• Author / Uploaded
• Hussein

Citation preview

The Structure of Stative Verbs

Linguistik Aktuell/Linguistics Today (LA) Linguistik Aktuell/Linguistics Today (LA) provides a platform for original monograph studies into synchronic and diachronic linguistics. Studies in LA confront empirical and theoretical problems as these are currently discussed in syntax, semantics, morphology, phonology, and systematic pragmatics with the aim to establish robust empirical generalizations within a universalistic perspective.

General Editors Werner Abraham

University of Vienna / Rijksuniversiteit Groningen

Elly van Gelderen

Arizona State University

Advisory Editorial Board Cedric Boeckx

Christer Platzack

Guglielmo Cinque

Ian Roberts

Günther Grewendorf

Lisa deMena Travis

Liliane Haegeman

Sten Vikner

Hubert Haider

C. Jan-Wouter Zwart

ICREA/Universitat Autònoma de Barcelona University of Venice

J.W. Goethe-University, Frankfurt University of Lille, France University of Salzburg

Terje Lohndal

University of Maryland

Volume 143 The Structure of Stative Verbs by Antonia Rothmayr

University of Lund

Cambridge University McGill University

University of Aarhus University of Groningen

The Structure of Stative Verbs Antonia Rothmayr University of Vienna

John Benjamins Publishing Company Amsterdam / Philadelphia

8

TM

The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.

Library of Congress Cataloging-in-Publication Data Rothmayr, Antonia.   The structure of stative verbs / Antonia Rothmayr. p. cm. (Linguistik Aktuell/Linguistics Today, issn 0166-0829 ; v. 143) Includes bibliographical references and index. 1.  Grammar, Comparative and general--Stative verb.  I. Title. P281.R673

2009

415'.6--dc22 isbn 978 90 272 5526 6 (hb; alk. paper) isbn 978 90 272 8946 9 (eb)

2009007785

© 2009 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa

Für Mama

Philosophieanfall: Die Zeit ist eine Zeit. Wenn sie zwei Zeiten ist, kann sie gleichzeitig sein. Heidi Heide (in: Podium 127/128 April 2003, 56)

Table of contents Acknowledgements List of tables chapter 1 Introduction chapter 2

xiii xv

1

Theoretical considerations 3 2.1 Background I: Event semantics and argument structure  3 2.1.1 Aspectual classes  3 2.1.2 The Davidsonian view  4 2.1.3 The Neo-Davidsonian view  5 2.1.4 The Post-Davidsonian view  5 2.1.5 Kimian and Davidsonian statives  6 2.2 Background II: Semantic form  8 2.3 The stative verb in argument structure theory  10 2.3.1 Projectionist views  11 2.3.1.1 Hale and Keyser: Argument structure  11 2.3.1.2 Levin and Rappaport: Structure and constant participants  14 2.3.2 Constructionist views  16 2.3.2.1 Marantz: distributed morphology  16 2.3.2.2 Kratzer I: Stage-level and individual-level predicates  16 2.3.2.3 Extension by Hallman  19 2.3.2.4 Kratzer II: Severing the external argument  21 2.3.2.5 Harley: Different types of v  22 2.3.2.6 Borer: Exoskeletal approach  23 2.3.2.7 Ramchand: A Post-Davidsonian view  25 2.3.3 Conclusion  27 2.4 A semantic definition: Stative verb  28 2.4.1 The ontology of Kimian and Davidsonian states  28 2.4.2 A filter that groups together D- and K-states  29 2.4.3 Syntactic tests  30 2.4.4 Infinitival complements of perception verbs  31

 The Structure of Stative Verbs

2.4.5.1 Manner adverbials  31 2.4.5 Tests based on adverbial modification  31 2.4.5.2 Locative modifiers  32 2.4.5.3 Degree readings of ein bisschen  33 2.4.6 Summary  34 2.5 Other kinds of stative expressions  35 2.5.1 Generic expressions  35 2.5.2 Perfect as a stativizer  36 2.5.3 Adjectival passive  36 chapter 3

Stative/eventive ambiguities 37 3.1 Introduction  37 3.2 Ambiguity due to instrumental alternation  37 3.2.1 Introduction  37 3.2.2 Tests for a Kimian stative reading  40 3.2.2.1 Manner adverbials  40 3.2.2.2 Event-related locative modifiers  42 3.2.2.3 Degree readings  42 3.3.3 Stative causation?  43 3.2.3.1 Counterfactual analysis of causation  43 3.2.3.2 Causation and event structure: A note on Kaufmann and Wunderlich  44 3.2.4 The structure of stative verbs of causation  47 3.2.4.1 The stative reading  47 3.2.4.2 Stative/eventive ambiguities  48 3.2.5 Conclusion  51 3.3 Object-experiencer verbs assigning accusative  52 3.3.1 Introduction  52 3.3.2 Object-experiencer verbs in previous work  52 3.3.2.1 A note on terminology  53 3.3.2.2 Stative and eventive object-experiencer verbs in Italian  53 3.3.2.3 Stative and eventive object-experiencer verbs in Spanish  56 3.3.2.4 Stative and eventive object-experiencer verbs in Finnish  56 3.3.2.5 Previous accounts of object-experiencer verbs  58 3.3.3 Tests for a Kimian stative reading  60 3.3.3.1 Manner adverbials  60 3.3.3.2 Locative modifiers  62 3.3.3.3 Degree readings  62 3.3.4 Properties particular to this verb class  63 3.3.4.1 Temporal modification  63



Table of contents 

3.4

3.5

3.6.

3.3.4.2 Modifiers that express the start of the target state  63 3.3.4.3 Modification of domains  64 3.3.5. The structure of object-experiencer verbs  65 3.3.5.1 The lexical-semantic structure  65 3.3.5.2 The syntactic structure  65 3.3.6 Conclusion  68 The threaten-class  68 3.4.1 Introduction  68 3.4.2 Tests for a Kimian stative reading  73 3.4.2.1 Manner adverbials   73 3.4.2.2 Locative modifiers   74 3.4.2.3 Degree readings  74 3.4.2.4 Summary  75 3.4.3 The modal component of threaten   75 3.4.3.1 The DO/MOD alternation  75 3.4.3.2 The modality of drohen  76 3.4.4 The structure of threaten-type verbs  77 3.4.4.1 The lexical-semantic structure  77 3.4.4.2 The syntactic structure  78 3.4.5 Conclusion  80 Dispositional verbs  80 3.5.1 Introduction  80 3.5.2 Stative/eventive ambiguities noted in previous work  80 3.5.3 Tests for a Kimian stative reading  82 3.5.3.1 Manner adverbials  82 3.5.3.2 Locative modifiers  84 3.5.3.3 Degree readings  87 3.5.4 Properties particular to this verb class  88 3.5.4.1 Temporal modification  89 3.5.4.2 Modifiers that express the start of the target state  91 3.5.4.3 Modification of domains  93 3.5.4.4 Evaluation  94 3.5.4.5 Semantic properties of dispositional verbs  94 3.5.5 The structure of dispositional verbs  95 3.5.5.1 Dative assignment  95 3.5.5.2 Lexical-semantic structure  97 3.5.5.3 Summary: Helfen-verbs  100 Perception verbs  100 3.6.1 Evidence for a Kimian stative reading  102 3.6.1.1 Manner adverbials  102 3.6.1.2 Locative modifiers  103 3.6.1.3 Degree readings  103



The Structure of Stative Verbs

3.7

3.6.2 The structure of perception verbs  105 3.6.3. Conclusion  106 Conclusion  107

chapter 4

Non-ambiguous statives 109 4.1 Subject-experiencer/possessor verbs  109 4.1.1 Introduction  109 4.1.1.1 Case assignment in Finnish  110 4.1.1.2 Inchoative subject-experiencers in Spanish  113 4.1.2 Tests for an underlying Kimian state  114 4.1.2.1 Manner adverbials  114 4.1.2 Locative modifiers  116 4.1.2.3 Degree readings  117 4.1.3 Are subject-experiencer verbs really stative? A comment on Rapp  117 4.1.4 Argument structure  121 4.1.5 The structure of subject-experiencer verbs  122 4.1.6 Conclusion  123 4.2 Dative-experiencer / possessor verbs  124 4.2.1 Evidence for a Kimian stative reading  124 4.2.1.1 Manner adverbials  124 4.2.1.2 Locative modifiers  126 4.2.1.3 Degree readings  127 4.2.1.4 Complement of perception verbs  129 4.2.2 The structure of dative-experiencer/possessor verbs  130 4.2.3 Conclusion  130 4.3 Measure verbs  131 4.3.1 Evidence for a Kimian stative reading  131 4.3.1.1 Manner adverbials  131 4.3.1.2 Locative modifiers  132 4.3.1.3 Degree readings  132 4.3.2 Further properties of measure verbs  134 4.3.3 The lack of an eventive reading  135 4.3.4 The structure of measure verbs  135 4.3.4.1 The lexical structure of measure verbs  135 4.3.4.2 The syntactic structure of measure verbs  139 4.3.5 Conclusion  139 4.4 PP-complement verbs  140 4.4.1 Remark  140 4.4.2 Evidence for a Kimian stative reading  140 4.4.2.1 Manner adverbials  140 4.4.2.2 Locative modifiers  141



