Not all direct objects are equal. Linguistic studies in the last two decades have shown a keen interest in some direct objects over others. The most prominent objects have been those that undergo scrambling (mainly in Germanic languages) and those that receive so-called differential object marking (DOM). Special attention within these two phenomena has been reserved for indefinite objects, as they receive different semantic interpretations depending on their (morpho)syntactic configuration. In his book, Luis López integrates these ingredients to develop a new account of the syntax–semantics interface. His main proposal is to pair syntactic positions with different modes of semantic composition. This is schematically shown in (1) (EA=external argument, DO=direct object, subscripts 1 and 2 indicate mode of composition, see text for explanation):

1. (1) [_vP EA v [_αP DO₂ α [_VP V DO₁]]]

López first makes a connection between scrambling and differential object marking. He proposes that unmarked objects stay within VP (DO₁ in (1)), but that marked objects scramble out of VP (to DO₂). The two syntactic positions are associated with different types of semantic composition. In DO₁ indefinite objects compose via Restrict (Chung & Ladusaw Reference Chung and Ladusaw2004), an operation resulting in a narrow scope reading of the object. Objects in position DO₂ compose via regular function application, but only after application of a choice function. As a result, scrambled objects can receive both wide and narrow scope readings (and also intermediate ones). This means that there is no rigid mapping in López' proposal from syntactic position to semantic interpretation. Instead, syntactic positions determine the available mode of composition, which in turn may have an effect on the interpretations of indefinite objects. The proposal is developed in the first three chapters of the book, mainly on the basis of Spanish, and extended in the final chapter to include other languages as well.

Chapter 1 presents the main ingredients of the proposal. The chapter starts with a few Spanish examples that establish an asymmetry between marked and unmarked objects in Spanish: marked objects can receive both a wide and narrow (specific and non-specific) interpretation, whereas unmarked objects are exclusively narrow scope/non-specific. This pattern is then shown to be problematic for a strict view on the syntax–semantics interface, as for instance represented by Diesing's (Reference Diesing1992) Mapping Hypothesis, in which syntactic positions are directly mapped onto a semantic interpretation. López maintains that there is a relation between syntactic position and interpretation, but that this mapping is indirect. The Spanish pattern is also used to motivate the adaptation of Chung & Ladusaw's theory of semantics. The fact that marked indefinite objects can take variable scope, even out of a syntactic island, suggests that they are interpreted by means of choice functions. The exclusive narrow scope interpretation of unmarked indefinites suggests the working of a compositional process along the lines of Restrict. López adds syntax to this picture: depending on syntactic position a different mode of composition is available, recall (1) above.

The second half of the chapter provides a detailed description of the use of accusative a in Spanish. López establishes firmly that marked indefinite objects can, but do not need to, have wide scope and, disentangling scope and specificity, that they can be interpreted as either specific or non-specific. He notes that the Spanish data seem parallel to the use of indefinite determiners in Maori, as analyzed by Chung & Ladusaw, and that this invites a similar analysis. In addition, López introduces a new set of contexts in which DOM is obligatory in Spanish: small clauses, clause union and object control structures. He concludes the chapter by showing that taken together, the full pattern of DOM in Spanish is not straightforwardly modeled in the dominant scale approach to DOM.

Chapter 2 develops the syntactic part of the proposal and its main goal is to establish that marked indefinite objects in Spanish occupy a scrambled position, in contrast to unmarked objects, and to explain this difference. López adopts a minimalist version of generative grammar with a realizational theory of morphology. The appearance of accusative a is taken to be the result of a process of Vocabulary Insertion that provides a phonetic realization when a syntactic category appears in the right context. In this way, López dissociates the use of a from semantic interpretation.

López assumes a structure for transitive verbs that includes in addition to VP both a vP and a functional projection αP, where α is identified as a ‘conglomerate’ (32) of applicative and aspectual properties. He establishes on the basis of c-command tests that unmarked objects stay in-situ, and that marked objects scramble to Spec,αP. The different positions of objects are related to Case requirements that depend on the size of nominal phrases. Unmarked in-situ objects are argued to be phrases smaller than KP, which can value Case through a process of incorporation: the highest nominal head is copied into the lexical verb which incorporates into v which is valued with an [accusative] feature. Marked objects are of category KP. Incorporation is blocked for these elements and they have to move to Spec,αP to satisfy their Case requirement. This position also provides the right syntactic context for K to be realized as a (given that other conditions including animacy are fulfilled as well).

