Ellipsis of a constituent whose head has moved out of it (“headless ellipsis”) is possible in some cases but not in others. Headless ellipsis is licensed only if the stranded head has not crossed a Spell-Out domain. The reason is that the silencing instruction responsible for ellipsis must be PF-visible on the head of the elided constituent, and PF-visibility is cut off at Spell-Out domain boundaries. A parallel effect is observed with remnants of head movement that are frozen for movement (“headless movement”). The two effects can possibly be united if ellipsis and copy deletion recruit the same silencing instruction at PF, hosted on the head of the deleted constituent. A third, mirror-image effect is observed with reprise fragments, which must be visibly headed. This time head movement removes the PF instruction that spares these fragments from ellipsis. Overall, these phenomena establish the significance of headedness for the syntax-PF interface.

Most accounts of sluicing assume it is derived by TP deletion at PF; however, C′ deletion is a viable option, which is even better equipped to explain the Sluicing–Comp Generalization. Nonetheless, this article shows that C′ deletion must be rejected, because it predicts scope enhancement for modals that raise to C in sluicing when in fact they take scope in T or below it. An implication is that some independent principle must block T-to-C movement under sluicing. The results further confirm that head movement must be a syntactic operation insofar as it affects semantic interpretation.

Standard semantic theories of Obligatory Control (OC) capture the obligatory de se reading of PRO but fail to explain why it agrees with the controller. Standard syntactic theories of OC explain the agreement but not the obligatory de se reading. A new synthesis is developed to solve this fundamental problem, in which the controller directly binds a variable in the edge of the complement. The associated semantics utilizes the idea that de se attitudes can be modelled as a special case of de re attitudes. The specific interaction of feature transmission and phase-based locality derives a striking universal asymmetry: Inflection on the embedded verb blocks OC in attitude complements but not in nonattitude complements. A semantic benefit is a straightforward account for unexpected binding between PRO and de re reflexives/pronouns.

Hebrew is standardly cited as a language exhibiting Verb-stranding VP-ellipsis (VSVPE). Systematic reassessment of the data demonstrates that all the alleged evidence for VSVPE is consistent with Argument Ellipsis (AE); furthermore, there are ample data that are only consistent with AE, and more revealingly, data that can only be explained if VSVPE is unavailable. Finally, the verb preceding the missing object need not match the antecedent verb, falsifying the “Verb Identity Requirement”. The conclusion that Hebrew employs AE (similarly to East Asian languages) but not VSVPE focuses attention both on the typology of AE and on the so-far hidden constraints against VSVPE derivations.

Nonfinite adjuncts present a puzzle in that they seem to allow both Obligatory Control (OC) and Non-Obligatory control (NOC) readings, which are normally in complementary distribution. Extending the analysis of complement control, I propose that adjuncts can either project as predicative clauses, or one layer further, as logophoric clauses; each projection is uniquely paired with its own type of control. This picture allows us to state in compact terms, though not to resolve, another puzzle that was noted in the 1980s: Rationale clauses resist implicit agent control when passivized. I show that the effect is systematic and extends to other environments (copular main clauses and temporal adjuncts). The emerging generalization is that passive adjuncts resist NOC.

Controlled null subjects (PRO) are semantically bound variables that bear morphological features. In certain environments of partial control, the morphological ϕ‐features (specifically, [person]) and the semantic value of PRO diverge. A natural explanation of the fact that the [person] feature of PRO is uninterpreted is that it is assigned at Phonological Form (PF). Given that this feature participates in agreement relations, we conclude that agreement must be (optionally or exclusively) a PF phenomenon.

“Hybrid” nouns are known for being able to trigger either syntactic or semantic agreement, the latter typically occurring outside the noun’s projection. We document and discuss a rare example of a Hebrew noun that triggers either syntactic or semantic agreement within the DP. To explain this and other unusual patterns of nominal agreement, we propose a configurational adaptation of the CONCORD-INDEX distinction, originated in Wechsler and Zlatić (2003). Morphologically-rooted (=CONCORD) features are hosted on the noun stem while semantically-rooted (=INDEX) features are hosted on Num, a higher functional head. Depending on where attributive adjectives attach, they may display either type of agreement. The observed and unobserved patterns of agreement follow from general principles of selection and syntactic locality.

A growingly popular analysis holds that the plural interpretation of PRO in partial control arises from associating a singular PRO with a null comitative phrase. Three novel arguments are presented to demonstrate the inadequacy of this analysis.

This book revives and reinterprets a persistent intuition running through much of the classical work: that the unitary appearance of Obligatory Control into complements conceals an underlying duality of structure and mechanism. I argue that control complements divide into two types: In attitude contexts, control is established by logophoric anchoring, while non-attitude contexts it boils down to predication. The distinction is also syntactically represented: Logophoric complements are constructed as a second tier above predicative complements. The theory derives the obligatory de se reading of PRO as a special kind of de re attitude without ascribing any inherent feature to PRO. At the same time, it provides a principled explanation, based on feature transmission, for the agreement properties of PRO, which are stipulated on competing semantic accounts. Finally, it derives a striking universal asymmetry: the fact that agreement on the embedded verb blocks control in attitude contexts but not in non-attitude contexts. This book is unique in being firmly grounded in both the formal semantic and the syntactic studies of control, offering an integrated view that will appeal to scholars in both areas. By bringing to bear current sophisticated grammatical analyses, it offers new insights into the classical problems of control theory.

The subject of nonfinite clauses is often missing, and yet is understood to refer to some linguistic or contextual referent (e.g. Bill preferred __ to remain silent is understood as Bill preferred that he himself would remain silent). This dependency is the subject matter of control theory. Extensive linguistic research into control constructions over the past five decades has unearthed a wealth of empirical findings in dozens of languages. Their proper classification and analysis, however, have been a matter of continuing debate within and across different theoretical schools. This comprehensive book pulls together, for the first time, all the important advances on the topic. Among the issues discussed are: the distinction between raising and control, obligatory and nonobligatory control, syntactic interactions with case, finiteness and nominalization, lexical determination of the controller, and phenomena like partial and implicit control. The critical discussions in this work will stimulate students and scholars to further explorations in this fascinating field.