“…In order to account for the irregularities and to provide a phonologically plausible explanation for the reduction, Rix () proposes that, instead of being reduced directly to the resulting vowels i and e , medial vowels were first reduced to schwa ( ə ); at a later stage, the reduced vowels were reinterpreted as the already existing neighbouring full vowels, which remained in the accented syllables. Rix's scheme is followed, for example, by Meiser (: 66f), but it was elaborated and extended by Nishimura (, , , ). In closed syllables and before r , ə always turns into e (e.g.…”