LINGUIST List 25.4826

Mon Dec 01 2014

Review: Syntax: Marelj, Siloni, Everaert (eds.) (2012)

Editor for this issue: Sara Couture <>

Date: 07-Oct-2014
From: Tomislav Socanac <>
Subject: The T
E-mail this message to a friend

Discuss this message

Book announced at

EDITOR: Martin Everaert
EDITOR: Marijana Marelj
EDITOR: Tal Siloni
TITLE: The Theta System
SUBTITLE: Argument Structure at the Interface
SERIES TITLE: Oxford Studies in Theoretical Linguistics 37
PUBLISHER: Oxford University Press
YEAR: 2012

REVIEWER: Tomislav Socanac, University of Geneva

Review's Editors: Malgorzata Cavar and Sara Couture


As the title of the volume suggests, “The Theta System: Argument Structure at the Interface” is primarily concerned with the Theta System (henceforth ThS), which was developed by Tanya Reinhart (2000, 2002 a.o.) in order to account for the properties of the lexicon and its interface with syntax. The volume consists of 13 chapters: an introduction, which provides a summary description of ThS, and 12 individual contributions dealing - for the most part - with various aspects and applications of Reinhart’s theory. The articles included in the volume approach the issues related to ThS from a wide array of linguistic perspectives, ranging from theoretical approaches (both formal and functional) to more experimental approaches, related to psycholinguistics or language acquisition.

In the introductory chapter, which is intended to facilitate readers’ access to the following articles, the editors of the volume provide us with a summary glance into ThS and some of the puzzles it helps to resolve. The main problem that Reinhart set out to address concerned the mapping of th(eta)-roles to syntactic positions, which is not entirely predictable from the roles’ labels (Agent, Theme, Instrument etc.). Reinhart thus proposed that th-roles as such should not be seen as atomic units, but rather that they can be decomposed into more primitive features. More precisely, th-roles are coded through two binary features: +/- c feature (cause change), which determines whether the argument is responsible for causing the event denoted by the predicate; and +/- m feature (mental state), which determines whether the argument is volitional or not. The various feature combinations (there are nine altogether) ultimately determine the thematic label of the argument and prepare the lexical entries for syntactic computation. The introduction ends with a brief overview of each individual contribution contained in the book.

In “The Linguistic Expression of Causation,” Ad Neeleman and Hans van de Koot set out to answer how the concept of causation is encoded in language. They claim that causation is not a primitive of the linguistic system and that predicates do not encode causing events either in their lexical semantics or syntax. Predicates can only emulate causation by combining two independent linguistic primitives: a crucial contributing factor (CCF) and the culmination of an event in an end state or a resultant activity. CCF is similar to the +c feature of Reinhart’s ThS but it differs in one crucial aspect: while +c can be interpreted in some contexts as encoding a causing event, CCF cannot; it only refers to the entity which is responsible for bringing about an end state. CCF is not inherent to causality because it can also be found in non-causal contexts, such as those involving maintenance (e.g. ‘John’s father supports him financially’). Therefore, CCF does not intrinsically encode causing events.

In “The Content of Semantic Roles: Predicate-Argument Structure in Language and Cognition,” Martin Haiden defends the existence of th-roles and their constituent features as described in Reinhart’s ThS. He first criticizes some of the alternative approaches that reject the notion of th-roles, such as those that view the lexicon as consisting of syntactic trees (as in Jackendoff, 1983), pointing out that such approaches are unable to account for the presence of more than one scale of prominence with certain types of verbs (causation/force, on the one hand, and intentionality/mental involvement, on the other). The author then goes on to offer a cognitive foundation for th-roles and the features underlying th-roles within ThS. He puts forward some cognitive experiments, based on the ‘Theory of Body Mechanism’ and the ‘Theory of Mind Mechanism’ (Leslie, 1994), which showed that our perception of causation can differ depending on whether the entities involved in the action are seen as volitional or not. Thus, both the c and the m features of the ThS are relevant for the cognitive system.

