LINGUIST List 28.981

Thu Feb 23 2017

Review: Dido; Tonga; Morphology; Syntax; Typology: Polinsky (2016)

Editor for this issue: Clare Harshey <>

Date: 25-Oct-2016
From: Robin Meyer <>
Subject: Deconstructing Ergativity
E-mail this message to a friend

Discuss this message

Book announced at

AUTHOR: Maria Polinsky
TITLE: Deconstructing Ergativity
SUBTITLE: Two Types of Ergative Languages and Their Features
SERIES TITLE: Oxford Studies in Comparative Syntax
PUBLISHER: Oxford University Press
YEAR: 2016

REVIEWER: Robin Meyer, University of Oxford

Reviews Editor: Helen Aristar-Dry


Since Dirr’s 1928 survey of 35 Caucasian languages, the term ‘ergative’ has been used to describe a morphosyntactic alignment pattern in which the subject of intransitive verbs (S) and objects of transitive verbs (O) are marked in one common way, but different marking is applied to the agent of transitive verbs (A); numerous descriptions of languages exhibiting ergative alignment to some extent have emerged over the course of the twentieth century (Verbeke 2013 on Indo-Aryan; Haig 2008 on Iranian; Grinevald and Peake 2012 on Mayan; Paul and Travis 2006 on Austronesian; to name but a few), but insufficient attention has been paid to the phenomenon of syntactical ergativity, displayed by a small subset of ergative languages, in which various coreferentiality constraints apply in complex sentences (e.g. constraints against S-A-coreference in co- or subordination, or against relativisation of ergative arguments; cf. Dixon 1994:143–180 for an overview).

Maria Polinsky’s monograph ‘Deconstructing Ergativity. Two Types of Ergative Languages and Their Features’ seeks to rectify this neglect in suggesting a new and systematic aetiology of syntactical ergativity: based on a set of diagnostic syntactic comparisons, it proposes that all languages exhibiting syntactic ergativity employ prepositional phrases (PP) as ergative subjects, contrasting with other morphologically ergative languages which have determiner phrase (DP) subjects. The contrast between DP-ergative and PP-ergative languages is explored in some depth, both theoretically in a transformational framework and by means of copious examples from, inter alia, Niuean, Samoan, Chuckhi, Q’anjob’al, Archi, Avar, and Circassian; ̥more than a third of the book is dedicated to a detailed discussion of two paradigm languages, Tongan (PP-ergative) and Tsez (DP-ergative).

The first part of the book establishes the necessary theoretical background and hypotheses for this investigation, the latter of which are then tested against corpus data. In the introduction (Chapter 1), ergative alignment and its manifestations in the languages of the world are discussed with a view to their syntactic peculiarities. Here, syntactically ergative languages are narrowly defined as the subset of morphologically ergative languages in which the ergative argument cannot undergo A-bar movement (e.g. in relativisation, wh-questions, or topicalisation) with a gap, whereas absolute arguments may. Forgoing a discussion of languages like Dyirbal, whose status as syntactically ergative is disputed, Polinsky sets out to provide a principled syntactic analysis and differentiation of ergative languages based on their constraints on A-bar movement. In this process, a number of other potential correlations between types of ergativity and other syntactic properties are proposed (see below, ch. 6). A brief appendix outlines strategies used by syntactically ergative languages to overcome this constraint, discussing, among others, antipassives, resumption of the moved argument, and nominalisation.

The specific reasoning for a differentiation between DP-ergative and PP-ergative languages is presented in Chapter 2. Owing to the diachronic origins of most ergative patterns in passive or possessive constructions, many transitive agents find expression in PPs. In syntactically ergative languages, some characteristics of these agentive PPs are maintained despite reanalysis or loss of an overt adposition; most prominent among these characteristics is the constraint against subextraction of the agent out of a PP in A-bar movement. Consequently, while the ergative is analysed as a structural case in DP-ergative languages, assigned by a functional head in vP (or higher), it is proposed that PP-ergative languages have an inherent ergative case in the external argument position of transitive verbs. Licensing here occurs in two places: the verbal head assigns theta role, the (potentially silent) adposition ergative case. Syntactic ergativity is, therefore, expected in ergative languages in which adposition stranding or pied-piping does not occur. Not all PP-ergative languages exhibit the latter constraint, however, making PP-ergative arguments a necessary but not sufficient for syntactic ergativity.

