When History Doesn’t Repeat Itself

Abstract

AbstractIn the burgeoning literature on Optimality Theory, two criticisms of rule-based models are frequently made. First, such models involve both rules and constraints, while OT has only the latter, and consequently escapes potential problems of duplication. Second, there is no necessary difference in complexity between a rule describing an extremely natural and common process, and its highly unnatural inverse, so that plausible and implausible processes cannot be distinguished by rule notation. Since many OT analyses now involve mechanisms that mimic rule applications or interactions, or even rules themselves, the first of these arguments has become less clear-cut, so that the second is increasingly important. However, although one side of the argument (that implausible or impossible changes can be described in rule form) is incontrovertible, the corollary is general left unspoken. In other words, it is not generally made explicit that OT cannot model such impossible or implausible changes, and without a demonstration of this sort, the superiority of OT in these terms cannot be asserted.In this chapter, I shall show that at least one highly implausible sound change can be modeled equally readily in OT and in rule-based terms. The change in question is Palatal Diphthongization in the West Saxon dialect of Old English, hypothesized on the basis of regular orthographic alternations between West Saxon digraphs <ie> and <ea>, with <e> and <ae> in other dialects, following consonant graphs signaling palatals. However, this putative change is both anomalous in its context (since it would involve the addition of a back offglide to a front monophthong, after a palatal, or front consonant), and inconsistent in its effects on the input segment (since the reflexes of the low monophthong would be raised, but not those of the mid one). The peculiarities of this development are such that it has typically been rejected by phonologists, in favor of either an alternative sound change analysis, or a purely orthographic interpretation.The unlikely version of Palatal Diphthongization therefore becomes an excellent candidate for an implausible or impossible change, and I shall show that this can be modeled both using rules and constraints. In the case of the rule-based model, this is no surprise: rules are simply post hoc, descriptive tools, employed to capture the facts of a particular development. If the facts are unclear, we cannot rely on rule formalism to adjudicate, but must turn to the phonetics, for instance, for information on the relative likelihood of changes. On the other hand, OT is strongly typologically based, and it is therefore a serious issue if a change that is generally agreed not to have happened can nonetheless be modeled in OT terms. It is true that some of the constraints proposed are ad hoc, but in the current state of play, it is hard to see how these are to be identified as such, or how they are to be excluded. If the superiority of OT over rule-based formalism is to be demonstrated, this issue is a priority for discussion and resolution.KeywordsOptimality Theoryrulessound changeconstraintsnaturalnessOld EnglishorthographyPalatal Diphthongization