Добавил:
Upload Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:
Phonetics_Lecture_2.doc
Скачиваний:
0
Добавлен:
01.05.2025
Размер:
189.95 Кб
Скачать

2.5.2. Autosegmental phonology

Autosegmental phonology (the term is a blend of autonomous and segmental) was to represent tone (and most languages of the world are tone languages) apart from segments (morphemes) with which tone can be associated.

That was a challenge to the 'linear' approach which treats tones as part of the segmental features defining each segment. The idea is that tone is represented on a separate tier, and that words get their tones by association. There are rules of association: tones are associated to vowels one to one and left to right; if any vowels receive no tone by this procedure, the tone to their left spreads to them; if tones are left over after all the vowels are associated with tones, then the remaining tones attach to vowels already associated with tones, producing contour tones. For example, if there are five tones high, low Jailing, rising, rising-falling in a tone language, and these could appear on morphemes of one, two, or three syllables, then simple tones (high, low) are represented as a single specification, contour tones are a sequence of simple tones: falling is the sequence high-low; rising-fallingis the sequence low-high-low. Thus a falling tone may be divided among the two syllables of disyllabic forms but realized on the single vowel of a monosyllabic form.

Autosegmental formalism also helps to describe those cases in tone languages when a segment is deleted but the tone remains to be associated with another word.

This theory was extended to other features spreading over segments, vowel harmony, for example, in Turkish. The vowels of stems are fixed either as [+back] or [-back]. Then the feature spreads over to suffixes. Thus each word is associated to a single value of [+back], for instance.

In most recent versions of the theory all features are claimed to be autosegmental, but arranged according to a strict hierarchy. The feature hierarchy is an area of ongoing research. In the Russian language, for example, palatalization/velarization may be viewed as a case of a feature [+palatalization] spreading over two adjacent segments, a consonant and a vowel: [CV+palatalized]. The exact disposition of other prosodic features in the hierarchy of features is not resolved yet. For English two examples may be of interest: an affricate may be represented in terms of autosegmental phonology as two values, [—cont] and [+cont[ associated to a single timing slot; a long vowel may be associated with two timing slots.

Autosegmental phonology has not had much influence on English phonology, perhaps because English has neither tone nor vowel harmony. Nevertheless, it has been very influential in the recent development of phonological theory.

2.5.3. Metrical phonology

The development of metrical phonology was motivated by the fact that stress was the only feature in generative phonology which was not binary, and the number of possible values was unlimited. As we have shown in 2.5.1, stress was assigned to words and phrases, cycle after cycle, according to stress subordination convention, and in principle there was no upper limit on the length of words and phrases with only one primary stress. For example: in ^black-board, black has primary stress and board has sec­ondary stress. If we expand this to lblack3board e2raser, black will retain primary stress, eraser now has secondary stress, but board is reduced to tertiary stress. Since there is no limit on the size of compound nouns pro­duced in this manner, there can be no upper limit on the possible values for the feature [stress].

Liberman and Prince (1977) found the solution for this problem by showing the relative degrees of prominence and keeping the selection of strong/weak stresses binary. They proposed metrical trees, with binary branches, such that every node of the tree would dominate a strong node (designated s) and a weak node (w). Such branching nodes could be embedded within other branching structures, thus capturing the facts of stress subordination without changing stress numbers. For example: blackboard has the metrical structure [sw]. Blackboard eraser has the metrical structure [[sw\ w]. The labelling within blackboard does not change just because it is embedded in a larger structure. Thus metrical relations remain binary and local. It will remain [sw] for blackboard and [ws] for sad plight (See 5.1), no matter how deeply they are embedded in a long sentence.

Liberman and Prince retained the feature [stress], reducing it to a binary feature in line with other features. This accounts for contrasts such as lgym2nastvs. ltem<ipest, where both are labelled [sw]. The difference is that gymnast has two syllables labeled [+stress], while in tempest only the first syllable is so marked. Later research in metrical structures (Selkirk 1980) eliminated the feature [stress] altogether by introducing the concept foot. A foot can be defined as a unit containing exactly one stressed syllable. Under this conception, gymnast has two feet, while tempest has only one.

Besides, metrical structures proved to be useful to describe a few segmental phenomena as predictable in terms of foot structure, such as the occurrence of aspiration in voiceless stops before a stressed vowel. In terms of metrical phonology the rule can be formulated as 'voiceless stops in English are aspirated at the beginning of a foot'.

Metrical phonology once again drew the attention to the key problems in English phonology, stress and rhythm, with their basic principle of strong/weak elements alternating in speech, earlier described by David Abercrombie and many other British authors. In Russia English rhythm was described in a less formalized way but within a much wider domain of text linguistics, in a number of stylistic varieties, with a hierarchy of rhythmic units starting from a syllable, through foot and intonation group called "syntagma", to supra-phrasal unities and the text (Antipova 1984). This approach bears much similarity to what is now described as "prosodic phonology".

Соседние файлы в предмете [НЕСОРТИРОВАННОЕ]