicative act",
    "linguistic act"
  ],
  "generated_at": "2026-11-03",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/speech-act/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/speech-act/",
      "json": "https://aiwiki.page/zh/speech-act.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "A speech act is an action performed in communicating: making an assertion, asking a question, issuing a warning, promising assistance, or pronouncing a verdict. The central insight of speech-act theory is that an utterance does not merely represent a situation; it can also undertake a commitment, solicit a response, establish a social relationship, or alter an institutional status. “The meeting is adjourned,” spoken by an authorized chair in the appropriate circumstances, does not simply describe an independently occurring event: it brings the meeting to a close.\n\nThe study of speech acts connects philosophy of language with pragmatics, semantics, and sociolinguistics. Its modern foundations were established by J. L. Austin and developed systematically by John Searle, although related ideas appeared earlier in philosophy and legal theory. Research examines what distinguishes one communicative action from another, what makes an action successful or defective, how speakers communicate actions indirectly, and how participants recognize and respond to them.\n\nDespite the word speech, speech acts are not confined to spoken language. A written resignation, a signed question in a sign language, and a digitally submitted acceptance can all perform communicative actions. Their interpretation depends not only on linguistic form but also on participants’ intentions, shared knowledge, conversational position, and, in some cases, recognized institutional authority."
    },
    {
      "heading": "Intellectual origins and the development of the theory",
      "text": "Speech-act theory emerged partly from dissatisfaction with accounts of language centered on statements and their truth or falsity. Assertions are important, but ordinary communication also includes requests, invitations, greetings, promises, permissions, and commands. Treating all meaningful utterances as descriptions obscures the different things people accomplish by producing them. Earlier philosophical discussions anticipated parts of this problem: Thomas Reid examined social operations such as promising, and Adolf Reinach’s 1913 work analyzed social acts and the obligations and claims arising from them.\n\nAustin developed the approach most influential in subsequent linguistics and philosophy. His 1955 William James Lectures at Harvard were published posthumously in 1962 as How to Do Things with Words. He initially contrasted constative utterances, which appear to report facts, with performative utterances, which perform actions. “I name this ship…” and “I promise to return it” exemplify the latter. Their most immediate assessment concerns whether the naming or promising has been appropriately accomplished, rather than simply whether the sentence is true.\n\nAustin subsequently questioned the sharpness of this distinction. Statements themselves are actions, and they can be inappropriate, unsupported, misleading, or otherwise defective. Conversely, performative utterances can involve factual commitments. His mature framework therefore analyzed dimensions of action present across ordinary utterances rather than assigning sentences to two wholly separate classes. This shift made the theory applicable to everyday statements as well as conspicuous ceremonies and formulas.\n\nSearle’s Speech Acts appeared in 1969 and developed a more systematic account based on rules, intentions, and the distinction between communicative force and propositional content. His 1976 article “A Classification of Illocutionary Acts” presented an influential five-part taxonomy. Other approaches emphasized different foundations. H. P. Grice’s 1957 paper “Meaning” analyzed communication through intentions and their recognition, while P. F. Strawson’s 1964 “Intention and Convention in Speech Acts” questioned how extensively communicative acts depend on conventions.\n\nThese positions produced a lasting theoretical division rather than a single unified doctrine. Some accounts emphasize conventional procedures and public rules; others emphasize the recognition of communicative intentions, conversational inference, or interactional organization. The broader intellectual setting also includes Ludwig Wittgenstein’s Philosophical Investigations, published in 1953, which emphasized the diversity of language use. Speech-act theory nevertheless has a more specific analytical aim: identifying actions performed in communication and explaining their conditions, force, and consequences."
