Abstract
The paper examines conditioned changes that occurred in Miyakoan (mostly Proto-Miyakoan) reflexes of Proto-Ryukyuan close vowels *i and *u after the unconditioned raising of Proto-Ryukyuan *e and *o had taken place. These changes in close vowels are interpreted here as chain shifts. The core assumption is that changes in *i and *u occurred in response to the raising of *e and *o in order to avoid or compensate for the functionally damaging merger of *i/*e and *u/*o.
The paper shows that there is a rather wide range of conditions under which *i and *u produced distinct reflexes in Miyakoan. Consequently, these vowels acted differently after stops, after sibilants, after nasals, in an onsetless/standalone position, after the flap, before the flap, and before nasals and other sonorants word-initially. At the same time, reflexes of both proto-vowels have been observed to maintain certain symmetry, meaning that in a similar environment, *i and *u generally underwent similar or analogical changes.
Thus, the conditions for identifying Miyakoan reflexes or *i and *u are listed and specified in this paper. Conversely, it is argued that unless one of these conditions has been met, one should reconstruct a Proto-Ryukyuan mid-vowel rather than a close vowel. Such specification may influence the comparative study of Ryukyuan languages to a significant degree, challenging a number of the so far established reconstructions (most notably Thorpe 1983).