The study also highlights the constructs of current linguistic theory, arguing for distinctive features and the notion 'onset' and against some of the claims of Optimality Theory and Usage-based accounts.
The importance of Henk Zeevat's new monograph cannot be overstated. [...] I recommend it to anyone who combines interests in language, logic, and computation [...]. David Beaver, University of Texas at Austin
AUTHOR: Portner, Paul H. TITLE: Modality SERIES TITLE: Oxford Surveys in Semantics and Pragmatics PUBLISHER: Oxford University Press YEAR: 2009
Zhiguo Xie, Department of Linguistics, Cornell University
Paul Portner's book is intended as a systematic survey of contemporary semantic theories of sentential modality, most of which were developed within the framework of possible world semantics. The book has five chapters, with Chapter 4 containing the biggest bulk of theoretical reviews and analyses. A self-contained monograph, it does not assume much familiarity on the reader's part with modal logic or any major theory of linguistic modality, so that those who have just started exploring modality will find the book very accessible and helpful. On the other hand, experts will find the book an excellent resource to turn to when they want to have a quick brush-up on one theory of modality or another. An authoritative figure on linguistic modality, Portner takes an unimposing position when presenting his own ideas (where applicable). Overall, the book is a great pleasure to read.
Chapter 1 contains a very brief introduction to modality and delineates the range of topics to be discussed in the upcoming four chapters. Of the three categories of modal forms -- sentential modality, sub-sentential modality and discourse modality, the discussion in the book is limited to the first category, with the other two left to another venue.
Chapter 2 provides a brief introduction to modal logic, which is an indispensible tool for the formal analysis of modality. Portner's discussion of modal logic is divided into two interrelated parts: a more philosophically-oriented one (§2.2) and a more semantically-oriented (§2.3). In the first part, such essential concepts as frame, model, valuation function, accessibility relation and validity are defined and applied to a simple yet well-designed modal game. In logic, the (in)validity of a sentence is due to its logic properties and devoid of specific content. Modal logic was designed to capture the logical properties of modal expressions. However, for this purpose the frame <W, R> has to be specified in greater details: W is a set of possible worlds and R is defined in terms of concepts like knowledge and rules. Portner defines modal frames in the model-theoretic framework he has concisely outlined and shows how modal logic leads to a better understanding of modality. Towards the end of the first part, Portner presents some axioms and systems of philosophical modal logic, as well as their relationship to frames and validity.
In the second part of the chapter, Portner summarizes a generalization underlying most current formal theories of modal expressions in natural language: the Simple Modal Logic Hypothesis. The hypothesis says that the meaning of every modal expression can be expressed in terms of whether it is a necessity or a possibility modal and in terms of its accessibility relation. It is hard to define a concept of weak necessity so as to capture some people's intuition that there ought to be more options than just necessity and possibility, because the set of possible worlds is standardly assumed to be infinite. A given accessibility relation may have different sub-types; and a given modal may be compatible with many different accessibility relations (perhaps indefinitely many). The most common varieties of modality include epistemic, deontic, ability, bouletic, teleological, and historical. Portner spells out the definitions of the accessibility relations corresponding to some of these modal types. The Simple Modal Logic Hypothesis does not make reference to the identity of the speaker and the time at which the modal sentence is used; that is, it fails to capture the indexicality of modals.
Chapter 3 aims to review major linguistic theories of modality: Angelika Kratzer's 'standard' ordering semantics of modality, dynamic logic theories of modality, as well as cognitive and functional theories of modality. Compared to the 'Modal Logic for Linguistics' as outlined in Chapter 2, in Kratzer's theories pragmatics plays a more significant role in the interpretation of modal expressions. In place of the accessibility relation, Kratzer (1977) uses the concept of conversational backgrounds, which is a function from worlds to sets of propositions. A modal sentence is interpreted with respect to a context and a conversational background. Necessity and possibility modals can be interpreted in terms of entailment and compatibility, respectively. The different readings of a certain modal are determined by the choice of conversational background, and not all kinds of conversational backgrounds can be employed to interpret each modal. The theory put forward in Kratzer (1977), however, does not make reference to comparison and ordering between different possible worlds. This leads to a number of empirical challenges.
