The semantic blindness of scalar implicatures
Magri (2009) points out that the computation of scalar implicatures appears to be insensitive ("blind") to contextual knowledge. This is indicated by the oddness of sentences like (1) and (2):
(1) Some Italians come from a warm country.
(2) John is sometimes tall.
Plausibly, these sound odd because their implicature-strengthened meaning clashes with our background knowledge that – in the case of (1) – all Italians come from the same country and – for (2) – that people's height is a stable property.
But how does the implicature-strengthened interpretation arise? The standard story is that we somehow conjoin (1) with the negation of its scalar alternative (3):
(3) All Italians come from a warm country.
This computation appears to be blind to our background knowledge that all Italians come from the same country. On the basis of this knowledge (and the knowledge that there are Italians), (1) and (3) are actually equivalent. Why, then, should an utterance of (1) convey the falsehood of (3)?
This is a nice point. But I think it can be strengthened.
Is it actually a contingent fact that all Italians come from the same country? If not, then the computation of scalar implicatures is not just blind to contextual knowledge, but blind to non-contingent semantic facts.
Take another example. It is arguably part of the meaning of 'dog' that the dogs form a natural kind. As a consequence, (4) entails (5):
(4) Some dogs are animals.
(5) All dogs are animals.
Yet (4) is just as odd as (1), and for the same reason. It suggests that (the speaker believes that) some but not all dogs are animals – an analytic falsehood.
Or take (6).
(6) Some people who are 2.00 metres tall are taller than me.
This is odd, because it suggests that some but not all people who are 2.00 metres tall are taller than me – another analytic falsehood.
What all this suggests is that the notions of entailment and consistency that commonly figures in theories of scalar implicatures needs to be finessed.
For example, Fox (2007) suggests that a scalar implicature conjoins the original sentence with the negation of its "innocently excludable alternatives", where S' is an innocently excludable alternative to S if it occurs in every subset of alternatives to S whose negation is consistent with S. This would rule out (5) as an innocently excludable alternative to (4), since its negation is not consistent with (4). So Fox's recipe can't account for the implicature in cases like (4) and (6).