An unintended utterance is not a speech act, the author argues, but its consequences still have to be answered for

Must a speech act (using words to do something: promise, order, request) always be intentional? If someone stretches an arm and the teacher reads it as a raised hand, or says “yes” without meaning to accept, has a speech act occurred, and how binding is it? This is the dispute between internalists, who treat the speaker's intention as a necessary condition, and externalists, who do not.

Pure philosophical argument with no empirical data. It starts from Mises's praxeology, which defines “action” as purposeful behaviour and separates it from mere behaviour such as sneezing or blinking, and lays that on top of the speech-act theories of Austin, Searle, Grice, Bach and Harnish, and Reinach's theory of social acts. Problem cases are sorted into four groups and argued one by one, in direct dialogue with Navarro-Reyes.

• A speech act is the intentional use of an existing convention (language, gesture, ritual) to reach a chosen social aim. It requires knowing the convention and being willing to follow it. Without intention there is no action and so no speech act. The author rejects the widely taught maxim that “one cannot not communicate”: without intention there is no communication, only a reader who mistakenly finds a message.
• Group one, accidents. A convention is evoked unintentionally (a stretched arm read as a raised hand). The author argues this is not a speech act but “speech behaviour” that happens to carry normative consequences, like a gun-cleaning accident that breaks a friend's television. Responsibility follows from the causal link (negligence, clumsiness), not from intention. This is where he parts company with Navarro-Reyes, who treats unintentional speech acts as a legitimate kind of speech act.
• Group two, abuses. Lying, promising without meaning to keep the promise, marrying for money. These are speech acts: intention is present and aimed at the goal (to deceive). The convention, not sincerity, determines the effect; a contract signed with no intention of honouring it still binds.
• Group three, failures. Intention and convention are present but the addressee does not take them up (a “No trespassing” sign that is ignored). Still a speech act, because action is striving, not succeeding, contrary to Austin's requirement that uptake be secured.
• Group four, inaction. The intention to produce an effect exists, but no convention is intentionally used (an empty milk bottle left on the table so a partner will buy more). It counts as a speech act only if that community, even a couple, shares a recognised convention that the empty bottle means “please buy milk”; otherwise the partner is reacting to a situation, like stepping around a puddle. Such cases must be judged one at a time.
• The conclusion has two legs. To perform a speech act, the speaker must intend to follow the conventions that make communication possible; yet the unintentional use of a convention can still produce valid effects, as when a diner orders by mistake but follows the procedure correctly and must pay. The convention makes the effect happen even where intention is absent.

Anyone who has to deal with “I didn't mean it” — teachers, online moderators, lawyers, public-relations staff — gets a tool for separating three questions: was there an intention to produce an effect, was there an intention to use a convention, and has the convention already taken effect. The article is also a compact rebuttal of the “one cannot not communicate” maxim taught in introductory communication courses.