Shifting Domains The notion of a cyclic derivation, defining (sub)-domains in a grammatical derivation to which rules apply dates from some of the earliest work in modern linguistics, and is a recurring theme in the work of Morris Halle (see, famously, Chomsky and Halle 1968). In phonology, it is recognized that not all morpheme concatenation triggers cyclic rule application, but that certain morphemes are designated triggers of cyclic rules (see, e.g., Halle and Vergnaud 1987 for one

more »

... A related idea pervades the history of syntax, holding that there are cyclic domains defined (at least in part) with reference to particular heads/projections, and that these cycles enforce * For feedback on some of the ideas presented here, we are grateful to audience members at ETI 1 (McGill), ZAS (Berlin), NELS 42 (Toronto), to seminar participants at UConn and MIT, and to the editors and reviewers for this volume. Bobaljik and Wurmbrand locality conditions on syntactic dependencies (see, e.g., Ross 1967; Chomsky 1973 Chomsky , 1986 . Within the intermodular perspective of Distributed Morphology (Halle and Marantz 1993) various authors have raised the question of how the domains (e.g., cycles, phases) of one module (syntax, morphology, semantics) interact with those of others (see also Scheer 2008 and related work). In this short paper, we explore one small aspect of this large puzzle. Specifically, we propose a general rubric that allows for some slippage in otherwise well-established locality domainscases in which a well-motivated cyclic domain appears to be suspended, allowing dependencies to span a larger structure than they normally may. To the extent that this is on the right track, it bolsters arguments that cyclic domains constrain the locality of operations across modules and thus constitute a deep property of grammar. Specifically, we suggest that the following Domain Suspension principle holds across modules, and present two applications, one from suppletion in morphology, the other from quantifier raising (QR), suggesting the potentially broad applicability of the principle. 1 (1) In the following configuration (linear order irrelevant), where the projection of Y would normally close off a domain, formation of such a domain is suspended just in case Y depends on X for its interpretation. 1 To keep within the scope (and page limits) of this short paper, we do not consider other domains here, including, for instance, interactions across phonological domains (as raised by reviewers). Suspension across domains Although (1) could be implemented in various ways, we conceive of suspension not as an operation, but as a condition restricting (or defining exceptions to) the algorithm(s) that determine(s) derivationally whether a given maximal projection will or will not constitute (or close off) a domain. Various terms in this general scheme (notably 'domain', 'interpretation') are relativized to some extent, to the module under consideration, accounting for a slight difference in the ways in which (1) plays out in the different components of grammar. For the cases to be considered here, the algorithms subject to Domain Suspension in the structure in (1) include: (2) a. Morphology: if X is a cyclic head, then Y n is a Spell-Out Domain, unless Y depends on X for its interpretation. b. Syntax: if Y n is the highest projection of a (potential) cyclic domain, then Y n constitutes a phase, unless Y depends on X for its interpretation. We illustrate these in turn. Optimal suspension: superlative suppletion The first case of (1) that we consider is in the morphology, and is drawn from the study of adjectival suppletion in Bobaljik 2012. We limit ourselves to a brief presentation here and refer the reader to the work cited for additional detail and important qualifications. POS CMPR SPRL GLOSS a. Persian bozorg bozorg-tar bozorg-tar-in 'big' b. Cimbrian šüa šüan-ar šüan-ar-ste 'pretty' c. Czech: mlad-ý mlad-ší nej-mlad-ší 'young' d. Hungarian: nagy nagy-obb leg-nagy-obb 'big' 8 As mentioned in fn. 5, the claim that English raising infinitives are phases does not entail that raising constructions cross-linguistically are phases. Rather, phasehood is dependent on the selectional properties between a verb and its complement. In Wurmbrand, Alexiadou, and Anagnostopoulou 2012, we suggest, for instance, that raising constructions in Greek, Romanian, and Spanish are value selected subjunctives or infinitives, and hence do not constitute phases, which, in contrast to English raising, allows phase-bound operations to apply across them. Furthermore, modal constructions, which can also involve raising (see Wurmbrand 1999; Bhatt 2000), show different QR properties than seem constructions in English, since modals originate in the tense domain and movement from the edge of the lower phase (Spec,vP) across a modal is allowed by Scope Economy. Suspension across domains interpretation, and the structure would not be interpretable. Thus, the only way the structure will converge is if the unvalued features are valued via Agree before LF. The features as specified above have the effect that a mutual dependency is established between certain verbs and corresponding types of complements. The unvalued features of the top clausal projection need to enter an Agree relation with a higher verb that has an uF: val. Similarly, the uninterpretable feature of the selecting verb also becomes dependent on a specific complement, as desired. Following Pesetsky and Torrego (2007) , uninterpretable features (whether valued or unvalued) need to be licensed, specifically, they need to be connected to a corresponding interpretable feature (cf. the proposal of the Thesis of Radical Interpretability, Brody 1997). The feature specification proposed allows us to address the question of why complements that are value-selected are not phases. There are two ways to implement this, and we will leave the choice between the two options open here. First, it could be assumed that valuation suspends phasehood. Since the unvalued features under consideration are interpretable features, these units would be interpretationally incomplete (before valuation takes place), and hence at the point where the clauses are completed, these units would not qualify as objects that are useable by the semantics. Alternatively, it could be assumed that the heads with the unvalued features undergo head-movement, which causes phase extension (see den Dikken 2007) or phase sliding (see Gallego 2005, 2010; Gallego and Uriagereka 2006), although note that these latter proposals, unlike Domain Suspension as we have defined it, do not obviously extend to the morphological cases discussed in section 1. Bobaljik, Jonathan D., and Susi Wurmbrand. 2005. The domain of agreement. Natural Language and Linguistic Theory 23, 809-865. Bošković, Željko. 1996. Selection and the categorial status of Infinitival Complements. Natural Language and Linguistic Theory 14, 269-304. Bošković, Željko. 1997. The syntax of nonfinite complementation: An economy approach. Cambridge, MA: MIT Press. Bošković, Željko. 2002. A-Movement and the EPP. Syntax 5, 167-218. Bošković, Željko. 2007. On the locality and motivation of Move and Agree: An even more minimal theory.