    },
    {
      "heading": "Locution, illocution, and perlocution",
      "text": "Austin distinguished three analytical dimensions of an utterance. A locutionary act consists in producing an intelligible linguistic expression with a certain sense and reference. An illocutionary act is the action performed in saying something, such as asserting, requesting, or warning. A perlocutionary act concerns effects achieved by saying it, such as persuading, alarming, reassuring, or inducing someone to act. These are usually dimensions of the same event, not three separate utterances.\n\nConsider “There is a dog behind you.” At the locutionary level, the speaker produces a sentence identifying a dog and locating it relative to the addressee. Its illocutionary force might be a warning, an answer to a question, or a piece of information. Its perlocutionary consequences might include frightening the addressee, causing a turn of the head, or prompting avoidance. The same wording can thus have different forces, and the same force can have different effects on different occasions.\n\nIllocutionary success and perlocutionary success must be distinguished. An addressee may understand a warning perfectly but ignore it. A request can be recognized without being granted. Conversely, a remark can frighten someone without having been intended or understood as a warning. Austin treated securing uptake, roughly the recognition of an utterance’s meaning and force, as important to the successful performance of many illocutionary acts. Recognition does not, however, amount to agreement, obedience, or belief.\n\nA further distinction separates force from propositional content. “You will leave,” “Will you leave?”, and “Please leave” may concern closely related circumstances while differing in what the speaker is doing. A common schematic notation is:\n\n\\[F(p)\\]\n\nHere \\(F\\) represents illocutionary force and \\(p\\) the propositional content. Thus, asserting that someone will leave is different from requesting that person to leave. The notation is an analytical abstraction: questions, greetings, and expressive formulas do not all fit a simple proposition-plus-force structure equally naturally.\n\nForce is indicated by multiple features rather than by sentence meaning alone. Explicit performative verbs such as “promise” and “warn” provide relatively direct evidence. Mood, word order, intonation, particles, address forms, and the surrounding activity also matter. An imperative can function as an order, invitation, permission, instruction, or good wish. Accordingly, neither grammatical sentence type nor a verb naming an action provides a complete account of the action performed on a particular occasion."
    },
    {
      "heading": "Felicity conditions, rules, and communicative commitments",
      "text": "Speech acts are assessed through felicity conditions: requirements governing whether an action is successfully and appropriately performed. Austin’s examples frequently involved established procedures. For a ceremonial declaration to take effect, there must be a recognized procedure, suitable participants and circumstances, and an adequate execution of the procedure. A person without the relevant authority cannot ordinarily adjourn a legislature merely by uttering the appropriate words.\n\nAustin distinguished misfires from abuses. In a misfire, the intended act fails to come into effect, as when a necessary procedure or authorization is missing. In an abuse, an act may be performed but be defective because a participant lacks the required intention or feeling, or subsequently fails to behave as expected. An insincere promise is therefore not necessarily no promise at all. The speaker can incur an obligation despite privately intending not to fulfill it. This distinction separates the existence of a public commitment from the speaker’s psychological sincerity.\n\nSearle organized conditions for acts such as promising into several types. A propositional-content condition restricts what the act concerns: a promise characteristically concerns a future action of the speaker. Preparatory conditions concern background requirements, such as the speaker’s ability to perform the action and circumstances in which the undertaking is relevant. A sincerity condition concerns the corresponding psychological state, such as intending to do what is promised. An essential condition specifies that the utterance counts as undertaking an obligation.\n\nThese conditions are theoretical analyses rather than universal checklists mechanically applied by participants. Their interpretation varies across act types and circumstances. “I promise that the sun will rise” can function as emphatic reassurance, although the speaker does not control the event. A threat, unlike a promise in its prototypical form, announces an action disadvantageous to the addressee. Such examples show why the occurrence of a performative verb cannot by itself determine the precise commitments created.\n\nSearle connected these analyses with constitutive rules, often represented schematically as “X counts as Y in context C.” Unlike rules that regulate an independently existing activity, constitutive rules help establish what an activity is: a particular move counts as a valid move within a game, or an authorized declaration counts as a change of institutional status. Not all speech acts require elaborate institutions, but the distinction explains how linguistic events can have consequences beyond their physical production.\n\nA continuing question is whether speech acts are best defined through intentions, conventional rules, or normative commitments. An assertion, for example, is often understood as making the speaker answerable for the truth of its content. A promise makes the speaker answerable for future conduct. These public responsibilities help explain why sincerity and successful performance can come apart."
