The syntax of the leftmost domain(s) of the clause and its interplay with discourse, the speaker’s coordinates, and those of the world in which a proposition is uttered or interpreted have been the focus of an immense amount of linguistic research in the last decades, in particular since Rizzi’s (Reference Rizzi and Haegeman1997) seminal paper on the fine structure of the CP. Some of the questions that have maintained the interest of the scientific community to date, and still have not been answered in a unitary way, include whether – and if so, to what extent – the makeup of the left periphery can be assumed to be cross-linguistically stable; what formal mechanisms (e.g. Move and Merge) lead to linearizations in this domain and how the projections situated there, whether they are positioned intra- or extrasententially, determine the licensing of the phenomena overtly realized in the clause; and how the interaction of syntax, pragmatics, and semantics can be effectively explained in a formal model.

Shigeru Miyagawa’s monograph Syntax in the treetops is an excellent and much-needed attempt to shed light on some of these long-burning issues. Looking at data from Japanese and other languages, the author explores the very role of syntax in connecting the contextual features and the semantics of an utterance, proceeding from the assumption that this module ‘provides the basic framework that makes the performance of a speech act and the conveyance of meaning possible’ (x). In particular, Miyagawa assumes with Krifka (Reference Krifka, Roeper and Speas2014) that the realization of any speech act involves the presence and activation of given ‘actors’ (at least a speaker and a hearer) endowed with given linguistic attitudes and correlates (an intention, the capacity to use language in a persuasive way, etc.) and that these factors are represented in the syntax, specifically in the left periphery. The explicit objective of this work is to demonstrate (and provide empirical evidence for) the existence of dedicated projections for the encoding of these coordinates in the left domain of the clause and to show how they work and correlate with syntactic phenomena.

Miyagawa’s monograph is organized in six chapters preceded by a foreword, preface, and list of abbreviations. The last chapter is followed by the section ‘Notes’, the references, a name index, and a subject index (234 pages in total).

In Chapter 1 (‘Setting the stage’, 1–36), the author discusses the notion of ‘root (clause)’ and its implications for the representation of speaker and addressee in the syntax. He starts from Emonds’ (Reference Emonds1970) definition of a root as a universal phonetically unrealized superordinate structure, which contains the speaker, the addressee, and a so-called ‘performative predicate’ (the latter being problematic for a number of reasons, as discussed in 1.2.2). As Miyagawa points out referring to recent work by Frey & Meinunger (Reference Frey, Meinunger, Molnár, Egerland and Winkler2019), this term has come to identify at least three types of phenomena variously labeled in the literature: (i) strongly root-sensitive phenomena, exclusively licensed in genuine main clauses (i.e. propositions realizing fully-fledged illocutionary acts), like hanging topicalization; (ii) weakly root-sensitive phenomena, licit in main and a restricted class of subordinate clauses, such as modal particles, and; (iii) phenomena that do not exhibit any root sensitivity whatsoever, e.g. right dislocation. In the following sections, the author presents selected cross-linguistic arguments from the literature of the last three decades in favor of a formal representation of the speaker’s and/or the addressee’s deictic coordinates in the syntactic structure: logophoricity in some African and native American languages, but also in English, where this feature is externalized by self-pronouns; allocutive agreement in Basque; sentential particles in Romanian; and temporal coordinates in English and Italian. Miyagawa then goes on to set the agenda for the following parts of the monograph: he will propose a revision of Krifka’s (Reference Krifka, Roeper and Speas2014) multi-layered model of the representation of speaker and addressee in the clausal architecture (1a) (in which, from bottom to top, the TP contains the clause with its truth value, JudgP encodes the speaker’s own attitude towards the proposition, CommitP is related to their liability for the content expressed in the utterance, and ActP is the locus of illocutionary force) as (1b) (where ActP is replaced by a revised version of Haegeman & Hill’s (Reference Haegeman, Hill, Folli, Sevdali and Truswell2014) – originally Speas & Tenny’s (Reference Speas, Tenny and Di Sciullo2003) – saP, which in turn contains the speaker’s and the interlocutor’s features, and JudgP is integrated into Rizzi’s (Reference Rizzi and Haegeman1997) Split CP).

In the last part, the author delineates the argumentation developed in each of the following chapters. Even though the premises made in this chapter are very clear and well presented, I find the choice of labeling the highest projection ‘S(peaker-)A(ddressee)P’ strategically less than optimal, especially because it is explicitly claimed to replace an (uncapitalized) ‘s(peech)a(ct)P’, to which a formally almost identical acronym corresponds. This is even more confusing since Haegeman & Hill (Reference Haegeman, Hill, Folli, Sevdali and Truswell2014) additionally propose a (capitalized) ‘SAP’ that is assumed to structurally dominate saP. Irrespective of whether this was a conscious ‘conservative’ decision or not, any other label – or just leaving the acronym and its correlates ‘untouched’ – would have been in my view more effective.

