Conversational Meaning:

                        Speech Acts and Intentions

What do people do when they use language?

Perform actions (promise, request, criticize, etc.)

Earlier approaches to meaning (semantics) restricted meaning to propositional/sentential meaning; what a sentence means (sense and reference; truth-conditional approaches) in isolation (sentence meaning, or literal meaning).

Ludwig Wittgenstein (1953) and John Austin (1962) proposed action-oriented approaches to language.

Austin's (1962) speech act theory

Impossible to determine the truth value of many utterances (e.g., "I promise to do it tonight")

Constatives truth value could be determined (e.g., It's raining out)

Performatives (e.g., "I apologize") used to perform some act (their occurrence changes the world in some way); not amenable to a truth-conditional analysis (although they can be infelicitous)

However, Austin eventually argues that constatives also are actions. (E.g.,assertions implicate a particular stance regarding the nature of the world; it is thus an action).

Austin abandoned the performative-constative distinction: all speech acts have a dimension of meaning (or propositional content) and a particular force.

Locutionary act: sense and reference; dimensions of language (phonetics, syntax, and semantics) with which linguists have traditionally been concerned.

Illocutionary act: conventional force associated with the uttering of the words in a particular context (e.g., I promise to do it tonight" (if performed felicitously) = illocutionary force of promise.

Perlocutionary act: effects the remark has on the hearer (recognition of act performed)

John Searle (1969): Speech Act Taxonomy and Felicity Conditions

Meeting felicity conditions constitutes the performance of that speech act; provides framework for comparing different speech acts.

Propositional Content; speech acts = form F(p), where F is the illocutionary force and p the propositional content.

Certain illocutionary forces specify what is acceptable in terms of propositional content (e.g., promises and requests require the specification of future courses of action)

Preparatory Condition; beliefs and desires of the interlocutors presupposed or implied in the performance of the act (e.g. requests: speaker believes the hearer has the ability to perform the requested act. Assertion: speaker believes the truth of the proposition and is not aware of the proposition).

Sincerity Condition; Speaker expresses a psychological attitude regarding the propositional content of the utterance (e.g., Promise; speaker is expressing intention to do the act that is being promised.)

Requisite psychological state the speaker must have in order to perform a particular speech act.

Essential Condition; specifies the particular illocutionary point of an utterance (e.g., "Please close the door" counts as a request for the hearer to shut the door)

Taxonomy: Illocutionary points (exhaustive and mutually exclusive):

Derived from a consideration of the possible relations between one's words and the world (as it is or could be). (other systems exist)

Illocutionary Point     Direction of Fit                     Examples

Directive                     World-to-words (hearer)         request, order

Assertive                     Words-to-world                       conclude, predict

Commissive                 World-to-words (speaker)        promise, warn

Expressive                     Null                                         thank, apologize

Declarative                     World-to-words and             declare war

                                           Words-to-world                 perform marriage

Five illocutionary points but many speech acts that one can perform; distinctions between speech acts within a type can be made on the basis of the felicity conditions.

Speech Acts and Intentions

Speech acts are intentional acts

Grice (1957) distinguished between signs and signals.

Signs convey information, but recognition of that information does not require a recognition of the speaker's intention to have that information recognized (e.g., falling asleep may be a sign that one is tired; but not intentionally communicative)

Signals convey non-natural meaning (or meaning-nn); communicative acts that achieve their ends by virtue of the hearer recognizing the speaker's intention to achieve those ends; the hearer's recognition of the speaker's intention fulfills the intention.

Intention = illocutionary force, not perlocutionary effects.

Some perlocutionary effects are tied to the illocutionary act; the illocutionary act can be viewed as a means of achieving a particular goal (a perlocutionary effect). But some perlocutionary effects may also (and simultaneously) be unintended. The inferences a hearer can make, the emotional reactions that can be elicited, the beliefs that can be formed or altered, and so on are infinite.

Problem: certain illocutionary verbs do not seem to be characterizable in terms of a reflexive intention; e.g., brag.

Recognizing Illocutionary Force

How do hearer's recognize illocutionary force?

Do they recognize illocutionary force?

Austin (1962) illocutionary force recognition is conventional and based on the performative verb, along with sentence mood and type.

Performative verbs name the action that they perform (e.g.,"I promise to do it"

Recognition of illocutionary force may be componential; comprehension involves recognition of the felicity conditions.

Amrhein (1992) - variations in the polarity of two components - speaker ability and speaker desire - underlie the comprehension of the four quasi-performative commissive verbs (promise, hope, guess, agree).

Comprehension of "promise" entails recognition of the speaker's desire and ability, "hope" entails recognition of the speaker's desire but not ability

Is illocutionary force recognized when performative verbs are not used?

Little research.

