20.12.2001, Phonological rules

The traditional approach in Phonology has always been to concentrate on

• basic units of phonology, e.g. distinctive features and phonemes,
• construction of larger units out of these basic units, e.g. syllables, words, sentences,
• description of the syntagmatic and paradigmatic phonological relations with which phonological structures are made,
• in particular the structures of syllables and words.

However, the discussion of these topics is full of paradoxes until one develops exact analytic criteria and distinguishes between different levels of phonological analysis such as the following:

Speech signal:

articulatory, acoustic and auditory correlates of linguistic units.

Phonetic:

segmentation of utterances into identifiable chunks by detailed phonetic criteria from one or all of the phonetic domains (articulatory, acoustic, auditory).

Phonemic:

segmentation into phones and classification of phones into phonemes according to the criteria of contrastiveness (either complementary distribution in phonetic contexts, or free variation, or both) and minimal phonetic similarity (i.e. using the minimum of phonetic features required to keep phonemes apart).

Morphophonemic:

further classification of phonemes into morphophonemes by taking morphological contexts (i.e. the contexts of sounds across boundaries between morphemes in inflected, derived and compound words) as well as phonetic contexts into account.

The relation between Morphophonemic, Phonemic and Phonetic levels is often thought of as three levels of representation linked by rules (morphophonological, phonological, and phonetic detail rules), as shown in the Figure.

There are two main kinds of phonological rule:

Structure-defining rules:

Structure-defining rules determine the construction of phonemes out of distinctive features, the construction of syllables or morphemes out of phonemes). These are sometimes called redundancy rules, since they formulate generalisations about structure - e.g.: "English syllables can only begin with 3 consonants if the first is an /s/", and enable redundancy in descriptions to be reduced to a minimum. Modern approaches use so-called inheritance hierarchies, in which all the generalisations are linked together. It is important to define the level of representation at which structure-defining rules are formulated.

Realisation rules:

Realisation rules are the "phonological rules proper", and relate the more "abstract" levels of representation to more "concrete" levels: Morphophonemics to Phonemics, and Phonemics to Phonetics. Realisation rules were developed in most detail in Generative Phonology (founded by Chomsky & Halle).

In Generative Phonology, rules were formulated not in terms of whole phonemes which were realised as allophones, but in terms of "bundles" of distinctive features, enabling them to deal with whole classes of phonemes at the same time. In fact, Generative Phonologists discovered that in many cases it is not necessary to propose an intermediate level of Phonemics, but to relate Morphophonemic representations with Phonetic representations via a chain of realisation rules. The interesting discovery was that there may be different kinds of rule chain for different types of sounds, so that the intermediate level of Phonemics between the "abstract level" and the "concrete" level does not need to be defined, and, effectively, Morphophonemes are mapped directly into Allophones.

Tasks:

1. Look for examples to illustrate Morphophonemic, Phonemic and Phonetic representations of the same words.
2. Look for examples to illustrate Structure-defining rules and Realisation rules.
3. Write phonological rules for the following cases, first using phonemes, then using distinctive features:
   • English plurals, as in cats, dogs, horses, using examples from the Christmas Pudding recipe.
   • The e-plural in German, as in Hund, Hunde.
   • The umlaut plural in German, as in Wolf, Wölfe, Mutter, Mütter.
4. What can you find out about the contributions of
   • Morris Halle,
   • Noam Chomsky,
   • Paul Postal and
   • Paul Kiparsky
   to the development of Generative Phonology?