The rise of generative phonology

In the 50s and 60s of the last century, with the computer revolution gaining, there emerged a necessity to develop man-machine interaction through speech synthesis, automatic speech recognition and text-to-speech analysis. The phonological theory was to explain how speech is actually produced and understood by humans. Besides, it had to be presented in a formalized way to be fed into the computer. Phonology again was viewed as part of lan­guage which is closely connected with syntax, semantics and, most of all, morphology. Chomsky (1964), a well-known American linguist, is a major critic of "taxonomic phonemics" (descriptivist phonology). His model was aimed at demonstrating how utterances are generated from the "deep structure" to the "surface structure" through the operation of certain rules.

The basic concepts of generative phonology are the underlying repre­sentation, the phonetic representation and the rules which are used to transform one into the other. For example, neutralization of word-final obstruents in Russian and German makes voiced and voiceless oral stops sound the same: Bund ('union') — bunt ('coloured'), столб— столп. However, for the speakers of the language these are different morphemes which are transmitted into the phonetic representations by the rule of word-final devoicing. The implication is that statements of allophonic distribution should be replaced by general phonological rules or processes that can explain the observed complexity of phonetic representations.

Applying the rules in a particular order is very important, too. In many varieties of English, for example, vowels are nasalized if a nasal consonant follows. The nasal consonant maybe retained, as in hand [haend] or absent phonetically, as in can't [kat]. In the latter case the nasalized [a] in can't creates a contrast with a non-nasalized a in cat. Nasal vowels may be as­sumed to be phonemic in English if the phonological analysis consists of constructing exactly two levels of analysis: the phonetic level as transcribed by the phonetician, and the phonemic level, constructed on the basis of contrasts and complementarity at the phonetic level. However, nasal vowels contrast with their non-nasal counterparts only before voiceless conso­nants. In other environments the nasalization is completely predictable. The analysis involves the two rules applied in this order:

a. Vowel Nasalization b. Nasal Consonant Deletion

V>[+nasal]/_[C+nasal] [C+nasal]>0/[V-tense]_[C-voicedJ

These two rules in this order account for the distribution of nasal vowels in English. The vowel of both hand and can't is nasalized by the rule. The nasal consonant is deleted in can't by the second rule, giving rise to the su­perficial [kat - kat] contrast. This order of rules cannot be reversed because otherwise the conditions for Rule 2 application will be destroyed. An interesting point here is that the nasal consonant causes a phonological change without being phonetically realized. Nevertheless, because it is transparently related to the modal can, its abstract nature is motivated. The underlying mental representation of [kant] can't is supported by the orthographic form.

The publication of Chomsky and Halle's "The Sound Pattern of English" (1968) is a landmark of both phonological theory and the phono­logical description of English, whose sources were distinctive feature the­ory and generative grammar. Generative grammar recognized the implications of syntactical and morphological representations for phonology. Chomsky and Halle tied phonology to syntax, claiming that the job of phonology is to interpret the surface syntactic structure phonetically. This surface syntactic structure in turn is derived by inserting lexical items into constituent structure trees, which may have to undergo various sorts of transformations before deriving the surface syntactic structure on which phonological rules can operate.

The paper evoked criticism for the underlying forms of morphemes and vowel alternations rules in sane — sanity, serene — serenity. We will not discuss it in detail but we could just say that the authors had to introduce a lot of rules to account for the historical vowel change and "irregular" forms of past tense, while their critics doubted and tested the psychological reality of these underlying representations in the minds of our contemporaries.

Much work on generative phonology is devoted to the analysis of English stress, both in words and phrases. An important principle of stress assignment is the phonological cycle. This is the idea that the same set of ordered rules can apply several times to the same form, first to the innermost morphological constituent, then to the next larger constituent, and so on, until the maximum domain is processed. They denote primary stress by [1 stress]. Lower degrees of stress are denoted by successively larger numbers, e.g. [2 stress] for secondary stress. According to Stress Subordination Convention whenever primary stress is assigned within a domain, all previously existing stresses within that domain are reduced by one degree. For example, in a word theatricality this system first assigns primary stress to ^theatre, then assigns primary stress in ²the^[atrical, at which point the originally assigned primary stress is reduced to secondary, and finally assigns primary stress in ³the²atri¹cality, where both previously assigned stresses are reduced by one degree.

The same conventions apply within phrases, so that ^гЫаск ^lboard (board that is black) appears with the indicated stresses. A phrase like had ^plight has that contour in isolation but in the following sentence it has the stress contour ^%sad ^xplight because the surface structure indicates that the word plighttermimtes no less than seven phrases to which the Nuclear Stress Rule (assigning primary stress to the rightmost member of a phrase) applies, so that primary stress is assigned to plight on seven successive cycles: My friend can't help being shocked at anyone who would fail to consider his sad plight.

But the internal relations of stress in sad plight are the same in the sentence as in isolation. Recent work on English stress has retained the principle of the cycle for word-internal stress, but uses different means for explaining the stress contour in phrases.

After "The Sound Pattern of English" by Chomsky and Halle a number of improvements, changes and complete reanalyses of various aspects of English phonology were offered. Three of these challenge the "linear" character of Chomsky's model: Autosegmental Phonology, Metrical Phonology and Prosodic Phonology. They belong to Nonlinear Phonology. Another model, Underspecification Theory, challenges the assumption that phonological representations are fully specified, and removes as much predictable information from underlying representations as possible. Finally, Lexical Phonology has the effect of enriching the formal apparatus of phonological rales and includes a specific proposal for integrating phonology with morphology.

Дата добавления: 2014-01-11