The title of Edwin Williams’ article “Combine” makes a direct reference to the author’s main theoretical claim: he proposes a new grammatical mechanism of Combine, which is presented as a unique generative operation in grammar. The Combine mechanism operates both in syntax and in derivational morphology, which means that they should both be seen as generative. The only difference between these two components in this context is related to a specific morphological parameter: the so-called ‘parameter M’ (M simply stands for morphology), which determines whether Combine operates on a root, a stem or an entire word. In syntax, the head A is not restricted to combining with the head B that is closest to it, but with the closest head B that satisfies its M-value, while in derivational morphology, the head A must combine with the closest head B regardless of its M-value.

In Hagit Borer’s article, “In the Event of a Nominal,” the author puts forward a study of different types of derived nominals. Borer’s main theoretical claim is that words should not be seen as indivisible, syntactically-atomic units, but that they contain internal syntax. The primary motivation for this claim is the fact that words of the same morpho-phonological complexity can exhibit radically different syntactic and semantic properties, depending on the type of nominal they appear in. In this context, Borer establishes a contrast between argument structure (AS) nominals (e.g. ‘the driving of a truck’) and synthetic compounds (SC) (e.g. ‘truck driving’). The former exhibit grammatical event properties, and contain a functional layer, whereas the latter do not. Moreover, the meaning of AS is strictly compositional, whereas the meaning of SC can be non-compositional as well. Borer concludes that these differences between the two types of nominals can only be accounted for if we assume the existence of an articulated syntax internal to words.

In “Lexicon Uniformity and the Causative Alternation,” Beth Levin and Malka Rappaport Hovav focus on the causative-anticausative alternation that certain verbs (e.g. ‘open’, ‘break’ etc.) exhibit in their argument structure (e.g. ‘He opened the door’ - ‘The door opened’). In this context, the authors disagree with Reinhart as to which of these two variants is basic and which is derived from the other. Reinhart claimed that the anticausative variant is derived from the basic transitive causative variant through the lexical operation of ‘decausativization’, whereby the external cause argument of the verb is removed. Levin and Rappaport Hovav make the opposite claim, i.e. that the anticausative variant is basic and the causative one derived. They identify a group of intransitive change-of-state verbs (e.g. ‘blossom’, ‘corrode’ etc.), which they define as basic, underived anticausative predicates. These verbs allow for a causative variant only under a precise condition - that of ‘direct causation’ - whereby there is no intermediate cause between the original causer and the caused eventuality. The authors then generalize this condition to apply to all cases of causative alternation.

The main claim of György Rákosi’s article, “In Defence of the Non-Causative Analysis of Anticausatives,” is that the anticausative variant of the causative alternation does not intrinsically contain a causal ingredient. The primary argument for the contrary assertion that such constructions do encode a cause is that they allow for the insertion of cause adjuncts (e.g. causal from-PPs in English or ablative PPs in Hungarian). Rákosi focuses on ablative cause adjuncts in Hungarian in order to show that they are not licensed by an underlying causal ingredient in anticausatives but, rather, that they introduce causation themselves. The main piece of evidence that Rákosi puts forward for this claim is that such ablative causes can also appear outside of anticausative constructions in Hungarian. Even though they differ in their syntactic realization (in anticausatives they are adjoined to VP as thematic adjuncts, whereas in other contexts they are merged in the left periphery), ablative causes introduce the same semantic contribution in both types of contexts - that of direct causation. By showing that anticausative constructions lack an inherent causal ingredient, Rákosi lends support to Reinhart’s decausativization account of the causative alternation.

