Optimality Theory anyone? Constraints on syllable structure.

[Joanna]

Hi everyone!

I am doing an OT analysis of [ə] epenthesis in a dialect. It appears that the only way to generate the right candidates is to include a constraint that would disallow "complex onset with open vowel". It also appears that this constraint gets very high ranking.
That is, a syllable like [pri] is allowed, but a syllable like [pra] will receive [ə] between [p] and [r], turning it into two syllables.

It works perfectly with it, and overall I can account for all cases of vowel epenthesis with just three markedness constraints, one of them being the one I just described.

What I'm worried about it that it looks too clumsy to me - such specifications just don't look elegant. I am quite new to OT and was wondering if anyone has read of similar constraints? Do you think a language could have a constraint like that?

Thanks!

[panini]

Besides constraints like *Low and *Complex, OT has "constraint conjunction", which penalizes the simultaneous violation of two constraints. A frequently cited example is Ito & Mester Japanese Morphophonemics. Markedness and Word Structure (p. 26-19). That is essentially what you are proposing, that you cannot both have a complex onset and a low vowel within a syllable.

[Joanna]

This is very helpful! Thanks a lot!

[Daniel]

A potential phonetic explanation might be that in an articulatory sense, [a] is more extreme than [i ], and therefore the heavy syllable (in that specific sense) would need to be broken up. So there is some limit to how heavy a syllable can be, that it can either have a low vowel or a consonant cluster, but not both.
(Just an observation, no background in this sort of phenomenon.)

[edit: fixed format]

[Joanna]

This is a very good point, thank you! I should definitely elaborate on syllable weight restrictions.

[Joanna]

So it appears that my dialect simply doesn't allow more than two moras in a syllable, if my analysis is correct.

I've just read in Kager (1999) - "The weight of a CVC syllable depends on whether or not its coda consonant is moraic. This varies from language
to language, and is determined by the ranking of Weight-by-Position". Still not sure how to work out whether a consonant is moraic or not, but working on it!

Thanks for all the hints again, most useful!

[panini]

One thing you should consider is whether there's another generalization lurking. For example, perhaps [ə] in an open syllable deletes before [a] but not other vowels. There are Arabic dialects with deletion of /a/ before /a/, which would be similar. Since /a/ tends to be longer, and /i u ə/ tend to be shorter, then the "flaw" of pəra from a rhythmic perspective is that it has a light-heavy profile, which is held to be bad for a trochaic system (if that's applicable to this language).

[Joanna]

Thanks again! I've revised my constraints and I think the epenthesis is purely consonant-cluster conditioned - it breaks all the consonant clusters into CV.CV.CV very nicely, except when there are two clusters in a word and then it prefers the cluster that is followed by a low vowel (here I use the constrain conjunction you suggested yesterday).

So basically it all comes down to [Low+Complex^ONS] >> COMPLEX >> DEP

The only problem now is that there is this one word in my data that goes /VC.CV.CV.CV/ (after epenthesis applies) and I'm not sure what to do about it yet.