The similarities between PPs and ergative subjects in syntactically ergative languages form the core of this argument; to corroborate this point, Chapter 3 details the diagnostic criteria according to which both syntagmata must be compared to establish a correlation, and in which they differ from DPs. Polinsky’s findings show that PPs in subject position are inaccessible to A-movement, cannot serve as pivots for clefts, tails of control chains, or binders for anaphora or depictives; where A-bar movement is permitted, resumptive pronouns are required at the extraction site. This negative definition, largely by lack of abilities or functionality, contrasts with the characteristics of DPs, which are not subject to such restrictions.

These qualities of PPs are tested against the evidence from syntactically ergative languages in Chapter 4 in order to show that an analysis of ergative agents as PPs is indeed appropriate. The specific cause of syntactic ergativity, it is argued, results from a constraint against stranding and pied-piping of non-overt (or null) prepositional heads of ergative agents. The discussion emphasises again that where an overt preposition heads the phrase containing an ergative agent, syntactic ergativity may but need not obtain, depending on restrictions on PPs in general. The permissibility of pied-piping PPs in particular is dependent on the phonological nature of the operator; the data and analysis presented suggest that only overt operators can license extraction and thus avoid syntactical ergativity at least partly. A syntactically ergative language therefore has to be a) morphologically ergative, and b) either lack an overt adposition licensing the ergative or, failing this, lack the ability to pied-pipe PPs.

A further concomitant characteristic of PP-ergative agents, namely their inability to bind anaphors and to occur in raising or control structures, is discussed succinctly in Chapter 5. True raising and control in these languages are either limited to intransitive embedded clauses (with absolutive subjects), or do not exist at all; syntactically similar patterns do, however, occur. As far as binding is concerned, dedicated anaphors do not seem to appear in this type of ergative languages; their function is fulfilled either by reinforced pronouns, or through reflexive or reciprocal marking on the verb itself. With this evidence, Polinsky concludes that the similarities between PPs in general and ergative subjects in a subset of morphologically ergative languages allow for the latter’s analysis as PP-agents.

While the discussion in the previous chapters focussed on arguing for a differentiation between DP-ergative and PP-ergative languages, Chapter 6 illustrates further consequences of operating with PP-ergative agents as regards word order, expletive subjects, and non-canonical subjects. Statistics suggest that there may be a non-trivial correlation between verb-initial languages and syntactic ergativity; Polinsky tentatively hypothesises that the raising of the ergative PP may be the result of its failure to otherwise satisfy subjecthood conditions (as per the extended projection principle). A brief discussion of expletive subjects offers the conclusion that their absence is expected in verb-initial syntactically ergative languages, but cannot make further predictions about other language types or specific correlations to the PP-nature of ergative agents; a similar caveat applies to the prediction about the absence of non-canonical subjects in these languages.

By way of contrast with the main topic of the argument thus far, Chapter 7 presents the flip-side of ergative languages: those with DP-ergatives. Distinguished most distinctly by their lack of syntactical ergativity, the languages cited here (among which Georgian, Walpiri, and Hindi/Urdu) show all the features lacking in PP-ergative languages, including (but not limited to) the extraction of the ergative agent with a gap (rather than a resumptive pronoun), the ability to bind anaphors, license depictives, float quantifiers, have discontinuous core arguments, and to be the controllee in control patterns.