    },
    {
      "heading": "Major classes of illocutionary acts",
      "text": "Austin proposed five broad classes—verdictives, exercitives, commissives, behabitives, and expositives—but acknowledged difficulties in their organization. Searle’s later classification became especially influential because it used several explicit criteria: the act’s illocutionary point, its direction of fit, and the psychological state it characteristically expresses. Illocutionary point concerns the act’s basic purpose; direction of fit concerns whether words are meant to match the world or the world is to be brought into conformity with words.\n\nAssertives, also called representatives, commit speakers to the truth of a proposition. Statements, reports, descriptions, and many predictions belong here. Their direction of fit is conventionally described as word-to-world: the utterance is supposed to represent how things are. The characteristic psychological state is belief, although speakers can assert what they do not believe. Assertives differ in strength and evidential posture: guessing, concluding, and insisting are not interchangeable simply because all present propositional content as true.\n\nDirectives attempt to get an addressee to do something. Requests, commands, invitations, and advice are typical examples. Their world-to-word direction of fit reflects the aim of bringing conduct into conformity with the utterance. Differences in authority, urgency, benefit, and freedom to refuse distinguish a command from a suggestion or an invitation. Commissives, including promises and undertakings, also have a world-to-word orientation, but characteristically commit the speaker rather than seek action from the addressee.\n\nExpressives convey an attitude toward a situation taken as given, as in thanking, apologizing, congratulating, or condoling. They do not characteristically aim either to describe a state of affairs or to bring it about. In “Congratulations on your promotion,” the promotion is normally treated as established background. Its status can be analyzed through presupposition, although not every expressive has the same linguistic structure or projects background assumptions in the same way.\n\nDeclarations bring about an institutional state through their successful performance. Appointing an official, declaring a meeting open, and pronouncing a sentence in an authorized legal setting are examples. Searle described declarations as having a double direction of fit: their successful utterance both represents and establishes the relevant institutional fact. Their effectiveness depends on recognized procedures and authority, not on linguistic wording alone.\n\nThe taxonomy is not an exhaustive inventory of ordinary action labels, and particular cases can resist a single classification. A warning may inform the recipient of danger while also seeking preventive action. Questions are often treated as directives seeking an answer, yet they can also test knowledge, display surprise, challenge a claim, or introduce a topic. Analysts therefore distinguish abstract act types from the multiple functions that one utterance may perform in an actual exchange."
    },
    {
      "heading": "Indirectness, inference, and the relation between form and force",
      "text": "An indirect speech act occurs when a speaker performs an action through an utterance whose more literal or conventional interpretation expresses another action. “Can you pass the salt?” is grammatically a question about ability, but at a dining table it commonly functions as a request. “It is cold in here” may be a report about temperature, a complaint, or a request to close a window. The intended action is not recoverable from sentence form alone.\n\nSearle’s influential treatment distinguished a primary illocutionary act, such as requesting, from a secondary act, such as asking whether the addressee can perform the requested action. The relationship is systematic: questions about ability, willingness, or future action can invoke conditions relevant to a request. Nevertheless, many familiar patterns are conventional enough to be recognized without a consciously reconstructed chain of reasoning. Their interpretation also varies across languages and social settings.\n\nGricean pragmatics explains indirect communication through assumptions about rational conversational conduct. Under the cooperative principle, an apparently irrelevant contribution may prompt an inference connecting it to the current activity. If a host says that dinner is ready after a guest asks whether there is time for another game, the statement may function as a negative answer or a suggestion to stop. Conversational implicature and speech-act force are related but distinct: an inferred proposition is not itself identical to the communicative action performed by conveying it.\n\nIndirectness has no simple one-to-one relationship with politeness. A question can preserve room for refusal, but it can also be confrontational: “Could you possibly stop interrupting?” may express irritation rather than deference. Conversely, “Come in” can be a welcoming invitation rather than an impolite command. Social distance, entitlement, urgency, prosody, and the participants’ relationship affect how a form is interpreted.\n\nIndirect acts must also be distinguished from figurative language. Verbal irony concerns an evaluative or distancing relationship to what is expressed, whereas indirectness concerns how an action is performed through another communicative form. An ironic compliment may function as criticism, but many indirect requests contain no irony or metaphor. Quotation likewise changes matters: saying “She told him, ‘I promise’” normally reports a promise rather than making that promise oneself.\n\nThe relationship between form and force is constrained rather than arbitrary. Imperatives strongly favor certain action families, performative formulas can explicitly mark an undertaking, and particular constructions become associated with recurring uses. However, contextual evidence can redirect interpretation. “Can you swim?” is an ordinary information question during an interview, a challenge beside a pool, or a preliminary to assigning a task. The sentence supplies resources for action, not a complete action description independent of its use."
