Abstract: In this paper we examine the ban on ¯ A-movement of the external argument of a transitive verb that holds in many morphologically ergative languages. We argue that the prohibition against movement of the ergative subject should not be derived from restrictions on the movement of the ergative DP. Rather, we suggest that movement of the ergative argument is per se unproblematic, but if it applies, it applies too early, and thereby creates problems for its absolutive co-argument, which does not receive structural case. In morphologically accusative languages, no such move- ment asymmetry arises because arguments move too late to trigger the fatal consequences that moving ergatives cause. We present a relational, co-argument-based analysis that implies a strictly derivational syntax in which the order of operations plays an important role in deriving properties of the grammar.

Abstract: This paper investigates a puzzling correlation between two seemingly disparate phenomena: the crosslinguistic distribution of reflexive possessives and definiteness marking. As observed in Reuland 2007, 2011 and supported here by additional crosslinguistic evidence, reflexive possessives are available only in languages that either lack definiteness marking or encode definiteness postnominally. Languages that have prenominal (article-like) definiteness marking, on the other hand, systematically lack reflexive possessives. I argue that such facts support a particular approach to reflexive binding—specifically, one that has the following properties: (i) binding domains are stated in terms of phases, (ii) in addition to CPs and vPs, DPs are phases, and (iii) DP is not universal. I closely examine another robust crosslinguistic correlation regarding definiteness marking—Boskovic's (2008) Left Branch Extraction generalization—and show how it directly follows from the key assumptions of the analysis. I situate my proposals within a broader context of the phase theory, arguing that the syntactic representation of (in)definiteness is the crucial factor in determining the phasehood status of nominal categories. I extend my analysis to the clausal domain and discuss it in the context of languages that allow reflexives in the subject position.

Abstract: In syntactic theory, `freezing' refers to the idea that a constituent extraposed to a non-canonical position is resistant to extraction of any of its subconstituents (What did Terry see a movie yesterday about ?). The unacceptability of such examples, compared to minimally different sentences without extraposition, has been claimed to be a result of a grammatical constraint on dependency formation, e.g. Ross' (1967) Frozen Structure Constraint. Here, we argue that the gradient pattern of acceptability associated with such examples is better explained in terms of processing complexity. Experimental evidence from a controlled acceptability task shows that the total penalty for extraction from extraposed constituents is predictable from the summed acceptability penalties independently linked to extraction and extraposition. As prior research links both of these displacement phenomena to increased processing costs, we conclude that a grammar of unbounded dependencies in English does not require a `freezing' constraint for the explanation of these cases.

Abstract: J. Lin (2010) argues against Sybesma's (2007) proposal that Mandarin Chinese has syntactic tense and contends that it does not need tense for the temporal interpretations of sentences. This paper shows that those arguments cannot be sustained. Specifically, this paper points out the following: (i) Lin's claim that Mandarin Chinese sentences do not show tense-related syntactic properties is wrong. (ii) The so-called nonexistent temporal interpretations of Mandarin Chinese sentences are merely a matter of choice of the reference time. (iii) Lin's argument against the existence of past tense in Mandarin Chinese is committed to the error of taking the English past tense as the only model for the tense system of Mandarin Chinese. (iv) The parallelism between Dutch and Mandarin Chinese that Sybesma observes is temporal in nature, despite the objections of Lin. (v) The temporal interpretations of Mandarin Chinese sentences can be accounted for straightforwardly by a pronominal tense system.

Abstract: In this paper, I argue against the standard analysis of so-called split topics in German as discontinuous noun phrases (van Riemsdijk 1989). Building in part on Fanselow 1988, I show that the construction rather involves two morphosyntactically autonomous nominal constituents that are predicatively related in underlying form. This predication is syntactically unstable, however. Merge of two XPs within a single argument or adjunct position yields a symmetric structure for which no label (“head”) can be detected by Minimal Search (“for any syntactic object {α,β},α is the head if α is a lexical item”; see Chomsky 2008). Therefore, one of the two noun phrases must move at the phase level in order to render the structure asymmetric; in case the stranded noun phrase is elliptical, the impression of a discontinuous constituent arises. By providing a principled explanation for split topicalization in these terms, the analysis furnishes evidence for an architecture in which Merge applies freely (pace recent claims to the contrary, e.g., in Kayne 2010), and as an asymmetrizing device when applying internally (as movement), in the spirit of Moro 2000 and Chomsky 2013.

Abstract: This article develops a constructionalist approach to “lexical” causatives, as in The sun melted the snow. It is argued that causation is a truly configurational meaning, arising as the interpretation of the syntactic combination of two verbal heads, the higher v representing the causing event (an unspecified dynamic vdo) and the lower v representing the resulting state named by the verbal root (a stative vbe). This structure contrasts with that of simple transitive activity verbs, which are monoeventive (vdo). A parallel contrast is established between bieventive inchoatives (vgo–vbe), as in The snow melted, and simple unaccusatives, as in The guests arrived (vgo). In this analysis, causatives and inchoatives both comprise two events and have an intersective nonderivational relationship. They share the lower resulting state; the type of the higher event distinguishes between the two. The analysis—developed with attention to Spanish data—can straightforwardly account for observed gaps in the causative alternation, the distribution of bare nouns, and scope ambiguity of adverbials and negation, and it sheds new light on the presence of reflexive morphology in inchoatives. The analysis implies that transitivity, as well as unaccusativity, can arise from two basic syntactic structures, on which distinctive verbal meanings are built. In this theory, no syntactic terminal or lexical verb expresses a relation between events; relations between events—such as causation, change of state, and resultatives—arise via semantic composition rules that interpret complex syntactic structures.