Chapter 3 discusses the modes of semantic composition available in different syntactic positions. In-situ objects are argued to compose via Restrict: they do not satisfy the argument structure but restrict the domain of referents. Combination of this operation with Kratzer's (Reference Kratzer, Rooryck and Zaring1996) Event Identification ensures a narrow scope reading of these objects: existential closure of the VP has to take place before it can be combined with vP. The variable scope behavior of marked objects is captured through the assumption that their K head is associated with a choice function. This choice function can be anchored to the speaker or subject of a sentence resulting in a specific reading.

Throughout Chapters 2 and 3, López shows how his proposal accounts for a wide variety of facts in Spanish, including the obligatory absence of accusative a on bare plurals, its obligatory presence in small clauses, and restrictions on existential and possessive have predicates. In Chapter 4, the theory developed in the preceding chapters is applied to similar phenomena in five other languages: differential object marking in Persian, Hindi, Kiswahili and Romanian, and object scrambling in German. This demonstrates the wider cross-linguistic applicability of the proposed account.

López' book presents a truly novel approach to the syntax, interpretation and marking of indefinite objects. It contains many interesting observations and new proposals for analysis. He convincingly shows that adding syntax to existing semantic analyses results in more restrictive models. His main contribution in my view is the dissociation of the occurence of accusative a from semantic interpretation. By linking both semantic interpretation and the presence of a to syntactic configuration, López provides a radically new perspective on differential object marking. His proposal to view accusative a as a postsyntactic insertion guided by the presence of the right context opens up a way of incorporating into a single account the many factors that appear to play a role in the presence of a. This seems an important step towards a better representation of the (seemingly) multifunctionality and sometimes irregular behavior of this differential marker (but see below for further discussion).

The book is written in a very clear and accessible style, which will make it a rewarding read also for readers who are not familiar with the latest developments in generative syntax. The description in Chapter 1 of the distribution of accusative a in Spanish presents the best overview to date and, in addition, uncovers some new interesting data on the obligatory use of this marker. It should serve as the starting point for any future work on this topic. López' argumentation can be characterized as pragmatic rather than dogmatic. In several places he makes theoretical choices but invites the reader to replace these with his own. This preempts the need to discuss the pros and cons of particular theoretical choices. Although this can be perceived as a weakness, I view it as a positive aspect of the book: it allows the reader to keep a steady focus on the main argument. Still, there are places where I would have welcomed some additional argumentation. The most prominent issue concerns the status of elements as a KP. I will elaborate on this in the next paragraph.

The syntactic category KP plays a decisive role in López' proposal: it determines the syntactic position of an object (KPs have to scramble for Case reasons), it determines its semantic interpretation (KPs translate as a choice function), and it partly determines the occurrence of accusative a (only KPs can be realized as a). The question when an element is a KP and when not becomes very relevant here. The situation is clear in the case of bare plurals. They do not project to DP/KP (only to #P) and hence it follows from López' account that they cannot occur with a, they take obligatory narrow scope, and they do not scramble. It remains unclear how this generalizes to other objects. Indefinite objects, for instance, are very volatile: they can occur in both positions, have both wide and narrow scope readings and occur with and without a. This seems to imply that indefinite objects are sometimes KPs and sometimes not. It remains unclear what motivates, or even helps, to identify, this category shift. Definite objects, by contrast, always require a when animate (Section 1.2.1). This suggests that they are always KPs. But KPs select DPs rather than vice versa. It is not immediately clear how the obligatory marking of definite objects follows from the proposed account. Finally, inanimates normally do not appear with a,Footnote ¹ but this does not exclude them as KPs. They just do not satisfy the context requirement for a to be inserted. Inanimate objects thus provide a good opportunity to investigate the consequence of dissociating a from semantic interpretation. If inanimate indefinite objects are indeed KPs, we would expect them to have syntactic and semantic behavior similar to marked animate indefinite objects. If they are not KPs, we would expect them to always have narrow scope. López does not provide the data to settle this question, but discussion of this and the other points raised above would have been a very welcome addition to his book and would have provided the opportunity to further strengthen the author's proposal. These observations are by no means intended to disqualify the book. López, beyond doubt, has written an extremely valuable contribution to the discussion of indefinite objects and differential object marking. Fortunately, he has also left us some room for future discussion.