Show me the magic

A couple months ago on her blog Ilani Ilani, Harvard linguistics student Bridget Samuels quoted the following from “Andrea Calabrese’s new manuscript, Markedness & Economy in a Derivational Model of Phonology, which you can download here.” (That’s a link to an index of “pubblications [sic] and work in progress” at the “Interdipartimental [sic] Center of Cognitive Studies on Language” at the Università di Siena; here‘s the direct link to the zipped .pdf file of Calabrese’s book manuscript.)

[A]n idiosyncratic and contradictory core, the product of history and its inescapable whims, will always remain. Linguists who deny this core and attempt to provide a synchronic explanation to all aspects of the phonology of a language– a common attitude, especially in OT– behave a little bit like individuals who, when faced with the painful contradictions of reality, retreat into magical thinking and try to give sense, through mysterious correspondences, to what is otherwise a broken, shattered and meaningless existence.

Let me start out by saying that, after downloading this manuscript and taking a look at some of what it covers, I have every reason to be interested in reading it. I’ve always liked Calabrese’s work; his dissertation influenced some of my thinking as I wrote my own dissertation. But there’s something truly shameful in tossing off a claim like the one quoted above.

There can be no doubt that Calabrese has done no sort of quantitative analysis to back up the claim that this “attitude [is] especially [common] in OT”. Aside from the fact that there’s no evidence provided to back it up, the same attitude was and is common enough in derivational phonology to take notice. There’s no denying that there was a whole lot of “magical thinking” going on in SPE, for instance, and in Ted Lightner’s work, and I’m sure Morris Halle still believes that the Great Vowel Shift is to be accounted for synchronically, and that (quoting Chomsky) “the rules deriving the alternants decide-decisive-decision […] are straightforward and natural at each step.” Highly abstract analyses like these all have their roots in (the beliefs of practitioners of) rule-based SPE, not in (the beliefs of practitioners of) OT.

As I commented on Bridget’s post, I can’t think of any particular piece of (influential) work in OT that uniquely or originally attempts to capture a synchronically spurious regularity like this that has a better diachronic explanation. I wonder whether Calabrese can actually cite one.

[ Update: Bridget has responded to my comment on her post, citing “a lot of OT attempts” to make sense of “epenthesis of synchronically arbitrary consonants”. ]

Even if a reliable quantitative analysis could be done that would tell you where the relevant attitude is more prevalent, in OT or in derivational phonology, what would the results tell us? Something about OT, something about derivational phonology, or something about (some subset of) the practitioners of these theories? Perhaps a little of all of these, but the results would certainly tell you nothing about OT as opposed to other phonological theories.

7 thoughts on “Show me the magic

  1. Nathan Sanders

    I have to agree with a (very small) portion of the Calabrese quote, that denying history and trying to provide a synchronic explanation for idiosyncratic interactions of long-dead sound changes is A Bad Thing. That was one of the primary points of my dissertation, in fact. It’s a shame he didn’t read it—he’d have found an OT analysis which does with historical facts what he seems to claim should be done with them (leaving them out of the synchronic phonology).

    But thumbing through his book, it seems that Calabrese isn’t following his own advice. For example, early in the first chapter, he claims that Polish raising of [o] to [u] before word-final voiced oral consonants requires a synchronic rule, despite admiting that it is both unnatural (just look at it!) and shrouded in historical opacity (there was lengthening, then raising, then shortening, with some other complications thrown in, such as the effects of nasal codas and word-final devoicing of obstruents).

    However, [o]-raising is in fact not a productive rule in Polish: there are numerous lexical exceptions, and native speakers fail to apply [o]-raising to made-up words that should raise. (Again, he should have read my dissertation, since this is the topic of Chapter 2.) Like irregular plurals in English, Polish speakers must learn, word by word, when to raise and when not to. That is, raising is a property of the lexicon, not of the productive, synchronic phonology. Any definition of “productive” which allows for lexical exceptions and failure to apply to new words seems like a strange (and somewhat useless) definition to me—what would it mean to be “unproductive”?

