how many units of meaning? of production?

[Carson Schutze]

Ann Peters said:

> Evidence for (1) would be that *don't* only occurs in a limited range of
> constructions, such as the imperative. So while we may see
> don't do that; don't touch that; don't drop it.
> we don't see full auxiliary use such as
> you like it but I don't (like it); I don't see it.
> Once the privileges of occurrence expand, then one can begin to claim that
> an adult-like analysis has been performed, and that in some sense "don't"
> now contains more than 1 unit of meaning.

That's exactly what I intended under my option 2) [though the imperative use
of "don't" might be a special case--there are linguistic arguments
suggesting it is different from the declarative]; sorry if I wasn't clear.
I'm curious, though, what exactly you think the child's analysis of "don't"
is at a stage when they use it only as an imperative--are both negation and
imperative parts of its meaning? In that case, you would at most be talking
about monomorphemicity in the weaker sense (my option 1), which would not by
itself explain the limited distribution of this form. Rather, you would need
the additional claim that "don't" cannot *also* represent the meaning
neg+indicative-aux. If instead you would propose that "don't" has only one
bit of meaning in it, what is that bit, and how does it capture the
distribution?

Ann also said:

> (2) later on, even though they may be analyzed, they may be *produced* by
> mature speakers as single, unanalyzed units.

This was a possibility I didn't raise, because I didn't think that the
production system was specifically at issue. I'm not certain I understand
what this proposal means, partly because I don't know what "analyzed" means
(does it refer to the comprehension part of the processor, or to the
grammar?), and partly because I again don't know which sense of "single,
unanalyzed unit" is intended. Here's one thing that might have been
intended: the grammar contains the information that don't = do + n't = aux +
neg, but in your (production?) lexicon there is an entry that says don't =
aux.neg, period, and in production you use only the latter. (This would be
analogous to the claim in the past-tense literature that high-frequency
regularly-inflected forms can be stored, even though their form could also
be computed by rule--it's just handy to store the result of this computation
rather than doing it over again each time you want to use the word.) One
could call this kind of production item monomorphemic in my weak sense if
desired, but as long as what is stored is identical to what the grammar
would compute compositionally, what I said about my option 1 still holds: on
its own it will not predict anything other than the adult distribution.

Brian MacWhinney wrote:

> When I referred to "initially" and "forever", I meant that a child
> may begin with a monomorphemic analysis initially and remain with that
> analysis throughout subsequent language learning.

That's what I understood, i.e. that speakers would maintain this analysis
throughout their lives. I was saying that linguists would find
monomorphemicity a lot less plausible "forever" than just "initially".

> It is interesting to think that back in
> 1967 Ursula Bellugi may have thought of "don't" as a single meaning
> primitive.

The claim wasn't that the adult meaning of "don't" was conceived of as a
meaning primitive by the child, but rather than the child had a different
meaning for "don't", essentially Neg with no Aux. The motivation for this
claim was, among other things, to explain why "don't" is apparently learned
earlier than many other Aux elements, in particular modals.