On Thu, 1 May 1997, JOEL MATTHEW PEARSON wrote:
On Thu, 1 May 1997, Spiff Tastic wrote:I'm not sure that you have to do all of the hand-waving to get h to be a feature of this or that. h is a perfectly nice way to begin a word in Tokana, so it does seem to have segmental status there. Why not in syllable codas as well?> Well! My Phonology teacher went into a shit-fit one day about metathesis.Yes yes yes! I use the term "h-metathesis" only as a convenient label: I don't actually believe that the "h" and the consonant switch places (except in the orthography). Basically, I believe your story, except that I think of the "h" as a preaspiration feature on the consonant (spread- glottis timed BEFORE rather than AFTER closure for maximum salience), rather than a breathy feature on the vowel...
> He was screaming about why we shouldn't "borrow metathesis from syntax"
> because it gives the theory too much "power" and ceases to be a good
> account for all the facts. As you have noticed, allowing metathesis as a
> phonological process doesn't account for the fact that changes like that
> are in fact highly rare.> It might be better to say that there is a constraint against consonantal aspiration,
> but it is tolerated in vowels. You might do something like this:> 1. Oh no! There is an aspirate consonant! [There is a constraint against this.]
> Quick -- de-link the glottal node from the root of the segment.> 2. But we want to be faithful to the underlying representation! What to
> do with this floating spread-glottis feature? Well, the constraint won't
> yell at us if we stick it on this vowel. ...
...'Course, I could be wrong. I'm no phonologist, and anyway, my job is to design Tokana, not subject it to theoretical analysis (although if someone else wanted to analyze it, I'd be tickled pink!). ...Kinda funny--I started out with an analysis and built a language. Well, I thought it was funny.
...Perhaps Mr. Elzinga (our resident generative phonologist) would have something to say on the featural status of preconsonantal "h" in Tokana. Do you think it's a feature of the preceding vowel or the following consonant? What is the status of 'preaspiration' in current theory? Anybody?Well, I do have something to say about the status of metathesis, and since you went and mentioned me by name ...
I personally have no problem with describing something as metathesis if it actually is metathesis. Earlier incarnations of generative phonology (Chomsky and Halle, The Sound Patterns of English) could do it fine-- just write a transformational rule:
Tokana H-Metathesis:Autosegmental phonology was somewhat better in providing more natural accounts of lots of phonological processes, particularly those dealing with tone. But autosegmental phonology, with its axiomatic constraint on the crossing of association lines wasn't able to cope with metathesis (hence the instructor's "shit-fit"). The usual strategy was then to explain away cases of metathesis as being 1) non-productive, therefore irrelevant to "Grammar", or 2) something other than real metathesis, like dispossessed floating features looking for a home. Using excuse (1) fails to explain actual language patterns (however restricted in scope), and (2) was way over-used--kind of the "when-all-you-have-is-a-hammer-everything- looks-like-a-nail" school of linguistics. (I just realized that this might be taken as critical of the proposed solutions already mentioned in this thread; believe me, there's no offense intended.)C]+[h
1 2 => 2 1
Current ideas about the status of phonological generalizations maintain that phonology can be described (and explained) as arising from the interaction of constraints on surface forms. These include constraints on syllable structure like:
ONSET: Syllables have onsets.and constraints governing the relationship between underlying and surface forms:
NO-CODA: Syllables don't have codas.
MAX: All of the segments in a morpheme should be realized.Constraints like these can be violated; after all, there are syllables without onsets as well as syllables with codas. A violation of MAX is simply deletion of a segment from the underlying form, and epenthesis, the addition of a segment not in the underlying form is a violation of DEP. These constraints can't be violated willy-nilly; by ranking them with respect to each other, violations of some constraints comes only because of pressure to satisfy higher-ranking ones.
DEP: Don't introduce segments that aren't in a morpheme.
CONTIGUITY: Segments in a morpheme should be next to each other.
So what about metathesis? Well, if we assume a constraint like CONTIGUITY, which insists that segments be next to each other if they belong in the same morpheme (and there seems to be good reason to assume such a constraint), then metathesis is the violation of that constraint. If a constraint is violated only so that a higher-ranking constraint can be satisfied, the question is what kind of pressure forces Ch to trade places in Tokana to hC. It might plausibly be a syllable structure thing--syllable codas prefer to be continuant if possible. I like that explanation, but I don't know right now.
Matt asks about opinions on the featural status of h; does it belong to the preceding vowel or the following consonant? I'm not sure that it has to belong to either. If h can occur word-initially as a segment, and if upon prefixation a stop comes to stand before it, I don't see any great problem with saying that the stop and the h literally trade places so that h is now the coda of a syllable and the stop is the onset of the following syllable.
sot + hi -> sohti 'its words'I'm not familiar enough with the phonetic literature to comment on pre- vs. postaspiration and the featural status of either, but Philip's comments on Icelandic sound awfully familiar in that context. Hmmm. I feel a spell of avoidance behavior coming on ... :-) (I should be in the phonetics lab entering Gosiute recordings into a database.)
...P.S.: By the way, what do you mean by "borrowing metathesis from syntax"? I was unaware that metathesis had any status in syntax...Just a guess here, but was the fact that a transformational rule has to be invoked to get metathesis to work out right the issue? Most linguists are familiar with these rules in syntax, but they could also occur in phonology (as my ad hoc Tokana H-Metathesis rule demonstrates). These rules have been abandoned in most frameworks since they were felt to be too powerful and unconstrained.
Copyright © 1997, Maurizio M. Gavioli,
Last updated: 20 Oct., 1997