In this book, Paul Elbourne presents a detailed analysis of definite descriptions in the tradition of Frege and Strawson, couched in a situation semantics. This synthesizes his earlier contributions to the topic in a unified form with new empirical and theoretical points. It is contrasted throughout with a Russellian view. A great number of issues concerning descriptions, including presuppositionality, the referential–attributive distinction, anaphoric interpretations, and domain restriction are discussed and analyzed in detail. The final chapter extends the proposal to pronouns, viewed as covert definite descriptions. The book also offers one of the most extensive introductions of a situation-semantic system to date. In the following, I summarize the chapters of the book, commenting on chapter-specific points throughout. Next, I evaluate the book as a whole and discuss some of the more general issues that arise.
While Elbourne does not aim to provide ‘an introduction to the study of definite descriptions’ (ix), Chapter 1 provides a brief introduction to some types of approaches. Apart from those central to the book, three other approaches are discussed: the definites-as-predicates view proposed by Fara, the familiarity view of Heim and Roberts, and Szabo's Neo-Russellian existential analysis. Elbourne argues against these at varying length, and, given the brevity of the discussion, with varying degrees of success. At times, the style is somewhat polemical, as acknowledged in the Preface (ix). Some of the points made against Fara's proposal, as well as the criticism of Roberts' theory, are not convincingly justified in the brief discussion allotted to them. The rhetoric of the rebuttals of these theories should thus be taken with a grain of salt, especially in light of the obvious need for a book-length defense of the Fregean proposal itself.
Chapter 2 provides an introduction to situation semantics, spelling out both the philosophical notion of Austinian topic situations as well as the formal details of a semantic system based on syntactically represented situation pronouns. The presentation is very clear and accessible. It includes a highly useful list of the properties a situation-semantic system should have. Two innovations are introduced: first, as is standard in situation-semantic proposals, tripartite quantification is taken to involve the introduction of minimal situations in the restrictor, which are extended to incorporate the information in the nuclear scope. Elbourne factors the nuclear scope part out into a separate morpheme, Q (27–28). The specific formulation of Q provides a key ingredient for what follows, and it is surprising that this is not really highlighted: it has two lambda-abstractors over situations, which make it possible to evaluate situation pronouns in the nuclear scope relative to the minimal situations quantified over in the restrictor. This amounts to a compositional reworking of the syncategorematic analysis of Heim (Reference Heim1990), something that Elbourne (Reference Elbourne2005: 57–58) had pronounced to be seemingly impossible. The second innovation is a variant of λ-Conversion that passes on domain conditions of certain functions embedded in a larger environment to the topmost lambda-binder (34).
The ground work for the analysis of a variety of phenomena with definite descriptions is laid out in Chapter 3. After a brief historical review of Frege's and Strawson's views, a situation-semantic meaning for the definite is introduced (47):
(1) 〚the〛 = λf <e,st>. λs: s ∈ D s & ∃!x f(x)(s) = 1. ιx f(x)(s) = 1
It comes with an existence and uniqueness presupposition that has to be met in the situation supplied via a situation pronoun; if met, it returns the relevant unique individual. Given the general setup from Chapter 2, which includes situation binders, this pronoun can either be interpreted referentially or be identified with the situation at which the entire sentence is to be evaluated. The latter involves the special λ-Conversion rule alluded to above: in short, when the situation pronoun is bound, the presupposition of the definite cannot be evaluated when the proposition expressed by the sentence is computed, because it depends on what Austinian topic situation it is used with. The presupposition of the definite is therefore passed on to the proposition as a whole.
Chapter 4 turns to a discussion of the presuppositional nature of definites. Elbourne presents his view in the spirit of the Frege–Strawson tradition (with some minor differences). Next, he goes on to show how his system handles projection behavior with possibility modals, conditionals, and disjunction. Several variants of the nuclear scope operator Q are introduced, though it is not always clear what motivates the variations. The discussion of presuppositions under negation introduces another covert modal operator (FIC, 72) to cover cases where the usual presupposition is not present at a global level (equivalent to Russellian narrow scope readings). The second part of the chapter deals with cases of ‘presupposition obviation’, i.e., sentences which yield straightforward ‘false’ judgments even though the presupposition of the definite is not met. The theory of von Fintel (Reference von Fintel, Reimer and Bezuidenhout2004) is reviewed in detail, with some considerations of possible changes and extensions. This section is one of the weaker parts of the book and seems at times long-winded, without introducing much that is new.