Julie Fadlon also examines the causative alternation (she uses the transitive-unaccusative terminology) in her article “Hidden Entries: a Psycholinguistic Study of Derivational Gaps.” She focuses on the problem of derivational gaps with such constructions, i.e. cases where the unaccusative does not have the corresponding transitive variant in a given language (e.g. ‘The castle collapsed’ - * ‘Mary collapsed the castle’). Reinhart analyzed such cases in terms of hidden lexical entries, proposing that the basic transitive entry is frozen in the lexicon and cannot be inserted in syntax. Fadlon provides evidence for this claim by using a psycholinguistic experiment in order to demonstrate the cognitive reality of hidden lexical entries. Her experiment tested the cognitive accessibility of an external cause (and hence of the transitive concept) in several types of unaccusative constructions. It showed that the external cause is more accessible with verbs whose transitive variant is argued to be a hidden lexical entry than with underived unaccusative verbs which lack a transitive variant altogether, thus proving the psychological reality of hidden lexical entries.

In “To Have the Empty Theta-Role,” Peter Ackema and Marijana Marelj explore the least studied option within ThS - the empty role option, i.e. the empty [ ] cluster which does not contain either the c or the m feature. They apply their analysis to the verb ‘have’, which they view as a light verb with no independent semantic content. As a result, the verb ‘have’, like all light verbs, assigns a single [ ] th-role, which only allows for the insertion of a semantically empty external argument (external merge being the default option within ThS). The authors propose that this external argument receives semantic content through a mechanism of theta-merger, whereby its empty th-role merges with another semantically meaningful th-role, whose properties differ depending on the semantic context in which the verb ‘have’ is used. This analysis allows the authors to account for various different types of uses of the verb ‘have’, including the auxiliary, possessor, experiencer and causative ‘have’.

In “Emission Verbs,” Joseph Potashnik focuses on a class of intransitive unergative predicates within the framework of ThS. Reinhart assumed that underived intransitive predicates do not undergo any lexical marking and hence merge externally by default. This assumption was necessitated by the existence of a class of emission verbs (e.g. ‘buzz’, ‘flash’, ‘flicker’ etc.), which Reinhart analyzed as theme unergatives. As any type of theme-predicate, these verbs would have to be marked with a [-c/-m] th-cluster and hence undergo internal merge (only an all [-] cluster obligatorily merges internally within ThS). This would be contrary to facts, since emission verbs are unergatives. Therefore, Reinhart assumed that this type of verbs has to be excepted from any lexical theta-marking. Potashnik finds this solution problematic on conceptual grounds and proposes an alternative analysis. He uses a syntactic test called ‘the caused-NP experiment’ in order to show that emission verbs contain an internal causal component (e.g. ‘The torch shone’ - ‘The torch caused the shine’). As a result, they should not be marked with a [-c/-m] cluster but rather with a [+c/-m] cluster. Since they contain a mixed [+/-] cluster, emission verbs can merge externally within ThS, which renders their exclusion from lexical marking unnecessary. In the second part of his paper, Potashnik accounts for the cases in which emission verbs may undergo transitive alternation.

In “Verbal Passives in English and Hebrew: a Comparative Study,” the author, Aya Meltzer-Asscher, is primarily interested in the properties of the demoted external argument in English and Hebrew passive constructions. Unlike unaccusatives, the demoted external argument in passives remains semantically accessible. Some have argued that this is because it remains present as a null argument in syntax, while others have claimed that it is only recuperated in the interpretative module. Meltzer-Asscher argues that both of these options exist in language, but they should be parameterized. Thus, in English passives the demoted argument is present in syntax, while in Hebrew it is represented only at the level of interpretation. This claim allows the author to explain some syntactic contrasts, as well as a number of thematic and distributional differences between the passive constructions in two languages.

In “An Event Semantics for the Theta System,” Alexis Dimitriadis provides a semantic implementation of ThS's primitives. The main contrast that Dimitriadis focuses on is the one between the lexicon, which involves unordered th-feature clusters, and the formal semantic representation, which involves ordered lambda forms. He concludes that there has to be a change in linguistic representation at some step in the derivation, which transforms an unordered linguistic object into an ordered one. He situates this step just before lexical items are inserted in syntax and calls it ‘assembly’. At the point of assembly, unordered th-feature clusters are assembled into ordered model-theoretic functions representing the verb's denotation. This demarcation allows the author to explain why certain arity operations (i.e. operations changing the verb's valency) are cross-linguistically restricted to the lexicon while others can occur in syntax as well.

