BOOK REVIEW
Kayne, Richard S. [1994] The Antisymmetry of Syntax. Linguistic Inquiry Monograph TwentyFive. The MIT Press: Cambridge.
Reviewed by Milena Slavcheva, Bulgarian Academy of Sciences
The monograph consists of ten chapters grouped in four parts. In Chapter 1, "Introduction and Proposal", R. Kayne formulates the basic statement of his theory, the Linear Correspondence Axiom (LCA). The formalized syntactic structure of natural languages contains several groups of relations that are represented differently in the different theories, thus determining the various kinds of approaches to syntactic phenomena. A fundamental relation is that between a hierarchical representation of sentential elements and their linear order, i.e., between phrase structure and word order. It is exactly here that R. Kayne's innovation lies. Contrary to the standard assumption that a given hierarchical representation is associated with more than one linear order, and the wide spread separate treatment of phrase structure and linear order, the author formulates the following hypothesis: "I will argue that phrase structure in fact always completely determines linear order and consequently that if two phrases differ in linear order, they must also differ in hierarchical structure." [p.3] This idea is developed as a formal representation where the following relations are considered: the relation of linear ordering of terminals, the dominance relation of nonterminals, the relation of ccommand of nonterminals, the dominance relation between nonterminals and terminals. Each one of the relations has specific properties. The linear ordering of terminals is transitive, total, and antisymmetric. The dominance relation on nonterminals is not a linear ordering, and it is transitive and antisymmetric, but is not total, i.e., there can be two nodes in a given phrase marker such that neither dominates the other. R.Kayne finds the point where all the properties of dominance and linear ordering coincide, that is when dominance is restricted to the set of nodes dominating a given node. In this sense dominance becomes locally total and consequently, a locally linear ordering. [p.4] The relation of ccommand of nonterminals is transitive, but neither antisymmetric (two sister nodes can ccommand each other and that is symmetry), nor total. R.Kayne first adds antisymmetry to ccommand and then restricts phrase structures to binarybranching ones, thus turning ccommand into a locally total, and hence into a locally linear relation. In this way, the author gets two locally linear relations on nonterminals, i.e., dominance and asymmetric ccommand, thus providing the possibility for "pursuing the intuition that there should be a very close match between the linear ordering relation on the set of terminals and some comparable relation on nonterminals. By comparable, I now mean locally linear." [p.5] R. Kayne takes the locally linear relation of asymmetric ccommand of nonterminals "to be the one that is closely matched to the linear ordering of the set of terminals." [p.5] The relation of dominance between nonterminals and terminals is responsible for transmitting the correspondence between linear ordering of nonterminals and terminals. This matching is the Linear Correspondence Axiom (LCA). The LCA is the central statement in the monograph, it is the basis for all syntactic representations in the rest of the chapters. After the formulation of his central proposal, the Linear Correspondence Axiom (LCA), in Chapter 2 R.Kayne illustrates the application of LCA within the phrase markers of Xbar theory providing some of the main postulates of that theory with explanations directly derived from the LCA. Chapter 3 treats the fundamental problem of adjunction. R.Kayne introduces some refinement to his theory of phrase structure in order to include specifiers and adjoined phrases into the phrase markers. That is achieved by the introduction of segments in phrase structures and the distinction between a segment and a category. In such a way, the author ensures antisymmetry, the main property of phrase structures necessary for the correspondence between the hierarchical structure of nonterminals and the linear order of terminals. R. Kayne considers adjunction as an operation involving heads, nonheads, clitics, specifiers, and formulates rules about what can be adjoined to what, in how many and what kind of steps. He gives several generalizations about the adjunction of syntactic units which are important for the further development of the theory in the following chapters: 1. A nonhead cannot be adjoined to a head. [p.19] 2. Multiple adjunction to a head is not allowed, thus, for example, in the adjunction of heads to heads "sequences of clitics must not be analyzed as successive adjunctions to the same head". [p.21] Empty functional heads are used as a device for building the phrase structures. 3. The adjunction of more than one nonhead to a given nonhead is impossible. [p.22] 4. Adjunction of a head to a nonhead is systematically unavailable. [p.32] Chapter 4, "Word Order", contains basic assumptions of the theory represented in the monograph. It is in this chapter that R.Kayne formulates in the form of linguistic universals the linear orderings of the main syntactic constituents. R.Kayne claims that a specifier and a complement are always on the opposite sides of the head. What is more, he concludes that "specifierheadcomplement, and not the reverse, is the only order available to the subcomponents of a phrase". [p.36] In the mapping between a phrase marker and its terminals, the ordering of terminals is understood as linear precedence, not subsequence. Another generalization about linear order concerns the adjunction of heads. "The present theory has as a necessary consequence that an adjoining head will invariably precede the head that it adjoins to." [p.38] It is also derived from here that a clitic always precedes the head that it adjoins to. In the same chapter, the author turns to the structure below the word level and tries to explore how the theory works at the morphemic level. The rules of the presented syntactic theory are applied to subword structure and the linear ordering of morphemes is considered. We can define the chapters from 1 to 4 as those