The main thesis of Chapter 5 concerns the referential–attributive distinction, first proposed by Donnellan. After critically reviewing Russellian approaches to the issue, Elbourne presents an analysis within his framework: referential uses involve referential resolution of the situation pronoun inside a definite description, whereas attributive uses involve the pronoun being bound. While this is an elegant integration of some of the core phenomena behind the distinction at hand, it is unclear whether it fully generalizes. One of the key observations by Donnellan was that a sentence with a referential definite could be true despite the fact that the property expressed by its noun phrase is not true of the individual referred to (e.g., The man with the martini is wealthy could be judged true even if the man referred to is drinking water out of a martini glass). This is not captured by Elbourne's proposal, since the situation introduced by the situation pronoun has to include the individual referred to, but nothing in such a situation will meet the description. Thus, the proposal seems incomplete. The last section of this chapter deals with predicative uses (as in Scott is the author of Waverly), which are analyzed as either involving a copula relating identical things to one another or a type shifter to the same effect.
Chapter 6 presents an analysis of anaphorically interpreted definites. The analysis of donkey-anaphoric uses of definite descriptions is the usual one provided in situation semantics, with the additional twist of using the nuclear scope Q-morpheme to interpret definites relative to the minimal situations in the restrictor. Additionally, the same treatment is also extended to cases that traditionally have been assumed to involve binding (under c-command) of an individual index, which yields a more unified theory than that of Elbourne (Reference Elbourne2005) (situational analyses of covariation under c-command have previously been considered in Kratzer (Reference Kratzer2009: 216) and Schwarz (Reference Schwarz2009: 121)).
The central issue of Chapter 7 is the de re/de dicto distinction, with additional discussion of the expressive power arguments by Cresswell and Kratzer. De re readings of definite descriptions result when their situation pronoun is referential or bound at the top level, identifying it with the topic situation. De dicto readings result when it is bound in the scope of an intensional operator, as illustrated, for example, by embedding under believe (135). The presupposition is argued to be trapped in the scope of the operator here, where it gives rise to a (global) presupposition that the attitude holder believes there to be a unique individual satisfying the noun phrase predicate. The last part of the chapter reviews the arguments by Cresswell and Kratzer for natural language to have the expressive power of quantification over worlds and situations in the object language, and shows how the system in the book derives the desired results (which is straightforward, given the availability of situation pronouns). Not much new is achieved here, but since these results are somewhat under-appreciated in the broader literature, the clear presentation of the issue is welcome and will hopefully help to remedy this situation.
In Chapter 8, Elbourne turns to an argument based on the global existence implications of definites in various embedding environments, which Heim, Kripke, and Elbourne himself have put forward previously as evidence in favor of a presuppositional treatment. Elbourne reviews the original argument, considers some Russellian objections to it, and proposes a novel variation that is immune to these objections. The final argument is based on the perceived inconsistency of the following two examples (examples (31) and (33), 155):
(2) I am unsure whether there is a ghost in my attic.
(3) I would like the ghost in my attic to be quiet tonight.
On a Russellian account, this should receive a perfectly consistent interpretation, since the definite can be interpreted in the scope of the attitude verb, as in the paraphrase ‘I would like there to be exactly one ghost in my attic’. Elbourne argues that his account can explain the inconsistency, though his discussion remains somewhat roundabout throughout much of the relevant sections. Finally, on page 159, we get a clear statement of the analysis, which is directly linked to the treatment of definites under believe in Chapter 7: a definite in the scope of an attitude verb (and with its situation pronoun bound there) gives rise to a presupposition that the attitude holder believes there to be exactly one entity of the right sort. Since the sentences under consideration are in the first person, that means that the speaker himself presupposes this, which is incompatible with the initial statement of ignorance. This indeed does seem to be a very convincing version of the argument, but note that we are not actually shown just how that presupposition relativized to the attitude holder comes about (see below).
The second to last of the main chapters is concerned with incompleteness, i.e., the question of how uniqueness can be reconciled with uses of definite descriptions for things that are not strictly unique (such as the table). Elbourne presents a brief review of five possible approaches, following Elbourne (Reference Elbourne2008a). Unlike in that paper, he argues in favor of an account in terms of syntactically represented situation pronouns, based on a new argument involving the unavailability of sloppy readings for down-stressed definites (for an argument in a rather similar spirit based on contrasts between relational and non-relational nouns, see Schwarz Reference Schwarz2009).