Table of contents 

4.5

4.4.2.3 Degree readings  141 4.4.2.4 Complement of perception verbs  142 4.4.3 The structure of verbs with a PP-complement  142 4.4.3.1 The lexical-semantic structure  142 4.4.3.2 The syntactic structure  144 4.4.4 Conclusion  144 Conclusion  145

chapter 5

Verbs of position Stative verbs of position  147 5.1.1 Tests for a Kimian stative reading  148 5.1.1.1 Manner adverbials  148 5.1.1.2 Locative modifiers  150 5.1.2 The structure of stative verbs of position  151 5.2 Verbs of body posture  153 5.2.1 Tests for an eventive reading  153 5.2.1.1 Manner adverbials  154 5.2.1.2 Locative modifiers  155 5.2.2 The structure of posture verbs  156 5.3 PP-argument versus PP-incorporation  157 5.4 Evidence from Tibeto-Burman  158 5.5 Conclusion  159

147

5.1

chapter 6

Verbs of internal causation 6.1 Non-agentive verbs of internal causation  161 6.1.1 The instrument role  163 6.1.1.1 Verbs of sound emission  164 6.1.1.2 Verbs of light emission  167 6.1.1.3 Verbs of smell and substance emission  168 6.1.1.4 Conclusion  170 6.1.2 The structure of verbs of emission  170 6.1.3 Conclusion  172

161

chapter 7

Event structure and theta features 7.1 Features within the vP  173 7.1.1 Features in v   173 7.1.2 Flavors of v  174 7.1.3 Dowty’s calculus from a minimalist perspective  175

173

 The Structure of Stative Verbs

7.2

7.3

Theta checking  176 7.2.1 Aspectual layout  177 7.2.2 Example verb structures  184 7.2.3 Checking and verb phases  187 7.2.4 The situation argument  195 7.2.4.1 The elimination of the Davidsonian event argument  196 Conclusion  197

chapter 8

Conclusion 8.1 Further verb classes  199 8.1.1 Modals  199 8.1.2 Sensation predicates  199 8.1.2.1 Manner adverbials  200 8.1.2.2 Locative modifiers  201 8.1.2 Degree readings  201 8.1.2.4 Conclusion  202 8.2 Conclusion  202

199

References

207

Author index

213

Subject index

215

Acknowledgements This monograph is a revised and extended version of the PhD dissertation I defended at the University of Vienna in 2006. The research for Chapters 1–4 was funded by a DOC-grant (Doktorandenprogramm) of the Austrian Academy of Sciences. The research for Chapters 5, 6 and 7 was not financially supported. I am indebted to my teachers Wolfgang U. Dressler, Claudia Maienborn and Martin Prinzhorn as well as to my editor Werner Abraham.

List of tables Table 2.1 Table 2.2 Table 2.3 Table 2.4 Table 2.5 Table 2.6 Table 3.1 Table 3.2 Table 3.3 Table 7.1 Table 7.2

Aspectual classes Analysis of event structure by Pustejovsky Kimian and Davidsonian statives Possible interpretations of AIVs Possible interpretations of AIVs Eventuality structure by Borer Types of modality Semantic properties of dispositional verbs Ambiguity of perception verbs Arguments licensed by do / cause Arguments licensed by predicate

3 6 7 20 20 24 76 95 101 187 188

chapter 1

Introduction When children first learn about grammar at school, they are often taught that there exist different types of words. In the German-speaking part of the world, kids learn to discriminate between Namen-Wörter (“name-words”, i.e., nouns), Wie-Wörter (“howwords”, i.e., adjectives), and Tun-Wörter (“do-words”, i.e., verbs). At least this is what I was taught about 20 years ago. But, how come that there are Tun-Wörter that do not express something one can do? Or, in more “adult-like” terms: how come there are verbs that express states rather than events? In what is to follow I will not answer these naive questions, but focus on the nature of stative verbs. What kind of stative verbs exist? Is there only a single class of verbs that classify as statives? Do they share some properties other than their stativity? How are stative verbs related to eventive ones? Are they derived from one another? And if so, by what kind of grammatical mechanism? In current linguistic theory “statives” or “stative verbs” are more often than not taken to be the most basic verbs, or the building blocks out of which more complex (eventive) verbs are formed. The first goal of this book is to sketch a picture of how stative verbs are seen in the literature, followed by a look on whether these claims are empirically justified. The answer will be that stative verbs are not simple at all: there exist different kinds of statives, simple and complex ones. The second goal of this monograph is to examine the different types of stative verbs in greater detail. Thereby I will come across a systematic pattern of stative/eventive ambiguities, which gives rise to the following questions: what are the grammatical mechanisms that trigger the eventive reading? What kind of predicates allow for the stative reading? Moreover, a second group of stative verbs does not display this kind of ambiguity. I will analyze the lexical-semantic structure of these verbs as well. The structure of the book is as follows. In Chapter 2 I will address the theoretical background. First, a short overview of the current picture of event structure is given, including a note on the distinction between two kinds of eventuality arguments, the Kimian and the Davidsonian one. After a brief introduction to the technical details of Semantic Form, I will take a look at how stative verbs are conceptualized within the research on argument structure. Next, a semantic definition of stative verb is given, following the distinction between Kimian and Davidsonian state expressions developed by Maienborn 2003. Finally, stative expressions other than verbs are mentioned. Chapter 3 deals with stative/eventive ambiguities. Verbs that may have both a Kimian stative and an eventive reading are analyzed in terms of their lexical-semantic



The Structure of Stative Verbs

structure. These include verbs that undergo the instrumental alternation, object-experiencer verbs, verbs of the threaten-class, dispositional verbs, and perception verbs. In Chapter 4, verbs that only allow for a Kimian stative reading are discussed. These include subject-experiencer/possessor verbs, dative-experiencer/possessor verbs, measure phrase verbs, verbs that require a PP-complement, and modal auxiliaries. Next, Davidsonian statives are examined. Is it the case that a third class of verbs apart from Kimian statives and Davidsonian events is necessary? Chapter 5 investigates the nature of verbs of position and Chapter 6 examines verbs of internal causation, both of which, according to Maienborn 2003, belong to the class of Davidsonian statives. Finally, Chapter 7 integrates the findings of the previous chapters and puts forward a novel account of event structure and argument licensing.

chapter 2

Theoretical considerations 2.1 Background I: Event semantics and argument structure In general verbs denote all kinds of actions, processes or events; sometimes they can even express states. Being part of the denotation, a process or a state, therefore, belongs to the semantics of a verb. In particular, stativity is a purely semantic notion.

