Ken writes: This is really an addendum to my semantic normativity post of a couple of days ago. I’m approaching the issue here from a different angle: babies.
On some ways of counting, Hugh has already spoken his first words. When he was a week old, for example, he said ‘hic’ (i.e. he hiccoughed) which is Latin for ‘this’. He’s also already said ‘athair’ (pronounced something like a-ha, with the first ‘a’ as in ‘cat’), which is Irish for father. What makes his achievement even more impressive is that neither of his parents speaks Irish.
OK. So do we have a genius on our hands or what? Clearly this isn’t what we standardly mean by a baby’s first words. The baby has to say something non-coincidentally meaningful. But I’m not sure what this comes to. One needs a different operational principle than: Reliably utters word w in the presence of the objects for which w stands, and only utters w in the presence of those objects. Even for words, like common nouns, that are meaningful primarily through standing for a class of things, this won’t work. We’d want a baby’s utterance of ‘where’s daddy?’ to count as correct when uttered in the absence of daddy. And we don’t want him continually uttering ‘daddy’ in daddy’s presence either (by the ‘children should be seen and not heard’ principle).
This indicates there is more to obeying a descriptive chracterisation of correctness conditions for certain words, like the one given for ‘horse’ earlier (‘if ‘horse’ means horse, then ‘horse’ correctly applies to an object, x, if and only if x is a horse’), than simply applying the word to all and only the instances of the kind to which it correctly applies. Properly following this rule is a matter of making appropriate utterances. Goodness knows what governs what counts as making appropriate utterances, but it looks a lot more normative and a lot less straight-forwardly descriptive than the statement of the conditions themselves. Maybe where the sceptic about the normativity of meaning goes wrong is in taking these characterisations to be facts like any other, without taking into due account the stage-setting that needs to be in place in order for these word-world relationships to exist at all.