In this book, Idan Landau offers a thorough exposition of control, based on previous and current approaches, along with a wealth of empirical data discussed in the literature. The book is a survey on control, its complexity in terms of the contexts where it occurs and the interpretations it is associated with. The underlying idea is that control is necessitated by the empty category PRO in its most recent instantiation, that is as an argument, which is subject to theta-role assignment and Case/case. Unlike other arguments, it is necessarily null.
The book is organized in eight chapters. Chapter 1 is a historical excursus on the theory of control, mainly in generative grammar. It distinguishes between control and raising and defines the interpretations attributed to control, namely obligatory, non-obligatory and arbitrary. Chapter 2 outlines the various approaches to control, in terms of predication, binding, argument structure (LFG), movement and Agree. Chapter 3 offers the empirical evidence for the existence of PRO, based on the fact that control contexts are clausal and therefore require a syntactic subject, given the EPP; further evidence is provided by secondary predication, case, number agreement and floating quantifiers. Chapter 4 brings in more recent views based on empirical data, which dissociate control/PRO from non-finiteness and show that case-marking is possible in control contexts (indirectly evidenced on the predicate); this evidence reinforces the argument nature of PRO. The next two sections concern two important questions, namely whether PRO is restricted to subject positions only (something which is no longer dictated by case conditions) and whether it is always null. Chapter 5 is the longest in the book (it extends over a hundred pages roughly), and presents approaches to core issues of control, such as the choice of controller, the role of locality (does minimal distance apply?), the crucial distinction between exhaustive, partial and split control, control by null (implicit) arguments, the interaction of control (PRO) with quantifiers (PRO-gate), and finally control in (and from) the DP. Chapter 6 is rather short, and discusses adjunct control (which adjuncts allow it or require it and which do not) and the mechanism behind it, given that in this context selection leading to some form of restructuring, or Agree, or straightforward movement is not relevant or not straightforward at least. Chapter 7 considers non-obligatory control (is it really control?). Finally, Chapter 8 concludes the discussion. The book includes a rich list of references and an index.
As Landau points out in the Preface, the book is meant as a survey of a multi-faceted and long-standing issue in the theory of grammar. Control has been a hot topic for almost fifty years, because it raises important questions about the contexts where it occurs (complement vs. non-complement clauses, (non-)finiteness, interaction with lexical semantics), the range of interpretations (obligatory vs. non-obligatory bound reference, arbitrary readings), and the nature of subjects. As such, it covers a number of key properties of control, including issues about clause structure, selection, syntactic and morpho-phonological realization of arguments. In other words, the discussion makes explicit the complexity of control as a phenomenon where syntax interacts with the two interfaces, LF and PF, for interpretation and realization, respectively. At the same time, the variety of factors that interfere with the phenomenon makes it difficult to have a clear-cut view of how exactly control arises. In this respect then, Landau's presentation does a great job in offering not only a history of the theory of control (from Equi-NP deletion, to the PRO-theorem, copies, Agree), but also a detailed presentation of the complexity of the phenomenon, bringing into attention discussions that have often been neglected or left aside in the recent literature.
The underlying assumption in the discussion is that PRO exists: it is a necessary property, or better say a syntactic entity, justified on syntactic (need for a subject), semantic (interpretation) and morpho-phonological (case or agreement on the predicate) grounds (see Chapter 3 in particular). In this respect, PRO although always null, has a ‘visible’ presence, manifested via anaphoric binding, agreement and/or case on some other element in the structure, as in Icelandic for example (see also Chapter 4). Interestingly though, when the corresponding position (i.e. that of PRO) is overtly realized, it always takes the form of an anaphor. One could then say that the presentation revolves around PRO and serves as a ‘hymn’ to PRO.
