The Syntax of Roots and the Roots of Syntax is a collection of thirteen papers from two workshops, held at the University of Southern California (20–21 February 2009) and the University of Stuttgart (10–12 June 2009). Edited by Artemis Alexiadou, Hagit Borer and Florian Schäfer, this volume starts with an introduction by the editors, discussing the morphological concept ‘root’ from several perspectives in generative linguistics and outlining the themes addressed by the volume, in particular the relation between the syntactic and semantic properties of roots. These properties of roots and other word parts fall only within the scope of Minimalist syntax; outside Minimalism, it is different. For example, cartographic syntax (Maximalism) is word-based, and more precisely, feature-based, that is words in the lexicon could be seen as particular combinations of features (see Mao & Meng Reference Mao and Fanjun2016: 921–924).

This volume, in our view, is not intended to produce a ready-made answer to the question of what defines a root, but to bring together various views on the syntactic and semantic properties of roots. More specifically, when addressing the relation between syntactic structure and syntactic terminals, i.e. ‘word-like elements but not words’ (Berwick & Chomsky Reference Berwick and Chomsky2016: 90), the papers mainly explore the syntactic and semantic properties of roots with a focus on the following three aspects: (i) the isolated syntactic properties of roots; (ii) the interaction of root ontology (semantic properties) with the syntax of roots; and (iii) the interaction of roots with ‘phases’, sequences of operations resulting in vP or CP-like structures.

The contributions by Víctor Acedo-Matellán & Jaume Mateu, Hagit Borer, Paolo Acquaviva, and Marijke De Belder, Noam Faust & Nicola Lampitelli concentrate on the isolated syntactic properties of roots. Acedo-Matellán & Mateu argue that the properties of roots depend on their syntactic position. That is, it is the structural position occupied by roots in the event or argument structure that determines their properties (including those properties which are related to meaning), rather than the conceptual meaning of roots contributing to syntactic operations. For example, Acedo-Matellán & Mateu propose that grammatical manner interpretation emerges whenever the root is adjoined to $v$ , which is against the contributions by Artemis Alexiadou and Antje Roßdeutscher etc. in this volume.

In the same vein, within the ‘Exo-skeletal approach’ (Borer Reference Borer2005), syntactic structure provides a skeleton or template into which lexical items are inserted, Borer proposes that the categorial properties of roots emerge in the context of particular functional structure, and as a result of that structure. Functors such as $v$ (verbalizer), $n$ (nominalizer) and D (determiner), etc. are used to define the categorial space. In detail, the determiner D projects a phrasal structure, the determiner phrase (DP), and determines the domain of its complement as noun (N). When a root enters into the corresponding position within such an extended projection, it need not merge with a category label and can be recognized as an N-like element, because the functor D defines the root as an N, such as Acquaviva, opposed to limiting the property of nominality directly to roots, argues that roots differ from nouns, and should not be stipulated as having the semantic function of nouns. Their function is to label the syntactic construction that corresponds to a noun; and lexical roots do not by themselves identify a concept, and what expresses a noun’s lexical semantics is always a syntactic construction.

De Belder et al. assume that, cross-linguistically, diminutives can realize both functional and derivational heads. And the latter, without category of its own, is responsible for what would traditionally be seen as derivational morphology. The category functional head (nominalizer) partitions two different structural domains. Above the functional head and within its functional projection, it is a SizeP (Size Phrase) that realizes functional material and denotes compositional meaning. Below the category head, there exists a lexical head, which merges with the root and forms a LexP (Lexical Phrase), showing non-compositionally or idiosyncratic lexical meanings. These two positions can be filled simultaneously, and independently of each other. For instance, xazarzir-on ‘a small piglet’ in modern Hebrew (MH) can be analysed as This proposal is not restricted to nouns, and can also be used to analyse a verbal head. For example, fischi-ette-are ‘whistle (pluractional)’ in MH can be analysed as

In contrast, contributions by Artemis Alexiadou, Edit Doron, Lisa Levinson, Malka Rappaport Hovav and Antje Roßdeutscher support the interaction of root ontology (semantic properties) with the syntax of roots. Alexiadou argues that prefixes, e.g. afto- (self) in Greek, mirror the relation between root ontology and its interaction with syntax. More generally, the root ontology influences syntactic limitations on merger (structure building) environment. For instance, result roots enter as complements of verbalizers, such as  while manner roots enter as modifiers of that is,  Meanwhile, Alexiadou admits that the structural positions restrict the prefixation. For example, afto- triggers a reflexive interpretation only when it combines with a non-active structure of a causative predicate.

