17.01.2002, Nonlinear phonologies

The traditional view of sound structure is based on the following assumption:

The absolute slicing hypothesis:

The basic units of sound structure are phonemes, which are characterised by distinctive features, and are realised by phones; phonemes (and phones) are concatenated one after the other with no overlap.

Autosegmental phonology

However, there are a number of properties of sound systems in the languages of the world which show that this assumption is too simple:

1. Co-articulation, assimilation: phones share features which are influenced by their immediate context; it is often simpler to regard the assimilating feature as being part of a separate, parallel stream of phonetic information which is associated with than one segment, rather than being primarily attached to individual phonemes and phones according to the absolute slicing hypothesis.
2. Vowel harmony: in some languages, features affect not only one segment, but whole words, for instance a word may contain only front vowels, or only back vowels. It is simpler to regard a harmony feature as being separate from the segments and basically associated with the word, spreading over the individual segments. Again, an argument against the absolute slicing hypothesis.
3. Tone: in tone languages, tone functions like a separate phoneme or like a morpheme, i.e. it has a lexical function. Tones are not associated with other phonemes directly, but with syllables or words, and may be associated with individual tones in different ways. Another argument against the absolute slicing hypothesis.
4. Intonation: the sentence-level intonational features of languages are, even in traditional approaches, separated from phonemes and are thus not covered by the absolute slicing hypothesis. A rather obvious argument against the absolute slicing hypothesis.

What remains is a ``skeleton'' of segment types which are highly underspecified in terms of distinctive features, and a set of parallel streams of phonetic information (usually called tiers) with autosegments, i.e. autonomous segments, of different sizes which are associated with the segmental skeleton by means of rules of association subject to certain well-formedness conditions.

Autosegmental phonology is one of the most influential approaches to phonology, and has been developed in a number of directions in linguistics, clinical linguistics, and speech technology.

Approaches to phonology which examine other units than just phonemes are sometimes referred to as prosodic phonologies.

Task:

Construct a lexicon containing definitions, examples etc. of the following terms, using the Clark & Yallop textbook and the web:

1. coarticulation
2. assimilation
3. vowel harmony
4. tone language
5. lexical tone
6. autosegmental phonology
   • well-formedness conditions
   • association rules
   • association lines
   • spreading
   • CV skeleton
7. metrical phonology
8. articulatory phonology (particularly the term gesture in this connection)
9. prosodic phonology

Make sure that you have examples (including graphics) which illustrate autosegmental structures as clearly as possible.