2.1.1 Aspectual classes Vendler 1957 was the first one to group verbs along the lines of event structure. He classified them into activities, accomplishments, achievements and states. The basic idea underlying this classification is how an event proceeds in time. For example, activity verbs such as pet the cat or smile have an actor who is doing something (i.e. petting the cat or smiling) for an unbounded time interval. Similarly, states such as know or love don’t include an endpoint in their basic denotation. In contrast to activities, states have no agent who is doing something (that is, there is no active knowing or loving). Achievements and accomplishments, on the other hand, include an endpoint. For example, an achievement like reach the top is not homogeneous like a state such as love. It denotes a sudden change from a state in which the subject is not at the top to a state in which the subject is there. Vendler’s verb classes are summarized in Table 2.1. Homogeneity can be captured in more formal terms (Dowty 1979, Krifka 1989, among others): homogeneous predicates such as love or know have the subinterval property. That is, if a predicate is true at a certain time interval, it is also true for any subpart of this interval. For example, if I have loved syntax ever since I read LGB for the first time, I also loved it at Christmas Eve last year. Activity verbs, as homogeneous as they may seem, do not have the subinterval property. Of course, down to a certain time-span the subinterval property holds for activities: imagine I am petting the cat for 30 minutes. There are subintervals – say of Table 2.1  Aspectual classes Activity Accomplishment Achievement State

run paint a picture recognize know



The Structure of Stative Verbs

about 2 seconds – where there is still some petting of the cat going on, but at a certain point the intervals get so small that it is physically impossible for an action to occur. At that point, the time-interval becomes so short that my hand is not rubbing the cat’s body anymore. Instead, just the tips of my fingers are on the cat’s head. This is not petting the cat anymore. Dowty 1979 already observes that not all verbs that are classified as statives behave alike. He mentions that verbs of position like sit and lie can be used in the progressive form, in contrast to all other kinds of stative verbs. Similarly, when classifying the various kinds of states and events, Bach 1986 observes that at least two kinds of statives must be distinguished, namely dynamic and static states. The typology of eventualities according to Bach 1986: 6 is depicted in (1). (1) eventualities wo states non-states 3 3 dynamic (a) static (b) processes (c) events 3 protracted (d) momentaneous 3 happenings (e) culminations (f) (2) a. b. c. d. e. f.

sit, stand, lie + loc be drunk, be in New York, own x, love x, resemble x walk, push a cart, be mean (agentive) build x, walk to Boston recognize, notice, flash once die, reach the top [Bach 1986: 6]

Within the literature dealing with event semantics there has been some confusion about the terminology. In the following, I will stick to the terms put forward by Bach, i.e., the term eventuality includes states and events, whereas the term event only refers to the eventive type of eventuality.

2.1.2 The Davidsonian view Events, but not eventualities, have also received attention within philosophy. In his seminal essay, Davidson 1967 proposes that action sentences contain a special variable referring to the event denoted by the verb. In other words, non-states (in the terms of Bach) contain an argument which refers to the ongoing event. Davidson substantiates his claim with the example in (3).



Chapter 2.  Theoretical considerations



(3) Jones buttered the toast slowly, deliberately, in the bathroom, with a knife, at midnight. [Davidson 1967: (1)] (4) ∃e [butter(Jones, toast, e) ∧ deliberately(e) ∧ in the bathroom(e) ...]

The structure given in (4) shows that the verb contains not only the subject (Jones) and the object (toast), but also the event argument e. It is this argument that anchors adverbials to the sentence. So, the manner adverbial deliberately does not modify the agent or the patient; rather, it modifies the event itself. Several linguists which conform to the Davidsonian paradigma have taken up this idea and argue that the event variable is present in argument structure (e.g. Kratzer 1995). According to this line of analysis, the event argument is the topmost element within the argument hierarchy, occupying the position of the external argument.

2.1.3 The Neo-Davidsonian view Parsons 1990 suggests a refinement of Davidson’s notion. He argues that not only adverbials, but all participants are introduced by predication of the event variable. Thus, thematic roles such as agent and patient are introduced with the help of such extra predicates. The structure of sentences like (3) is therefore extended to (5).

(5) ∃e [butter(e) ∧ Agent(Jones, e) ∧ Patient(toast, e) ∧ deliberately(e) ∧ in the bathroom(e) ...]

Linguists who assume that an event variable is present in argument structure do not focus on the distinction between the Davidsonian and the Neo-Davidsonian view. For them, it is only important that the event argument is present both in argument structureand in syntactic structure.

2.1.4 The Post-Davidsonian view Contra to the assumptions of linguists following Davidson or Parsons, Post-Davidsonian scholars (e.g. Pustejovsky 1991, Hale and Keyser 1993) argue that the event argument is not present in syntactic structure. Rather, syntactic structure as a whole is taken to represent the event. Hale and Keyser 1993 argue that the event semantics of a verb does not come from the properties of its event variable; rather, the semantics is read off from the parts of the verb’s structure: every subpart of a verbal structure denotes a subevent. These subevents are composed by implication: a matrix event corresponding to a VP implicates the subevent denoted by the complement VP, i.e. the subevent is part of the matrix event.





The Structure of Stative Verbs

(6) a.

VP wo NP V’ wo Vmatrix VP 3 Vsubordinated XP

b. subeventmatrix ⇒ subeventsubordinated Depending on the category of the complement, different verb classes emerge. For example, V subcategorizing a preposition corresponds to a locatum verb like saddle. Similarly, an AP gives rise to a change of state verb (e.g. narrow), whereas a NP results in a verb of creation. Therefore, the kind of aspectual class a verb belongs to is determined by the category of its complement. Pustejovsky 1991 argues that events are not unstructured entities, but objects made up from different subevents. For example, a state consists merely of a single event, whereas a process develops through time via several subevents, each occurring one after the other. Moreover, a transition includes an initial state, several transitory subevents and a final event which is different from the initial one. Table 2.2  Analysis of event structure by Pustejovsky State S | E

Process

Transition

| P

T

6 el ...en

2 El...   ...¬E2

[Pustejovsky 1991: 56 (13)]

In sum, the common ground of Post-Davidsonian accounts on event structure is that events are complex entities which consist of one or more subevents. In particular, these subevents can be associated with different levels of the verbal projection, as suggested by Hale and Keyser 1993 and Ramchand 2008. Since each layer of the verbal structure is associated with a particular thematic participant (for example, the little v projection is said to be responsible for the introduction of the agent), an intimate relation between subevents and arguments is established.

2.1.5 Kimian and Davidsonian statives The event argument, as introduced by Davidson 1967, has not only been used for the analysis of action sentences, but also for a variety of purposes in the recent literature.



Chapter 2.  Theoretical considerations

For example, Diesing 1992 and Kratzer 1995 argue that the difference between stagelevel (SLP) and individual-level (ILP) predicates can be captured in terms of presence and absence of the event variable. The theta-grid of a SLP, which denotes a temporary predicate, includes an event argument, but the theta-grid of an ILP lacks it. In this way, the different behavior of SLPs and ILPs regarding weak and strong determiners can be attributed to the difference in the theta-grid. Kratzer illustrates this claim with the help of the following examples. SLPs like hit contain an event argument (“location” in her terms) as in (7a), and ILPs like know lack it as in (7b). (7) a. hit 〈location, agent, theme〉 b. know 〈experiencer, theme〉 [Kratzer 1995: 136] The event argument appears in several other analyses of grammatical problems, in particular, it is used in the analyis of stage-level copular constructions by many authors. However, Maienborn 20031 argues that the event argument as defined by Davidson should be restricted to action sentences or to predicates expressing an event.2 She takes a closer look at different stative expressions and concludes that there are at least two different kinds of statives. In addition to the class of pseudo-stative verbs (verbs of position and a group containing sleep, wait, glow and stick, among others), there exists the class of Kimian state verbs (Maienborn 2003: 55). These include copular constructions and various stative verbs such as weigh, know and resemble. Davidsonian and Kimian statives differ with respect to the nature of the Davidsonian argument. Whereas Davidsonian statives contain an event argument – in this case a stative one – Kimian statives contain an ontologically different argument. This argument, called a Kimian state,3 does not denote an event. It refers to a property being instantiated at a particular time, i.e., a Kimian state is the object of a property realized at a particular time (see also the discussion in Engelberg 2000). Table 2.3  Kimian and Davidsonian statives Davidsonian statives sit, stand, lie sleep, wait, gleam

Kimian statives be intelligent, be tired weigh, know, resemble

1. As the original book (Maienborn 2003) is written in German, I will cite the English shortversion (Maienborn 2005b) whenever possible. 2. See Katz 1995 for an early argument that statives do not contain the Davidsonian event argument. 3.

This idea goes back to Kim 1969 and Kim 1976, hence the term “Kimian” state.





The Structure of Stative Verbs

(8) Kimian states (K-states): K-states are abstract objects for the exemplification of a property P at a holder x and a time t. [Maienborn 2005b: (47)] Maienborn 2003 shows that not only Kimian stative verbs, but also both stage-level (be sick) and individual-level (be intelligent) copular constructions express a Kimian state. The stage-level/individual-level distinction is, according to this line of research, not due to the inherent structure of the predicates; rather, it is due to the discourse structure (see Maienborn 2005a for an account of the Spanish ser/estar-distinction). In other words, the stage-level copular construction which is often associated with Spanish estar is related to a situation that is already present in discourse. The individuallevel predicate, on the other hand, is not.

