Reviewed by Haitao Liu, Communication University of China

Although the literature on phrase structure is much larger than that on dependency relations, the discussion of dependency relations predates generative grammar. This book adds to the much needed literature on dependency relations. In the foreword, the editors define a dependency grammar as ‘A sentence […] associated with a formal object depicting its internal organization called the syntactic structure [which is] a set of lexical units of this sentence linked together by syntactic relations’ (xiii). It is on the basis of this definition that this volume investigates several important topics of contemporary syntactic research.

In the forward, Igor A. Mel’čuk and Alain Polguère introduce dependency syntax and present the four articles contained in this volume. They argue that the syntactic structure of a sentence has four defining properties: (i) connectedness of the syntactic structure, (ii) directedness of syntactic relations, (iii) strict hierarchical organization of the syntactic structure, and (iv) meaningfulness of syntactic relations. This preparatory knowledge is helpful to understand the other articles in this volume.

After introducing some basic concepts, Igor A. Mel’čuk sketches dependency theory with a demonstration of the existence of three types of dependency relations between two wordforms (xix): (i) semantic dependency; (ii) syntactic dependency, which controls the passive valence of the phrase as well as the mutual linear positioning of its wordforms; and (iii) morphological dependency, in which one wordform controls the inflectional values of the other. Focusing on the syntactic level, M proposes three criteria for establishing a syntactic dependency between two wordforms (word 1 and word 2) in a sentence: the first of these three criteria establishes the presence of a syntactic link between word 1 and word 2, based on determining their mutual linear arrangement and possible prosodic unity; the second establishes the direction of the syntactic link, based on the passive syntactic valence of the phrase word 1-word 2, its external morphological links, and its semantic content; and the third establishes the specific type of the surface-syntactic relation r that holds between two wordforms based on semantic contrast, syntactic substitutability of the dependent subtree, and repeatability of r. These criteria are necessary for a deeper characterization of syntactic dependency and for building a dependency syntax for a language. This article also proposes an illustrative list of surface-syntactic relations for English.

Bridging the gap between dependency and a phrase structure approach to syntax, Sylvain Kahane presents a formal dependency grammar compared with a head-driven phrase structure grammar (HPSG). K provides a lexicalist model of extraction, in which HPSG formalism is used to implement a pure dependency interpretation of this phenomenon. The author argues that the modeling of extraction belongs to the syntax-semantic interface, whereas phrases are only entities of syntax, and therefore, a lexical-based approach to extraction is more economical than a phrase-based approach. K’s study also demonstrates that HPSG formalism can support a dependency approach to syntax very well.

Lidija Iordanskaja and Igor A. Mel’čuk investigate how to establish an inventory of surface-syntactic relations (in general) and valence-controlled surface-syntactic dependents of the verb in French (in particular). The central idea of this article is to fit the techniques for establishing surface-syntactic relations (SSyntRels) in a language into the accepted theoretical and typological frame, which is used for establishing the inventories of other linguistic units. Based on the commonality of the syntactic properties of dependents and three tests, the authors list sixteen SSyntRels. For each SSyntRel described, the authors also supply: (i) the properties of the SSyntRel, (ii) formal types of its dependents, (iii) linguistic comments, and (iv) a justification in the form of a comparison with other SSyntRels.

Jasmina Milićević describes a well-known word order problem of Serbian syntax: the linear placement of clitics. After introducing this syntactic phenomenon, the author proposes a set of rules that allow for synthesizing Serbian sentences that contain clitics within the meaning-text model. Linear positioning of a clitic cluster is carried out by first establishing the linear order of all the constituents of the clause and then processing of all the constituents to determine which ones can or must host the cluster.