A conversational background can be used to define a partial ordering on a set of possible worlds. In Kratzer (1981)'s revised analysis of modality, the interpretation of a modal sentence is doubly relative to two conversational backgrounds: the modal base and the ordering source. Ordering semantics can capture graded modality, comparative modality, the Samaritan paradox and Chisholm's paradox, as well as subtle differences in modal meanings. However, Kratzer's version of ordering semantics appears to fall short of complex expressions of probability and possibility, and the so-called weak necessity modals like 'ought'. To handle this problem, von Fintel and Iatridou (2006) propose that the meaning of 'ought' involves two distinct ordering sources.
The second part of the chapter concerns the dynamic semantics of modality. The dynamic view of meaning takes the meaning of a sentence as an information change potential. Modal operators can be defined in the dynamic logic system. A sentence of the form 'diamond β' checks whether β is compatible with an information state. If the information update is not empty, then the information state is left unchanged. If it is not, it leads to a defective information state. A sentence of the form 'box φ' checks whether the current information state already contains the information in φ. Dynamic logic is useful for tackling non-epistemic modals and such modality-related topics as the logic of expectation often embodied in sentences containing the words 'normally' and 'presumably'.
Despite the fact that functional and cognitive approaches to modality often do not aim to provide a 'serious' semantic theory, Portner devotes the last section of the chapter to the two frameworks. In the cognitive semantic approach to modality, it has been argued that deontic modal meanings are metaphorical extensions of the force dynamical concepts, and the same analysis presumably can be extended to the domain of epistemic modality. However, crucially the view cannot explain the non-actuality of modality, a core property inspiring possible world semantic analysis of modality in the first place. Neither can it provide analysis of the nature of force dynamics of authority, desire, evidence, etc., which relate to hypothetical situations. In addition, Portner shows that the cognitive-functional criticism against formal semantic analyses of modality either does not hold or simply arises from misunderstanding. Furthermore, functional and cognitive approaches to (inter)subjectivity of modality do not have as much advantage as they have traditionally claimed over formal approaches.
Chapter 4 is the most essential part of the book, in which Portner provides reviews, remarks and analyses of a number of phenomena which fall into the broad category of sentential modality. In Portner's classification, there are three primary categories of sentential modality: epistemic, priority (including deontic, bouletic and teleological) and dynamic (including volitional and quantificational). Epistemic modals have often been taken not to fit into standard theories of modal semantics. First, epistemic modals do not contribute to the ordinary, truth conditional semantics of sentences containing them, but affect the type of speech acts performed by the sentences in which they occur. Second, the truth or falsity of an epistemic modality shows a certain degree of relativity to such factors as who evaluates it.
Portner starts his review of the semantics of epistemic modality in §4.2 by examining Papafragou (2006), who identifies two empirical puzzles to argue that epistemic modals lack an ordinary truth-conditional semantics. One has to do with the limitations on the embedding of epistemic modals; the other with patterns of assent, dissent and questioning towards epistemic modal sentences. Portner shows that the non-truth-conditional analysis is not as strong as Papafragou and others had thought. Dynamic logic does not come to their rescue, as it does not explain all the important facts which motivate the non-truth-conditional perspective. In contrast, probabilistic semantics analyzes epistemic modals as involving performativity, in that it identifies three types of speech act which are presumably relevant to epistemic modals. Portner shows that it does not help Papafragou in any direct way.
The nature of subjectivity plays an essential role in the analysis of epistemic modals. Papafragou's claim, that subjectivity arises when the speaker is the sole member of the relevant group that an epistemic modal quantifies over, does not make any reference to the dichotomy of modal base and ordering source. Portner suggests two distinct notions of subjectivity, each based on one of the two types of conversational backgrounds. In addition, epistemic modals are not evidentials, which typically have been analyzed as non-truth-conditional. Analyses of performativity of epistemic modals fall into two types: those that claim an epistemic modal sentence expresses a reduced level of commitment and those that claim an epistemic modal sentence brings up or 'proffers' a possibility. Portner posits that a sentence of the form 'might S' both asserts the modal proposition and adds to the common propositional space the unmodalized proposition S.
In epistemic modal contexts, relativism is the dependency of the semantic meaning of a modal sentence on the context of evaluation. There are two main arguments offered in the literature for the relativist view. One concerns embedding an epistemic modal under a verb like 'say'. Stephenson's (2007) proposal, that the identity of a 'judge' affects the meaning of several kinds of sentences, is helpful in dismissing this argument. The second major argument, the presence of an eavesdropper, can receive alternative explanation as well. In Portner's view, either the epistemic modal or the associated non-modal proposition can be judged for truth or falsity, and certain considerations of relevance can determine which one to use in a particular case.