Abstract: Do the grammars of English and German contain a ban on moving the lower of two wh-phrases (Superiority), or is the lower acceptability due simply to the complexity of processing the longer dependency that results when the lower wh-phrase is moved? The results of four acceptability-judgment studies suggest that a pure processing account is inadequate. Crossing wh-dependencies lower the acceptability of both German and English questions but with a significantly larger penalty in English than in German (experiment 1). The larger penalty in English cannot be attributed to greater sensitivity to violations in English, because relative clause island violations result in similar effects in the two languages (experiment 2). A pure processing account might claim long dependencies are easier to process in German than in English because of richer case, but a control experiment did not support this possibility (experiment 4). We suggest that moving the lower of two wh-phrases is banned in the grammar of English but not in the grammar of German. This predicts that there should be a penalty for crossing dependencies in English even in helpful (Bolinger) contexts, as confirmed in experiment 3, and even in short easy-to-process sentences, as confirmed by simple six-word sentences in Clifton, Fanselow & Frazier 2006. Finally, if German grammar does not contain a ban on crossing, it is not surprising that the penalty in German is smaller than in English or that like animacy of the two wh-phrases plays a larger role in German than in English because feature similarity generally gives rise to difficulty in processing, whereas in English a grammatical ban on crossing will reduce acceptability regardless of whether there is processing difficulty.

Abstract: We propose a Case-driven account of the “reanalysis” operation assumed by some researchers to underlie pseudopassivization. Contra previous analyses, we argue that reanalyzed prepositions do not form a unit with the verb at any level of representation. Rather, reanalyzed prepositions raise to a v/V-medial Agr head, and the complement of the preposition raises to the specifier of this complex head to receive Case. This places the erstwhile complement in essentially the same structural configuration as an ordinary direct object, so that passivization proceeds normally. Preposition stranding, although it does not require reanalysis as such, is tied to one of the prerequisites for reanalysis: the presence of independent Agr projections. This explains why all languages with pseudopassives also permit preposition stranding. Reanalysis also accounts for certain instances of anaphoric binding in the apparent absence of c-command.

Abstract: This paper presents an analysis of data in which imperative verbs appear in nonroot clauses, namely relative and finite verbal complement clauses. Although crosslinguistically rare, such embedding is well attested in languages such as Ancient Greek and Slovenian, and it is impossible in languages such as English. An analysis is presented in which rich imperative morphology is related to the possibility of embedding imperatives in CP contexts, which is formalized in terms of the independently motivated theory of Feature Transfer (Chomsky 2008). According to the analysis presented here, the difference between Ancient Greek and English imperatives is one of subcategorization, such that a ϕ-complete C is able to subcategorize for imperative T in Ancient Greek but not in English. Rich imperative morphology, defined as having overt forms beyond the second person, is taken to be a necessary condition for the learner to posit that ϕ-complete C can select imperatives, as in Ancient Greek but not in English. The analysis suggests a view of the grammar in which sentential force is encoded in the syntax, given that morphologically imperative verbs are disassociated with directive force in Ancient Greek.

Abstract: Grafting is a syntactic device developed and defended by van Riemsdijk (2000 and subsequent work) by which two independent input trees are connected via external remerger of internal elements, thus sharing a constituent. Grafting clearly violates standard assumptions about phrase structure. Still, several structures found in natural language are best analyzed in terms of grafting. The present paper discusses one such construction: complex numerals. It shows that the characteristic multidominance of grafting explains the word order, case assignment, plural, paucal, and singular distribution associated with complex numeral combinations better than a coordination approach and is more compatible with more number (or counting) systems found in the world's languages.

Abstract: Drawing on data concerning the long-distance dependency of the Mandarin bare reflexive ziji ‘self’ in embedded comparatives, this remark compares two types of analyses of Mandarin comparatives—the Reduction Analysis and the Direct Analysis. The discussion shows that the long-distance dependency of ziji in an embedded comparative can be easily captured in the Reduction Analysis, where a deletion operation is involved in the derivation; however, the Direct Analysis, where the constituent size of the standard of comparison is transparent, encounters challenges when facing the same data.

Abstract: This paper discusses the following type of right-node-raising construction (Ross 1967): Coordinate structures in which elements that are interpreted in only one conjunct have moved across material that is interpreted as part of both conjuncts. I show that there are constructions of this type that simultaneously exhibit noncanonical plural agreement with two singular DPs and a ban on collective predicates in the same environment, which suggests contradictory conclusions. I argue that these properties follow from an approach to right-node raising that allows multiple instances of sharing within the same tree (see Gracanin-Yuksek 2007, 2013; de Vries 2009). I propose an analysis of such plural agreement that treats it on analogy with agreement with conjoined singular DPs. I argue that, for the constructions with noncanonical plural agreement that I discuss, the present analysis is superior over a backward-deletion analysis and a (non-multidominance-based) across-the-board movement analysis.

Abstract: This paper explores the syntactic and semantic properties of movement and adjunction and proposes a new analysis of these two phenomena according to which they are closely related to each other. The basic pieces of grammatical machinery that carry out movement also carry out adjunction. Building on independently appealing ideas concerning neo-Davidsonian semantic composition and “move as re-merge,” the proposed framework naturally accounts for (i) adjuncts' puzzling status “either inside or outside” a maximal projection, (ii) adjuncts' ability to escape reconstruction, and (iii) the prohibition on extraction from adjuncts.