The concept of grammatical relations (GRs) – subject, object, indirect object, etc. – has a long, illustrious history in the description and analysis of languages. For example, subject, as opposed to predicate, can be traced back at least to Aristotle. Grammars of languages often refer to GRs in the description of case, agreement, and word order. Valence is defined in terms of the number and type of GRs; and voice mechanisms, such as passives and applicatives, are defined in terms of changes in GRs. Various constructions, such as relative clauses and reflexives, make reference to accessibility hierarchies of GRs. In sum, GRs are a central concept in language description and analysis. In fact, the twentieth century was the heyday of the GR for two reasons: first, descriptions of far-flung languages both made use of these concepts and pushed them further than ever before, and second, various linguistic theories, scrambling to keep up with the emerging data, made either direct or indirect use of GRs in their architecture. However, the more the empirical window was opened, the more the core conceptualization of GRs became fragmented, resulting in a lack of consensus about what GRs are and how they should be implemented in linguistic theory.

In this book, Patrick Farrell does an excellent job of surveying the issues pertaining to GRs. He explicates the various concepts and constructions with a rich selection of complex data, and thus the book lives up to its billing as an ideal introduction to the field for graduate students. Examples are drawn from several dozen languages, including Choctaw, Dyirbal, French, Halkomelem, Icelandic, Jarawara, Kamaiurá, Kinyarwanda, Seri, Southern Tiwa, Tagalog, and Farrell's own work on Brazilian Portuguese. Some languages are revisited for various points, and this allows students to build up a picture of each language over the course of the book. Thus, Farrell gives good cross-linguistic coverage without sacrificing descriptive depth. Personally, I am very pleased to be one of the authors whose work he draws on for examples, but this also makes me aware that the analyses that he cites are often somewhat dated. Much work has been done on many of these languages since the references he cites; there is a notable lack of citations from 2000 onwards. Thus, my recommendation for use of this book must come with a caveat that recent work on the languages discussed should also be consulted.

The first 111 pages (chapters 1, ‘Introduction’, and 2, ‘Grammatical relations across languages’) give an insightful survey of the various grammatical relations, making excellent use of illustrative data. Chapter 2 starts with a primer on three types of language (accusative, ergative, and split-intransitive) and the major voice mechanisms (passive, antipassive, inverse, and applicative). Two important constructions in the GR literature, raising and causatives, are not covered. Next, Farrell details some of the complications that arise in languages when the properties usually assigned to a GR are split across more than one noun phrase (NP) in a clause. GRs show several kinds of fragmentation – for example, the notion of subject in ergative languages, the split of intransitive subjects under the unaccusative hypothesis, and the mismatch of case and agreement in dative subject constructions. To aid in understanding the data, Farrell uses the theory-independent and analysis-neutral labels ‘quasi-subject’ and ‘quasi-object’ to refer to an NP that has a restricted subset of the morphosyntactic properties of a subject or object (97). He distinguishes quasi-subjects from oblique subjects (also known as quirky-case subjects or dative subjects), in which the NP exhibits all subject properties except case, and this is a welcome clarification. However, the term ‘quasi-object’ is unfortunately already in use for an oblique NP that lies on the hierarchy in between indirect objects and true obliques, for example the French adverbial pronouns y ‘there’ and en ‘some, any’, which, other than cliticization, have no object privileges (Postal Reference Postal, Postal and Joseph1990). Trying to come up with a theory-neutral set of labels and definitions for GRs is a difficult but laudable task, one that provides an ongoing challenge to typologists.

The next 86 pages (chapters 3–5) delve into the way grammatical relations are treated in several twentieth-century theories. For each theory, Farrell shows how the major language types and voice mechanisms are treated. The first theory Farrell profiles is, of course, ‘Relational Grammar’ (chapter 3), which is built totally on the idea of grammatical relations as primitive constructs, allowing easy comparison across languages with different case, agreement, and word orders. One shortcoming of the now-moribund Relational Grammar is that it presumed certain mappings of semantic roles to initial GRs that were never elaborated. Farrell notes that Relational Grammar brought the concept of unaccusativity to the fore, but met with difficulty when various tests for unaccusativity split among different verbs, suggesting that more fine-tuned semantics was at play. The second theory Farrell profiles, ‘Role and Reference Grammar’ (chapter 4), thoroughly develops the relationship between verb meaning and grammatical relations. By adding a large set of semantic roles as a level in the grammar, the grammatical relations themselves can be constrained to a small set of macro-roles. Another failing of Relational Grammar was not developing precise mechanisms for mapping GRs to the morphology and configurational syntax. This is the forte of the third theory Farrell profiles, ‘Transformational Grammar’ (chapter 5), which Farrell tracks from the (Extended) Standard Theory via Government and Binding to Minimalism. In contrast, making reference to GRs, especially to object and indirect object relations, has proven difficult in this theory. The drive to capture generalizations that refer to GRs has led to increasingly abstract structures that have ended up being almost equivalent to the Relational Grammar notion of GRs as primitives.

For each theory, Farrell gives a set of brief case studies. For Relational Grammar, these include Philippine voice, A (transitive subject)/O (transitive object) reversal in Jarawara, and Icelandic dative subjects; for Role and Reference Grammar, passive and switch reference in Seri and quasi-objects in Indonesian and Brazilian Portuguese; and for Transformational Grammar, Halkomelem applicatives. Each case study homes in on the crux of an issue. Sometimes the reader must flip back to previous chapters to find the fuller range of data and discussion. It is also not made clear how the particular studies contribute to the flow of the discussion. Nevertheless, each case study provides a challenge for a particular theory. Sometimes Farrell gives a novel solution, though sometimes he leaves the issue unresolved.

According to Farrell,

Overall, the book achieves this goal in an accessible and efficient fashion. However, the book lacks a conclusion. Some more explicit comparison of the various theories would have been welcome. Comparative comments are made only in passing. Farrell could have taken one recurrent theme, for example, Icelandic dative subjects, and summarized the similarities and differences of the treatments in the three types of theories. The elements needed for such a comparison are all included in the book, though, so this exercise can be left up to the reader. It also would have been interesting to read Farrell's viewpoint on the role of grammatical relations from a twenty-first century perspective. Probably no scholar has studied GRs more thoroughly, especially as regards the oblique edge of the system, and thus Farrell is in a good position to advise us on the elements that a theory of language must have in order to handle GR phenomena insightfully. I hope that we will hear from him on this topic soon.