Friday, May 30, 2008

Challenge IV: The big picture

No, the picture doesn't mean what you think. Just hold on a second.

"Finish what you start." That’s what the old men always told us when I was little. I feel compelled to check off Challenge IV in this Quixotic quest:
4. The benefits of OT over SPE “don’t look all that great”.
This one triggered the first nastiness, the quip that it “requires a lot of (willful) ignorance of a huge amount of important work”.

As far as I know, everyone, even outsiders to the theory wars like me, agrees that phonology in the late 1980s needed major reworking, and everyone agrees that constraint-based work has provided real insights. Of the list that Eric gives, conspiracies stand out as the big example for me.

But the complaint you can hear from people who've grown disillusioned with OT (or never accepted it) is that a huge amount of the work in the framework has tended to theory-internal tinkering, rather than research aimed at more general progress on how sounds work. The failure to deal adequately with opacity, for such people, is symptomatic of that. I've been surprised at how many people say such things. One of the best-known phonologists around has apparently bemoaned the failure of OT to produce “deep descriptions of languages”. And others (though not all) have the sense — again, this is reporting something I've heard — that while OT works well for higher-level prosodic stuff, like reduplication, it stumbles too often on issues of segmental phonology that traditional approaches handled easily. So, I think every non-OTer I know sees OT as having brought valuable discussion and progress, but the question of just how much has remained virtually untouched.

Now we get to the image above: Many of us outsiders smelled a troubling air of triumphalism in OT's salad days. I remember vividly hearing young phonologists — people who had and probably still have contributed little or nothing to the field — mocking Kiparsky for pursuing stratal OT.

But we are finally seeing a change. It's good to see Eric sounding so positive about Kiparsky's and Bermúdez-Otero's stratal work (see Challenge I). McCarthy's embrace of Harmonic Serialism will be a positive move to many. Eric's own Phonology piece grapples directly with comparing traditional approaches to opacity with OT approaches, acknowledging some strengths of the former. Maybe we're seeing some rapprochement, and with it more discussion of the grounding of phonological theory, instead of talking-heads-style yelling and posturing. The shocking part, of course, is how rare this has been of late — or so it seems to me. Hell, if it continues, I might do a little phonology some day.

Anyway, I'll post again later to clear up some lingering questions.

Challenge IV
Frankly, this one’s a real puzzler to me. In my view, it requires a lot of (willful) ignorance of a huge amount of important work in the 70s and 80s to think that OT doesn’t make significant progress in many areas (duplication, conspiracies, top-down and bottom-up effects, the emergence of the unmarked, …) where SPE essentially foundered. Yes, one significant consequence of all this progress was at the very least a re-evaluation of the ubiquity, diversity, and explanatory analysis of opacity — and this re-evaluation has led to a lot of interesting work and new developments. Some of it may involve what some consider to be “lots of gymnastics”, but as I’ve noted above, similar (and well-accepted) developments in SPE are arguably no different in this regard.


cassaday rassmussen said...

Yeah! I get to be the first one (I think)...

I think Mr. V and Eric might just be able to hug after this post and Eric's comment on the Opacity Throwdown III.

I've found this whole exchange to be very illuminating because apparently many issues in phonology are much more settled than I thought them to be. Derivations are ok. Morphology is ok. Abstractness is ok. If everyone is on the same page for these things then I'm extremely pleased and tickled with rose petals...

I have some questions to answer so in no particular order.

(g) Ed, I checked the Moreton and Smolensky WCCFL paper and you don't really get my point. Yes, the Moreton and Smolensky paper checks the typology of chain shifts but this is a narrow typological check. I'm more interested in asking the question of what happens to say the famous *NC Pater typology if you're allowed to locally conjoin *NC with say *t or NoCoda or other logically possible but not necessarily sensible conjunction of well attested and accepted constraints.

I can compromise and admit that maybe there is some work on the the typological effects of conjoined constraints if Ed can admit that the existing corpus of work on this topic is incomplete and doesn't justify that the addition of conjoined constraints is all rosey... more of an open question that requires a lot more work to demonstrate that OT is just stretching and not contorting...

(q) To Eric's question, the EC or the SCC... that is a good here are some thoughts on these matters

The Elsewhere Condition

This principle apparently goes all the way back to Panini, can be related to Occam's Razor and seems to resonate with general principles of simplicity in the scientific method. If the EC is actually derived from these general types of principles then I can see it being applied in language acquisition (see Charles Yang's work) and even Pinker's 'rules and words' type of two processes approach is in line with the EC in that the more specific word blocks the application of the general rule. Maybe I'm just blinded by my own human cognition but it appears to me that the EC is so general in the world of science that it would be more of a surprise if any cognitive component did not follow it in someway. Science in general would be really screwed if the EC or its general analogs didn't exist.

The Strong Cycle Condition

So it looks like I'm throwing the SCC under the bus. I like the EC so much that I'm willing to do this but I have some reservations about it though. Basically, the SCC looks a lot like what those crazy minimalist syntacticians are calling the 'No Tampering Condition' and 'phases' in that once a part of a derivation is computed, it is more efficient to not change anything that isn't new. This where my reluctance arises because it feels like the SCC could possibly be derived from the syntax or possibly general principles of computation.

This is the dilemma to this question that I have. Both the EC and SCC are suspect if they only exist in the phonology but it appears that there is cross modular (and possibly cross cognitive) evidence for these two principles. Ah, so I now see the answer...I'd get rid of both of them from the phonology proper and derive them from more general cognitive processes... so ranking of these things is

*SCC > *EC > deriving both from general cognitive processes

or maybe I have that backwards...

(]) It seems that Eric has come full circle. Maybe I haven't been paying attention... if the only thing OT has been replacing or arguing against is

i) that phonological processes result from string rewrite rules and (ii) that those rewrite rules are crucially ordered (within a component; that is, extrinsically)

then I'm really lost. Mr. V has offered up the examples of Tiberian Hebrew (dead) and Canadian Raising (live) which have the characteristics that demonstrate both of the above aspects of phonology are necessary.

If all other issues/aspects/delusions of phonology are in play and can be equal meaning there is no cost for either theory to invoke them then it is extremely difficult to do theory comparison. It also magnifies crucial test cases to be make it or break it kinds of deals. I agree with Mr. V that the opaque interaction of post-cyclic rules in both Tiberian Hebrew and Canadian Raising are these litmus tests which demonstrate the role for extrinsically ordered rules.


Ed Keer said...

So cassaday we both get to be misunderstood. Whoohoo! Anyone who proposes conjoining *NC with NoCoda will have to provide the justification for this universal constraint. A thorough justification would include the typological consequences. That's how one does OT. Or should anyway.

If this doesn't happen it seems more like a problem with the theorists, not the theory. I'm sure there were people doing bad phonolgy with rules too. Or were all phonologists back in the day as upright and serious as yourself?

Anonymous said...

Awwww. I've been waiting around to see a knife fight, and this sounds all 'we CAN all get along.'