Distributed Morphology Today: Morphemes for Morris Halle

9. Morphemes and Morphophonological Loci

9 9.1 Introduction Some questions in linguistics have persisted through hosts of theoretical changes. The conflict between affixless and morpheme-based theories raises questions of this type. In its contemporary incarnation, at least two significant objections raised against affixless theories are that they (i) render the interface between syntax and morphology opaque, and (ii) have serious difficulties with the analysis of blocking (e.g., Halle 1990; Noyer 1992; Marantz 1992; Halle and Marantz 1993; Embick 2000; Embick and Halle 2005; Embick and Marantz 2008). Nevertheless , the tension between morpheme-based and affixless theories is as relevant as ever (see section 9.3). My objective here is to develop a further line of argument in favor of morphemes and against affixless theories, one that also opens up new questions in the study of morphophonology. Nonaffixal morphological alternations—that is, phonological alternations that are morphologically triggered or targeted—are often taken to provide evidence for affixless theories. In this chapter I develop an argument for the opposite conclusion. The argument is based on the observation that morphophonological changes behave as if they have a morphological locus—that is, they operate in a way that is expected if they are linked directly to a morpheme that has a position (hierarchically and linearly) within a complex word, and act in a way that is (phonologically or morphologically) local to that morpheme . This aspect of nonaffixal morphology is a component of a broader theory of morphophonological locality, one based on morphemes and the principles governing their composition into complex objects (see Embick 2010, 2012). Crucially, to the extent that the correct theory of morphophonological loci follows from a morpheme-based theory of morphology, significant generalizations about morphophonology are missed in affixless frameworks. Ideas along these lines have been advanced in different forms in the literature . In my view, however, these points have neither been fully appreciated, Morphemes and Morphophonological Loci David Embick 152 Chapter 9 nor developed in sufficient detail. I first outline a generalized theory of morphophonological loci in section 9.2, and then illustrate difficulties for affixless theories in section 9.3; section 9.4 concludes. 9.2 A Morpheme-Based Theory of Loci The empirical focus of this chapter is on different types of evidently nonaffixal alternations, of the types illustrated in (1)–(3).1 German umlaut is vowel fronting triggered by several morphemes that have nothing in common, as far as the synchronic grammar is concerned (see, e.g., Lieber 1980, 1987; Kiparsky 1996; Wiese 1996a, 1996b; Embick and Halle 2005): (1) Basic Umlauted Gloss Morphosyntactic feature lauf-en läuf-t ‘run’ 3sg present verb Huhn Hühn-er ‘hen’ plural Vater Väter-chen ‘father’ diminutive Europa europä-isch ‘Europe’ adjective formation hoch höch-st ‘high’ superlative In the Arawakan language Terena, first person singular is realized by progressive nasalization (from left to right), with (simplifying somewhat) the spread stopped by obstruents, which become prenasalized (see Akinlabi 2011 and references cited there): (2) 3sg 1sg Gloss arıne ãr̃ı̃nẽ ‘sickness’ emoʔu ẽmõʔũ ‘boss’ owoku õw̃õn gu ‘house’ ıwuʔı∫o ı̃w̃ũʔn ʒo ‘to ride’ takı n daki ‘arm’ paho n baho ‘mouth’ In the Ethiopian Semitic language Chaha, verbs suffixed with the third singular masculine object marker (3sg.masc.obj) show labialization of the rightmost labializable consonant (Banksira 2000; Rose 2007). The -n morpheme is analyzed as a “case” affix that precedes 3sg.masc.obj, so that the middle column is derived from -“CASE”-3sg.masc.obj (3sg.masc.obj position is marked with △): (3) Without obj 3sg. masc. obj Gloss kәtәfә kәtәfw ә-n-△ ‘chop’ nәkәsә nәkw әsә-n-△ ‘bite’ k’әsәrә k’w әsәrә-n-△ ‘erect’ Morphemes and Morphophonological Loci 153 My primary claim is that essential generalizations about the locality of alternations like those seen in (1)–(3) follow directly in a morpheme-based theory, but not in an affixless theory. An initial statement of the observation to be explained in this way is given in (4): (4) Morphophonological Locus (ML) A morphophonological rule triggered by morpheme X behaves as if the effects of the rule are local to the position of X. The wording in (4) assumes that there are morphologically conditioned phonological rules—that is, that the identity of morphemes is available in the phonology, such that phonological processes may be triggered by certain...