    Pushing this further, I’m of the belief that all cases of opacity and other “unnatural” phonological generalizations (those that truly cannot be grounded in articulation, acoustics, or cognition) are not fully productive, applying only to specified existing stems or at a limited set of particular morpheme boundaries. I’m still trying to find an unquestionable counter-example…

    Back to Calabrese, he’s just following a path long set in phonology of “attempt[ing] to provide a synchronic explanation to all aspects of the phonology of a language”. I’m just a bit surprised that he’s trying to lay the blame for this mentality at the feet of OT, when as you point out, it dates back at least to SPE. Worse, however, is that it is still present in Calabrese’s own work, work in which he decries this very mentality, as quoted above.

    “Painful contradictions” and “magical thinking”, indeed!

  2. Eric Bakovic

    I’m so glad that Nathan commented on this post, not only because he points to his very relevant work on this overall topic, but also because he took the time to find that there’s arguably a residue of magical thinking in Calabrese’s own work, mere pages from the original quoted text above.

    (Personally, though, I’m not yet willing to give up on at least some kinds of opacity.)

  3. Pingback: phonoloblog » MJ and OT

  4. Travis Bradley

    At least the quote from Calabrese acknowledges that OT exists. A certain “famous Spanish phonologist X” from MIT (to borrow an expression from Eric) labels his own account of Spanish tap and trill as "the generative consensus" and "the standard analysis" without even considering any OT accounts as viable alternatives. Personally, I find the opinion expressed by José Ignacio Hualde to be much more "fair and balanced" if you will:

    Although [the Spanish phonologist in question–TGB] refers to the one-phoneme analysis as “the generative consensus” and “the standard analysis”, from the references above [Alarcos 1965, Quilis 1993, Bakovic 1994, Bonet and Mascaró 1997, Bradley 2001, Padgett 2003, D’Introno et al. 1995] it should be clear that there is really no consensus on this analytical matter (Hualde 2004:387).

    Of the works that Hualde cites as supporting the "two-phoneme" analysis, three are cast within some version of OT, namely Bakovic 1994, Bradley 2001, and Padgett 2003, and to these we can also add Heather Robinson’s (2001) contribution to the RuLing Papers 2.

  5. Eric Bakovic

    This just in: according to this Linguist List posting, Calabrese’s book has just been published (in hardback for $130+; no thanks, I’ll stick with the zipped .pdf manuscript). Here’s the summary blurb:

    This book proposes a new model of phonology that integrates rules and repairs triggered by markedness constraints in a classical derivational model. In developing this theory, the book offers new solutions to many long-standing problems involving syllabic and segmental phonology with analyses of natural language data, both well-known and and relatively unknown. The book also includes a new treatment of Palatalization and Affrication processes, a novel theory of feature visibility as an alternative to feature underspecification and an extensive critique of Optimality Theory.

    I invite anyone to review this book for us. As a reward, you’ll get a free subscription to phonoloblog

  6. Gabriel Poliquin

    This is a bit of a late response to the discussion above, but I happen to be interested in these issues too. I side with Eric in not giving up all of opacity. The thing to do is not throw the baby with the bathwater, and keeping a non-ideological mind about things like this. If you can explain a case of opacity with diachronic data AND show that it is no longer productive, you’ve got a good case to say that it shouldn’t be accounted for synchronically. Nathan does this excellently in his dissertation. But if you can show the opposite to both those things: you’ve got a case of opacity for which there is no evidence of historical rule layering AND it is fully productive, this is a good case that you’re dealing with a synchronic case of opacity, which should be accounted for. Whether you do this in a rule-based or constraint-based framework really doesn’t matter at this point. I agree with Eric and Nathan that no framework is guilty of being more or less careful regarding this topic. On the one hand, right there’s the synchronic account of the Vowel Shift in SPE, on the other, there’s Sympathy in OT. . .

    Check out my upcoming dissertation (and related papers) for what seems to be a true case of opacity. . .

  7. Pingback: phonoloblog»Blog Archive » New phonology books

Leave a Reply

Your email address will not be published.