2.2 Background II: Semantic form In the following, I will analyze stative verbs in the framework of the two-level approach to meaning (c.f. Bierwisch 1982, Bierwisch 1987 and Bierwisch and Lang 1987; see also Kaufmann 1995a, Wunderlich 1997). The core idea of this approach is the separation between linguistic information and conceptual knowledge. Regarding lexical entries, the former is specified at the level of Semantic Form (SF), whereas the latter is represented at the level of Conceptual Structure (CS). At the level of Semantic Form, a lexical entry is specified with respect to grammatically relevant information, e.g. the number of arguments of a verb and its decompositional structure. At the level of Conceptual Structure, on the other hand, world knowledge comes into play. To illustrate, certain verbs of speech such as gossip, talk, swear or preach may differ with respect to the social context of the action, while their grammatical information is the same. Hence, these verbs differ with respect to their associated conceptual structure but not with respect to their SF-representation. Semantic Form as developed by Bierwisch 1987: 94 functions as the interface between linguistic knowledge and world knowledge. In other words, Wunderlich 1997: 29 suggests that “[t]he notion of SF assumed here may best be compared with the notion of Logical Form (LF) extended to sublexical structures”. So, both LF and SF take part in the computation of the meaning of a sentence. Regarding the relation between LF and SF, Bierwisch and Lang 1987: 665 argue that at LF the meaning between syntactic entities is computed, whereas SF determines the meaning of words themselves. Hence, both LF and SF are part of grammar. The SF-representation of a lexical entry specifies the number of arguments required. In order to do this, every argument position corresponds to a λ-abstractor. In



Chapter 2.  Theoretical considerations

other words, thematic roles are rephrased as λ-operators which bind the corresponding variable at SF. For exampe, a verb like eat is represented as in (9).

(9) λy λx [eat(x, y)]

The two-level approach to meaning is compatible with recent minimalist views on argument realization (e.g. Butler 2004). Butler builds on Adger and Ramchand 2005, who develop a feature valuation mechanism for wh-dependencies. According to Adger and Ramchand, there exists a syntactically visible feature which corresponds to predicate abstraction in semantics. They term this feature [Λ] and take it to correspond to a λ-operator. The variable bound by this operator is reflected by a syntactic feature as well, this time an unvalued feature [Id]. In this way, the semantic operation of λ-abstraction has a syntactic correlate: [Id] gets its value by agreeing with [Λ]. Agree (c.f. Chomsky 2001) is subject to syntactic restrictions such as locality conditions. In sum, the correspondence between syntax and semantics can be summarized as in (10). (10) [Λ ... Id] → λx ... x [Adger and Ramchand 2005:173 (38)] Butler 2004: 64-68 extends this proposal to the mechansim of argument introduction. He argues that lexical entries of verbs (roots in his terms) come with as many unvalued [Id]-features as they have arguments. In order to be interpretable at the interface, these [Id]-features, must be valued by a corresponding [Λ]-feature, as proposed by Adger and Ramchand 2005. Butler claims that the [Λ]-features are introduced into syntactic structure by little v-heads. Every [Λ]-feature binds a single [Id]-feature by assigning a value to it. In this way, as many λ-abstractions are generated as there are [Id]-features on the root. By agreeing with the [λ]-features, a predicate is created that requires as many arguments as there are [λ]-features (or λ-operators, respectively). In sum, we end up with a representation similar to an SF-entry. At this point in the derivation, the predicates created by the λ-abstractions must be saturated. Butler assumes that the relevant arguments are introduced into the specifiers of the little v-heads, which express sub-situations. The macro-situation is expressed (or anchored) at the top v-layer, which Butler takes to correspond roughly to the Davidsonian event argument in the sense of Kratzer 1995. To illustrate this mechanism, the initial derivation of Arthur laugh is given in (11). The [Id]-feature of the verb laugh has already received its value Λ via agreement with the [Λ]-feature on v. The uninterpretable feature [uId] on v has been checked. Moreover, the argument Arthur has been merged into the specifier of v in order to satisfy the requirements of the predicate λx.laugh(x).





The Structure of Stative Verbs

(11) a. b.

vP wo Arthur v’ 3 v[Λ, uid: Λ] VP 6 laugh [id: Λ] λ.[laugh(δ)] (Arthur) = Arthur laugh [Butler 2004: 68 (61)]

In conclusion, the idea that roots select their arguments via some features corresponds directly to the representation of Semantic Form where the number of arguments of lexical entries is specified via λ-abstractors. In what is to follow, I will therefore concentrate on the Semantic Form of stative verbs, keeping in mind that this format is directly compatible with recent Minimalist developments.

2.3 The stative verb in argument structure theory In this section, I would like to take a look on how different theories of argument structure judge the status of stative verbs. Research on argument structure investigates what kind of thematic roles are required by different kinds of verbs (starting with Fillmore 1968, Gruber 1976). Furthermore, the hierarchical relation among the arguments and various argument alternations such as the passive and the middle formation are central to this topic (cf. Grimshaw 1990, Baker 1997). The review is oriented along the lines of the projectionist / constructionist distinction that was put forward in Rappaport Hovav and Levin 1998 and more recently in Levin and Rappaport Hovav 2005. Projectionist views (e.g. Bresnan 2001, Levin and Rappaport Hovav 1995, Williams 1981, among many others) assume that argument structure is semantically anchored within the lexical structure of the verb. In other words, there is something in the meaning of a verb that requires the arguments to be the way they are (with respect to their number, hierarchy and Case-status). The morphosyntactic properties of the arguments are therefore a reflection of the lexical meaning of the verb. Constructional views (e.g. Borer 2005, Goldberg 1995, Harley 1995, van Hout 1998, Jackendoff 1990, McClure 1995, Marantz 1997, Ritter and Rosen 1998), on the other hand, assume that all words, not only verbs, consist of category-neutral roots that are combined with functional categories in order to add event-based meanings. Crucially, arguments are not licensed by the root, but from the functional heads. Therefore, constructional views assume a close relation between event structure and the realization of arguments.



Chapter 2.  Theoretical considerations

Therefore, the study of stative verbs provides a new argument for the distinction between projectionist and constructional views. From a constructionist perspective, the ideal stative verb is a simple predicate with no further structure. Because there is no event structure within stative verbs, it is expected that there is no “complex” argument structure as well. Stative verbs, then, have often been associated (see below for a discussion of the claims that have been made within this framework) with a stative little v projection that turns the category-neutral root into a verb and licenses a subject which is interpreted as the holder of the state. Projectionist theories, on the other hand, do not pose such a strong requirement. For such a view, it is possible to have a single type of eventuality associated with different patterns of argument realization. Both approaches require a kind of predicate decomposition in order to spell out their assumptions about the different structures of verbs. Decompositional analyses have been put forward by e.g. Dowty 1979, Hale and Keyser 1993, Parsons 1990, Rappaport Hovav and Levin 1998, von Stechow 1995, van Valin 1990, among many others. Although these analyses may differ greatly with respect to their theoretical assumptions, the authors converge on the claim that verbs must be decomposed into smaller building blocks in order to capture their behavior, both with respect to event structure and with respect to argument realization. Despite the central nature of the structure of stative verbs to both approaches, these verbs have been somewhat neglected by scholars of both frameworks. Nevertheless, implicit assumptions are spelled out at various points within the literature.

2.3.1 Projectionist views 2.3.1.1 Hale and Keyser: Argument structure In their seminal paper, Hale and Keyser 1993 put forward a concise theory of how the relation between theta roles, argument structure and aspect can be captured. They claim that theta roles must be defined in structural terms: a DP occupying a particular structural position is always mapped to a fixed theta role; for example, the specifier of a verbal projection that takes another verbal projection as its complement is always interpreted as the agent argument. All the other theta roles are derived by particular structural conditions along the same lines. Tying together phrase structure positions and thematic roles in this way, their theory implements the requirement posed by the Uniformity of Theta Assignment Hypothesis (UTAH; cf. Baker 1988). In addition to their treatment of theta roles, Hale and Keyser argue that the event semantics of a verb does not come from the properties of its event variable; rather, the semantics is read off from the parts of the verb’s structure: every subpart of a verbal structure denotes a subevent. These subevents are composed by implication: the matrix event corresponding to the VP implicates the subevent denoted by the complement VP, i.e., the subevent is part of the matrix event.





The Structure of Stative Verbs

(12) a.

