Reflection: On understandings of implication – implicature versus entailment

One sense of implication in English is something that necessarily follows as a logical consequence of some event or matter. Natural language representations can thus have logical properties, that is, the ability to act as input into logical inference rules and to enter into entailment or contradiction relations. For example, one logical implication of the statement that John drinks juice (p) is that John drinks liquid (q). This is termed an entailment, which is defined as a semantic relation between two propositions (p and q) where p entails q if and only if the truth of p guarantees the truth of q (Levinson 1983). In other words, if it is true that John drinks juice then it must be true that he drinks liquid. A second sense of implication in English is something that is expressed indirectly rather than said.

For example, if someone asks me Does John drink juice?, and I respond He doesn’t drink any liquid, I implicate (but do not say1) that John doesn’t drink juice. However, while these two senses are formally separated into the notions of entailment (in logic and semantics) and implicature (in pragmatics) they are not always easy to keep apart in practice. In the example above, what is implicated is also entailed (i.e. John doesn’t drink juice is a necessary, logical consequence of the claim that John doesn’t drink any liquid). However, such entailments can be “removed” through the addition of further contextual information. For example, if I add except for the odd tot of juice to my prior statement, He doesn’t drink any liquid, the implication (and thus implicature) that John doesn’t drink juice does not arise, and the entailment is thus removed (Haugh 2013b). Thus, while implicatures may have logical properties, it is important to distinguish between folk or pseudo-logical analyses of meaning (in pragmatics), and formal logical analyses of meaning (in semantics). The latter are constrained by a strictly defined and fixed set of assumptions, while the former are not.

What is implicated was further divided into those that are conventionally implicated and those are that are non-conventionally implicated, with conversational implicatures constituting a subset of the latter (Grice [1975]1989: 25–26). This resulted in three main representations within the category of what is implicated:

1. Conventional implicature

2. Conversational implicature

3. Non-conventional, non-conversational implicature

Grice had little to say about non-conventional non-conversational implicatures, although he suggested in passing that these might encompass other dimensions of pragmatic meaning, such as politeness (ibid.: 28) or irony (ibid.: 53). This line of thought was developed in much more detail by Leech (1983), as we discuss further.