Control and Restructuring is Thomas Grano’s revised doctoral dissertation (University of Chicago, 2012). The book investigates the phenomenon of control structures in which the implicit subject of the embedded nonfinite clause can be construed as co-referential with the subject of the embedding clause. Drawing on data from English, Mandarin Chinese and Modern Greek, Grano explores the relationship that complement control bears to restructuring and to the meanings of the embedding predicates.

Complement control falls into exhaustive control and partial control. The split between exhaustive control and partial control was identified by Landau (Reference Landau1999, Reference Landau2000) as follows:

Grano proposes that ‘exhaustive control predicates realize inflectional-layer functional heads that instantiate monoclausal raising structures, whereas partial control predicates realize lexical verbs that instantiate biclausal PRO-control structures’ (211–212). He argues that the principles operating in the syntax of control and restructuring are largely universal, related to the semantics of the embedding predicates.

The first chapter offers a brief overview. Chapters 2 and 3 present the book’s central thesis. Chapter 2, ‘Exhaustive control as functional restructuring’, articulates the book’s core syntactic proposal, i.e. a functional restructuring approach to exhaustive control. It begins with empirical and theoretical backgrounds on exhaustive and partial control and provides its empirical relationship to the distribution of restructuring, which is ‘a phenomenon whereby apparently biclausal structures act as transparent domains for ordinarily clause-bound processes’ (11; see Wurmbrand Reference Wurmbrand2001 for more detail). Grano generalizes the control-restructuring correlation as follows: exhaustive control predicates are restructuring predicates; partial control predicates are non-restructuring predicates (15). This relationship can be further illustrated via Cinque’s (Reference Cinque2004) proposal that exhaustive control predicates are realized as functional heads in the inflectional layer of the clause, giving rise to monoclausal raising structures, whereas partial control predicates instantiate lexical heads and thereby give rise to biclausal PRO-control structures. The distinction between exhaustive control predicates and partial control predicates can be independently observed in their different syntactic behaviors in terms of finite complementation, overt embedded subjects, subject orientation and licensing of polarity items.

In Chapter 2, Grano elaborates the obligatoriness of restructuring and argues that an adequate theory should be able to explain the asymmetry between exhaustive control verbs and partial control verbs in terms of restructuring. Regarding the distribution of PRO, Grano makes two predictions. First, the embedded subject position under an exhaustive control predicate is always non–case-marked, since exhaustive control predicates instantiate raising structures, which prohibit movement from a case-marked position to another case-marked position. Second, the embedded subject position under a partial control predicate is always case-marked, since partial control predicates instantiate biclausal PRO-control structures and PRO has case just like an ordinary DP. In terms of the interpretation of PRO, Grano holds that exhaustive control is a semantic identity effect, thus consistent with a movement relation between the controller and the controllee, whereas partial control is a semantic nonidentity effect and lends support for an agreement relation between the controller and the controllee. The partial control property of PRO results from its status as a bound pronoun.

Chapter 3, ‘The semantics of restructuring: Verb meaning, variable binding and clausal architecture’, explores the semantic components of the functional restructuring approach to exhaustive control. Inspired by Cinque’s (Reference Cinque2004) cartographic approach to IP structure, Grano argues that exhaustive control predicates contain as part of their meaning a variable that must be obligatorily bound by a subject. Non-restructuring predicates must be placed above Tense for their meanings. Considering that subjects are interpreted no higher than [Spec,TP], restructuring above Tense leads to failed variable binding and hence ungrammaticality. The fact that only control predicates that correspond semantically to heads below Tense in Cinque’s (Reference Cinque2004) hierarchy admit restructuring can naturally follow from the variable binding approach to exhaustive control, since restructuring above Tense will result in its variable unbound.

In Chapter 4, ‘Explaining the mixed properties of want’, Grano deals with the recalcitrant want, which is assumed to be a restructuring predicate but allows for partial control, apparently against Grano’s theory. He proposes that want is a restructuring verb which serves as a functional head in the inflectional layer of the clause. Nevertheless, want can combine with a silent predicate HAVE and project a biclausal structure, which makes available an overt embedded subject and partial control. Only if there are no overt restructuring effects and in combination with HAVE, can want exhibit partial control properties.

In addition, want is distributionally similar to the predicate need in that both allow an embedded subject, disallow finite complementation and tend to be a restructuring predicate. However, need does not support partial control. Grano provides a semantic account for this contrast and suggests that want is the only pure attitude predicate among all the predicates corresponding to the heads below Tense in Cinque’s hierarchy. Because want relates a proposition to an arbitrary attitude holder, the argument introduced by HAVE can directly provide the value for this attitude holder via raising and variable binding. The other predicates (including need) cannot accommodate the extra argument introduced by HAVE.

Chapter 5, ‘Control, restructuring, and temporal interpretation’, investigates the temporal properties of infinitive tense. Grano generalizes three temporal relations between two temporally interpreted expressions: simultaneity (e₁ overlaps with e₂), posteriority (e₁ follows e₂) and anteriority (e₁ precedes e₂). He argues for two constraints on the temporal interpretation of infinitives in exhaustive or partial control predicates. The first constraint is that the infinitive of a partial control predicate allowing for simultaneity must be stative. The second one is that exhaustive control predicates cannot admit anteriority.

