Phonology: case studies

The analysis of the palatalization pattern within Optimality Theory (OT)

From the perspective of OT, a complementary distribution always involves an interplay between context-free segment markedness and context-sensitive markedness of a sequence or representational array.

The contextual class of sounds in a complementary distribution is generally more marked than the default class. In the case of palatalization, it is certainly the case that palatal stops are less common in the world's languages than velar stops, and that palatal stops usually occur in languages that also have velars. We can express this typological fact with a segmental markedness contraint:

    *PAL-C: If a segment is [+high, +cons], then it is also [+back].

This constraint penalizes dorsal consonants that are front. It penalizes them regardless of where they occur in the representation.

The contextual class of sounds is marked, but there is a sequence involving the default class of sounds that is also marked. In the case of palatalization, that is a sequence of a velar followed by a front vocoid. The avoidance of this sequence can be expressed by the following context-sensitive markedness constraint:

    PAL: A [+back] consonant must not occur before a [-back] vocoid.

This constraint penalizes sequences of a back consonant followed by a front vowel.

These two markedness constraints conflict with any faithfulness constraint that opposes a change that could get rid of the marked structure. For example, one way to get rid of a palatal consonant would be to delete it. This would violate the faithfulness constraint MAX-IO (McCarthy and Prince 1999: 294):

    MAX-IO: Every element of the input must have a correspondent in the     output.

This constraint penalizes segment deletion. Another way to get rid of a palatal consonant would be to change its features so that it wasn't palatal or it wasn't a consonant. Such changes would violate constraints of the IDENT family (McCarthy and Prince 1999: 294, Padgett 1995):

    IDENT-[high]: Corresponding segments must have identical     specifications for [high].

    IDENT-PL: Corresponding segments must have identical specifications for     the place features.

    IDENT-[cons]: Corresponding segments must have identical     specifications for [cons].

Finally, the sequence of a velar followed by a front vowel could be avoided by switching their order. This would violate the constraint LINEARITY (McCarthy and Prince 1999: 296), which opposes metathesis.

Now that we have the relevant constraints we can consider the required ranking. To capture the fact that palatal consonants are generally avoided (except in one particular contexts), the segmental markedness constraint (*PAL-C) must dominate a faithfulness constraint. The two classes differ in [back] and [cor], so the dominated faithfulness constraint must be the one that opposes to these features: IDENT-PL. All the other conflicting faithfulness constraints must dominate this one:

    *PAL-C, MAX, IDENT-high, IDENT-cons >> IDENT-PL

This ranking is illustrated in the following tableau.

Input: care	*PAL-C	MAX	IDENT-high	IDENT-cons	IDENT-PL
care	*!
are		*!
tare			*!
jare				*!
→ kare					**

The ranking so far rules out palatal consonants altogether. Since such consonants do occur in Kinyarwanda, there must be a constraint dominating *PAL-C, and in this case it is the context-sensitive constraint PAL:

    PAL >> *PAL-C >> IDENT-PL

This ranking insures that palatal consonants can occur just in case their occurrence would eliminate a sequence of a velar consonant followed by a front vowel, the sequence opposed by PAL. This is illustrated in the following tableau:

Input: umugezi	PAL	MAX	LINEARITY	*PAL-C	IDENT-PL
umugezi	*!
umuezi		*!
ugumezi			*!**
→ umuɟezi				*	**

The complete ranking is thus expressed by the following Hasse diagram:

Generalizing, the following schematic ranking will always yield a complementary distribution:

References

McCarthy, John J. and Alan Prince (1999). Faithfulness and identity in Prosodic Morphology. In René Kager, Harry van der Hulst, & Wim Zonneveld (eds.) The prosody-morphology interface. Cambridge: Cambridge University Press. 218-309.

Padgett, J. (1995). Feature classes. In J. Beckman, L. Dickey and S. Urbanczyk (eds.) Papers in Optimality Theory. University of Massachusetts Occasional Papers 18. GLSA, Amherst. 385-420.