Given the discussion presented in the book, it does become quite obvious that the PRO-implementation has a number of advantages with respect to the movement theory of control for example, as in Hornstein (Reference Hornstein1999, and elsewhere; the whole list of references is provided in the book). The strongest argument is that the interpretation associated with PRO is that of a bound element, something which cannot necessarily be forced on a copy. This was the main argument against Equi-NP deletion and the line of theorizing that (correctly) guided research in the Government and Binding Theory framework in the 1980s. Still what remained unclear at that point was the peculiar PRO-theorem (no governing category) and the mechanism that would determine PRO's controller. Part of the problem then was due to the definition of government, or to be more precise the notion of government as such. Nevertheless, the basic distinction between obligatory control as an instance of binding and non-obligatory or arbitrary control as an instance of lack of binding sufficed to give a broad description of the distribution of PRO. On the other hand, equating PRO to a NP-copy has the advantage of predicting in a straightforward way the choice of the controller (since controller and controlee are one and the same element), but cannot easily capture the following facts: that the reading associated with the controlee is a bound one (not simply obligatory coreference in other words), that lexical semantics of the selecting predicate play a role (something which is obvious with the distinction between exhaustive and partial control), that Case requirements imputed to non-finiteness are not empirically supported without stipulations, and that non-obligatory or arbitrary control cannot be accommodated in terms of movement (instead in this approach they are attributed to another empty category, namely pro). These two general approaches do not differ just in terms of technical implementation but further give rise to different conceptions of how grammar works. In particular, for the movement theory of control, all instances of ‘coreference’ in the broad sense are derived by movement (see Hornstein Reference Hornstein2009: Chapter 6). Thus interpretation arises from movement. The PRO approach, on the other hand, in whichever way it is implemented, takes interpretation as the core property which may or may not be manifested in terms of displacement (‘movement’). As such it can be compatible with either a derivational or a representational view of grammar. In this respect, the discussion in Chapter 5 where the choice of controller and the contexts of PRO are discussed is quite illuminating.
In short, the postulation of PRO has certain advantages in terms of distribution and interpretation. At the same time, the nature of PRO as such remains a mystery, in the sense that PRO may be an anaphoric element (so why not anaphoric agreement, as in Borer (Reference Borer, Jaeggli and Safir1989)?) but its presence cannot be independently attested. The anaphoric interpretation determines the interpretation and how this is derived in association with other elements in the clause structure, but does not suffice to derive or force its null phonological properties. So in the best case, PRO is a mnemonic for a set of interpretations (see Manzini Reference Manzini and Grohmann2009). At this point, I should comment that the conclusions drawn at the end of Chapter 3 (78), although based on empirical evidence presented before, still remain problematic. To be more precise, Landau argues that ‘these sentences demonstrate three points at the same time: (i) clauses must have structural subjects (the EPP); (ii) PRO cannot be an expletive; (iii) PRO exists’ (78). Point (i) is an independent requirement, although the exact realization of the syntactic subject as a DP, full pronoun, or even an agreement affix without pro is something that requires further discussion. Point (ii) also follows from the empirical data in the sense that the interpretation assigned to PRO cannot be the equivalent of an expletive, partly expected since it is bound and necessarily an argument (so unlike pro which can be an expletive subject, given its pure pronominal nature). What is unclear is how point (iii) arises as a conclusion or follows from (i) and (ii).
It becomes quite evident from the discussion in the book that in the last decade at least, research around control has received a revived interest, which has diverged research from the distribution of PRO (see PRO-less accounts for example) and has brought other important aspects into attention. Of those, the distinction between complement and non-complement clauses has been of the greatest importance. It has allowed researchers to investigate the properties of control predicates, their potential interaction with (non-)finiteness and restructuring (for a more recent view see Grano Reference Grano2012). As Landau shows (Chapter 6) the issue of control in adjunct (non-complement) clauses still has open ends, which challenge any theory of control and trigger a closer examination of adjunct clauses and the relation to the matrix predicate (or the matrix clause more generally).
To conclude, the book is an invaluable guide for anyone interested in control. It is what one would call a ‘critical survey’. It offers a complete coverage of the contexts where control is attested (obligatorily or not), the problems that have been brought forward, as well as open issues which still require an explanation. At the same time, it provides a wealth of references, which quite often have gone unnoticed. The presentation is clear, detailed (sometimes too detailed for a survey), meticulous, and helps the reader understand why control is still an interesting topic, and why new questions can always arise. Landau is an expert on the topic; his own work has helped researchers view control from a new angle, and this is evident in the book. The way the discussion is built makes the book a useful companion for students but also for researchers. It is also relevant to anyone interested in the structure of grammar in general beyond the topic of control.