While both DP- and PP-ergative languages can have different origins, it is possible for PP-ergatives to be reanalysed, and for the language to change subsequently into the DP-ergative type; such diachronic developments and other questions regarding the relationships between ergative language types are discussed in Chapter 8. Niuean, a Polynesian language closely related to syntactically ergative Tongan (see Chapter 10), is discussed as an instance of change in progress: the reanalysis of the ergative adposition as a case marker has triggered syntactic realignment, most obvious in the ability of ergative agents to undergo A-bar movement; other DP-ergative features, such as anaphor binding, however, have not yet been implemented. Another case of transition is that of Adyghe, a Caucasian language, which has progressed further towards DP-ergative status; only resumption of ergative arguments in relative clauses remains as a putative sign of their former PP-ergative status. The conclusions drawn from these languages are that, unsurprisingly, alignment change does not happen all at once, and that the prepositional head of the ergative agent must have been lost, either through sound change or reanalysis, for this development to commence.

Before proceeding to give a closer overview of two paradigm ergative languages, Polinsky considers alternatives to her explanation of syntactical ergativity, independent of the notion of a PP-DP-dichotomy. Her approach, it is argued, is preferable to COMP-trace (e.g. Perlmutter 1971), criterial freezing (Wexler and Culicover 1980), phase-based (Coon et al. 2014), and processing analyses in being relatively theory-neutral and requiring the fewest assumptions. In particular, the processing approach, suggesting that structures imposing a heavy load on language processing are avoided, is tentatively rejected; experimental data on the acquisition of extraction in ergative languages, a supposedly resource intensive and thus dispreferred process, contradict expectations, but are as yet too few to be authoritative.

The two long chapters, which form Part 2 of the monograph, each deal with a paradigm language: Tongan, representing PP-ergative language (Chapter 10), and Tsez for the DP-ergative languages (Chapter 11). In both instances, an introduction to the basic morphological and syntactic structures of the language is given. In the case of Tongan, Polinsky proceeds to systematically discuss in some detail the manifestations of all the characteristics of PP-ergative languages outlined above; it is noted that only in two cases do ergative and absolutive subjects behave alike, namely in their ability to be associated with subject clitics, and to be expressed as null pronominals. It is further observed that, apart from the A-bar movement of absolutives, there are no other movement operations in Tongan.

Tsez, presented in a more concise fashion, shows all the signs of a DP-ergative language in not being subject to the restrictions imposed on languages of the PP-ergative type. Here the presence of all the structures discussed in previous chapters is exemplified, and discussed on a more theoretical level, e.g. as concerns the nature of Tsez clause structure.


Polinsky provides a new and systematic approach to describing ergative languages that offers attractively simple categories, clearly defined and empirically testable conditions, and data from a set of relevant languages to back up the suggestions made. The dichotomy between PP-ergative and DP-ergative languages proposed is essentially theory-neutral and makes a number of interesting observations awaiting further testing.

As such, this monograph will be of interest to anyone working on ergative languages, particularly on syntactical ergativity, and to typologists in general. Certain parts (esp. Chapter 8) may also be of interest to historical linguists. The analytical methods used and proposals made concerning specific languages may further be relevant for linguists working in transformational syntax or on Polynesian and Caucasian languages.

The book’s overall goal, that is to provide a principled analysis and aetiology of syntactic ergativity and to outline the structural diversity of ergative languages, has clearly been achieved. Yet, owing to the survey’s purposed limitation to a small number of languages and the narrow definition of syntactic ergativity (both given above), the proposed characteristics of PP-ergative languages in particular will require further testing. While the core argument, viz. the existence of two types of ergative languages, is presented clearly, consistently, and with great attention to detail, the fact that PP-ergatives are defined largely negatively by the absence of specific characteristics—as admitted by Polinsky (p. 56)—raises the question whether the similarities between certain ergative agents and PPs are sufficient for an unambiguous definition of this group, and how languages transitioning from one type to the other ought to be classified. Similarly, a brief discussion of languages such as Dyirbal, which have been explicitly excluded from initial consideration owing to debates about their status as ergative languages, and their place in this theory would be of interest, and might have helped to either corroborate or show issues with the dichotomy proposed.

