Retrospect and p[r]ospect
 

Natural Phonology and Morphology, in Wurzel's (1989: 196-7) words,
investigate the language systems or the components of language systems that are the result of processes of change and that are themselves in the process of changing. To neglect change and changeability in language is equivalent to idealizing away a property that determines its very essence, something constitutive of human language as a whole. Thus, it is not surprising that, given a purely synchronic approach, many grammatical facts defy explanation which, from a historical point of view, are quite explainable.

I have tried to show how Lexical Phonology helps us understand the connections between the phonological past and the phonological present. The dependency between the two is necessarily bidirectional. A constrained, rule-based, derivational model makes testable predictions on the course sound changes follow, as they develop from low-level variation into postlexical and then lexical rules. This is not the same as the recapitulation of history found in SGP, where sound changes and phonological rules were essentially identical: instead, variant pathways into the grammar are determined by issues of learnability and by the constraints on the model. The consequence is that present-day rules, like the Vowel Shift Rules and [r]-Insertion, have altered consider ably with respect to their predecessors, the Great Vowel Shift and [r] Deletion (the latter being, as I showed above, a label for a number of interrelated phonetically and phonologically motivated changes which together set the scene for Modern English linking and intrusive [r]). In all these cases, an appropriately constrained Lexical Phonology helps deter mine the analysis we select in interesting ways, blocking particular options and enforcing others. Apparently arbitrary present-day processes can equally be shown to be explicable in diachronic perspective.

The constraints on Lexical Phonology, namely the Derived Environment Condition, the Alternation Condition, construed as a condition on learnability, and Structure Preservation, are also both synchronically and diachronically relevant and active. The DEC limits the application of Level 1 rules, and determines the point at which alternations must be regarded as fossilized, and hence incorporated into the underlying representations: recall from Applying the constraints: the Modern English Vowel Shift Rule the ex-Vowel Shift alternations, such as food ~ fodder, profound ~ profundity, which the model rules out for Present-Day English, and the related fact that irregular verb past tenses could be derived productively only for the relatively recently developed keep ~ kept type. The Alternation Condition makes a phonology with a rule on Level 1, controlled by DEC, more highly valued from the point of view of learnability than a less constrained model with the same rule on Level 2: this implies (Giegerich in press) that Level 2 phonology will be rather limited or in some languages may not exist at all. This has a further consequence for Structure Preservation, which itself appears to play a diachronic role in encouraging the further lexical diffusion of alternations derived via a Level 1 rule, leading to incipient underlying contrast for SSE /Λi/ ~ /a:i/. Violations of Structure Preservation tend to occur with Level 2 rules, which in this model are low-valued in any case, although they may represent a temporary, new lexicalization of a rule on its way to Level 1. If there are so few truly Level 2 rules (as opposed to actual Level 1 rules which have been ordered on Level 2 in the past to defuse the operation of the DEC), it is no wonder that phonologists have been unable to decide on the role of SP on Level 2.

Representational decisions in this model also have strong derivational consequences. The lack of underspecification means a theory of possible underlying representations is required, and the application of both DEC and the Alternation Condition encourages a restrictive approach: hence the rather concrete underliers adopted here, which typically correspond to the lexical representation of non-alternating forms or the underived member of alternating sets. Rejecting underspecification also reveals the operation of the constraints of Lexical Phonology, making them harder to bypass; and leads to a more realistic approach to dialect variation by allowing historical changes to accumulate at the underlying level, not only in the rules. Historical analyses, based on contemporary evidence, also provide us with maps of variation in English dialects for processes like SVLR and [r]-Insertion; diachronically, each step determines the next, while synchronically, each step is maintained in some variety. Finally, if we incorporate a version of Articulatory Phonology into the model, as argued tentatively in connection with English [r], low-level variation can be modelled using limited gestural manipulations; but when these are interpreted cross-generationally as segment insertion or deletion, the resulting synchronic rule may have to become lexical. Here we find another new perspective on Structure Preservation: it seems on this view that late postlexical, low-level processes really can create new information, but not randomly: new information can arise only if it is motivated in its context. When that context changes, or the insertion or deletion is learned as categorical, truly new gestures may be introduced, contravening Structure Preservation, which is therefore a useful diagnostic of a rule that is lexicalizing, but has not yet penetrated into the underlying representations.

In short, the phonological model presented here attempts to connect synchrony, diachrony, variation and phonological theory in a non-fortuitous way, and to show that all are mutually informing. In devel oping phonological theory, we must consider issues beyond synchronic, variety-specific alternation and distribution on the one hand, and universal constraints on the other. Given a summary as apt as Johansson's (1973: 67), I am happy to relinquish my authorial right to the last word:
It is the object of linguistics to determine how abstract phonology should be and, ultimately, the psychological reality of phonological descriptions. If this problem is to be solved, more evidence of the kind provided by linking and intrusive /r/ must be found. The search for such evidence, which is available in historical change ... first language acquisition, interference in second language learning ... etc., would seem to be a more relevant task than constructing elegant but untestable phonological descriptions.

مواضيع ذات صلة

Lexical Phonology and English /r/

Modelling the past

Using the past to explain the present

Synchronic arbitrariness and diachronic transparency The problem

Alternative analyses Harris (1994)

Alternative analyses McCarthy (1991)

Alternative analyses Scobbie (1992), Donegan (1993)

Alternative analyses

Rule inversion and [r]-Insertion

Non-rhotic /r/: an insertion analysis An orthoepical interlude

English /r/: a brief outline

English /r/ Introduction