VP 3 NP V’ 3 Vmatrix VP 3 Vsubordinated XP

b. subeventmatrix ⇒ subeventsubordinated Depending on the category of the complement, different verb classes emerge. For example, V subcategorizing a preposition boils down to a locatum verb like saddle. Similarly, an AP gives rise to a change-of-state verb (e.g. narrow), whereas a NP results in a verb of creation. Therefore, the aspectual class a verb belongs to is determined by the category of its complement. The authors only treat eventive verbs in their 1993 paper, but they elaborate on stative verbs in Hale and Keyser 2002: first, they discuss the difference between deadjectival verbs (clear, narrow, redden, darken; Hale and Keyser 2002: 205) in (13a) and predicative adjectives as in (13b). (13) a. b.

The sky cleared. [Hale and Keyser 2002: 206] We found [the sky clear]. [Hale and Keyser 2002: 206 (2a)]

The authors argue that the fundamental difference between deadjectival verbs and predicative adjectives is categorial. A verb is active due to its verbal category, a predicative adjective is stative due to its category δ. The head δ corresponds to an extended projection of A that hosts the usual degree morphology (i.e., comparative and superlative morphemes). Both verbal and adjectival structures share the property that they require a specifier and a complement. They differ with respectto the nature of the denoted event: the verbal structure is active, whereas the adjectival one is stative (see (14), from Hale and Keyser 2002: 206 f. (1) and (4)). (14)

V 2 DP V 2 V A | clear

δ 3 DP δ 6 2 the sky δ A | clear



Chapter 2.  Theoretical considerations

Next, Hale and Keyser discuss subject experiencer verbs as in (15a). The authors argue that these verbs are derived from possessive constructions such as (15b). (15) a. b.

John respected the truth. [Hale and Keyser 2002: 208 (6e)] John got the truth (to be) with his respect. [Hale and Keyser 2002: 208]

The structure of verbs like respect (and know, admire, like etc.) resembles the structure in (14), but now there is a prepositional head instead of a verbal one (Hale and Keyser 2002: 208 (7)).4 A special conflation mechanism allows for P to appear with inflectional morphology. (16)

P 3 DP P 6 2 the truth P N | respect

Finally, the authors take “true” stative verbs (in their terms) such as cost and weigh to correspond directly to copular constructions. Therefore, the argument of such verbs does not count as a direct object but rather as a measure phrase that is part of the predicate. In sum, these three ways are the only ones to form stative expressions: the extended projection of A, a subset of P and copular constructions. According to the theory of Hale and Keyser, the property of stativity can be reduced to the nature of the lexical category. Hence, a predicate is active if it contains a lexical item of category V; it is stative if categories A and P are present. Besides the non-standard conflation mechanism, Hale and Keyser’s answer to the question why statives fail to undergo the passive alternation remains stipulative. According to the authors, passive formation is ruled out, since stative verbs are equivalent to copular constructions. However, they do not explain why the passive is unavailable for copular constructions. The authors present no arguments supporting for their assumption that stative verbs are equivalent to copular constructions. Similarly, Hale and Keyser’s treatment of Case assignment in statives remains speculative. They claim that the DP that is part of the predicate does not get assigned accusative case. Rather, it is nominative case that is transmitted via the copula. This cannot be true once German verbs are looked at. Take (17). (17) Der Computer kostet einen Haufen Geld. The computer nom costs a bunch acc money. ‘The computer costs a bunch of money.’ 4. The authors assume that subject experiencer verbs are ambiguous between a stative and a nonstative use. The latter have an additional V head that takes the structure in (16) as its complement.





The Structure of Stative Verbs

In (17), the subject bears nominative case and the object einen Haufen clearly shows accusative morphology. 2.3.1.2 Levin and Rappaport: Structure and constant participants A somewhat different perspective on argument structure is developed by Levin and Rappaport Hovav 1995. Instead of relating syntactic structure to event semantics, these authors establish linking principles derived from the lexical semantics of verbs. Rappaport Hovav and Levin 1998 and Levin 1999 deal with the question of objecthood. The authors distinguish two kinds of arguments: first, there are structure arguments. They are DPs that fill a slot in the event structure of a verb. These are obligatory in order to render a sentence grammatical. Second, there are constant participants. A constant in the sense of Rappaport and Levin hosts the ideosyncratic meaning of a verb, i.e., it can host other participants of the event, participants that are not licensed by the event structure itself. Rappaport Hovav and Levin 1998 illustrate the distinction between structure and constant arguments along the following lines: activity verbs such as run host only one structural argument. Their event structure is given in (18). (18) [ x act ] [Rappaport Hovav and Levin 1998: 108] The activity of running involves minimally only an actor, which is assigned to the structural position x. The runner is therefore a structure argument. In contrast, verbs of surface contact like sweep the floor minimally involve an actor and a surface. Again, the actor is mapped into the structural position. The argument denoting the surface cannot be licensed by the event structure anymore, since there is no free variable left. Therefore, this argument is licensed by the constant alone. Constant arguments, the way Rappaport and Levin understand them, may remain implicit as long as they can be recovered semantically. While it is not possible to omit a structure argument, it is allowed to do so for constant arguments. Levin and Rappaport propose the event structure in (19) for stative verbs, which suggests that a state holds for a certain argument. (19) [ x ] [Rappaport Hovav and Levin 1998: 108] This structure is not only suggested for stative verbs, but also for verbs such as blossom, which express an internally caused state. These verbs show a systematic ambiguity between a stative and an eventive reading. The former expresses that a flower is in the state of blossoming as in (20a), whereas the latter describes the change-of-state until blossoming is reached, see (20b).



Chapter 2.  Theoretical considerations

(20) a. The amaryllis blossomed for ten days. b. The tree blossomed in a day. [Rappaport Hovav and Levin 1998: 125 (48)] The structure of eventive change-of-state verbs like those in (20b) is created through the addition of the become-operator. (21) [ become [ x ] ] [Rappaport Hovav and Levin 1998: 126 (59)] The authors note that “it is a widespread phenomenon that stative verbs have achievement interpretations” (Rappaport Hovav and Levin 1998: 126). In contrast to internally caused verbs, verbs that are externally caused, such as break, never show this stative/eventive ambiguity. In particular, “a sentence like The vase broke can never mean that the vase was in a state of being broken” (Rappaport Hovav and Levin 1998: 125). Rappaport and Levin also discuss the two other main aspectual classes: achievements and accomplishments. Whereas achievements have the uniform structure in (22), accomplishments may or may not have an agentive component, see (23). (22) [ become [ x ] ] [Rappaport Hovav and Levin 1998: 108] (23) a. [ [ x act] cause [ become [ y ] ] ] b. [x cause [ become [ y ] ] ] [Rappaport Hovav and Levin 1998: 108] Whereas those accomplishments that involve an intentional action correspond to verbs which express an externally caused state (e.g. break, dry, harden, melt, open; Rappaport Hovav and Levin 1998: 109), the other type of accomplishment may be realized asone of two major types. On the one hand, the structure in (23b) may be instantiated with a “placeable object” (butter, oil, paper, tile, wax; ibid.: 109). On the other hand, the state predicate may be realized as a pure location predicate,i.e., , responsible for verbs such as bag, box, cage, crate, garage and pocket (ibid.: 109). Note that according to the verb structures discussed by Rappaport and Levin, cause only occurs together with a complement that contains the become-operator. The authors do not discuss whether it is possible for cause to occur together with stative predicates that do not contain become. Hence, a stative verb containing cause is in principle compatible with a decompositional account such as Rappaport Hovav and Levin 1998. In sum, the structure for stative verbs, as far as the proposal of Levin and Rappaport is concerned, only requires that there is a single (uniform) structure argument. Different classes of stative verbs are not expected within the domain of lexical semantics.





