Modal Sentences, authored by Anastasia Giannakidou and Alda Mari, presents a novel modal theory capable of accounting for a broad range of modal sentences. Building upon standard modal theory (Kratzer Reference Kratzer, Eikmeyer and Rieser1981), the theory integrates key insights from the authors’ previous work on topics like polarity (e.g. Giannakidou Reference Giannakidou1998) and bias (e.g. Giannakidou & Mari Reference Giannakidou and Mari2018).

This book consists of eight chapters, which can be further divided into two parts. The first part – the first three chapters – equips readers with necessary descriptive facts and theoretical background concerning the modal expressions discussed in this book. On the descriptive side, some of the modal phenomena surveyed, notably modal spread (i.e. the co-occurrence of modal adverbs and modal verbs), are essential to the proposed theory.

On the theoretical side, the authors introduce and develop a series of notions to characterize different aspects of modal sentences. First, they employ nonveridicality to capture the objective aspect of modal sentences, i.e. they do not entail truth. Second, they draw on veridical commitment to capture the subjective aspect of modal sentences, i.e. they reflect linguistic agents’ uncertainty. The authors’ version of commitment is defined as a complex process of truth assessment by linguistic agents. Crucially, unlike existing versions, this version does not place emphasis on the existence of a conversation context. Consequently, it applies to the propositional content rather than to the speech act. Equally novel is the veridical commitment strength scale, devised to distinguish different modal sentences. For example, sentences containing possibly are less committed than those containing must. The last of the authors’ innovations brought to bear in this book is nonveridical equilibrium, which proves useful for describing modal bias. In the state of nonveridical equilibrium, linguistic agents hold no bias toward p or not p. These notions enable the authors to provide a more nuanced and precise characterization of modal sentences.

The main modifications to the standard theory are summarized as follows. First, the modal base is relative to an individual anchor (e.g. the speaker), to better capture the subjective aspect of epistemic modals, and to a tense parameter, to better capture the interaction between modals and tense. Second, the ordering source is absent in the theory. Its restriction function is replaced by a partition of the modal base M(i) into ideal worlds and non-ideal worlds. Its ranking function, meanwhile, is replaced by a new component termed meta-evaluation. It governs the ranking relation between ideal worlds and non-ideal worlds. For example, it can rank ideal worlds higher than non-ideal worlds. It is overtly realized by modal adverbs such as probably and possibly. These modal adverbs contribute to the strength of modal bias. For example, adding maybe to sentences containing the universal modal must weakens the bias conveyed. This treatment enables the authors to capture more flexible patterns of modal bias, since bias is not completely related to modal force (necessity or possibility). Syntactically, the authors posit meta-evaluation as an additional layer above the ModalP, which is motivated by modal spread. Consequently, the semantic core of modals in this theory is much simpler, which predicts that modals can have more flexible uses. Although the authors clearly take great care to make this part accessible to more readers, the formulas they provide are not entirely error-free. For example, in (77), the authors use ‘z ≤_g(w) u’ to mean u is closer to the ideal than z, yet in (78) they use the same expression to convey that z is closer to the ideal than u.

The second part, comprising the remaining chapters, demonstrates how the proposed theory is applied to a variety of modality-related phenomena. These phenomena are cleverly chosen for distinct purposes. The authors focus on the uses of flexible epistemic necessity modals, including the category future (FUT) and unitary modals such as Salish K’a, to justify positing a separate meta-evaluation function. Traditional theories have difficulty explaining why the unitary modals can alternate between necessity and possibility readings, and why FUT in some languages can be non-biased. The present theory addresses these challenges elegantly by assuming that all necessity modals share the same semantic core but differ in which types of meta-evaluation function they can combine with. For example, Salish K’a can combine with an empty or non-empty meta-evaluation function. When the function is empty, the modal conveys possibility. Although the theory can account for these phenomena, some aspect of it remains debatable. For example, it is doubtful whether necessity can be weakened to possibility purely by manipulating meta-evaluation. The co-occurrence of K’a and an empty meta-evaluation function can equally well be interpreted as a weak necessity modal or as a possibility modal.

