synthetic/periphrastic alternation

[Heidi Harley]

... hi all! I haven't read any of the relevant stuff by stump&ackerman,
or you, John, but I'm having trouble perceiving the issue raised by the
allomorphy/suppletion cases you're describing below.

You wrote:

 In what is called the first-past tense, for example, negated
 verbs are much like "didn't forms" in English, with an auxiliary
 verb bearing inflection and negation and the verb root standing
 apart.  The problem is that there are two forms of the root, one
 of which is used in the 3pl and the other elsewhere.  The problem
 is to account for the dependence of root allomorphy (or
 readjustment, or suffixation, or whatever) on agreement, which is
 attached to a different word.

I don't see why, within DM, it should matter whether a conditioning
feature for allomorphy/suppletion/readjustment, etc., is within the
same phonological word or not. What should matter is that the feature
be sufficiently syntactically local for the relevant insertion or
readjustment rule to 'see'. Allomorphy/suppletion/readjustment of a
verb stem in the context of a subject agreement feature, doesn't seem
too strange to me, whether the verb stem is actually affixed to the
tense/agr head or not, and even if a negation head intervenes
structurally.

What such facts DO bear on, I think, is the issue of how big the
locality domain that a VI can be sensitive to is. Is it just string- or
tree-adjacent features/material? Is it any features anywhere in its
extended projection (or phase)? Is it any features upwards in the tree
but not downwards? Is it features of heads but not features of phrases?
etc. (See, e.g., JDB's 2000 Itelmen allomorphy paper for the Maryland
Morphology Mayfest). This is clearly an empirical issue, whose answer
should be determined by looking cross-linguistically at facts like
these -- but I don't see anything about the architecture of DM that
will preclude a treatment of these facts. Indeed, since DM is an
explicitly post-syntactic theory of morphology, I would think that we'd
have a better shot at it than pre-syntactic theories, which might have
a problem getting featural info from one item to affect an item in a
different phonological word before the syntax has introduced the first
item's featural info into the structure.

But perhaps I'm missing something? Is there some part of the problem I'm
not percieving?

yrs suppletively, hh