The unattestedness/overgeneration "cop-out"

Thanks to Eric for setting up this blog–I think it is a great idea. And of course I am pleased to find out that someone read something I have written! So, I will write a little bit to respond to the useful comments and challenges to some of my work that have been mentioned.

I think that all objections to my discussion of unattestedness and overgeneration ARE justified. We (me, Mark Hale, and others with similar viewpoints) need to provide some results or at least get others to do so, if we want to claim that phonologists don’t need to explain every kind of unattestedness phonologically. I think some of Ohala’s work does this–and in fact the idea was his long before it was ours.

But I think there are some things we can agree on.

1. Obviously this issue is not tied to OT vs. derivations. If the factorial typology of a proposed constraint set yields some bizarre grammars, I am not willing to invoke that as an argument against OT. In the same way, the complaint in chapter 9 of SPE about the bizarre grammars you get by reversing signs in all your rules is not an argument against that model. I think a theory of possible rules has to abstract away from the details of the value of alpha in an expression alphF (this is the idea of substance free phonology). I hoped, in the LingReview paper on quantification and in my paper in LI on feature-filling/changing, as well as in a recent paper on locality (with Fred Mailhot, available on request), to make a positive contribution to a theory of possible rules. (Critics of rules are justified in critiquing the lack of work in this area for many years—I wish more people would work on it. Bert Vaux and Andrew Nevins have done a lot of work that converges with ours.)
Probably some of my results are trivially adoptable within OT—I have seen constraints using quantifiers, so this is a point of agreement.

2. The fact that no phonology makes use of sounds in the supersonic range does not need to be explained phonologically. (Or, at least, not as a computational property of phonological systems.)

Bob Kennedy mentions my talk on stress two LSAs ago. I can’t remember enough about it or about his question to know if the system could actually generate stress every fourth syllable, so I can’ t respond. I also do not know whether any other proposed model gets all and only the set of attested patterns in a principled manner. However I tried to provide some concrete explanations for recurrent and non-occurring patterns in my talk at the last LSA (Boston) which was a reworked version of a BLS talk from a year and a half ago. Here is a copy of the latest version Immediate constituent analysis in audition… — the BLS proceedings version is a mess, so please don’t hold me to what I said there. In this highly speculative paper I try to argue that evidence from an auditory illusion can give us insight into things that happen in phonology. First of all, the mere existence of auditory illusions tells us that a phonology that depended on veridical auditory perception would not be usable, since we do not have veridical perception. Second, the illusion gives insight into the nature of segments. And most relevant to our discussion, I claim that the illusion shows why certain patterns of assimilation are predicted to be particularly common, based on the nature of audition.

Finally, I take a stab at the syntacto-centrism of the field, with an argument that recursion is not just a property of syntax. So, us phonologists might at least find something to bond over in that.

One person I respect a lot told me that the ideas in this paper were very important. Another said that it has all been said before, but better. Some of you will just think it is wrong. All types of comments are appreciated. In any case, the illusion I describe (discovered by Nakajima et al) is cool.

Charles Reiss

PS Since I have mentioned substance free phonology, let me recommend our paper in the OUP Phonological Knowledge volume rather than the LI squib. The former is a revision of the latter and is longer and hopefully more coherent.