The theory proposed so far has not been applied to tensed propositions. To complete the framework, the authors examine the interaction between modals and tense/aspect. Specifically, they demonstrate how and why tense can affect modal uses. Crucially, they propose a Now-anchoring rule, which is triggered by modals, to analyze the meanings of various modal–tense combinations. This rule substitutes any free tense variable in TP with the utterance time t _u, which ultimately comes from modals. It applies to NONPAST (i.e. tense for predication and future orientation), which has a free variable, but not to PAST, which lacks a free variable. With this rule, the connection between modals and tense can be established. Consequently, it is not necessary to lexically stipulate which modal readings go with which tense. Finally, in contrast to existing studies, the authors attribute actuality entailment to the interaction between modals and past tense rather than to that between modals and perfective aspect. This assumption has the advantage of explaining why actuality entailment does not arise with epistemic modals.

The next two groups of phenomena to be discussed serve to demonstrate how the theory can be applied to modals or modal sentences of different modal quality (epistemic or non-epistemic). The authors do not focus on typical epistemic modals but rather on evidential markers. The goal is to offer a unified account of evidentials and epistemic modals. The authors report some similarities between the two. For example, different types of evidentials, like epistemic modals of different modal force, yield different strengths of veridical commitment. In addition to offering a unified account, the authors challenge classical views on the precondition for using must and establish a novel one that can license both must and evidentials. They argue that it is not the directness of evidence but rather the lack of knowledge that licenses must and evidentials. Both seeing and hearing are direct, but only hearing or seeing with obstacles can license must and evidentials. This assumption is supported by empirical data and is more convincing than classical views.

In terms of directive modal sentences such as deontic and imperative sentences, the authors offer a new account of deontic modals, which involves three parameters – deontic source (e.g. law), deontic authority (e.g. the speaker) and the deontic subject (e.g. the addressee). Deontic modals also express bias, which is implemented by ranking in the meta-evaluation. In the case of the deontic necessity modal must, worlds where its prejacent proposition holds are preferred. Given the similarities between epistemic modals and deontic modals, the two analyses are highly unified, which renders the authors’ account more elegant than other approaches. The new analysis is then extended to imperatives. The authors capitalize on the similarities between deontic modals and imperatives. Their modal approach to imperatives captures properties of imperatives that competing approaches have difficulty explaining, such as the fact that imperatives can be strengthened or weakened by particles.

Finally, the authors turn to interrogative sentences to demonstrate the theory’s broad empirical coverage. This discussion can be taken as a further consequence of the theory. At first glance, questions appear unrelated to modal sentences, since they need not contain modals. With the theoretical notions introduced in the book, the authors reveal the close relation between questions and modal sentences. For example, questions share the core property of modal sentences – nonveridicality. Additionally, questions differ in the strength of veridical commitments they express just like modals. For example, while non-biased information-seeking questions, akin to possibility modals, express nonveridical equilibrium and signal trivial veridical commitment, biased questions, similar to necessity modals, signal partial veridical commitment. Based on these observations, the authors treat questions as modal sentences and demonstrate how the same mechanism used to analyze modals can be applied to biased polar questions (e.g. high negation questions and ‘really’ questions) and reflective questions.

Overall, this book makes a significant contribution not only to the descriptions of modal sentences but also to the linguistic theory of modality. It presents a wide range of crosslinguistic data on modal sentences, enabling readers to appreciate the commonalities and differences regarding the properties of modal sentences across languages. For example, the authors illustrate the major patterns of evidentials by presenting data from seven languages (e.g. Japanese, Quechua and Bulgarian). In terms of theory, the work can be considered as an example of constructing a new theory out of a standard theory by decomposing one of its components. As discussed above, the authors introduce meta-evaluation to realize the function of ranking, which is an essential component of the standard modal theory. They make full use of this new device to explain the flexibility of modal force exhibited by various types of modal expressions such as FUT and the bias expressed by various sentences such as biased questions. Given the range of phenomena explained, it is fair to say that the new theory constitutes a successful extension of the standard theory.

Beyond its empirical and theoretical contributions, the book’s readability and formatting are commendable. Works in formal semantics tend to be dense with complex semantic notions and formulas. However, the notions and formulas introduced in this book are surprisingly intuitive and accessible. It is entirely conceivable that researchers without a technical background can also grasp the gist of the book. Moreover, the book is formatted like a textbook: each chapter concludes with questions and further readings, and the book closes with a glossary of key notions. This way of formatting appears to serve two purposes. First, it can help readers consolidate their understanding of the novel notions introduced in the book. Second, it can guide readers to explore relevant topics on modality. Most of the questions at the end of each chapter are open questions, which are valuable prompts for new ideas.