syncretism w/o paradigms

[John Frampton]

Andrew Nevins wrote:

> Hello, and thanks Martha -- the paper (handout!) was from LASSO,
> is "When 'we' dis-agrees in Circumfixes" and
> can be downloaded from http://web.mit.edu/anevins/www/lasso.pdf
>
> Basically it seems to be to be the null hypothesis that
> impoverishment happens AT a TERMINAL precisely when THAT
> TERMINAL is being WORKED on. It seems like a bizarre
> state of affairs to do impoverishment globally on the
> whole tree BEFORE doing ANY VI.

I admire Andrew's strong feelings about what would and would not
be bizarre about the way the brain does linguistic computations.
That aside, he misunderstood my own hunch about the way things
work. My hunch is that spellout is not a direct translation of
syntactic nodes (i.e. bundles of syntactic features) to
phonology.  There is an intermediate step in which syntactic
nodes are mapped to morphological nodes (i.e. bundles of
morphological features) -- this an aspect of linearization.  The
mapping may preserve many syntactic features, but the mapping can
delete (i.e. fail to map) certain features (i.e. impoverishment)
and can add other features (i.e. case features).  I agree that it
makes a certain amount of sense if impoverishment happens at the
point that syntactic node is "being worked on", for me that would
be the point that it is mapped into a bundle of morphological
features and put into a linear structure.  The main point is that
vocabulary insertion does not interact directly with
impoverishment.

I have no particularly evidence for the correctness of my hunch,
my inquiry was directed in seeing if there was significant
counterevidence.

I looked at the analysis of Algonquin in the handout mentioned
above and am not convinced that it provides much evidence that my
hunch is wrong.  There may be other facts about Algonquin
morphology that bear on the issue, but on the basis of what you
have in the handout, it seems to me that the following works
fine:

   1) Group is deleted in the context of Author.
   2) Author is deleted in the context of Addressee.

-wa is the Group suffix; the Minimal suffix is null; and /-nan/
is the default Individuation suffix.

The two impoverishments must be ordered --- whether for a reason
of the kind Andrew discusses or for some other reason.  It could
be simply that pairs of rules of this form are automatically
ordered in the only way that makes sense.  I think that this
could most likely be formalized without difficulty.

- John Frampton