In Chapter 2 (‘The SAP and the politeness ϕ-feature’, 37–88), Miyagawa provides a substantial amount of evidence, mainly from Japanese (honorific affixes -mas- and -des-), but also in a comparative perspective with Basque, Magahi, Korean, Thai, and Tamil, supporting the idea that the addressee is syntactically represented in SAP. In a nutshell, the author proposes, on the basis of the function and syntactic position of -mas-, that the allocutive ϕ-feature, which this element overtly lexicalizes, begins in a low position between negation and the vP (where -mas- is obligatorily spelled out) and automatically moves all the way up to SAP via every head on the way, thereby providing additional evidence for head-to-head movement.

Chapter 3 (‘The SAP, CommitP, and sentence final particles’, 89–126) investigates structures involving the CommitP layer. The author convincingly shows that in Japanese, the sentence-final particles ne and yo are to be conceived as lexicalizations of the A(ddressee)-component of SAP and CommitP, respectively. The former, indeed, is used to ask for confirmation of a claim, while the latter expresses certainty and seems to behave like a German modal particle (e.g. Hanako-wa ki-mas-u ne?/yo!, lit. Hanako-top come-hon-prs ne/yo, ‘Hanako will come, right?’/Hanako will come!, 110). The two particles can co-occur in one and the same sentence, but only in the order yo > ne, which suggests that they are situated in different loci that Miyagawa identifies with the two above-mentioned projections – given the structure in (1b): ([SAP ne [CommitP yo [CP]]]).

In Chapter 4 (‘Is the judgment phrase needed? A view from topicalization’, 127–162), Miyagawa discusses comparative data concerning the relation between the information-structural status and the syntactic position of topics in English, German, Japanese, and Spanish and argues against JudgP as part of the clausal spine of languages; in particular, he contends that what other authors (Krifka (Reference Krifka, Roeper and Speas2014), but also Frey & Meinunger (Reference Frey, Meinunger, Molnár, Egerland and Winkler2019) in relation to weakly and strongly root-sensitive topicalization in German) call ‘JudgP’ is not a permanent, i.e. universal and always present, projection but rather an ‘occasional’ position found in some contexts of some languages, which is better included into an extended C-domain à la Rizzi (Reference Rizzi and Haegeman1997).

Chapter 5 (‘Questions and the commitment phrase’, 163–195) is devoted to the relation between the intermediate treetop projection CommitP assumed by the author and (the expressive component of) interrogativity, in particular to the issue of what exactly is being committed by whom when a question is asked. Miyagawa concludes that in an interrogative question, ‘the speaker commits to the goal of having the addressee commit to p, while the addressee commits to (some sort of) p’ (195). On the basis of the syntactic behavior and semanto-pragmatic properties of the Japanese question particle no, exhaustivity (which is part of the commitment domain) is contended not to be automatically included in the expressive component of questions but rather to instantiate a feature that needs to be overtly realized by/in the C-area. Miyagawa argues that exhaustivity and non-exhaustivity do not require two different commitments. Interestingly, the data discussed in this chapter are strongly reminiscent of one hitherto neglected particle of German (one special type of fully optional middle-field so) that in fact seems to play a similar role as Japanese no, but ‘the other way round’: questions in general, which in German do not involve any particular particle, can be interpreted, depending on the context and on a number of other factors, as exhaustivity- or non-exhaustivity-oriented, but when the particle so does surface in the clause, the sentence is read as requiring some kind of non-exhaustive answer, as exemplarily illustrated in (2), which also shows that so, just like no, is apparently licit in any wh-context, but does not seem to be ideally compatible with why-questions (191):

Assuming that so is a (modal) particle and admitting the existence of a left-peripheral projection responsible for licensing it (cf. e.g. Coniglio Reference Coniglio2011), the optionality of this element can be taken to suggest that at least in German (and perhaps differently from Japanese), even non-exhaustivity requires overt marking under specific conditions. This would imply, then, that it cannot be excluded that in some language and under certain premises, exhaustivity and non-exhaustivity may correspond to distinct mechanisms and different technical implementations. This observation could pave the way for further investigations from a comparative perspective.

Chapter 6 (‘Concluding thoughts: the uniqueness of human language’, 197–200) reiterates the focus, objectives, research questions, and major findings of the treatise and briefly discusses them in relation to each other. In a last note (199–200), the author underlines the exceptional nature of human language as opposed to the communication systems of other animals, whose core – at least with respect to this work’s topic – he identifies in the existence of the projection CommitP.

Miyagawa’s monograph is a compelling, very well-written, and extremely insightful look at the syntax-discourse interface from a generative perspective. The structure of the book, in particular the sequence of the chapters, is well thought out and effectively leads the reader through the author’s argumentation, which is built in a coherent manner and systematically embedded in the ongoing discussions on the relation between grammar and semanto-pragmatics in the literature. I find the idea of repeating the proposed makeup of the left periphery for the reader’s convenience very appealing – and helpful, in order to recapitulate the points made in the previous chapters, in case the book is not read in full, but selectively.

For the reasons elucidated above, this monograph not only is a brilliant contribution to the research but can also be fruitfully employed as a resource for teaching syntax at the graduate level.