This book is about semantic theories of modality. Paul Portner defines modality as ‘the linguistic phenomenon whereby grammar allows one to say things about, or on the basis of, situations which need not be real’ (1). This definition is rather vague and permits a large number of linguistic phenomena to be classified as modal – such as, for example, generic, habitual and individual-level predicates as well as tense and aspect – which not all semanticists would necessarily want to treat under the heading of modality. Portner discusses some aspects of the question of what counts as a semantic theory in the context of presenting cognitive and functional approaches to modality. For him, a semantic theory is a formal theory (a system of ideas) explaining certain facts about the meaning of linguistic elements, facts which should also be able to make predictions about other semantic facts. On this view, most work on semantics qualifies as being descriptive rather than theoretical. It is rather odd, however, to refer, as Portner does, to scholars such as John Lyons or Frank Palmer as representatives of descriptive semantics. It should also be made clear that proposing a formalism for certain semantic phenomena does not automatically render an approach more theoretical.
After an introductory chapter, the book proper starts in chapter 2 with a brief description of ‘Modal logic’, which is essentially based on Hughes & Cresswell's (Reference Hughes and Cresswell1968) classic introduction. Portner points out quite correctly that although modal logic has given many important insights into the semantics of modality, ‘its goal is not to provide a semantic analysis of natural language’ (29). He believes, however, that a linguistically realistic version of modal logic can be developed. As a first step, he proposes to revise traditional modal logic by making the accessibility relation depend on the context of utterance and by allowing multiple accessibility relations. The former allows the inclusion of indexical concepts, while the latter makes it possible to associate different modal meanings with different accessibility relations. Both extensions are very similar to what we find in Angelika Kratzer's work.
Chapter 3 provides a survey of the ‘Major linguistic theories of modality’. Since Portner has a rather restricted view of what counts as a semantic theory, only two major theories of modality are discussed. One is Angelika Kratzer's approach, developed in several stages and with several alterations (cf. Kratzer Reference Kratzer1977, Reference Kratzer1978, Reference Kratzer, Eikmeyer and Rieser1981, Reference Kratzer1986, Reference Kratzer, von Stechow and Wunderlich1991a, Reference Kratzer, von Stechow and Wunderlichb), and the other one is Dynamic Semantics (cf. Groenendijk & Stokhof Reference Groenendijk, Stokhof, Kálmán and Pólos1990, Reference Groenendijk and Stokhof1991). For Portner, functional and cognitive approaches to modality do not provide a coherent theory of modality, and he sees their importance mainly in having introduced the concepts of subjectivity and intersubjectivity.
Kratzer's work is discussed extensively in this book. Portner first shows why the original version of her theory of modality (Kratzer Reference Kratzer1977) is unsatisfactory: it does not account for graded and comparative modality, it cannot handle the paradoxes of deontic logic, and it does not do justice to certain problems raised by conditionals. Tracing the sources of these well-known problems and sketching their history in modal logic, Portner then discusses how these shortcomings are overcome in Kratzer's later work. He also shows that the idea of preference ordering is not alien to modal logic and identifies some weaknesses in Kratzer's more recent theory. One weakness concerns her treatment of graded modality, which cannot account for all distinctions in probability, and Portner suggests a solution within Kratzer's framework.
As for Dynamic Semantics, the basic idea is that the meaning of a sentence is its potential to update the context in which it is used. This update may concern either the knowledge (‘the information state’) of some participants in the conversation or the pragmatic presuppositions of the conversation. The set of mutually accepted (pragmatically presupposed) propositions is the common ground. Since the several versions of Dynamic Semantics have so far been applied only to very few modal expressions, it is as yet unclear whether Dynamic Semantics provides an adequate theory of modality.
Although functional linguists do not propose a full-fledged semantic theory of modality, their contribution to our understanding of modality is important. One of the major contributions to modality in functional linguistics is the discussion of subjectivity and intersubjectivity. The problem of subjectivity (on which more below) has been around for quite some time, with the distinction between subjective and objective modality first introduced in Lyons (Reference Lyons1977). Portner's characterization of the cognitive approach is based mainly on Fauconnier (Reference Fauconnier1985), Talmy (Reference Talmy1988) and Sweetser (Reference Sweetser1990). He also mentions some of the criticisms levelled by cognitivists against accounts of modality in terms of possible-worlds semantics. Portner acknowledges the merits of the Force-Dynamics theory of modality but claims that it falls short of accounting for the central issues in modality.
Chapter 4 is devoted to ‘Sentential modality’. Sentential modality is the expression of modal meaning at the level of the entire sentence. It is expressed, among other things, by modal auxiliaries and sentence adverbs. Their description has always been a core issue in modality. Portner proposes a partly new terminology for the classification of sentential modality and distinguishes between (i) epistemic, (ii) priority and (iii) dynamic modality. The idea behind the term ‘priority’ is that in view of certain circumstances, i.e. rules, wishes or goals, the speaker may prefer one possibility over the other. Priority modals are further subdivided into deontic, bouletic and teleological modal expressions. Dynamic modality, too, has to do with certain circumstances and is subdivided into two subgroups: (i) volitional modality (ability, opportunity, disposition) and (ii) quantificational modality, which involves the existential or the universal quantifier. Of the three types of volitional modality, Portner discusses only ability at greater length – probably because its modal status is rather controversial. The inclusion of quantification under modality seems to be motivated by the fact that the English modals can and will may bring about the existential and universal reading, respectively. However, this is not a general phenomenon; not all languages behave like English. We may wonder if there are other, more convincing arguments which support the claim that quantification is related to modality.