The Structure of Stative Verbs

2.3.2 Constructionist views 2.3.2.1 Marantz: Distributed Morphology Extending the idea that the verbal phrase consists of several “layers” or verbal heads (c.f. Larson 1988, Travis 1994, Harley 1995, Pesetsky 1995, Kratzer 1996, among others) in order to host the arguments, analyses conforming to the framework of Distrubuted Morphology require that words consist of a category neutral root and a functional head. This functional head turns the root into a lexical category such as verb, noun or adjective (see, for example Halle and Marantz 1993, Marantz 1997, Marantz 2001). This idea is schematized in (24). (24)

vP 2 v root

With respect to verbs, this functional head also expresses the type of eventuality the verb denotes. For example, there are different kinds of little v-projections in order to capture causers, agents, and holder of states (see below). 2.3.2.2 Kratzer I: Stage-level and individual-level predicates Kratzer 1995 discusses the difference between stage-level and individual-level predicates. This terminology, which was first introduced by Carlson 1977, classifies predicates into two groups. Kratzer argues that stage-level predicates contain an additional argument in their thematic grid. Individual-level predicates do not have this argument. This additional argument is a spatiotemporal variable similar to the Davidsonian event variable (cf. Davidson 1967). All statives belong to the group of individual-level predicates. Kratzer distinguishes between two kinds of individual-level predicates (Kratzer 1995: 136): on the one hand, there are individual-level predicates that are not unaccusative such as know and be altruistic. These predicates have an external argument, bearing the experiencer or the theme role. On the other hand, predicates such as belong to and be known to are analyzed as individual-level predicates that are unaccusatives. Hence, they do not have an external argument. (25) Individual-level predicates that are not unaccusatives a. know 〈experiencer, theme〉 b. altruistic 〈theme〉 [Kratzer 1995: 136 (27)] (26) Individual-level predicates that are unaccusatives a. belong 〈theme, goal〉 b. be known to 〈theme, experiencer〉 [Kratzer 1995: 136 (28)]



Chapter 2.  Theoretical considerations 

Following Diesing 1992, Kratzer bases her argument on the phenomenon of quantifier split: in German main clauses, the subject may move to the sentence initial position, leaving its quantifier behind. This kind of operation is only possible for stage-level predicates and for unaccusative individual-level predicates as in (27a). Non-unaccusative individual-level predicates are ungrammatical, see (27b). (27) a.

Esel gehören ihm viele. donkeys belong to him many ‘As for donkeys, there are many that belong to him.’ [Kratzer 1995: 137 (31c)]

b. *Bürger wissen das viele. citizens know this many ‘As for citizens, many of them know this.’ [Kratzer 1995: 137 (29c)] This difference is replicated with negative quantifiers. Kratzer argues that negative quantifiers in German must be analyzed as nicht (not) plus a bare plural NP (ibid.: 145). As nicht is assumed to be base generated next to VP, subjects with negative quantifiers must therefore be internal to VP. Because subjects of individual-level predicates that are not unaccusatives must leave the VP, these are ungrammatical with negative quantifiers (see (28a)). (28) a. * ... weil das keine Kandidaten wissen. ... since this no candidates know ‘... since no candidates know this.’ [Kratzer 1995: 146 (52)] b.

... weil ihr keine Fahrräder gehören. ... since to her no bicycles belong ‘... since she owns no bicycles’ [Kratzer 1995: 145 (49)]

The verb belong corresponds to German gehören, which has no external argument but a dative possessor. Note that the verb besitzen (possess/own) belongs to the class of not unaccusative indiviudual-level predicates, since it passes alltests for the external argument. In sum, both kinds of individual-level predicates lack the Davidsonian argument. As far as the nature of the Davidsonian argument is concerned, Kratzer argues (ibid.: 154 ff.) that it does not appear within the VP at S-structure, it shows up only at LF; Davidsonian arguments are introduced by the argument structure of certain, namely stage-level, predicates. Additionally, locatives and tense may introduce or license the Davidsonian variable as well. It remains unclear whether Chomsky’s Principle of Full Interpretation (cf. Chomsky 1995) is violated here, as this principle requires that all elements that end up in the



The Structure of Stative Verbs

syntactic structure are already present in the numeration. The principle prohibits that new elements are added during the derivation. Furthermore, Kratzer distinguishes between verbs that select well-behaved indefinite objects and verbs that have ill-behaved ones. (29) verbs selecting well-behaved indefinite objects: contain, require, be close to, have (30) verbs selecting ill-behaved indefinite objects: like, appreciate, know, be responsible for, fit Well-behaved indefinite objects are always interpreted existentially. According to Kratzer, these objects remain within the VP and are therefore subject to existential closure. In contrast, ill-behaved indefinite objects are not always interpreted existentially, it is possible that they receive a generic reading. These two types of objects behave differently when occurring in when-clauses. Well-behaved but not ill-behaved objects are grammatical in this type of construction. (31) a. * When this proof contains a mistake, Mary will point it out to us. [Kratzer 1995: 148 (57b)] b. When Sue likes a movie, she recommends it to everyone. [Kratzer 1995: 151 (62a)] Only sentence (31b) is grammatical because the indefinite object is saved from existential closure by scrambling out of VP. Kratzer shows that only ill-behaved objects allow for scrambling (see (33)). (32) a. * ... falls ein Beweis einen Fehler nicht enthält. ... if a proof a mistake not contains [Kratzer 1995: 152 (64a)] b. * ... weil diese Zeitung einen schlechten Artikel meistens enthält. ... since this paper a bad article usually contains [Kratzer 1995: 152 (68b)] (33) a. ... falls ein Kritiker einen Film nicht mag. ... if a critic a movie not likes [Kratzer 1995: 152 (65a)] b. ... weil ein Pianist eine Sonate immer auswendig kann. ... since a pianist a sonata always by heart knows [Kratzer 1995: 152 (71b)] In (33), the direct object must have been scrambled out of VP, since it occurs to the left of the adverbial immer, which marks the upper boundary of the VP. The English ill-behaved verbs receive a generic interpretation, because they are not caught by existential closure.



Chapter 2.  Theoretical considerations 

(34) no scrambling: enthalten (contain), erfordern (require), grenzen (border), haben (have) (35) scrambling: mögen (like), schätzen (appreciate), beherrschen (master), verantwortlich sein (be responsible), schwerfallen (struggle), passen (fit) 2.3.2.3 Extension by Hallman Hallman 2004 extends Diesing’s claim that there are different subject positions for stage-level and for individual-level predicates. He argues that the stage/individuallevel contrast is repeated predicate-internally, i.e., objects show different behavior along the lines of this distinction as well. Hallman distinguishes between accusative intensional verbs (AIV) such as seek, owe and hunt and transitive subject-experiencer verbs (SEV) (fear, love, hate etc.).While members of the former class don’t show any exceptional behaviour, members of the latter correspond to Diesing’s class of verbs with ill-behaved objects. Hallman’s main point is “that quite generally weak objects of SEVs are interpreted according to the individual-level pattern and weak objects of AIVs are interpreted according to the stage-level pattern” (Hallman 2004: 720). Stage-level subjects and objects of AIVs allow for a nonspecific intersective reading of weak determiners, but not for a specific intersective nor a generic reading. Milsark 1974 defines weak determiners as those that may occur in the existential construction. Strong determiners, in contrast, are ungrammatical in this environment. (36) Weak determiners a. There is a unicorn in the garden. b. There are many/several/two/few unicorns in the garden. (37) Strong determiners ?? There is the/every/most unicorn in the garden. Furthermore, Keenan 1996 shows that weak determiners are intersective. In order to determine the meaning of a DP containing a weak determiner, the intersection of two sets must be taken into consideration. To illustrate this point, Hallman (Hallman 2004: 709) uses the example “a fireman is available”. In order for this sentence to be true, the intersection of the set of firemen with the set of available things must be non-empty. In contrast, strong determiners are non-intersective. Hallman 2004 shows that weak determiners, when occurring within stage-level subjects or objects of AIVs, can only have a nonspecific reading.5 It is not possible to interpret these determiners with a specific nor a generic reading. This is illustrated in (38) and Table 2.4. 5. It has been noted in the literature that opaque verbs like seek do not imply the existence of their object. However, these verbs allow for both a specific and a non-specific reading. (1) Die Irmi sucht eine CD. The Irmi looks for a CD. ‘Irmi is looking for a CD.’



The Structure of Stative Verbs

Table 2.4  Possible interpretations of AIVs Possible interpretations of AIVs nonspecific yes

specific

generic

no

no

(38) Stage-level subjects / objects of AIVs a. A fireman is available. [Hallman 2004: 712 (16a)] b. Three firemen are available. (nonspecific) [Hallman 2004: 713 (19a)] c. John is looking for a sorcerer. [Hallman 2004: 721 (39a)] d. John is looking for three sorcerers. [Hallman 2004: 726 (56a)] The opposite pattern occurs with weak subjects of individual-level predicates and weak objects of SEVs. They easily allow for a specific intersective and for a generic reading, but prohibit the nonspecific intersective interpretation. (39) Individual-level subjects / objects of SEVs a. A fireman is altruistic. [Hallman 2004: 712 (16b)] b. Three firemen are altruistic. [Hallman 2004: 713 (19b)] John fears a sorcerer. [Hallman 2004: 721 (39b)] d. John fears three sorcerers. [Hallman 2004: 726 (56b)] In order to account for this pattern, Hallman suggests the following structure for subject and object positions: external to the little v projection there is a position for strong subjects (DSUBJ), followed by the position for individual-level subjects (dsubj) and the Table 2.5  Possible interpretations of AIVs Possible interpretations of SEVs nonspecific specific no

yes

generic yes

In (1), Irmi may either look for a CD no matter which one, or she might have a particular one in mind. See Zimmermann 1993 for further discussion.