Portner's review and discussion of issues in the semantics of priority modals concern the argument structure of priority modals and the connection between priority modals and imperatives. Some deontic modals are control predicates and some others are raising predicates. The subject sometimes plays a different role in the semantics of priority modals than in the semantics of epistemic modals. Utterance of priority modal sentences can put the addressee under an obligation, and in this sense priority modals are performative. Based on Lewis's discussion of performativity of deontic expressions, Portner (2007), among others, develops the idea that a priority modal sentence expresses a proposition within the possible world semantics, and at the same time adds the proposition (property) it expresses to the To-Do-List of the addressee. Imperatives and priority modals are thus closely related.
The line between volitional and quantificational modals is murky. Assuming 'can' also can function as a quantificational modal, the subject of a volitional modal in English, which is a control predicate, is invariably an agent or a cause. As for volitional modality, Portner's discussion focuses on ability modals. Common to the major available analyses of the semantics of ability modality is the idea of combining existential quantification, which corresponds to the idea that the agent chooses some action, with universal quantification, which corresponds to the idea that the action guarantees an outcome.
Perfective sentences containing ability modals and certain priority modals entail that the event described in the sentences actually occurred, giving rise to actuality entailment. Bhatt (1998) argues that 'able to' in a perfective context is an implicative predicate, and that in imperfective context a higher Gen(eric) operator removes the actuality entailment. Hacquard (2006) criticizes Bhatt's analysis and derives the actuality entailment of ability modals in perfective contexts through the interaction of event semantics and modal semantics. With Hacquard, the perfective aspect undergoes movement to a higher position binding the event argument of the verb. Through event identification across worlds, the event associated with the verb exists in both the accessible worlds and the actual world, hence actuality entailment arises. Portner points out that Hacquard's analysis cannot rule out actuality entailment for the imperfective cases. Rather, he identifies the actuality entailment as a special case of performativity.
Quantificational modals function as adverbs of quantification and modals simultaneously. They quantify over counting situations (both actual and possible ones). The semantics of quantificational 'will' is identical to that of 'always', except for the accessibility relation between situations and worlds in the former.
In Chapter 5 Portner undertakes the task of examining the relations between modality and several other intensional categories including temporality, conditionals and evidentiality. Modal sentences sometimes contain tense. In some cases the tense is grammatically marked but does not contribute to semantic interpretation (e.g. 'should'), while in some other cases it is grammatically present and semantically real (e.g. 'could'). The scopal relation between an epistemic modal and the perfect shows a certain degree of flexibility, which leads to an epistemic/metaphysical ambiguity. In such cases there are two sources of 'pastness', one from the temporal perspective and the other from the temporal orientation. How to shift the temporal perspective and temporal orientation for a theory of past interpretation of modals is an open issue. The many varieties of modals add to the problem.
As for the future interpretation of modal sentences, some analyses hold that tense is not involved in the future readings, while other analyses agree that futurity is due to an interaction between the modal meaning and some general principle (aspectual semantics, modal type, etc). With regard to 'will', one school of views hold that 'will' is not a future tense but just a modal, whereas others argue it is never a modal and has a purely temporal meaning. Both proposals have their own problems.
Some simple present tense sentences express genericity or a future situation, both being a kind of modality. A past tense can express modality in terms of genericity or unreality (counterfactuality). The progressive expresses modality with a unique combination of accessibility relation and temporal meaning. The perfect expresses some modal meaning as well, though the analyses available in the literature differ considerably with regard to what kind of modality it expresses. Towards the end of the book, Portner points out some major issues concerning the relationship between modality on one hand and conditionals, mood, as well as evidentiality on the other hand. He does not delve into the details of these topics, leaving them to a later venue.
Why is this book an important piece of work? First, as far as I know it provides the first comprehensive, up-do-date survey of a wide range of modals and related phenomena. All types of sentential modality (epistemic, priority and dynamic) receive the extensive discussion they deserve, which integrates across such topics as tense, aspect, subjectivity, and performativity. The wide of range of issues addressed in the book alone makes it a great reference book for those who work on modality.
