This book consists of an introduction plus eight chapters dealing in detail with morphological and/or syntactic ergativity in sixteen Amazonian languages: two isolates, six Panoan, three Jê, three Cariban, one Tacanan, and one Katukinan language. It is the outcome of ‘Manifestations of Ergativity in Amazonia’, a three-year project funded by the Centre National de la Recherche Scientifique (France), for which working papers are available online at http://celia.cnrs.fr/FichExt/Documents%20de%20travail/Ergativite/Introductions_ergativite.htm.

The volume is divided into two parts, entitled ‘Well-established systems: Morphological ergativity’ and ‘Recent diachronic innovations: Syntactic ergativity’. The book provides a wealth of fascinating descriptive detail on ergativity and is a welcome addition to the literature. In what follows I give a brief overview of each contribution, followed by discussion of some of the issues raised.

In the introduction, ‘Manifestations of ergativity in Amazonia’, the editors outline the distribution of ergativity in Amazonia. They discuss three issues recurrent throughout the book: the definition of ergativity, ergativity and grammatical relations, and diachrony. Regarding the first two issues, they raise problems with defining ergativity in terms of ‘subject’ and ‘object’ or likewise, S (intransitive subject), A (transitive subject), and O (object) (see e.g. Dixon Reference Dixon1994). In relation to the third issue, they present (13) the diachronic scenario which is presumed throughout the volume and motivates the division of the book into the two parts, namely, that today's morphology is yesterday's syntax. Ergativity begins, they aver, in syntax, through reanalysis of biclausal constructions (particularly nominalisations) and marked voice constructions (like passives). The introduction includes a useful summary of the chapters, although a map would have been useful for non-specialist readers.

David W. Fleck's chapter, ‘Ergativity in the Mayoruna branch of the Panoan family’, describes ergative patterning in five languages, and argues that it is expanding. This is especially so in Matses, where not only is ergative case-marking expanding into the grammatical system, but also a typologically highly unusual ergative-absolutive system of pronominal enclitics is emerging. A plausible case is made that ergative marking in negative clauses in Mayoruna languages arose when a biclausal negative construction was reinterpreted as monoclausal.

Pilar M. Valenzuela's contribution, ‘Ergativity in Shipibo-Konibo, a Panoan language of the Ucayali’, deals with another Panoan language. The ergative case marker is apparently cognate with the ergative marker in Mayoruna languages. However, it shows more allomorphy; it is distributed all the way across the animacy hierarchy, though there is split according to aspect (84). Unlike Mayoruna languages, Shipibo-Konibo shows pivots, which pattern mostly on a nominative-accusative basis.

Since Camp (Reference Camp1985), Cavineña has been notorious in the typological literature as a language with a complex system of split ergative marking based on a multiplicity of conditioning factors. Antoine Guillaume, in ‘How ergative is Cavineña?’, summarises his arguments (Guillaume Reference Guillaume2006) against this analysis, which he shows is flawed because it fails to draw a distinction between free and enclitic pronouns (107). He demonstrates convincingly that the perceived split is a result of the operation of morphophonemic processes applying just to the bound pronouns. Guillaume also examines whether there is any evidence of syntactic ergativity by searching for coreference constraints in clause combinations.

Bruna Franchetto's contribution, ‘The ergativity effect in Kuikuro (Southern Carib, Brazil)’, provides an overview of ergative orientation in Kuikuro morphology and syntax. Ergative orientation pervades the morphology, but there is no evidence of a pivot, and thus Kuikuro is syntactically neutral. There is a lengthy discussion of the ergative marker heke, which is marred by poor terminology and poor expression. For instance, the gloss ‘perspective’ seems inappropriate, and the suggestion that it marks ‘the distance between two points in space, one of which being the perspective from which the distance is conceived’ (135) is confusing: examples and comments elsewhere in the paper (e.g. 142) indicate that heke marks the reference point, not distance from it. Other examples suggest a connection with focus, on which Franchetto herself comments (138) (see also McGregor Reference McGregor, Josephson and Söhrman2008, Reference McGregor2009).

The chapter by Spike Gildea & Flávia de Castro Alves, ‘Nominative-absolutive: Counter-universal split ergativity in Jê and Cariban’, discusses some unusual split ‘ergative’ patterns. One, which is restricted to future-tense and negative clauses, is person-based: full noun phrases take nominative case-marking, but pronouns follow an ergative pattern. The other split involves the TAM (tense/aspect/mood) hierarchy, and is less convincing as an exception to the general pattern associating ergativity with past tense, perfective aspect, realis mood, and positive polarity, because the ergative patterning is not found in case marking, but in the bound cross-referencing pronominals. Ergative patterning is certainly extremely rare in bound pronominals; however, I am not aware of studies that show that within this domain there is a strong tendency for the ergative patterning to be distributed in the same way as it is in case marking. Thus, for instance, Bickel & Witzlack-Makarevich (Reference Bickel and Witzlack-Makarevich2008: 5) argue that ‘the generalization beyond case marking has no empirical support: tested against a world-wide database on alignment splits in agreement systems, there is no trend for such systems to follow the predictions [of the animacy hierarchy (Silverstein Reference Silverstein and Dixon1976)]’. Indeed, to the extent that it exists, case-marking in these clauses is as expected, that is, nominative, not ergative.