Chapter 2.  Theoretical considerations 

position for strong objects (DOBJ). Stage-level subjects (δSUBJ) originate within νP. VP accomodates the position for individual-level objects (dOBJ) and, finally, the locus of stage-level objects (δOBJ). (40) [DSUBJ ... [dSUBJ ... [DOBJ ... [VPδSUBJ v [dOBJ ... [VPδOBJ V]]]]]] [Hallman 2004: 742 (92)] Thus, the picture is repeated for subjects and for objects. The sequence for both is strong NPs – weak NPs with individual-level interpretation – weak NPs with stagelevel interpretation. Both structures are connected with a twist: the highest position for objects preceeds the lowest position for subjects. In sum, Hallman relates the different positions with different functional layers of the clause: first, theta heads such as v and V are responsible for the stage-level interpretation of both subjects and objects. Second, functional heads that license aspectual structure (T licenses outer aspect and Asp licenses inner aspect) host the NPs with individual-level interpretation. Third, Hallman assumes that there are distinct agreement phrases (AgrSP and AgrOP) which are responsible for strong NPs. 2.3.2.4 Kratzer II: Severing the external argument In Kratzer 1996, a different possible relation between syntactic structure and the Davidsonian argument is explored. In this paper, Kratzer argues that all external arguments must be introduced by a functional head, Voice in her terms, which is generally assumed to correspond to little v. Furthermore, it is assumed that “[t]he Event Argument is not syntactically realized at all” (Kratzer 1996: 133 fn.12). The external argument, in contrast to internal arguments, does not add to any particular interpretation of the predicate. For example, internal arguments but not external ones may be part of idiomatic expression, a point originally made by Marantz 1984. Thus, the meaning of the verb take in (41) is dependent on its particular object. (41) a. b. c. d.

Take a leap Take a leak Take a piss Take a break [Marantz 1997: 207 (5a-d)]

In order to compute the semantics for the combination of the external argument and the predicate, Kratzer introduces a new operation called Event Identification which relates the external argument to the predicate via the identification of the two event variables of the predicates. Although the event argument is not present in the syntax anymore, it serves a crucial role for combining the verb with the external argument. Depending on the type of the predicate, the Voice head may either introduce the agent of an action or, in case of a stative predicate, the holder of the state (Kratzer 1996: 123). Only if the two predicates have compatible Aktionsarten, event identification may take place.



The Structure of Stative Verbs

Both action and stative predicates possess the Voice projection, which licenses external arguments and checks structural Case. These two kinds of Voice head are the only active Voice heads. However, Kratzer assumes that there may be non-active Voice headsas well. In particular, “all other verbs start out with non-active voice markers, and hence without an external argument” (Kratzer 1996: 123). These verbs may get an external argument by incorporating a preposition into the non-active Voice head. In this way, Kratzer implements the unaccusative analysis of object-experiencer verbs originally put forward by Belletti and Rizzi 1988. It follows from Kratzer’s claims that non-external arguments are always introduced by the predicate itself. In sum, all verbs are headed by the Voice or little v projection. In order to derive the different verb classes with respect to argument structure, little v comes in several types: active, stative and non-active, where both active and stative Voice heads introduce external arguments. In contrast, non-active Voice may only license an external argument via preposition incorporation and raising of a former internal PP-argument. Within the account of Kratzer 1996 it is not possible to have a bare VP projection. 2.3.2.5 Harley: Different types of v In her dissertation, Heidi Harley (Harley 1995) discusses the typology of the little v projection. She agrees with Kratzer 1996 that all external arguments are introduced by this functional head. Similarly, Harley conforms to the claim that little v comes in two different types: an active type for external arguments and transitive constructions, and a non-active type for passive or unaccusative verbs. The non-active type doesn’t license an external argument. Harley redefines Kratzer’s account. She argues that the little v head corresponds to an Event head which comes in two types: either it is causative and selects an external argument, or it is a “be”-head which does not select an external argument and is used for stative expressions. The event argument (or the Voice head) is the border where l-syntax (lexical syntax in the sense of Hale and Kratzer 1993) starts; above, there is clausal syntax. L-syntax includes vP and all projections below. So, for Harley there is a separation between clausal syntax and lexical syntax. Below vP there is a basic element that forms a verb together with the Event head. Basing her ideas on the work of Hale 1995, Harley argues that bases come in three different forms, depending on the number of arguments they take. Items selecting no arguments are nouns, items selecting one argument boil down to adjectives, and elements which require two arguments are taken to be prepositions. Therefore, v combines with either N, A or P to form a verb. The causative nature of v is further developed in Folli and Harley 2005. Little v comes now in one of three types: caus, do or become. The main point is that the causative semantics is separated from the agentive interpretation. Little v may have causative interpretation without an agentive one. Only the occurrence of do poses an



Chapter 2.  Theoretical considerations 

animacy restriction on the external argument. See also Kallulli 2006 for an account regarding dative causers in unaccusative constructions that expands this idea to a more general level. 2.3.2.6 Borer: Exoskeletal approach Hagit Borer endorses a radical constructionist view regarding the lexicon/syntax interface (Borer 1994, Borer 2005). For her, the lexicon contains the encyclopedia which is a list of pairings between sound and meaning. These listemes do not bear any grammatical information. Additionally, there is a functional lexicon accommodating grammatical formatives. These are either f-morphs or certain head features. While f-morphs correspond to free grammatical morphemes such as the English future auxiliary will, head features have an abstract phonological shape. For example, the English past tense marker is a head feature, indicated by Borer as 〈pst〉. Functional heads are characterized as operator-variable pairs. The variables must be assigned a range, and functional heads contain a categorial label which restricts the set of elements, the functional operators, who are to bind the variable. The assignment of the range can take place in different ways: either by direct range assignment or by indirect range assignment. The former comes down to merging a grammatical formative, i.e., either merging an f-morph or a head feature directly. The latter, on the other hand, allows for a quantificational adverb to bind the open variable of the functional head. In addition, spec-head agreement may bind the variable as well, i.e., some element in the specifier of the functional head may bind the open value. For instance, in English possessors in SpecDP may assign definiteness to the D which they possess. This mechanism of indirect range assignment is not available in traditional versions of generative grammar such as X-bar theory. Grammatical formatives may bind more than one open variable at once if its semantic properties are compatible. In sum, the semantics of an item is computed by the combination of the conceptual value of a listeme and the semantics of the corresponding syntactic structure. As syntactic structure plays the central part in the theory of Borer, she assumes that it is this structure which corresponds to Universal Grammar. For her, functional structure is innate and universal. Borer’s main argument for pushing a constructionist perspective on argument structure comes from variable behaviour verbs. A verb such as drop may be used transitive (Irmi dropped the stone.) or unaccusative (The stone dropped.). She claims that some unaccusative verbs may be changed into non-unaccusatives. Because variable behavior verbs exist, it would be redundant to have a distinct entry in the lexicon for each of the variants. Borer stresses the fact that the lexicon is non-redundant. Therefore, the different variants appear depending on the syntactic configuration where the verbs are inserted. Thus, arguments are not licensed via the lexical entry, nor within the lexicon. Rather, arguments are licensed via the syntax. In particular, arguments are licensed in the specifiers of functional projections. For example, what is usually called the external