In the last article of the volume,“Children Acquire Unaccusatives and A-Movement Very Early,” João Costa and Naama Friedmann put forward the results of a number of experiments they have conducted in order to assess the acquisition and production of unaccusatives and unergatives by children acquiring Hebrew and European Portuguese. The conclusions they come up with go against some of the previous studies which have argued that children at the early stages of language acquisition lack the notion of A-movement and, as a consequence, misanalyze unaccusatives projecting a preverbal subject as unergatives. This would go against ThS, because the latter implies that the thematic properties of unaccusatives and their transitive variants are universal and innate, thus predicting that children should map unaccusatives correctly from the outset. On the other hand, Costa and Friedmann’s study supports ThS, because their results indicate that children as young as two years old have already acquired A-movement and make a correct distinction between unaccusatives and unergatives.


Through analyses centered on Reinhart’s ThS, the authors contributing to this volume give us a comprehensive glance into the area of lexical semantics and its interface with syntax, because they not only focus on Reinhart’s system but also situate it in relation to other influential theories dealing with this area of grammar. The brief overview of ThS provided by the editors in the introductory chapter is very useful and it considerably facilitates the reading and the comprehension of the following articles. It is nevertheless unavoidable that, given the wide range of theoretical perspectives contained in the volume, the readers will have more difficulties with certain articles depending on their own theoretical background. This, of course, should not be seen as a shortcoming of the book itself.

Despite containing a wide variety of linguistic approaches, the volume on the whole is largely coherent with regards to its subject matter, with most contributors conforming to the following outline set out by the editors in the introduction: “The goal of this volume is to present the Theta System, examine its underpinnings, explore its advantages, and suggest further developments and improvements” (15). There are, nonetheless, a few articles that do not entirely conform to this objective and that are somewhat less relevant for the overall theme of the volume. This can be said of Williams’ and Borer’s contributions, because, while the authors deal with the issues related to the lexicon, they do not touch upon ThS.

One of the main strengths of the book is the fact that most contributions contained in it are largely data-driven. Depending on their domain of study, the authors put a great emphasis either on experimental data or on linguistic examples in order to back up their theoretical claims. Most of the empirical examples come from English, but languages such as Hebrew, Hungarian, Dutch or Portuguese are represented as well. The balance between theoretical claims and data is generally very satisfactory. The authors rarely advance major theoretical proposals without thoroughly backing them up with evidence, without shying away from data that appears more problematic for their theory and attempting to address them as well. The only slight criticism in this context could be leveled against Williams’ and Rákosi’s articles, because each of these authors only briefly mentions the most problematic data that go against their theories (long distance movement and se-anticausatives, respectively), without addressing them in any detail.

Nevertheless, these few critical remarks do not undermine the general merits of the volume. The book will be a valuable and interesting read for any scholar or advanced student interested in lexical semantics and in the relation between structure and meaning in general.


Jackendoff, Ray. 1983. Semantics and Cognition. Cambridge, Mass.: MIT Press.

Leslie, Alan M. 1994. ToMM, ToBy, and Agency: Core architecture and domain specificity. in L.A. Hirschfeld and S.A. Gelman (eds.), Mapping the Mind: Domain Specificity and Culture. New York: Cambridge University Press. 119-48.

Reinhart, Tanya. 2000. The Theta system: Syntactic realization of verbal concepts. Uil-OTS Working Papers. University of Utrecht.

Reinhart, Tanya. 2002. The Theta system: an overview. Theoretical Linguistics 28 (3). 229-90.


Tomislav Socanac is a PhD candidate in Linguistics at the University of Geneva. The subject of his dissertation is the syntax and the semantics of subjunctive mood in Slavic languages. His main areas of interest are formal syntax and the syntax-semantics interface.

Page Updated: 01-Dec-2014