But, Holtgraves & Ashley (2000):

"I appreciate your help yesterday" primes 'thanks'

Alternative approaches:

Performative hypothesis (Gazdar, 1979); sentences have a performative clause as the highest clause in deep structure; can be deleted which produces implicit performatives (e.g., "I say to you that it is raining out" becomes transformed into "It's raining out.") Illocutionary force = performative clause, and this clause is true simply by being uttered (hence its truth or falsity is not an issue).

Evidence: sentences may contain adverbs that appear to be modifying the implicit performative clause as in "Frankly, I don't give a damn" (derived from "I say to you frankly that I don't give a damn").


(1) I apologize for what I did.

(2) I apologized for what I did.

performative treats these two utterances as the same (differing only in tense), but (1) constitutes the performance of a specific action in a way that (2) does not.

AI/Intentionalist view: (e.g., Cohen & Levesque, 1990). Illocutionary force recognition not required. People assume others have goals, plans, etc.

Bill: Small coke and fries, please.

Counter person: Would you like ketchup with that?

Bill: Yes, please.

Indirect Speech Acts

People frequently mean more than they say; intended illocutionary force not same as literal illocutionary force. (E.g., "It's warm in here")

Inferential Approaches. Assume: (1) an indirect remark has both a literal illocutionary force and an indirect illocutionary force, and (2) inferential processing required to recognize indirect.

Conversational Postulates (Gordon and Lakoff ,1975). (inference rules) for converting literal meaning blocked by context (e.g, "Can you x?" translates into "I request you to do x")

Links mechanisms for performing indirect requests with the felicity conditions underlying requests; people perform indirect requests by questioning hearer-based preparatory conditions such as ability, willingness, etc., and by asserting speaker-based preparatory conditions.

Request Form                                 Preparatory Condition

Can you shut the door?                 Question hearer's ability

I want you to shut the door.         Assert speaker's desire

Would you shut the door?             Question hearer's willingness

Did you shut the door?                 Question whether or not act

                                                            has been performed

Can be extended to other speech acts (but more difficult)

May explain indirect requests in many different languages (Brown & Levinson, 1987).

Searle (1979) ISAs involve two distinct speech acts; literal and intended.

Literal recognized and then rejected because it violates conversational maxim (see below). Intended indirect is then computed. Felicity conditions guide ISA recognition.

But, what about ISAs not based on felicity (I'm thirsty)

Grice's Theory of Conversational Implicature

Very influential approach.

Concerned with how people mean more than they say

Assume: interlocutors abide by the cooperative principle (CP):

"Make your conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged"

Conversational Maxims:

1. Quantity - Make your contribution as informative as required (i.e., do not be either over-informative or under-informative).

2. Quality - Try to make your contribution true; one for which you have evidence.

3. Manner - Be clear. That is, avoid ambiguity, obscurity, etc.

4. Relation - Make your contribution relevant for the exchange.

Conversationalists rarely abide by these maxims - But CP adherence is ASSUMED; remarks are interpreted AS IF they were relevant, truthful, etc.

Hearers interpret remark so as to maintain adherence to the conversational maxim (Pass the salt)

Speakers will sometimes intentionally flout or violate a maxim; But hearer assumes speaker cooperativeness and generates a conversational implicature that makes sense of the violation. Figurative language, e.g., tautologies such "Boys will be boys" or "War is war."


What implicature will a speaker make? (Many possible)

What counts as a violation? Individual and cultural variability; Kennan; Malagasy routinely withold information (quantity violation)

How many maxims? One? (Relevance; Sperber & Wilson).

Idiomatic Approaches. Intended indirect meanings are assumed to be idiomatic, or noncompositional. (E.g., Can you pass the salt?" is as an idiomatic expression).

Recognition and rejection of literal meaning not required.

Linguistic evidence: The constraints on where "please" can be inserted are identical to those for direct requests (e.g., one can't say "Could you shut please the door?"). Underlying semantic structure of indirect requests = semantic structure of direct equivalents.

May be difficult to distinguish between literal and indirect meanings (Gibbs). Some utterances may not have a literal meaning (e.g., How about a beer?). Often, literal meaning is rarely, if ever, intended (e.g., Can you pass the salt?). Often, literal meaning is not context-free (e.g., I've eaten)

Comprehending Indirect Speech Acts: Psychololinguistic Evidence

Inferential approaches assume addressees:

(1) always first recognize the literal meaning of a remark prior to comprehending the indirect meaning,

(2) search for an indirect interpretation only after deciding that the literal reading is defective (i.e., it violates a conversational maxim), (3) generate additional inferences in order to comprehend meaning

Support for inferential:

Clark (1979) - telephone requests (e.g., "Can you tell me what time you close?"); replies frequently addressed both the literal and the indirect meaning of the request (e.g., "Sure, 8 p.m.")

