|Title:||English Voicing Assimilation: Input-to-Output [+/-voice] and Output-to-Input [+/-voice]|
|Authors:||Awwad Al-Ahmadi Al-Harbi|
|Abstract:||An explicit formal account for obstruent voicing alternations in English
when onset-specific faithfulness is irrelevant is provided within the framework of Optimality Theory. We will argue that we must break the identity constraint that specifically targets obstruent [voice] into [+voice] and [-voice] constraints and elaborate these identity constraints somewhat, so that we have a way of stating that in regular voicing assimilation an input obstruent maps to an output one, while in case of variation in voicing alternation in output an output obstruent maps to an input one. Our analysis shows that variation data can be modeled by a grammar with constraints on output-input relations. OT's usual alternatives to variation, free ranking and tied violations, have proved inadequate to handle the English facts. The respective patterns provide evidence against the claim that [voice] is a privative feature. The English language data show that [voice] is a binary valued feature and underspecification is a third option in English.