Phonology: case studies
By Scott Myers and Megan Crowhurst
Department of Linguistics at the University of Texas
Derivational analysis of the vowel harmony pattern
A neutralization distribution involves two classes of sounds: a restricted class that can occur various places but not in one particular context, and an unrestricted class that can occur in that particular context as well as others. Given such a distribution, there are the following rules for analysis in a derivational framework:
- Since the two classes of sounds are in contrastive distribution, both belong to the set of phonemes and occur in underlying representations.
- Alternating morphemes must have a member of the restricted class in its underlying representation, since a member of the unrestricted class would not change according to context.
- There must be a phonological rule changing members of the restricted class to the unrestricted class in the context where the restricted class can't occur.
In the case of Kinyarwanda vowel harmony, both vowels must be included among the phonemes (/i, e/). They must both be included in underlying representations, in order to represent contrasts such as [kuɟeɾaʰ] vs. [kuɟiɾḁ] (from /kugeɾa/ and /kugiɾa/, respectively).
The suffixes that alternate between mid and high, however, must have the high vowel in their underlying representation, since that is the restricted class: /-i:ʃ-, -iɾ-/.
There must be a phonological rule that changes the restricted sounds (high vowels) to the unrestricted ones (mid vowels) in the context where the former cannot occur (where the preceding vowel is mid):
V → [-high]/ V Co _____
[-high, -low]
To the left of the arrow, the target specifies what class of sounds is subject to this rule, i.e. vowels. To the right of the arrow is the difference between the two alternating sounds: the mid vowels are identical to the corresponding high vowels except that they are [-high]. In the context, the preceding vowel must be mid, i.e. [-high, -low], and there can be any number of consonants intervening (C with a subscript zero denotes a string of C's of any length, including zero).
The analysis is illustrated in the following derivations:
| Underlying representation | ||||
| Vowel harmony | -------- | -------- | ||
| Surface representation |
Both neutralization distribution and complementary distribution (e.g.
- V is a shorthand for [+syll] or [+syllabic].
- [+syll] or V denotes the set of syllabic sounds. This includes vowels and syllabic consonants.
- [-syll] or C denotes the complement set of nonsyllabic sounds.
- [+high] denotes the set of sounds produced with the tongue body raised close to the palate. This set includes high vowels, and velar and palatal consonants.
- [-high] denotes the complement set, including mid and low vowels, nondorsal consonants such as coronals or labials, and the further back dorsal places of articulation, uvular and pharyngeal.
- C is a shorthand for [-syll]. It denotes the set of nonsyllabic sounds.
- The 0 subscript indicates that there can be any number of C's, from zero (the lower bound) to infinity.
- C with the subscript 0 indicates that the intervening consonants do not matter - what matters is the closest preceding vowel.
- [+low] denotes the set of sounds produced with low tongue position, close to as low as the tongue can get. This set includes low vowels, pharyngeal consonants, and pharyngealized consonants.
- [-low] denotes the complement set. It includes mid and high vowels, palatal, velar, and uvular consonants, and consonants made without significant tongue body constriction such as coronals and labials.