Particularised and generalised implicature

In classic examples, a man may suggest the absence of a close relationship, and between events may suggest sequence, if may suggest the additional restriction ‘only if’, and some may suggest ‘not all’. I don’t believe… is often understood as a tentative belief in the negative proposition. These are typical readings, not a mechanical transfer of negation or universal rules; theoretical analyses of individual cases differ.

The word ‘typically’ matters: these inferences are not logical entailments. They arise by default and disappear when the speaker or context blocks them. They are therefore cancellable, as discussed below.

A particularised implicature has no trigger in the form alone. The listener sees that the response does not answer the question directly, assumes that the other person is nevertheless cooperating, and looks for a connection within their shared knowledge. That connection produces the inference.

A particularised inference relies on contextual assumptions: classes may take up an evening; a manager may be expected to reply promptly in this situation; lights being off in a pharmacy may indicate that it is closed. These are shared or recoverable assumptions, not guaranteed truths: the lights could have been switched off for another reason.

Changing the preceding question helps reveal an inference’s dependence on the situation. If the statement about the laptop stops functioning as a refusal after a different question, that supports a particularised analysis. An inference’s survival or disappearance in one test does not automatically classify every case.

The test also works when the wording stays identical: the same sentence in different conversations can be a refusal or simply information. This shows that refusal was not part of the words themselves.

An implicature can be cancelled without contradiction because it is not part of what is stated: ‘Some of the samples were contaminated — I haven’t checked the rest’ does not contradict itself. A contradiction arises only when what is stated is denied, rather than what is typically inferred.

Cancellation can be explicit, through a continuation with but, though or in fact, or contextual, when the circumstances show that the inference does not apply. A particularised inference also disappears when an added fact breaks the link with the question, such as ‘but I have already packed’.

Cancellability helps distinguish a typical additional inference from literal content. However, the test must be interpreted in context: an unnatural continuation does not always imply a logical contradiction, and changing a word’s meaning or correcting what was said is not ordinary cancellation of an implicature.

Distinguishing the two types is useful, but the boundary is not always sharp. A generalised inference also depends on the situation: when completeness is not relevant to the conversation, it weakens or disappears. Particularised inferences can be strong, virtually unavoidable, or weak, merely possible.

Researchers explain the boundary between the two types in different ways. For learning purposes, identify the literal content, the possible inference and the context that supports or blocks it. Do not assign an exact frequency ratio to the two types without research.

Knowing about implicatures helps not only with understanding but also with choosing words. When an unwanted generalised inference could mislead, especially in reports, letters and contracts, block it in advance: state the exact quantity, the limits of your knowledge or the actual relationship.

A particularised inference, by contrast, lets a speaker refuse politely or hint, but leaves them vulnerable to misunderstanding: the listener may lack the necessary knowledge. When the stakes are high and knowledge is not shared, it is better to be direct.

If you are unsure whether you have interpreted an implicature correctly, clarify it with a neutral question naming your inference, rather than an accusation.