representing, so to say, the static part of the syntactic model. Word order is extremely abstract and fixed, thus the main goal of the author is achieved: the formulation of a theory of syntax that is as restrictive as possible. The range of possible phrase structures, as statements of Universal Grammar, is reduced to a minimal set. But the word order variations of natural languages must be expressed. How is that done? Once the underlying order is fixed, there come into use different combinations of movements. Here again the basic principle of restrictiveness is observed. In the rest of the chapters R.Kayne tries to find the most plausible movement rules. As a natural consequence to the uniquely imposed specifierheadcomplement order, movements are also highly restricted to given positions in the role of suitable landing sites for moving constituents. The principles of structure, word order and movement require the introduction of a great number of abstract heads as an important device for constructing the phrases of the formal description according to the basic assumptions of the theory. In chapter 5 R.Kayne makes the important conclusion that the Linear Correspondence Axiom applies to all syntactic representations having in mind that in the tradition of transformational grammar there are several levels of representation such as Dstructure, LF, and PF. In accordance with the universal word order of heads and complements, R.Kayne claims that head movement is always leftward. Structures in different languages, including the so called headfinal languages, are derived in such a way that this generalization is preserved. In fact R.Kayne concludes that "no movement rule can adjoin anything to the right of anything" [p.71], since rightward adjunction is generally prohibited in the theory. Chapters 6, 7, 8, and 9 are dedicated to the application of the theory represented in the previous chapters to different syntactic phenomena. Arguments are given for the acceptance of one phrase structure or another as the formal description of certain facts of natural languages. Syntactic structures from English, French, Spanish, Italian, Hungarian, Japanese, Dutch are analyzed. Chapter 6 deals with coordination. The role of coordinating conjunctions in the phrase markers is considered. A principled account is given for the coordination of heads, including clitics, for the coordination with "with", for right node raising. All considerations obey the basic statement that rightadjunction is forbidden. Chapter 7 treats complementation. Here the basic assumptions determining the representation of structures are the following: a head is not permitted to have more than one complement; small clause analysis is included as a possible representation of phrase structure; heavy NP shift is reinterpreted involving leftward, not rightward movement, with a special consideration of what exactly is moved; rightdislocation constructions are also reinterpreted according to the rules of the theory entirely prohibiting any type of rightward movement. Relatives and possessives are the topic of Chapter 8, namely, postnominal possessives in English, relative clauses in English, Nfinal relative clauses, reduced relatives and adjectives, nonrestrictive relatives. The chapter deals with very concrete issues, all of them illustrating how the Linear Correspondence Axiom and the accompanying principles work in practice. Chapter 9, entitled "Extraposition", describes relative clause extraposition, result clauses and comparatives. Relative clause extrapositon is reanalyzed as relative clause stranding. R.Kayne reveals the advantages of this analysis and proves its compatibility with his theory. That is achieved through the detailed analysis of concrete syntactic structures. Result clauses are represented in terms of LF raising of some sentential elements, for example "so". Comparatives are considered to display double behaviour: some comparative sentences are an illustration of stranding, and others of LF movement, for example, of "more". Chapter 9 is a conclusion where R. Kayne sums up the basic ideas of his theory and points out the advantages of the proposed representations.
R.Kayne's monograph treats a fundamental issue in formal linguistic theories: word order. He first poses his original underlying approach to the problem: he assumes that the linear order of terminals "turns out to be more fundamental to syntax than is normally thought" [p.131], and he claims that it is present at all levels of syntactic representation. He not only makes his general statements, but also gives solutions to word order issues in a number of concrete cases applying his basic proposals about the place of word order in the derivation of syntactic structures. In the monograph, syntactic representation is understood in terms of the research framework of transformational grammar (let's use this name for the well known framework) with its specific levels of structure and operations. The high degree of restrictiveness, an important part of which is antisymmetry, makes R.Kayne's theory extremely configurational with a great number of movement rules and abstract functional heads. The theory is an interesting, clearly defined approach to linguistic structure within the tradition of the big trend of transformational grammar. Undoubtedly, it is a contribution to the armoury of competing formal approaches to natural language. Let me finish with the last sentence of R.Kayne's monograph, revealing the flavour of his formal linguistic approach: "To a significant extent, the LCAbased theory of syntax proposed here allows us to have the all too infrequent pleasure of seeing the theory choose the analysis." [p.132] ===============================================================
Reviewer:
Milena Slavcheva  researcher
Bulgarian Academy of Sciences Linguistic Modelling Laboratory 25A, Acad.G.Bonchev St. 1113 Sofia, Bulgaria phone: (+359 2) 713 2812 email: slavchev@bgearn.bitnet
A proponent of the nontransformational approach to natural language, especially of HPSG. Writing a dissertation on the valency frames of Bulgarian verbs within the framework of HPSG. Having a lot of experience in modelling morphological knowledge and the creation of a large grammatical computer dictionary of Bulgarian.
