The Consequence of Rule Ordering in Haya Tonology

In the 1970's a major debate took place on the question of rule ordering in phonology. One group argued that the specific ordering of phonological rules, if needed at all, was always intrinsic, being predictable on the basis of universal principles. The second group, following in the tradition of Chomsky and Halle and the SOUND PATTERN OF ENGLISH, responded that these principles did not work, and that rule ordering is extrinsic, having to be stipulated in the phonologies of a number of languages. In the course of this debate, the proponents of extrinsic rule ordering sometimes argued that the analyses forced by the universal, intrinsic approach lacked insight, missed generalizations or simply did not work. Curiously, although positions were taken against extrinsic rule ordering and in favor of either simultaneous or random sequential ordering, no one to my knowledge argued in parallel fashion that the extrinsic approach lacked insight, missed generalizations, or simply did not work. In this paper I would like to present one such possible case. I shall attempt to demonstrate that in the lexical tonology of Haya, an Eastern Bantu language spoken in Tanzania, extrinsic rule ordering simply gets in the way. In section 1 I present the relevant tonal data, showing that a classical autosegmental analysis utilizing extrinsic rule ordering runs into serious problems. After showing, in section 2, that various alternative solutions involving rule ordering still fail to overcome these problems, I then consider in section 3 two possible analyses: one with simultaneous application of the three lexical tone rules in question, the other exploiting morphemic planes. I will conclude that this may be one language where simultaneous rule application is warranted. The data come from the lexical tonology of Haya, a subject that was covered in some detail in Hyman and Byarushengo (1984). For reasons of simplicity, I shall present only the underlying and lexical representations of Haya verb forms. It should be borne in mind that the forms cited in this study are subject to subsequent postlexical tone rules that are described in the Hyman and Byarushengo paper