    },
    {
      "heading": "Speech acts in conversation and shared context",
      "text": "Classical speech-act analyses often begin with an isolated utterance and ask which action it performs. Conversation analysis instead gives particular weight to sequences: what an utterance responds to, what response it makes relevant, and how subsequent conduct displays understanding. These approaches overlap, but their preferred units differ. An utterance can be ambiguous in isolation while becoming recognizable as an answer, rejection, or repair in a developing exchange.\n\nAn adjacency pair consists of linked actions such as question–answer, greeting–greeting, or offer–acceptance. The first part makes a restricted range of second actions relevant, although it does not guarantee that one will occur immediately. A question can receive a request for clarification before an answer. An invitation can be followed by a question about timing before acceptance. The absence, delay, or design of a response may itself become socially interpretable.\n\nSequence position can change action identity. “Are you busy?” may be a genuine inquiry about someone’s activity, but it can also initiate a preliminary sequence before a request. The recipient may answer “What do you need?”, displaying recognition of that projected request. Similarly, “Yes” may answer a polar question, accept an offer, acknowledge hearing, or confirm a proposed understanding. Identifying its function requires more than classifying the word’s lexical meaning.\n\nInterpretation depends on common ground: information treated as shared for the purposes of the interaction. Deictic expressions such as “here,” “that,” and “tomorrow” require contextual anchoring, while action interpretation also draws on roles, current activities, and known plans. A laboratory supervisor saying “The samples are still outside” may be reporting a problem or prompting their retrieval. Which reading is warranted depends partly on who is responsible for the samples and what has already been discussed.\n\nParticipants can negotiate or contest the force of an utterance. “Was that a request?” seeks clarification; “I was only suggesting” revises or defends an earlier contribution; “You promised” attributes a commitment that the original speaker may dispute. Subsequent responses provide evidence of understanding without being infallible records of an original private intention. Communication is therefore both prospective, in projecting responses, and retrospective, in allowing later turns to clarify what an earlier turn accomplished.\n\nEmbodied conduct also matters. Gaze can select an addressee, pointing can identify the object of a request, and a pause or facial movement can shape how a response is understood. In sign languages, manual signs and nonmanual features jointly contribute to grammatical and pragmatic interpretation. Speech acts are consequently better understood as communicative actions realized through available modalities than as actions inherently dependent on sound."
    },
    {
      "heading": "Institutional authority, law, and performativity",
      "text": "Some speech acts create publicly recognized changes because they are embedded in institutions. Courts, legislatures, universities, workplaces, and religious organizations maintain procedures through which particular participants can issue decisions, confer statuses, or establish obligations. The operative conditions are specific: who may act, through which medium, at what stage, and with what required formalities. An identical sentence spoken outside those conditions may merely imitate or report the institutional act.\n\nIn law, linguistic actions include offers, acceptances, testimony, judgments, notices, and certain forms of consent. Their legal consequences depend on the applicable legal system and circumstances, not simply on their classification in ordinary-language philosophy. A statement described conversationally as a promise does not automatically constitute an enforceable contract. Conversely, an obligation may arise without the explicit phrase “I promise.” Speech-act analysis clarifies the structure of such acts but does not replace legal rules governing their validity or effects.\n\nDocuments illustrate the difference between linguistic content and authorized performance. A draft appointment letter, an unsigned form, and a properly issued appointment may contain similar wording while having different institutional consequences. Digital systems add questions of authentication, submission, receipt, and recorded agreement. Clicking an acceptance control can participate in an institutional act when the surrounding procedure gives it that role. The visible words or interface action alone do not explain the entire transaction.\n\nThe concept of performativity developed beyond Austin’s original analysis. Jacques Derrida’s 1972 essay “Signature Event Context” emphasized the repeatability of signs and questioned whether ordinary communication could be sharply separated from quotation and other displaced uses. Judith Butler’s Gender Trouble (1990) and Bodies That Matter (1993) extended performativity into an account of how repeated, norm-governed practices contribute to the constitution of social identities. This broader usage should not be equated with the narrower claim that an explicit performative sentence performs a named act.\n\nSpeech-act theory also provides vocabulary for analyzing communicative power. Authority can affect whether an utterance counts as an order, a ruling, or a dismissal, and social conditions can affect whether an attempted action receives recognition. These questions require separating at least three issues: whether an act was performed, whether it was legitimate or valid, and what effects it produced. A command may be understood yet unauthorized; a declaration may be institutionally valid yet contested; an attempted intervention may influence conduct without acquiring the status its speaker sought.\n\nThe distinction between linguistic action and consequences is particularly important in contested cases. Harm, offense, commitment, and institutional validity are not interchangeable categories. An analysis must identify the act, its circumstances, the relevant standards, and the evidence for its effects rather than treating all consequential language as a single kind of performative."