To appreciate the book’s argument and presentation in their entirety, a thorough grounding in (and acceptance of) transformational grammar is advantageous, since analyses rely strongly on the tenets propounded and terminology used therein. Although each chapter offers a more general summary of the topics discussed, a less theoretically inclined audience may have appreciated more descriptive analyses at the side of theoretical considerations. In particular, some readers may take issue with the use and proposal of phonologically zero or non-overt elements, e.g. the prepositional heads or operators mentioned (esp. Chapter 4). Nonetheless, since terms and concepts potentially unfamiliar to the reader are explained and referenced clearly, the overall argumentation remains comprehensible, even when expressed in framework-specific terms.

As far as the structure of the overall argument is concerned, some choices may be rather surprising. For instance, a discussion of previous approaches to and explanations of syntactic ergativity is provided only in Chapter 9; it may have served as a smoother point of departure than the immediate proposal of a new theory. The proposal itself (Chapter 2) is arrived at somewhat abruptly, with too little context, and mainly on the basis of theoretical considerations; an initial presentation of the data used in chs. 4 and 5 would introduce the reader to the issues arising more naturally, as well as advocating a data-driven approach to linguistic analysis. In spite of these choices, the book shows great internal coherence and, with the caveats mentioned, a clear line of argumentation.

The detailed discussion of Tongan (Chapter 10), which takes up almost a third of the monograph, lacks focus in places; it provides too much information specific to the language discussed, but only partly relevant to the discussion of ergativity. The section on deriving Tongan clause structure (10.3), for example, explores the issue of word order, termed a secondary correlate of some PP-ergative languages in Chapter 6, in more detail than necessary here; similarly, a briefer discussion of raising-like structures in Tongan would not have diminished the argument’s force.

Polinsky’s proposal will have to stand the test not of time but rather of new data; her approach is appealing and will provide both theoretical and historical linguists as well as researchers working on specific languages with a new theory to explore. Especially the insights gained from applying the proposed criteria to supposedly syntactical languages not included in the present definition, and to those languages in the process of alignment change should reveal whether the latter are indeed sufficient.


Coon, Jessica, Pedro Mateo Pedro & Omar Preminger. 2014. The role of case in A-bar extraction asymmetries: Evidence from Mayan. Linguistic Variation 14(2). 179–242.

Dirr, Adolf. 1928. Einführung in das Studium der kaukasischen Sprachen. Leipzig: Verlag der Asia Major.

Dixon, Robert M. W. 1994. Ergativity (Cambridge Studies in Linguistics 69). Cambridge: Cambridge University Press.

Grinevald, Colette & Marc Peake. 2012. Ergativity and Voice in Mayan languages: a functional-typological approach. In Authier, Gilles & Katharina Haude (eds.), Ergativity, Valency and Voice (Empirical Approaches to Language Typology 48), 15–50. Berlin: De Gruyter Mouton.

Haig, Geoffrey L. J. 2008. Alignment Change in Iranian Languages: A Construction Grammar Approach (Empirical Approaches to Language Typology 37). Berlin: De Gruyter Mouton.

Paul, Ileana & Lisa Travis. 2006. Ergativity in Austronesian Languages. In Johns, Alana, Diane Massam & Juvenal Ndayiragije (eds.), Ergativity: Emerging Issues (Studies in Natural Language and Linguistic Theory 65), 315–335. Dordrecht: Springer.

Perlmutter, David M. 1971. Deep and surface constraints in syntax. New York: Rinehart & Winston.

Verbeke, Saartje. 2013. Alignment and Ergativity in New Indo-Aryan Languages (Empirical Approaches to Language Typology 51). Berlin: De Gruyter Mouton.

Wexler, Kenneth & Peter W. Culicover. 1980. Formal principles of language acquisition. Cambridge, MA: MIT Press.


Robin Meyer is a doctoral student at the University of Oxford, working on diachronic syntax and language contact. His research focus lies on Old and Middle Iranian languages and Armenian, but he also works on Greek and Latin. His dissertation explores contact-induced language change in early Classical Armenian, specifically as regards alignment change.

Page Updated: 23-Feb-2017