Optimality Theory (OT) is perhaps the newest and most exciting new branch of phonological research. The premise of OT is that there is a set of universal constraints which hold in every language, but each of which can be "violated" if doing so is the only way to satisfy a more important constraint. These contraints are presented in the literature paired with possible outputs for a given input form, in "tableaux" such as the one seen in Figure 1, in which a * in the table shows a violation, and ! indicates a violation too high to allow the form through. (See Prince and Smolensky (1993) for more detail on OT.)

Figure 1: a typical tableau The question that remains unanswered throughout the OT literature, however, is the following. If these things are clearly "tables," why call them tableaux? I propose to answer this question within an OT framework, as follows. First, we need to postulate the following two constraints: Snooty : Be as pretentious as possible when writing an academic treatise.

* French : Avoid using French. (For those unfamiliar with constraints: they are always written in capital letters, a * in front of one indicates "avoid" or "do not use," and the constraints are always abbreviated in semi-comprehensible ways. A better name for "* French " might well be "*FRCH" or such, but we'll let it go.) Given the input form "table," then, we have the following meta-OT tableau (an OT pragmatics, perhaps?). Here, we see that the ranking " Snooty >> * French " will give us the right output form:

Figure 2: the translation from "table" to "tableau" Is this the right way to think about the issues raised here? I think it is, for two reasons: Extendability. Consider the following passage from Fodor (1983, p.89, emphasis mine): ...of its acoustic properties (mutatis mutandis, the linguistic properties of written tokens are uniquely determined by certain properties of their shapes). "Acoustic" properties, according to this usage, are ipso facto transducer-detectable.... Don't worry about what this quote means.(1) Fodor gives us evidence for another constraint, either Latin , "Use Latin wherever possible," or possibly * Latin , "Don't use Latin if you can help it." His overuse of Latin phrases suggests that this new constraint will be needed to describe his dialect, in which * French is mercifully higher. (Compare Bill Reynolds' dissertation (1994), with its use of supra, infra, et cetera.) Universality. An informal survey of a number of the world's languages(2) is enough to show that * French is a near universal constraint; almost all languages have this ranked at or near the top of their constraint hierarchy. There are a few exceptions--English, for instance, which has some borrowings from French; certain creoles based on French (in Louisiana, Canada, and other places); and, at a guess, French. In these languages, * French is ranked too low to have any effect on the language. One might suspect, as well, that second-language acquisition--both acquisition of French, and acquisition of other languages by native speakers of French--involves reranking the * French constraint (higher, for native speakers, and lower, for people learning French). Bibliography to follow, once I get my act together. 1 Frankly, I don't really know what it means. I didn't understand most Fodor (1983) when I read it, and I only read the third chapter. But that's not the issue here. 2 For instance, Chinese, Lardil, Yawelmani, Sanskrit, and that other one, you know, it's spoken in one or two villages, there's only one researcher who's ever looked at it, so we're trusting their data. Further Reflections on Optimality Theory Lance Nathan, 2004 An anonymous reviewer(3) raised an interesting question for this analysis, namely: what is the plural of "tableau"? More properly, this is two related questions, the first of which concerns the orthography and the second of which concerns the pronunciation. When written, the plural might be either "tableaux" or "tableaus". Consensus seems to favor the former (comparing Google hits, a resource only barely available on the first writing of this paper, shows that "optimality theory" tableaux outnumbers "optimality theory" tableaus by a factor of eight), as does usage in the papers of Prince, Hayes, and so forth. This is the expected result: using the French plural violates * French , but satisfies Snooty , as it demonstrates the author's pedantic superiority in forming foreign-derived plurals.(4) Spoken language is another matter.(5) Written papers, a prime source of data for the previous question, are of no help here, which leaves us only to speculate until someone feels brave enough to attend enough talks on OT to collect data. If the word is pronounced like the singular "tableau", then Snooty once again wins out over * French . If the word is pronounced with a [-z] on the end, however, one of two things seems to be happening. First, the speaker may be part of the minority linguistic community which writes "tableaus", for which the [-z] pronunciation is the only sensible one. This should not affect the analysis. Second, the speaker may write "tableaux" but say "tableaus". In this case, Snooty is being violated in favor of * French : the French pronunciation is being avoided, at the risk of sounding less pedantic. This leads to an apparent ranking paradox (which will not, let us be clear, be solved with Sympathy Theory). But all is not lost for the analysis, as there are other factors which may be at work here. One is that Snooty and * French are not the only two constraints in the system (indeed, if they were, we'd pick a ranking and pack up our offices, as there would be no work left to do). There may be another constraint somewhere in the ranking that affects the spoken plural of "tableau". It's easy to postulate some--for instance, Cmprhnsbl , "Be comprehensible to your audience", would ensure that "tableaux" is pronounced with a final [-z] to distinguish it from "tableau". This is clearly not the answer, as Cmprhnsbl tends to be ranked quite low in most linguistics talks.