Doron investigates Hebrew adjectival passive participles constructed in the causative template. She distinguishes between dynamic roots (which form resultative participles) and stative roots (which form stative and resultative participles), and demonstrates that the only non-active form of causative-template verbs is the passive, a voice that obligatorily licenses an (implicit) external argument. Thus, the properties derived from roots influence their syntactic realization.

Levinson specifically explores the connections between semantic and morpho- syntactic properties of roots, and shows that interaction between interpretation and morphosyntax can be derived from the semantic types of roots, which means that a root is not semantically vacuous in and of itself. For instance, a truly ‘root creation verb’ means that the theme is the material from which the created object is created. To build the English verb braid from a root  we need to conflate the meaning ‘make … into a braid’ into the word. As a result, the created individual (argument) is denoted by the root rather than by the DP, and the semantic type of braid turns out to be <e, t> (from entity to truth value). In other words, genuine root creation verbs do not license the double object construction. Therefore, the semantic restrictions on composition lead to different structural environments for different types of roots.

With reference to a scalar analysis of incremental theme verbs, change of state (COS) verbs and direct motion (DM) verbs, Rappaport Hovav argues that it is possible to isolate and explicate the grammatically relevant meaning components that are associated with a verbal root and determined in the lexicon. For example, for ascend, a DM verb with the location of the theme argument moving along a path, the stative scalar attribute is lexicalized by a preposition. More precisely, the path is specified as a rising one. Thus, we cannot say ‘The plane ascended from 3000 to 1000 meters’ on the grounds that lexicalized meaning components are entailed in the use of the verb.

Starting with the question ‘How can the semantics of verbs be constructed from their roots?’, Roßdeutscher argues that the lexical meaning of roots can determine whether certain functional heads such as $v$ (verbalizer), and $a$ (adjectivizer) can select them as arguments, because roots have a semantics that is the source of argument structure. To cite eventive or manner roots like  as examples, they are simple event types, and thus directly selected by $v$ .

Ángel J. Gallego and Jean Lowenstamm discuss the interaction of phases with roots. To explore the setting of argument structure on the basis of phase theory, Gallego, follows Chomsky’s (Reference Chomsky, Freidin, Otero and Zubizarreta2008) definition of ‘feature inheritance’, wherein phase heads such as C[omplementizer] or v [light verb] provide non-phase heads such as T[ense] with all their properties, i.e. category labels and phi-features ([ $u$ Person] [ $u$ Number] and [ $u$ Gender]). Within this framework, Gallego proposes that argument structure is established after the categorial label merges or combines with a root. Furthermore, only the roots c-commanded by light verbs (v) with unvalued phi-features take syntactic arguments in terms of featural valuation, and the conceptual arguments are connected with the conceptual contexts of roots. For example, a DP such as the city is semantically related to the conceptual content of the root $\surd$ (DE)STROY in all cases (destroy, destructive and destruction).

Lowenstamm takes as a departure point his own observation that Distributed Morphology fails to analyse cyclically the phenomenon of stress shift in English (e.g. átom, atómic and atomícity), and assumes that domains (phases) for the application of phonological rules should be determined for roots, such that stress shift would apply cyclically to the most deeply embedded category-less root, or the next highest root, beyond which rule application is never cyclic and meaning is compositional. For example, suppose that affixes are roots, and atomic could be analysed as a complex root: In Lowenstamm’s system, is regarded as [u $\surd$ ] (Class One affix, stress shifter). Thus,  can be stressed as the most deeply embedded root, as required by the corresponding phonological rule.

The volume provides readers with comprehensive and insightful viewpoints about the syntactic and semantic properties of roots, which contributes to elaborating the properties of the syntax–semantics interface and the morphology–syntax interface. For example, Borer’s proposal has the potential to stimulate research on morphology–syntax mapping, and Gallego’s model may inspire similar research on the syntax–semantics interface. However, this volume also has some shortcomings. Specifically, some explicit discussion about how to integrate these views about roots into Chomsky’s (Reference Chomsky and Kenstowicz2001: 10) definition of lexical item (which combines syntactic, semantic and phonological features) would make this volume more attractive. On the whole, however, this volume offers a valuable contribution to the debate on the nature of roots and to the investigation of their interaction with syntactic structure. Offering different views on the relation between roots and syntactic structure, it paves the way to deepening our understanding of the nature of ‘atoms of computation which are used to yield structured expressions’ (Berwick & Chomsky Reference Berwick and Chomsky2016: 66), and of the interaction between different linguistic subsystems.