In lieu of an abstract, here is a brief excerpt of the content:

6 6.1 In the Beginning Was the Q The origins of Distributed Morphology can be traced to an argument Morris Halle and I had when I arrived (back) at MIT to teach in the fall of 1990. I came with “lexicalist” assumptions about morphology, as worked out for example in Lieber 1992—not the notion that words were built in the lexicon but rather the notion that lexical items, identified by their phonology, brought syntactic and morphological features with them into the derivation. Morris was working out a proposal that morphemes with suppletive (phonologically unrelated) allomorphs , like the English past-tense morpheme, were “abstract,” as he put it, finding their phonology after the syntax had done its work. Such morphemes he proposed to call “Q” in the syntax, taking over the variable for complex symbols used by Chomsky in Aspects of the Theory of Syntax. On the other hand, morphemes without suppletive allomorphy like the English progressive -ing were “concrete” and came into the syntax with their phonology, here consistent with lexicalist approaches. Although I was not convinced that we should abandon lexicalist assumptions about morphemes and adopt Qs, I was adamant about one point—if some morphemes were abstract, the grammar would be more coherent if all morphemes were Qs (To put it differently, if the only difference between abstract and concrete morphemes was in the multiplicity of spell-out, then concrete morphemes should be treated as abstract morphemes that just happen to have a single realization; i.e., if being a Q did not determine behavior in the syntax, Q-ness should not be a syntactic property of morphemes.) Thus “Distributed Morphology” (DM) in its current form, as associated with Halle and Marantz 1993, 1994, was from its start about the interpretation of abstract morphemes —the building blocks of syntax that found their interpretation in form and meaning in the interpretive components of the grammar.1 Let history show that Morris was right, but that I made him even more right; to paraphrase a folktale famous in linguistic circles, it was Qs all the way down. Locality Domains for Contextual Allomorphy across the Interfaces Alec Marantz 96 Chapter 6 In the Halle and Marantz 1993 theory, Vocabulary Insertion (VI) could be contextually determined, but we did not discuss locality conditions on the size of conditioning contexts for VI that might be principled consequences of the architecture of the grammar. Nor did we envision a melding of DM with Chomsky’s Economy framework, critically discussed at the end of that paper. In the following years, with the advent of the Minimalist Program (MP), DM found a natural union with Chomsky’s approach, and by the turn of the twentyfirst century, the linking of cyclic phase-based syntax within the MP with the architecture of DM was being actively explored. A number of related lines of research on constraints on Vocabulary Insertion sketched out reasonable theories of constraints on contextual allomorphy—that is, choice between Vocabulary Items (VIs = allomorphs) determined solely by context, where the competing allomorphs realize (are the exponents of) the same set of features on a terminal node from the syntax (cf. competing realizations— allomorphs—of the English past tense, all of which spell out just past-tense features on a terminal node). Bobaljik (2000) proposed a mechanism of root-out VI that guaranteed the downward context could see phonological information and specific VIs while the upward context could only refer to grammatical features (where “up” and “down” are relative to the tree-structure representation of a word). Bobaljik’s proposal proved influential as a productive working assumption for research in this area. In the most detailed discussion of the relationship between the syntactic and phonological structure of words to date, Embick (2010) proposed additional locality constraints on contextual allomorphy specifically related to the phonological structure of a word, making reference, for example, to phonological adjacency as defined by concatenation of morphemes. While researchers within the DM/MP universe explored different assumptions about the nature of cyclic spell-out, all work assumed that context would be constrained by phases—that is, only material within a spell-out domain defined by phase heads could be visible as context for VI. Within the MP, some attention has been paid to the issue of whether semantic and phonological interpretation happen at the same time in a derivation, so to speak—that is, whether the spell-out domains for LF and PF are the same. The most natural assumption within...

Share