Re: Small Derivational Idea

On Feb 24, 2009, at 6∞22 PM, Alex Fink wrote:�> Leti has two classes of
verbs. In class II, which is unmarked, the�> subject�> prefixes are subject
to the metathesis and apocope binding�> processes that�> take place at most
word-internal juncture: so a verb like [sOpla]�> 'sail'�> with 1st singular
prefix [u-] becomes [swOpla]. In class I, which�> verbs can�> get into for
phonological or morphological or lexical (i.e. no)�> reasons, the�> binding
processes are blocked: so [Beli] 'buy' has 1sg. [uBeli].�> Voila,�> infix
and prefix as allomorphs.��I don't think that's the best way to analyze
that, but if you did,�formally�the two are now distinct, and have little if
any relation to one�another.�That seems quite obviously to not be the case,
so why have the formalism�encode that?��> Well, if that's what disturbs
you, why not say the masculine ending�> is /-o/�> in _Carlos_ as in every
other word, and it just happens to have an�> extra�> morpheme /-s/ tacked
on after that? (Meaning, I dunno, '(masculine�> proper�> name)', but
happens not to appear on most masculine proper names.)��That is *precisely*
what the formalism forces you to do: to posit�some sort�of meaning for the
/-s/ there. And doing that is patently absurd.�(Well,�depending on what
your goals are. If your goals are to fill out the�framework's�predictions,
then, by all means, mess with the data however you want to�make it so. If
your goal is to explain language, though, this is�troubling,�at the very
least.)��> But more fundamentally, you seem to be objecting to the
morphemic�> way of�> thinking here because it formally lets you do
implausible things,�> doesn't�> constrain you to the realm of plausibility,
wrt. allomorphs looking�> similar.�> Could I not level the very same charge
against WP, indeed even�> more so,�> given that it's a more general
framework? Maybe not in this�> example, but�> you could certainly write
down a pattern which was normally a�> suffix but�> happened to be an infix
just for stems ending in /-men/ or /-tifuli/.��I never said anything about
WP, of course, but if you're talking�about that�formalism, yes, it simply
doesn't work, without constraints, and�doesn't suit�a linguist. Why?
Because it can do *everything*--even represent�language�morphemically. The
goal of frameworks in linguistics is to model which�natural languages can
exist, and not model those that can't (or to�predict�that they won't be
able to). Bochnerian WP, as it is, grossly�overpredicts,�as it's limited
only by imagination. In order for it to be a�workable linguistic�framework,
it must be constrained in some way, and before I left grad.�school a few
colleagues were coming up with some interesting ideas on�just how to do this.
I don't think it will come to anything, but�it's a start.��> I think it's
too much to ask that one's theory of morphology rule�> out such�>
absurdities by purely formal means.��Unfortunately, that really is the goal
of formal theoretical�linguistics: to�formally exclude that which can't
exist, while formally explaining that�which can.��> We get progressively
more absurd as we�> pass from one masculine marker to two dissimilar ones to
three to a�> kapillion, but there certainly must be natlangs that have
two�> dissimilar�> ones by an accident of history, maybe even three: so how
is a theory�> supposed to draw the bright line and say, okay, you can have
N�> allomorphs,�> any more and it's the loony bin?��The question is if
this makes sense for Spanish, and it really doesn't.�With an analogical
model, you can get an explanation, by showing�that it simply doesn't make
sense (i.e. it doesn't fit with the patterns�of the language). The problem
isn't with the number, at all--look at�how many cases Tsez has. The point is,
does this explanation make�sense given the context of the specific
language--not does it make�sense given everything that every language that
has ever existed�has ever done.��> I would suspect that for every one who
simply accepts the idea of a�> morpheme�> and proceeds to do that, there's
another three who haven't even�> gotten that�> far in linguistic
background, and are doing that because it's the�> easy thing�> to think of
and implement. If you want to make a uniform way to�> mark, say,�>
masculines, I'd wager that you're much likelier to hit on�> sticking /-men/
on�> the end of your string than something with several different�>
alloforms or�> paradigms just because it's objectively simpler, and it
didn't�> occur to you�> to throw in any complications...��I agree it is
simpler--for those who have a language like English as�their only�language
resource. This isn't the case for all conlangers.��> For that matter there
are conlangers of the breed who turn to the�> hobby to�> make something not
beset by all the irregularities and complexities�> that�> plague natlangs.
I don't think it's right to call them morphemists�> either.��Why on earth
should regularity be confined to morphemes?�Plus, once you accept that
affixes need not to be treated as�morphemes, it simplifies things quite a
bit. It's not like there's�anything wrong with affixes, or with anything
else. But, for�example, this pattern is totally regular--artificial, but
regular:��batol "cow"�lotab "cows"��donat "dog"�tanod
"dogs"��keef "horse"�feek "horses"��If you want to explain this with
morphemes, you have to have�an invisible "switcharoo" suffixes that is added
to the plural�to switch the phonemes around. The
morpheme-based�explanation, in fact, is much more complicated than the
simple�observation that you flip the word around to form the plural�(or
perhaps flip the plural around to form the singular). This�pattern is totally
regular and pretty simple. Why is it that separating�meanings out and
stringing them together is the default "easiest"�and "simplest" way of doing
things? That seems to me to be�a culturally-dependent idea at
best.��-David�*******************************************************************�"A
male love inevivi i'ala'i oku i ue pokulu'ume o heki a."�"No eternal reward
will forgive us now for wasting the dawn."��-Jim
Morrison��http://dedalvs.conlang.org/�

Reply