Politeness: request politeness based on the remark's literal rather than indirect meaning. E.g.: "Could you shut the door?" is more polite than "I want you to shut the door"

Memory for politeness: people spontaneously remembered the politeness wording of remarks at better than chance levels.

Above is not direct evidence (e.g., politeness recognized simultaneously with (rather than prior to) the recognition of the conveyed, or indirect, meaning)

Support for idiomatic:

Most psycholinguistic research on figurative language fails to support inferential.

People do not take more time to understand the meaning of figurative expressions (e.g., "He spilled the beans") than they do literal equivalent expressions (Gibbs, 1980)

Literal and figurative meanings are assessed simultaneously, and in some cases even in a reversed order.

In Kosinski's (1971) novel Being There, the president of the United States at one point asks Mr. Gardiner his opinion of the economy, and the latter responds with:

"In a garden ..... growth has its season. There are spring and summer, but there are also fall and winter. And then spring and summer again. As long as the roots are not severed, all is well and all will be well" (p. 45).

Nonliteral meanings activated even when the literal meaning is acceptable in context (Gildea & Glucksberg, 1983)

Indirect requests facilitated subsequent sentence verification judgments of indirect meanings but not literal paraphrases (Gibbs. 1983)

Evidences is mixed; inferential vs. idiomatic depends on several factors

Particularized vs. generalized implicatures

Generalized implicatures are context independent:

Much of the figurative language appears to produce generalized implicatures (e.g., "He spilled the beans")

The nonliteral meaning of many metaphors is not optional; even when context supports a literal reading, nonliteral meaning is still activated

Also, nonliteral meaning is derived primarily from the words in the utterance. Metaphors (e.g., My job is a jail) represent mapping process; prototypical name (e.g., jail) is used to represent an unnamed superordinate category without a name (things confining).

Particularized implicatures are context (discourse) dependent:

Little empirical research (psycholinguists believe idiomatic)

Relation maxim violations are particularlized:

        Bob: What did you think of my presentation?

        Andy: It's hard to give a good presentation.

Comprehension of these types of replies involves an inference process. The conveyed indirect meaning (I don't like your presentation) is optional; the indirect meaning is not activated if the context supports a literal reading.

Particularized implicatures are not the result of any particular feature of an utterance (there are an infinite number of utterances one could use to violate the relevance maxim), but rather a feature of the placement of the remark in a conversational sequence.

Particularized-generalized distinction is not a property of utterances per se, but rather a property of use.

Figures of speech might also yield particularized implicatures.

Tom asks Bob if he should apply for a job with Bob's firm, and Bob replies "My job is a jail". Tom will recognize the metaphorical meaning (a generalized implicature) and some type of particularized implicature


Conventional means for performing a speech act means that the literal meaning of the utterance is pro forma, not to be taken seriously. Conventional indirect requests generally have the following features: (1) They can be performed by asserting or questioning the felicity conditions that underlie requests

(2) The utterance contains the request-based propositional content (e.g., 'shut the door' in "Could you shut the door?"),

(3) and the preverbal insertion of "please" is allowed (e.g., "Could you please shut the door?").

People do not respond to the literal meaning of conventional indirect requests and their indirect meanings are recognized in a direct fashion,

Nonconventional requests.

Negative state remarks: A speaker can perform a request by asserting or (questioning) the existence of a negative state (or a state that the hearer can infer is negative) if there is some action that the hearer can perform in order to alter the negative state (e.g., "It's warm in here" or "I'm thirsty")

Extension to replies to requests: Because requests project to a next turn, the felicity conditions underlying a request should remain relevant for that turn; same conditions can be denied as a means of refusing to comply with the request. (E.g."Can you loan me $20?"; ("I don't get paid until to Friday").


Strengths of SA theory:

Placement of language within the context of social activities.

Clear for declaratives (I declare war on Ohio), but necessary for other speech acts

To understand a directive, requires an understanding of not just language as an abstract system, but an understanding of peoples' relations with one another; that people can and do perform actions for one another, that people have desires for others to perform such actions, that people have various rights and obligations, and so on.

Research needed on comprehension(emphasis has been on production)

E.g., speaker status; effects production but comprehension too

Research needed on perlocutionary effects (speech acts will differ)

Research needed on why people speak indirectly (politeness)

Major weakness:

The unit of analysis in speech act theory is a speaker's single utterance; conversational utterances usually incomplete, fragmented, elliptical, etc. Also, single speech acts can be performed over a stretch of dialogue

Al: Hi, ya busy?

Bob: Sorta, what's up?

Al: I'm having problems with my computer.

Bob: I could look at it this evening.

Al: Great, thanks