Chapter 10 turns to pronouns, which, following his earlier work, Elbourne argues to be definite descriptions in disguise. However, he reconsiders his earlier claims that they always involve noun phrase ellipsis (narrowly understood), while maintaining the notion that they generally involve a covert noun phrase. An interesting question that this raises – which is not explored in detail – is just how speakers are able to recover the precise content of the deleted noun phrase. As with the analysis of overt definite descriptions, Elbourne provides a more unified account of both referential and covarying interpretations of pronouns than in previous work: situation pronouns do all the work, with no appeal to additional individual indices for referential and (what are traditionally seen as) syntactically bound cases. The remaining sections largely focus on providing evidence for the presence of phonologically null noun phrases with pronouns. This includes gender marking on referential pronouns without an antecedent, descriptive indexicals (Nunberg Reference Nunberg1993), and – as Elbourne suggests, most convincingly – so-called Voldemort phrases. These involve pronouns with relative clauses such as he who must not be named (thus the label). Further issues discussed include details of ellipsis, in particular with regard to certain sloppy readings, anaphora to facts, and problems with focus. In the penultimate section, psycholinguistic evidence based on priming is presented as evidence for the presence of noun phrase material not only in conceptual or semantic terms, but also at a phonological level. Overall, this chapter constitutes a rich and interesting collection of data and theoretical analysis that provides a comprehensive account of many important facts concerning pronouns.
The concluding chapter sums up the findings in the book in very brief terms. Turning to an overall evaluation of the book, its greatest strength is the combination of a precise formal system with a wide range of central phenomena from the extensive literature on definite descriptions, along with many novel observations and arguments. The book will be of great interest to linguistic semanticists and philosophers of language alike, and one of the virtues of its clear and transparent presentation is the way that it makes results from linguistic semantics accessible to a more philosophical audience (and vice versa, at least to some extent). Its comprehensive approach constitutes a rather impressive endeavor, and the commitment to explicit and detailed formalization is laudable. Given that the overall analysis is crucially couched in a situation semantics, the book also serves as a detailed introduction to both the conceptual motivation and the formal implementation of such a system.
There are, however, some shortcomings of the book. First, while great efforts are made to show that the proposal captures available interpretations, problems of over-generation are not considered as thoroughly. Covert formal operators are at times introduced a bit too hastily without further motivation. For example, not all possible placements of the various situation binders are consistently considered. Furthermore, a plethora of variants of the Q-morpheme that introduces the scope of quantificational constructions is introduced. These are not always innocent, or in any case inconsequential, in that they introduce variation in the ability to evaluate situation pronouns relative to the quantificational restrictor (e.g., the version Q M for modal adverbial quantifiers does not do that; this predicts, e.g., that necessarily, in contrast with always, does not allow for sage-plant examples (discussed below)). That does not necessarily suggest any deep problems, but a more principled discussion of the variations would have been welcome. Relatedly, the appeal to a fictional operator, FIC, that is freely available, seems like a very powerful tool whose consequences need to be carefully considered in order to ensure that there is no serious over-generation.
Furthermore, in the treatment of the presuppositions of definites in intensional contexts, the formalism is not spelled out in sufficient detail. The empirical claim – which seems appropriate and is fairly standard in the literature – is that presupposition triggers under attitude verbs give rise to the global presupposition that the attitude holder believes the presupposition introduced by the trigger. However, Chapter 7 does not formally show how the presupposition gets relativized to the attitude holder, beyond assuming that this involves situation-binding in the scope of the attitude verb. One might consider that it results from universal quantification over worlds by the attitude verb. But that would not extend to attitude constructions involving existential quantification:
(4) I consider it possible that the ghost in my attic will be quiet tonight.
This still seems to have the presupposition that the speaker believes there to be a ghost in the attic (and is thus inconsistent with (2) above). Furthermore, Elbourne makes parallel points for possibility modals:
(5) Possibly the knave stole the tarts.
Thus, the issue is not confined to attitude verbs. In footnote 8 (on page 59), we are told that a structure with a situation binder in the scope of possibly, which would be parallel to the analysis of the attitude case, will be uninterpretable. But it remains unclear why, precisely because other propositional operators such as attitude verbs (135) and necessarily (65) are analyzed in this way.