In ‘Ergativity in Trumai’, Raquel Guirardello-Damian explores alignment patterns in the isolate Trumai (Brazil). Both ergative-absolutive and nominative-accusative patterns are present. The former predominate and are found in case-marking of nouns and pronouns, person marking on verbs, and word order. Nominative alignment is restricted to posture auxiliary verbs, indicating ‘associated posture’ (as in, for example, ‘cry while lying’). Guirardello-Damian argues that there is no evidence for the grammatical categories subject and object in Trumai.

Francesc Queixalós's chapter, ‘Grammatical relations in Katukina-Kanamari’, describes the morphological and syntactic domains showing ergative patterning in this Katukina language. The Agent is case-marked by an apparent ergative (also genitive). Queixalós claims that morpho-syntactic evidence argues for ranking the Undergoer above the Agent, and for its alignment with the intransitive subject (259). Thus the absolutive grouping has more subject-like properties than the ergative group (which the author likens to object; 260), at least in the basic ‘ergative clause’ pattern. By contrast, in the derived ‘accusative clause’, it is the Agent that shows subject properties, and the patient that shows the object properties.

The final chapter, ‘The intransitive basis of Movima clause structure’, by Katharina Haude, deals with an unclassified endangered language of lowland Bolivia. An unusual pattern of split alignment is found: the split is conditioned by the relative animacy of the participants. In transitive clauses, alignment depends on a salience hierarchy; the NP denoting the non-salient participant is treated the same as the sole argument of an intransitive clause. In direct marking on the verb, the salient participant is the actor and the non-salient participant is the Undergoer, whereas in inverse marking the roles are reversed. Thus, direct marking patterns ergatively, while inverse marking patterns accusatively.

A number of themes recur throughout the book, of which the following are particularly worthy of comment (and further exploration). First, there is general dissatisfaction with the S, A, O theory that has been prominent in typology in recent decades (see, for example, Dixon Reference Dixon1994), although all authors use some variant of it, often with the qualification that it is useful for ‘expository convenience’ (7).

Second, many contributors rightly find problems with classifying constructions as ergative. I wholeheartedly agree with the following observation by Spike Gildea & Flávia de Castro Alves:

Nonetheless, they, and other authors still seem to want to classify constructions according to some alignment pattern, rather than identify different alignment patterns in different structural components of constructions (as suggested, for example, by McGregor Reference McGregor2009). To assume that the loci of ergativity are to be found in constructions seems no less problematic than to locate them in entire languages.

Third, in some circumstances S and O may be treated alike, while A does not come into the picture. A number of contributions observe that there are problems with calling such patterns ergative. I fully concur, and agree with the editors' suggestion in the introduction that such patterning is better called absolutive.

Fourth, many contributors could have paid more serious attention to constructions. As recognised by various contributors, possession of two arguments does not necessarily make for a transitive construction, which means that the intransitive–transitive dichotomy is inadequate. Nonetheless, in a number of places distinct bivalent constructions are not recognised.

Fifth, concerning the diachronic presumption that morphological ergativity always derives from syntactic ergativity, it is unfortunate that none of the papers in this volume actually presents reconstructions which demonstrate the viability of this pathway of emergence of the ergative. Rather, the arguments are based exclusively on constructional parallels. While I agree that processes of grammaticalisation must be understood in their constructional context (see Hopper & Traugott Reference Hopper and Traugott2003), this does not mean that it is pointless to seek historical sources of ergative markers; surely reconstruction of both constructions and morphemes should be attempted. The evidence presented in Franchetto's chapter, for instance, tends to support the idea that the ergative marker derives from a focus marker. (Although she comments that there seems to be a recurrent semantic theme underlying the perspectival and ergative uses of heke, she fails to make clear what precisely she is proposing.) Nor does the mechanism for the emergence of ergativity proposed by the editors shed light on the origin of ergative case marking in what Fleck refers to as ‘extended intransitive’ verbs (48). Ultimately, evidence adduced in his chapter shows that the editors' diachronic scenario runs into difficulties.

A few other criticisms are in order. A perhaps minor, though very important issue is whether we have zero-marking or no marking of the absolutive in a particular language. Some of the authors in this volume (for example, Fleck and Haude) confuse zero and nothing, and fail to motivate alleged zeros. The volume would have also benefited from more careful attention to style and expression. While the majority of the infelicities do not affect understanding and could have easily been rectified by a copy editor, there are some that do cause problems in interpretation. For instance, while I have some idea as to how to interpret ‘[t]he non-agentive nominalization is construed adding the non-agentive nominalizer’ (137, fn. 12), I am not sure that my interpretation matches the intended meaning. And what does it mean to ‘conjugate two hypotheses’ (149)? Moreover, in some chapters, descriptive detail comes at the expense of thematic coherence. Whilst I love to see detail, in an article in a thematic book one expects that descriptive detail will be directly harnessed to the overall theme. This is not always the case in the chapters here, and some of them seem meandering. The authors could have expended more effort on signposting the text and explaining the relevance of detailed observations. Finally, some unusual phenomena emerge in the contributions, which are accorded little prominence. An example is optional ergative marking (see McGregor & Verstraete Reference McGregor and Verstraete2010), which seems to appear in certain environments in Matses (48–50), Shipibo-Konibo (76, 83–84), and Suyá (188, 189). Unfortunately, almost nothing is said about the possible motivations for usage or non-usage of the ergative.

This volume is an important addition to the literature on ergativity in the world's languages, providing as it does detailed information on ergativity (both morphological and syntactic) in sixteen Amazonian languages. One appreciates the level of fine descriptive detail given in the articles; this represents the primary contribution of the book. However, I have numerous disagreements with interpretations proposed by the authors (far too many to deal with in this review), and found little that satisfied me either in terms of theory or diachrony.