Grano further argues that the temporal properties of infinitives provide further support for his functional restructuring approach to the two distinct complement control structures, given that one of the consequences of the core analysis is that partial control structures have two TPs whereas exhaustive control structures have only one TP.

Chapters 6 and 7 expand the empirical coverage by applying the core analysis of this book to Mandarin Chinese and Modern Greek respectively. Chinese is argued to have no finite vs. nonfinite distinction (see Hu, Pan & Xu Reference Hu, Pan and Xu2001) and Modern Greek lacks nonfinite complementation. Both languages are typologically quite different from English. Neither of them exhibits typical restructuring properties.

Chapter 6, ‘Control and restructuring in Mandarin Chinese’, is concerned with the control in Mandarin Chinese. The syntax of control in Chinese is inevitably bound up with the unresolved issue of whether Mandarin Chinese exhibits a finite–nonfinite distinction. Grano holds that the much debated finite–nonfinite distinction can be reinterpreted as reflecting the monoclausal–biclausal split. He first discusses the distribution and interpretation of embedded subjects, holding that renwei ‘believe’ class predicates fall into partial control predicates, and shefa ‘try’ class predicates fall into exhaustive control predicates, according to whether they allow overt embedded subjects. He further proposes that the former allows overt embedded subjects and can have a biclausal analysis, whereas the latter does not allow overt embedded subjects and can only have a monoclausal analysis. There are exceptions to the generalization that the shefa-class predicates disallow overt embedded subjects, which, Grano argues, can be explained via resumption. The overt embedded subjects in shefa-class predicates must be identified by the matrix argument and separated from the predicates by an adjunct.

Grano then probes into the distribution of embedded modal and aspectual morphemes. As argued by Hu et al. (Reference Hu, Pan and Xu2001), the future/modal marker yao, not hui, is acceptable in Mandarin Chinese controlled complements, whereas in noncontrolled complements, both yao and hui are acceptable. In a monoclausal structure, the incompatibility between shefa-class predicates and hui involves constraints on the co-occurrence of functional heads. It is assumed that yao is similar to woll, an abstract root, which is argued to be the base of English will and would.

The aspectual le and guo can be allowed in controlled complements, just as in noncontrolled complements. Grano holds that this phenomenon does not argue against the monoclausal analysis of the controlled complements in shefa-class predicates. First, le and guo are not purely aspectual in meaning; they also have a temporal meaning. Occurrence of le and guo in the controlled complement does not entail that the event time associated with the embedded verb precedes that of the matrix verb. Second, the adverb congqian exhibits the same-clause constraint with the experiential aspect marker guo. However, the same-clause requirement can be suspended in controlled complements, which can be explained by the monoclausal functional restructuring approach. Third, the same conclusion is supported by the interaction between aspect and negation in Mandarin. In controlled complements, when negation and perfective aspect le co-occur, the perfective allmorph you must appear overtly at the matrix clause, in contrast to noncontrolled complements. This asymmetry is due to the syntax of le and you, crucially relying on the hypothesis that shefa-class (exhaustive control) is monoclausal. In addition, focus movement shows that control predicates are parts of a monoclausal structure, since focus movement, which is clause bound, cannot occur in renwei-class predicates, but can occur in shefa-class predicates.

Chapter 7, ‘Control and restructuring in Greek’, applies the book’s core analysis to Modern Greek. Following Spyropoulos (Reference Spyropoulos, Davies and Dubinsky2007), Grano classifies three types of Greek embedding predicates with respect to the restrictions they impose on the embedded tense. He argues that just like English and Mandarin Chinese, Greek exhaustive control predicates instantiate monoclausal structures, evidently supported by facts concerning emphatic polarity item licensing and inverse scope. The ban on overt embedded subjects and embedded past tense follows from there being only one argument structure and one tense in a monoclausal domain. Conversely, the existence of a disjoint embedded subject and an independent embedded tense both indicate biclausality.

However, overt embedded subjects with ‘try’ and ‘manage’ can occur in Greek. Grano proposes that similar to the account of English want, Greek ‘try’ and ‘manage’ instantiate monoclausal structures but they also introduce an extra argument via the silent main verb HAVE. The crucial difference between Greek ‘try’/‘manage’ and ‘want’ is that when the former embeds an overt subject, there is a hidden causative component that links the matrix subject to the embedded event as a causative agent. He further proposes that the causative component syntacticize as the silent predicate CAUSE(x), consequently reducing cross-linguistic variation in the acceptability of overt embedded subjects with ‘try’/‘manage’ to cross-linguistic variation in the availability of CAUSE(x). He shows that even internally to Greek, the distribution of CAUSE(x) is limited: it is only a ‘last resort’ strategy.

Chapter 8 concludes the book, briefly summarizing its main contents from empirical, syntactic, semantic, syntax–semantics interface and cross-linguistic perspectives.

To conclude, Grano explores the control phenomena in terms of restructuring from the cartographic approach, based on data from English, Mandarin Chinese and Modern Greek. He argues that the distinction between exhaustive control and partial control can be made with reference to monoclausality/biclausality. The former is in accordance with movement theory of control, whereas the latter supports PRO theories of control. Thus Grano successfully combines these two seemingly controversial analyses. Though its conclusion is still awaiting further verification, Grano’s book will definitely exert great influence on the study of control.