While modality has been classified in many different ways, it is not quite clear what advantages Portner's new classification has over any of the earlier classifications known from the literature, and the brief comparison of his model with some other classifications in chapter 4 fails to convince this reader of the merits of Portner's new classificatory system.
A separate section in chapter 4 (4.2) addresses ‘Issues in the semantics of epistemic modality’. The discussion focuses on (i) the non-truth conditional analysis of epistemic modality, (ii) the nature of subjectivity and (iii) the relationship between epistemic modality and evidentiality – topics which will receive further attention below.
Finally, chapter 5 provides a brief survey of the interrelationship between ‘Modality and other intensional categories’. It discusses the relationship between modality and time as well as modality and mood, and provides an overview of some other topics related to modality, such as conditionals, aspect and evidentiality. Conditionals are uncontroversially related to modality. As for aspect, Portner restricts his discussion to the English progressive and perfect, neither of which needs to be analyzed in terms of modality. In this chapter, he also briefly refers to the notion of ‘sentence mood’, which has received much attention in previous work on modality, especially in German linguistics (see Rosengren Reference Rosengren1992). It is regrettable that Portner does not discuss the problems connected with this notion in greater detail.
Let us now return to the problem of subjectivity. It should first be noted that, as argued in Lyons (Reference Lyons1977), only subjective epistemic modals are claimed to fall outside of truth-functional semantics. Objective epistemic modals are of course truth-functional. The meaning of modals, as pointed out by Portner himself in several places in the book, is not always clear-cut, which is also true of epistemic modals. If there are no structural differences, how do we know whether the interpretation of a modal is objective or subjective? If we want to settle the issue whether epistemic modals contribute to truth conditions, we must look for a language in which the difference between objective and subjective modality correlates with structural differences. Hungarian is such a language (Kiefer Reference Kiefer1983). If one wants to say that the garden is one of the possible places where Bill can be, the modal must be stressed and precede the locative, as is illustrated in (1).
(1) Bill LEHET a kertben. Bill be.can the garden.in ‘Bill can be in the garden.’
If, on the other hand, one has evidence to the effect that the garden is the place where Bill is most likely to be, the locative must be stressed and precede the modal, cf. (2).
(2) Bill A KERTBEN lehet. Bill the garden.in be.can ‘Bill may be in the garden’
Only the sentence in (2), not the sentence in (1), can be paraphrased using ‘perhaps’; and only (1), but not (2), represents a proposition. The sentence in (2) cannot be negated, questioned nor embedded under factives, and it cannot occur in a conditional clause. Thus, it can be concluded that (1) expresses objective possibility and (2) expresses subjective epistemic possibility, and, moreover, that this distinction correlates with structural differences.
It is not quite clear, however, whether the above distinction corresponds to what Portner has in mind. It is certainly not identical to his definitions of subjectivity and objectivity: objective modality need not be based on shared knowledge, and subjective modality is not necessarily scalar. To be sure, the definitions proposed by Portner are not uninteresting; however, it remains to be seen how they will fit into a more complete theory of modality.
There are (at least) two promising non-truth-functional accounts of epistemic modality: one comes from Dynamic Semantics and the other from probabilistic semantics. Portner's discussion of these two approaches is very interesting and revealing because he shows both the merits and the shortcomings of these approaches in considerable detail.
Evidentiality, too, has often been discussed in connection with modality. If linguists argue that evidentials and epistemic modals are the same thing, this view seems warranted only in the case of subjective epistemic modality. The two notions are very close to each other, but it can still be maintained, following Palmer (Reference Palmer1986), that two systems can be distinguished across languages: one based on necessity and possibility, and the other on the source of evidence that the speaker has for what she or he is saying. The latter view is characterized by Portner as follows: ‘evidentiality and epistemic modality together constitute a broader class, and … within this class, what we call epistemic modals and evidentials are two common types’ (264). Unfortunately, Portner does not provide a more detailed discussion of this idea.
The section on evidentiality in chapter 4 is followed by a brief discussion of the performativity of epistemic modals. Portner proposes that when one uses a sentence containing an epistemic modal, one both asserts a modal proposition and makes the non-modal proposition a shared possibility. Consequently, the performativity of epistemic modals is conceived of as affecting the discourse context. At the same time, Portner admits that ‘the theories in this area must be developed in greater detail’ (177).
Volitional modality includes ability, which, once again, is not always considered to belong to modality. Portner's arguments to the contrary are not entirely convincing. The sentence John can swim is not a statement about non-actual situations but rather about the fact that John knows how to swim or John is able to swim. Ability seems to implicate possibility: if John can swim, there must be situations in which he is or was swimming. The fact that in Germanic languages, possibility and ability can be expressed by the same word is just a historical accident. Slavic, for example, has a separate word for ‘can’ as a modal (e.g. Polish móc) and for ‘can’ expressing ability (e.g. Polish umieć). The same is true for Finno-Ugric (Kiefer Reference Kiefer1988). In these languages, the two verbs also have very different morphosyntactic properties. While we can readily agree with Portner that the expressions of ability should not be excluded from ‘the field of true modals’ (197) on the grounds that they do no have the properties of subjectivity and performativity, there are quite a number of other reasons to do so, as noted above.
To sum up, in spite of my critical remarks, this is a most welcome, challenging and insightful book, which provides a good starting point for anyone looking for a current overview of issues in and approaches to modality. It is an invaluable source of information for both students and senior researchers in linguistic semantics.
