Reviewed by: Argument realization Artemis Alexiadou Argument realization. By Beth Levin and Malka Rappaport Hovav. Cambridge: Cambridge University Press, 2005. Pp. 286. ISBN 0521663768. &dollar37.99. Verbs as argument-taking elements show a very complex set of properties, as the examples in 1 illustrate. While the participants in the events described in 1a and 1b are expressed in a parallel fashion, break and hit show divergent behavior in 1c and 1d. (1) a. Jane broke the vase. b. Carla hit the door. c. The vase broke. d. *The door hit. Patterns like this form the core of the area of argument realization, that is, the study of the possible expressions of the arguments of a verb. Ever since Fillmore 1970, it has been known that verbs like break and hit are representatives of a larger set of verbs. Studies on the theory of argument realization aim at identifying the factors that underlie patterns such as these, and researchers have followed a number of different paths in an attempt to find an answer to account for the divergent behavior of the different verb classes. Some have focused more on the contribution of lexical meaning, others on the impact of the syntactic constructions verbs appear in. All of the above issues (and many more) are addressed in Argument realization, the most comprehensive study to date of the relationship between verbs and their arguments. Two main aspects make this book invaluable. The first one is that Beth Levin and Malka Rappaport Hovav succeed extremely well in providing an excellent overview of the rich literature on the subject matter. The second one, and in my view the more important one, is that they provide a bridge between lexical semantics and syntactic research, putting together the results of work from a variety of fields and in a variety of theoretical frameworks. By doing this, the authors have produced a reference book for all issues concerning argument realization that nevertheless shows very clearly (i) where the differences lie between the various approaches, (ii) the different predictions these make, and (iii) how different insights can be combined with one another in order to determine the grammatically relevant aspects of verb meaning. Because of this, the book is rich in information and covers a lot of territory, without, however, leaving the reader scattered. One does not find omissions or oversimplifications. The authors provide useful bibliographic references, which help students and researchers to deepen their knowledge on the specific issues discussed in the survey. The book is an achievement of true scholarship and a priceless contribution to the study of the subject matter. The book contains seven chapters, an introduction, and a postscript. The introduction formulates the questions that any theory of argument realization has to address. L&RH state their aim in concrete terms: this book attempts to provide a synthesis and evaluation of the results of both syntactically oriented approaches and approaches developed within lexical semantics. From then on the whole exposition is directed toward one of the main conclusions, which is emphasized again in the postscript, namely that if we want to understand how verbal behavior works, we have to consider both the idiosyncratic element of verb meaning, the root, and the constructions in which this root appears. Ch. 1 offers the descriptive generalizations that emerge from investigations of argument realization, as well as the methodological issues that arise. The chapter offers the background for what is to follow. Here it becomes clear that, in attempting to locate the grammatically relevant aspects of verb meaning, studying a variety of languages and constructions is necessary. Ch. 2 discusses semantic-role lists. The chapter is more concerned with specifying the problems faced by representations of lexical semantics that take the form of semantic-role lists and it reviews attempts to overcome these problems. Semantic-role lists have the form illustrated in 2, based on Fillmore's (1970) proposal that the difference between hit and break verbs, observed in 1, relate to the different semantic-role lists associated with each verb class. (2) a. break: agent, instrument, object b. hit: agent, instrument, place [End Page 649] For a given verb, the choice of subject and object depends on its...