    },
    {
      "heading": "Cross-cultural variation, acquisition, and empirical research",
      "text": "Languages provide different resources for performing and recognizing speech acts. These include mood systems, particles, honorifics, address terms, formulaic expressions, and prosodic patterns. Social conventions also affect who may make particular requests, how refusals are delivered, and which circumstances call for apology or thanks. Broad action categories can be useful for comparison, but their boundaries and preferred realizations should not be assumed identical across communities.\n\nCross-cultural pragmatics has examined requests and apologies in particular detail. The Cross-Cultural Speech Act Realization Project, associated with Shoshana Blum-Kulka, Juliane House, and Gabriele Kasper, produced a major edited volume in 1989. Research in this tradition distinguished components such as a request’s head act, internal modification, and supporting moves. “Could you possibly send the document? The deadline is tomorrow” contains a request form and an accompanying account that helps explain its urgency.\n\nSuch studies require careful interpretation. A greater frequency of direct forms in one dataset does not establish that an entire language or population is less polite. Participants’ relationships, the task, the imposition, the communicative medium, and the sampling method all affect results. Register and setting can produce substantial differences within a language. A workplace directive, a family request, and an emergency instruction should not be treated as interchangeable evidence of a single national style.\n\nSpeech acts are also central to language acquisition. Children learn not only vocabulary and grammar but also how to attract attention, request objects, answer questions, repair misunderstanding, and manage commitments. Early communicative actions can involve gestures and vocalizations before fluent sentences appear. Their classification nevertheless depends on evidence of communicative intention and interactional uptake; not every behavior that causes an adult to act is thereby a request.\n\nIn second-language acquisition, grammatical proficiency does not guarantee appropriate interpretation or production of speech acts. Learners may transfer familiar conventions, interpret an indirect request literally, or use a formula whose social force differs from its closest expression in another language. Research distinguishes pragmalinguistic resources—the forms available for performing actions—from sociopragmatic judgments about relationships, obligations, and appropriateness. The distinction is analytical, since actual communication normally combines both.\n\nMethods include recordings of naturally occurring interaction, role plays, interviews, and discourse-completion tasks in which participants supply a response to a described situation. Each captures different evidence. Elicited tasks permit controlled comparisons but may not reproduce turn timing or embodied conduct. Corpus linguistics supports investigation across larger collections, although annotators must still infer action from context. Reliable analysis requires explicit category definitions and attention to cases where a single contribution performs several actions."
    },
    {
      "heading": "Formal models and computational applications",
      "text": "Formal approaches seek to represent the conditions, commitments, and consequences of speech acts precisely. Searle and Daniel Vanderveken’s Foundations of Illocutionary Logic (1985) developed a framework in which illocutionary forces differ along dimensions including point, strength, mode of achievement, content conditions, preparatory conditions, and sincerity conditions. Such frameworks aim to explain relationships among acts: for example, why ordering and requesting belong to a related family while differing in authority and mode of performance.\n\nSpeech acts also intersect with dynamic semantics, which treats utterances partly through their effects on a conversational context. An assertion can propose adding information to the common ground; a question can introduce an issue to be resolved; a directive can propose a change in future conduct or obligations. These models differ in their technical implementations, but they share the view that understanding an utterance involves tracking a change in a structured communicative situation rather than assigning truth conditions alone.\n\nComputational research often uses the related category of dialogue act. A dialogue-act scheme labels contributions by their interactional function, such as informing, questioning, confirming, acknowledging, or requesting clarification. Dialogue acts are not simply a renamed version of Searle’s five classes. They commonly include functions important for managing interaction, such as turn management, feedback, and repair, and they may assign several functions to one segment.\n\nThe ISO 24617-2 standard for dialogue-act annotation was first published in 2012. It provides a multidimensional framework for describing communicative functions and their dependence on other contributions. This approach accommodates the fact that an utterance can advance a task while simultaneously acknowledging a previous turn or managing the exchange. Annotation therefore requires decisions about segmentation, contextual dependencies, and which functional dimensions are relevant.\n\nIn artificial intelligence, dialogue-act recognition can support customer-service systems, spoken interfaces, meeting analysis, and conversational agents. A system must distinguish “Cancel my booking” from “Did you cancel my booking?” and from “Do not cancel my booking,” even though all contain similar vocabulary. Earlier systems frequently relied on hand-built rules or statistical classifiers; later systems use machine learning methods, including neural models and large language models, to incorporate wider linguistic context.\n\nPersistent difficulties arise when surface form underdetermines action. Sarcasm, indirect requests, ambiguous acknowledgments, and institution-specific terminology can defeat a classifier that uses only the current sentence. High-quality interpretation may require preceding turns, prosody, speaker roles, and the state of an ongoing task. A further distinction separates recognizing a user’s action from being authorized to carry it out: correctly identifying a request to transfer funds does not establish permission to execute the transfer. Computational systems must therefore represent communicative intent, task state, and operational authorization as related but distinct matters."
    }
  ],
  "links": [
    "philosophy-of-language",
    "pragmatics",
    "semantics",
    "sociolinguistics",
    "sign-language",
    "intonation",
    "cooperative-principle",
    "conversational-implicature",
    "verbal-irony",
    "law",
    "performativity",
    "register-linguistics",
    "language-acquisition",
    "second-language-acquisition",
    "corpus-linguistics",
    "artificial-intelligence",
    "machine-learning",
    "large-language-model"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}