The lack of clarity and explicitness in this regard is unfortunate, given that it is absolutely key for a number of arguments throughout to trap presuppositions under (certain) propositional operators: in the analysis of presuppositions in modal and conditional contexts (59, 65), the analysis of attitudes (135), as well as the account of existence entailments (167–168). All this does not undermine the empirical argument. But given the centrality of the issue, we would really want to see just how the relevant interpretations come about. On a related note, the interconnections between the various issues considered could have been explored more deeply. One example is the issue of domain restriction in attitude contexts, which fails to be incorporated since attitudes are analyzed as quantifying over entire worlds.
Another set of issues with the book arises with regard to the omission of some central topics. Perhaps most prominently, this is the case with possible problems for the uniqueness component of the analysis. The only issue with uniqueness discussed in the book concerns incompleteness and domain restriction. But the literature is full of other potential problems, e.g., in relation to ‘weak definites’ in the sense(s) of Poesio, Barker, and Carlson, as well as in the analysis of donkey sentences and their weak and strong readings. A related, and highly surprising, omission concerns bishop and sage plant examples. The former involve introducing additional entities meeting the noun phrase description in the nuclear scope, as in Every woman who bought a sage plant here bought eight others along with it (Elbourne discusses similar issues with covariation under c-command, but draws no connection). These are handled straightforwardly by Elbourne's theory, because of his formulation of the Q-morpheme that allows definites in the nuclear scope to be evaluated relative to the restrictor situations, thus avoiding a violation of uniqueness. It would have been useful to highlight this. Bishop sentences (as in If a bishop meets another bishop, he blesses him), on the other hand, cannot be captured on the present analysis, because they involve multiple individuals meeting the noun phrase description in the restrictor situations, and are thus predicted to violate uniqueness. Elbourne (Reference Elbourne2005) proposed to analyze these in terms of domain restriction, but this only works if there is a non-situational mechanism for domain restriction available. Thus, the proposal in the present book cannot capture these, which at least should have been acknowledged. Another problematic data point (which is only alluded to in a footnote referring to a footnote in Elbourne Reference Elbourne2008b) is one Elbourne himself has raised, namely the contrast between definite descriptions (the cat of Mary's) and Saxon genitives (Mary's cat), which Elbourne (Reference Elbourne2005: 144) argued to speak against an analysis of binding under c-command in terms of situation variables.
More broadly, there are cases of lack of engagement with relevant literature. One such case concerns the sensitivity of pronouns to salience, as laid out in Craige Roberts' work. And even the book's central antagonist, a Russellian approach, is only construed within the limits of existing proposals. One obvious variation would be a Russellian account couched in a situation semantics. Many of the book's arguments indeed are driven by details of the situation-semantic setup – why should a Russellian not be allowed to take advantage of this without buying into a Fregean theory of definites? Such considerations would have allowed us to see more clearly to what extent the success of the proposed analysis relies on such a view of definites, and to what extent the situation-semantic framework does the crucial work.
There are various other issues whose discussion would have enriched the book substantially, but whose omission is perhaps more naturally justifiable in terms of space constraints. First, referential interpretations of situation pronouns naturally call for a theory of how reference is resolved at this level, parallel to reference resolution for personal pronouns. Secondly, the implications of the role of situation pronouns for various of the key analyses at a more general level will need to be assessed. For example, the analysis of presuppositional phenomena with definites, including projection, rests crucially on the presence of situation pronouns. If this is a general theory of presuppositions, does this mean that all triggers must come with a situation pronoun argument, and if so, is that desirable? As a final case in point, it seems important to mention relatively recent developments in the discussion of definites cross-linguistically (for a brief overview, see Schwarz Reference Schwarz2013), which offer both a richer perspective on the range of interpretative options of different forms of definite articles and important issues for the interaction of definites with surrounding linguistic structure, e.g. relative clauses (Wiltschko Reference Wiltschko2013). Finally, recent evidence from sign language has revived an analysis of pronouns based on individual indices (Schlenker Reference Schlenker2011).
These criticisms notwithstanding, this book is an extremly valuable piece of scholarship on definite descriptions, and will be a standard piece of reference for the foreseeable future. Given its clarity of exposition and formal explicitness, it will greatly facilitate and enhance future discussions – both in fleshing out a situation-semantic perspective further and in critically evaluating it in relation to other approaches.