Theoretical breadth and depth are another merit easily witnessed in the book. Breadth is easily seen in Chapters 2-3, where Portner explains modal logic and major theories of modality. His unbiased review covers not only formal semantic approaches to modality, but also the modal logic, dynamic, functional and cognitive frameworks of modality. Portner provides clear reviews of the many proposals that logicians, philosophers of language and semanticists have made on various aspects of modality over the past forty years or so. He demonstrates great expertise in summarizing complex arguments and presenting them to the reader in a less technical manner. The assessment and comparison he gives of the different ideas are precise and accessible. When the mechanisms and frameworks he examines and/or employs may be difficult for beginners to understand, he works out the details to a considerable extent. For example, on page 68 he uses an illustrative figure to go through the abstract definition for 'necessity modal' in ordering semantics.
And, even more importantly, Portner keeps his critical eyes wide open through the book. He makes another valuable contribution to the project of investigation into modality through pointing out various issues with analyses put forward in the literature. Here is just one example that illustrates this merit. In §5.1.2. Portner suggests that Kissine's (2008) explanation that 'will' is never a modal does not rule out another equally possible approach and thus what Kissine shows does not guarantee his conclusion. There are a lot of other such examples in the book. By doing so, Portner threshes out many open issues in the semantics of modality that await further research. So the book is a valuable resource especially for beginning students of modality to identify potential topics that they can work on.
I am not suggesting that the book is impeccable. It contains a number of typos (for example, on pages 43, 42, 44, 50, 86, 87, 94, 95, 125, 200, 205, 207, 208, 211, 261). In addition, Portner cites Pancheva's (published) work on pages 245-46, but he does not include the reference in the bibliography. Though these are just small points, some readers of the book may find it more or less irritating.
Content-wise, I find the discussion on a few topics covered in the book somewhat confusing and not as easy to follow as the rest of the book. For example, on pages 210-211, Portner tries to argue that Hacquard's aspect-based analysis of ability modality cannot explain the absence of actuality entailment when an ability modal appears alongside the imperfective. However, his argument involves an explicit qualification 'if the actual world is a normal world'. I am not sure if the actuality entailment under such a qualification is still the same actuality entailment that Hacquard is after. On a separate topic, Portner uses modals embedding under conditionals as an argument for the claim that deontic modals have both a truth conditional contribution and a performative contribution. I suspect that the same evidence can be used to support the exactly opposite claim, e.g. one along the lines of Schwager (2006). In addition, the definition for temporal orientation he offers on page 228 makes reference to the temporal relationship between the speech time and the event time. If I understand it correctly, the definition is at odds with what Condoravdi (2002) uses, on which Portner claims to base his definition. Following Portner's definition, the counterfactual reading for 'non-root modal + have' as classified by Condoravdi does not have the future orientation, but stays open with respect to temporal orientation.
Bhatt, Rajesh. 1998. Covert Modality in Non-Finite Contexts. PhD Dissertation. University of Pennsylvania. Condoravdi, Cleo. 2002. Temporal Interpretation of Modals: Modals for the Present and for the Past. In David Beaver et al (eds.). The Construction of Meaning: 59-88. Stanford, CA: CSLI Publications. von Fintel, Kai & Sabine, Iatridou. 2006. How to Say Ought in Foreign: The Composition of Weak Necessity Modals. Paper presented at the Fall 2006 Workshop in Philosophy and Linguistics at the University of Michigan. Hacquard, Valentine. 2006. Aspects of Modality. PhD Dissertation. MIT. Kissine, Mikhail. 2008. Why 'Will' Is Not a Modal. Natural Language Semantics 16: 129-155. Kratzer, Angelika. 1977. What 'Must' and 'Can' Must and Can Mean. Linguistics and Philosophy 1: 337-355. Kratzer, Angelika. 1981. The Notional Category of Modality. in H. J. Eikmeyer, and H. Rieser (eds.). Words, Worlds, and Contexts: 38-74. Berlin: de Gruyter. Papafragou, Anna. 2006. Epistemic Modality and Truth Conditions. Lingua 116: 1688-1702. Portner, Paul. 2007. Imperatives and Modals. Natural Language Semantics 15: 351-383. Schwager, M. 2006. Conditionalized Imperatives. SALT 16: 241-258. Stephenson, Tamina. 2007. Towards a Theory of Subjective Meaning. PhD Dissertation. MIT.
ABOUT THE REVIEWER
ABOUT THE REVIEWER:
Zhiguo Xie is a PhD candidate in linguistics at Cornell University. His
broad research interest is in crosslinguistic semantic variation and
universality. He has worked on modality, questions, NPIs and verb duplication.