Edited by Michela Cennamo and Claudia Fabrizio
[Current Issues in Linguistic Theory 348] 2019
► pp. 11–26
According to Pedersen (1913) and Thurneysen (1946) Old Irish had three different consonant qualities: i-colored/palatalized (before original front vowel), u-colored/labiovelarized (before original u-vowel), and neutral (elsewhere). The first two are normally indicated by preconsonantal i and u respectively, functioning as diacritics.
Greene (1962, 1973, 1976) rejects this view, claiming that systems with three different consonant qualities are rare; “u-color” has little functional load; later Irish only distinguishes i- and neutral quality; and u constitutes the second element of a true diphthong.
None of these arguments and claims are cogent. Rarity does not equal impossibility; functional load is an unreliable criterion; many languages have all three “colors”; orthography and the parallel evidence of Avestan argue for treating u as marker of labiovelarization; Greene’s “diphthongal” account is historically problematic; and finally, certain phonological and analogical changes can only be explained by accepting an account that interprets u as marker of labiovelarization.