• Chaha assignment (on last week’s material) due tomorrow; next assignment posted tomorrow • Next reading: Hayes 1989 (study questions due Tuesday) • Meet with me a second time about project by end of this week Overview: Last time we looked at a model of phonology-morphology interaction, Lexical Phonology and Morphology. Today we turn to conceptual issues for the phonology-morphology interface: What regulates morphological affiliation? How broad is the candidate set? 1 What regulates morphological affiliation in OT? Original idea (McCarthy & Prince 1993): “Consistency of Exponence. No changes in the exponence of a phonologically-specified morpheme [i.e., not RED] are permitted ” (p. 21) → epenthetic segments have no morphological affiliation (also, nothing can actually be deleted, only underparsed—this is the containment theory of faithfulness rather than the correspondence theory) 2 But how does reduplication really work? Problems some of you identified in your Samoan homeworks: • /REDaffix+alofa / → [alolofa]: 2 violations of MAX-BR (fa) or 3 (a,fa)? What counts as the base? • Does /REDaffix+alofa / → [alofa] violate AFFIX=SYLLABLE? How do we know whether the output contains an instance of the morpheme REDaffix? • Any other problems you remember... 3 A better theory: Walker & Feng 2004 • There’s an input-output correspondence relation between phonological entities (segments, autosegments, maybe moras...) • But there’s a second indexing for morpheme affiliation (I used superscripts)—imperfections in this indexing are regulate by constraint too Walker & Feng’s Zoque ex. (Mixe-Zoque from Mexico, nearly extinct; data orig. Wonderly 1) input morph. must have output corr. morph a b N 1 + s 2 ɨ b 3 k b