8.2 Scalar implicatures and generalized conversational implicatures

Scalar Implicatures and Generalized Conversational Implicatures

Scalar implicatures explain how choosing a weaker word (like "some") leads listeners to infer that a stronger word (like "all") doesn't apply. This is one of the most systematic ways we communicate more than we literally say.

Generalized conversational implicatures go broader: they're inferences that arise automatically from common linguistic patterns, without needing any special context. Together, these concepts show how much of everyday communication depends on what's not said.

Scalar Implicatures and Horn Scales

Scalar implicatures in communication

A scalar implicature is a conversational implicature that arises when a speaker uses a weaker term from a scale of alternatives. The hearer infers that the stronger alternatives on that scale don't hold.

For example, if someone says "Some students passed the exam," you'll naturally infer that not all students passed. The speaker could have said "all" but chose "some" instead. Since speakers are assumed to be informative, the hearer concludes the stronger claim isn't true.

The literal meaning of "some" is actually compatible with "all" (logically, "some" just means "at least one"). The "not all" part isn't part of the sentence's semantics. It's an implicature, derived from the speaker's choice of a weaker term.

Horn scales and scalar implicatures

A Horn scale is a set of linguistic alternatives ordered by informational strength, where each term entails the ones below it. Some classic examples:

• ⟨some, most, all⟩
• ⟨warm, hot, boiling⟩
• ⟨possible, likely, certain⟩
• ⟨or, and⟩
• ⟨good, excellent⟩

The mechanism works like this:

1. The speaker uses a term from somewhere on the scale (e.g., "warm").
2. The hearer recognizes that stronger alternatives exist on the same scale (e.g., "hot," "boiling").
3. The hearer reasons: if the stronger term were true, a cooperative speaker would have used it.
4. Therefore, the hearer infers the stronger term does not apply.

This reasoning is grounded in Grice's maxim of Quantity: speakers should be as informative as required, but not more so. Using a weaker term when a stronger one is true would violate this maxim, so hearers assume the stronger term was avoided for a reason.

Generalized Conversational Implicatures

Patterns of generalized conversational implicatures

A generalized conversational implicature (GCI) is an inference that arises by default from certain linguistic forms or patterns, without needing any special context to trigger it. Scalar implicatures are one type of GCI, but the category is broader.

Here are some common patterns:

• Conjunctions ("and"): "P and Q" typically implicates a temporal or causal sequence. "She picked up her keys and left the house" implicates she picked up the keys first, then left. Logically, "and" is symmetric (P and Q = Q and P), but pragmatically we read order into it.
• Possessives: "John's car" implicates that John owns the car, even though the possessive could technically describe other relationships (the car John borrowed, the car John is standing next to).
• Indefinite articles: "A student called" implicates the student isn't someone already familiar in the conversation. The indefinite "a" signals new or unidentified information.

A crucial property of all GCIs: they are defeasible, meaning they can be cancelled by additional context. You can say "Some students passed; in fact, all of them did," and the "not all" implicature simply disappears. This cancellability is what distinguishes implicatures from entailments.

Pragmatic reasoning for implicatures

The reasoning behind both scalar and generalized implicatures traces back to Grice's cooperative principle and its four maxims:

1. Quantity: Be as informative as required, but not more.
2. Quality: Be truthful; don't say what you lack evidence for.
3. Relation: Be relevant.
4. Manner: Be clear, brief, and orderly.

Hearers assume speakers are following these maxims. When a speaker chooses a particular word or construction, the hearer considers what alternatives the speaker could have used but didn't. If a stronger or more informative alternative was available and the speaker skipped it, the hearer infers there's a reason.

For scalar implicatures, the reasoning leans heavily on Quantity: the speaker didn't use the stronger term, so it must not apply. For other GCIs, Manner and Relation also play a role. Saying "She got dressed and went to work" in that order (rather than reversing it) follows the maxim of Manner (be orderly), which is why hearers read temporal sequence into it.