(6) But it may be a step in the right direction. Clearly there is more work to be done here. Those who are singularly interested in the topic are encouraged to explore it on their own, or to assist my own explorations by getting me tenure so that I have time for this sort of thing again. 3 Which means, in this case, that I don't know which first-year grad student wrote it on the whiteboard. 4 Left for further research is the interaction of this sort of snootiness with the above postulated Latin constraint. Do the same people who write "tableaux" also write "agendae" (4300 Google hits, all of them wrong)? To say nothing of "viri" or, worse, "virii" for "viruses". Classical plurals are tricky, and the data gets muddled. Er, get muddled. 5 See also the above footnote, with respect to the pronunciation of "alumni" with an [a i ] at the end instead of an [i]. Whether "phi features" are referred to as "fie" features or "fee" features is a related topic, though outside the scope of this paper (cf "Ways for Papers to Take Scope", elsewhere in this volume). 6 Not in content, though that's hardly immune to incomprehensibility, but in spoken form. Hence the trouble distinguishing, when heard, "Case" and "case"; "D-linking" and "delinking"; "PRO" and "pro"; and so forth. Author's notes, October 2004 The original author's note to this piece read: Nothing above should be taken too seriously, by any means. In fact, I have the highest respect for Fodor, Reynolds, and the formulators of Optimality Theory, and no comment above should be taken as a serious slur on their work. That comment is no less true today than it was five years ago. This piece is, I stress again, primarily humorous. For instance, if I didn't understand the Fodor piece, the failing was mine, not his (and, truth be told, the passage above isn't that hard, though it's made worse by the lack of context). True, it was written in a moment frustration at not being able to remember which one was mutatis mutandis and which one was ceteris paribus, but, honestly, these things are in any good Collegiate dictionary, and not even in the foreign phrase section. You can't fault people for using them. All the same, humor often grows out of a kernel of truth. (Often. I won't swear that that's true for Gilligan's Island , but then, I won't swear that Gilligan's Island is humor.) Is there a core of truth in this paper? Certainly others seem to have found one; one colleague wrote to me, having stumbled across this page somehow (quoted without permission, but also anonymously): I'm curious. French is a real criticism of OT. Why do you continue to work in that theory if you already knew better in 1999? Unless I've missed something truly important, OT is not a falsifiable hypothesis. It can predict anything. Thus, it is not a scientific theory. Additionally, it bears no relationship to any plausible model of brain function. These are real concerns, especially if French is a real criticism of OT. However, in the interest of full credit where credit is certainly due, I had already read Jason Eisner's 1997 handout, "What Constraints Should OT Allow?". In it, he suggests that there are any number of "clearly bad constraints", such as Palindromic : The candidate reads the same backwards as forwards and MatchesOutputOfSPE : The output matches the result of applying Chomsky & Halle (1968) to the input. French is not entirely unlike these constraints. The goal of Eisner's paper is to rule out such constraints (more properly, it's to rule out the ones that seem bad without being egregious, such as Generalized Alignment). Do I think that OT is not a falsifiable hypothesis? I fear that it can be; that OT done poorly is subject to exactly the criticism leveled above. But then, anything done poorly is subject to that criticism; rule-based phonology is hardly immune to the writing of spurious rules. OT is a tool. You can slice bread with a knife, or you can slice your finger off; you can drive to work in your car, or you can drive into a telephone pole. All tools should be used sensibly. The kernel of truth to this "paper" is that OT, like everything else, must be approached sensibly, with care taken to ensure that nonsense isn't written and that the theory remains, as much as possible, predictive and falsifiable. In my opinion, Eisner's Primitive Optimality Theory is a solid step in that direction, and the reader is referred to my own phonology generals paper for further thoughts on how to use OT sensibly. Or, to put it another way: if you think this is really a knockdown argument against Optimality Theory, you should read the Metalleus papers or some of the work in Studies out in Left Field . Just because you can do something silly with a theory doesn't make it wrong. And now that you've read this, please feel free to browse my serious academic papers.