The Structure of Stative Verbs

argument (Williams 1981) is now termed the originator, denoting a special position in the tree. In the next few paragraphs, let’s see how the syntax of stativity is captured within Borer’s theory. In general, Borer agrees that her theory needs to accommodate the syntax of stative verbs. Regarding this point, she claims that “it is also clear that the substantiation of such a proposal must rely on a full articulation of the structure of stative predicates, a task beyond the scope of this work”. (Borer 2005: Chapter 18). Borer commits herself to a Davidsonian approach (Davidson 1967) regarding the representation of event structure in syntax. Therefore, she assumes that a special event projection (EP) hosts the Davidsonian argument. The event phrase is a functional projection, the range assignment to its variable corresponds to a mapping from predicates to events. Range assignment happens by every element that occupies the specifier of the EP. A phrase marker, according to Borer 2005, consists of the event projection (EP), heading the clause. Below, TP is generated, subcategorizing for an aspect phrase (ASPQ), which is a functional projection responsible for telic interpretation; likewise, accusative case is checked in its specifier. DPs which move to ASPQ must have a quantized interpretation in order for telicity to arise. Finally, the bare VP is the complement of ASPq. Regarding statives, Borer argues that they contain an event projection, although no special stative structure. Similarly, there is no dedicated atelic functional structure. In general, atelic structures are characterized by either not projecting an ASPQ projection, or by not projecting a contentful ASPQ projection. Stativity and Quantity (the predicate that is responsible for telicity) are both predicates of the event argument, i.e., the event argument may come in different types - stative for atelic events and quantized for telic events. Therefore, both kinds of expressions must project the event phrase layer. In contrast, activites lack this projection. Activities, then, are the default case that arises when this functional layer is missing. In Chapter 18, Borer postulates a distinct and elaborate structure for statives. She argues that statives have a special functional stative event structure (SP) that is below the EP layer. Compared to telic verbs, ASPQ and SP occupy the same position. Thus, there is either ASPQ or SP, or nothing in case of atelic activities. SP, in principle, may function as an adjecivizer, since some adjectives are stative, too. SP is taken to be higher than the projection that licenses direct objects, but Borer leaves the question open where SP is with respect to PT. Table 2.6  Eventuality structure by Borer Stative: Eventive, atelic (activity) Eventive, telic:

[EP 〈e〉E ... [SP [EP 〈e〉E [TP [EP 〈e〉E [TP [ASPQ [Borer 2005:ex. (25a-c)]

[VP/AP [VP [VP

]]]]] ]]]] ]]]]



Chapter 2.  Theoretical considerations 

It is possible, for example in the case of statives, that EP does not license the originator role. Borer criticizes the accounts which establish a battery of different little vs in order to account for the different behaviour of verbs (cf. Harley 1995, Kratzer 1996, Folli and Harley 2005). She argues that specifying as many different little v projections as there are different argument structures boils down to the lexicalist’s point of view that every kind of argument structure must receive a different syntactic type of projection. This is, according to Borer, quite redundant. 2.3.2.7 Ramchand: A Post-Davidsonian view Extending the theory of event decomposition of Pustejovsky 1991 and Hale and Keyser 1993, Ramchand argues that the Davidsonian account of event structure must be overcome. She develops a post-Davidsonian view: the event argument itself is not part of syntactic structure. Rather, the event of a verbal expression is determined by its parts. These parts, which correspond to subtrees, are either characterized as a process or as a stative event. Following Hale and Keyser, Ramchand argues that the nature of theta roles is entirely determined by their position in syntactic structure. The position of the arguments within syntactic structure determines their role in argument structure. At the same time, the position within syntactic structure determines their function with respect to the aspectual interpretation. Ramchand suggests five different participants instead of the traditional thematic roles. First, there is a causer or initiator role that is assigned in SpecvP. Then, there are two internal arguments, the undergoer, which is connected to the termination of an event in case it is quantized, and the resultee, which is part of the resultant state. Both are assigned in the specifier of VP. Finally, there is a non-aspectual argument called the rheme. Rhematic material is not assigned in a specifier position, it is always assigned in complement position. While aspectual arguments are restricted to DPs, non-aspectual ones can be DPs, APs or PPs. Providing no function for the aspectual interpretation, rhemes are parts of the description of the predicate. The structure of a telic verb thus is as follows: a causing event represented by vP initiates a process event. The process event is represented by VP, the complement of vP. It terminates with a resultant event that is expressed by the resultant phrase RP, which is the complement of VP. Note that the causing event and the resultant event are both states. Only the process event is characterized as a process.



The Structure of Stative Verbs

(42) vP 3 Initiator v’ 3 v VP 3 Undergoer V’ 3 V RP 3 Resultee R’ 2 R XP The structure of a stative verb is not as elaborate as the one in (42). It only consists of a stative event and a rheme, i.e., its structure is made up only of the little v projection. The specifier of vP, which is usually interpreted as a causer, is now understood as the holder of the state. The rhematic material occupying the complement position further specifies the predicate. It is important that statives have no VP projection, since they do not denote aprocess. Similarly, they lack a RP projection. Their structure is depicted in (43) [cf. Ramchand 2008: 55 (34)]. (43)

vP 3 DP v’ Holder 2 v DP/NP Rheme

The structure in (43) predicts that the class of stative verbs behaves completely homogeneously. A difference within the group of statives is not expected. Although Ramchand allows the rheme to vary within DP, AP and PP, no systematic subgroups are predicted. As we will see in the remainder of this thesis, the class of stative verbs must be divided into several subclasses, each of them having a distinct lexical-semantic representation. In particular, two main groups of stative verbs will emerge: on the one hand, there are those verbs that allow for a stative reading only. On the other hand, there are verbs that display a systematic stative/eventive ambiguity. In the following chapters, I will argue that the ambiguity is due to the optional presence of the tscdo and/or the become-operator. Since the structure in (43) is the only one available for stative verbs in the framework of Ramchand 2008, the analysis suggested by her is too simple. It predicts that there is onlya single, homogeneous class of stative verbs. Moreover, each of those two groups that I am going to present in this thesis consists of several classes of verbs that have a distinct argument structure. For example,



Chapter 2.  Theoretical considerations 

the group of ambiguous verbs contains object-experiencer verbs, verbs that undergo the instrumental alternation, perception verbs and “dispositional”-verbs (like help). It is not possible to capture this variety of different verbs nor the systematic ambiguity with the structure in (43). Finally, the group of verbs that allows for a stative reading only includes – among others – measure verbs, subject-experiencer verbs like love and object-experiencer verbs that assign dative case (e.g. appeal to). It is very likely that dative DPs occupy a different structural position than common rhematic material. In particular, dative experiencers behave differently from complements of measure verbs such as cost 5 euro, thus posing another problem for Ramchand’s analysis.

2.3.3 Conclusion In sum, the various approaches to argument realization, as different as they may be, converge into three core questions. Levin and Rappaport Hovav 2005 summarize these as follows: (44) “[C]urrent versions of all approaches – projectionist, constructional, and neoconstructionist – share certain features. [...] [A]ll theories of event structure distinguish between the verb root and the structural aspect of meaning or event structure.[...] All approaches must assume some mechanism which makes the various event structures, whether encoded syntactically or not, available for combination with the root in a productive and compositional way. All approaches agree that to a certain degree argument projection is determined compositionally[.]” [Levin and Rappaport Hovav 2005: 192] Regarding stative verbs, no clear picture has emerged yet. Rather, each view contains more or less implicit assumptions about statives. Some ideas that are assumed by several scholars include the following: – Stative predicates are the smallest and simplest building-blocks of event strucutre. – Stative verbs have a less complex structure than eventive verbs. – Stative verbs are very similar to copular constructions. – Stative verbs contain only a single (structural) argument which is the holder of the state. However, up to now these claims have not been studied in a thorough way nor is there sufficient empirical evidence in order to substantiate this claim. In what is to follow in the subsequent chapters, I will strongly question the assumptions on stative verbs that have been put forward in the literature. The fact that there are two main groups of stative verbs (those that allow for a systematic stative/ eventive ambiguity and those that don’t), each of them again containing different types of statives, poses severe problems for the previous analyses.



The Structure of Stative Verbs

Stative verbs are not the smallest and simplest building-blocks of event structure. The systematic stative/eventive ambiguity shows that stative verbs may have quite a complex underlying structure (take, e.g. the threaten-type verbs in Section 3.4), contrasting the two “simple” mechanisms of rendering a verb eventive (the insertion of do and become). Apart from the lack of the do and the become-operator, there is no property of argument structure that stative verbs do not have. Stative verbs may contain a causeoperator (as with verbs of the instrumental alternation, object-experiencer verbs (accusative), and dispositional verbs), and they may contain an external argument (as in subject-experiencer verbs). Not all stative verbs are similar to copular constructions. If this similarity holds at all, it is only measure verbs (Section 4.3) and verbs that select for a PP-complement (Section 4.4) that express a single property of the subject. Before I continue with the investigation of the behavior of different kinds of stative verbs, I will provide a working definition of stative verb in the next section.

2.4 A semantic definition: Stative verb 2.4.1 The ontology of Kimian and Davidsonian states After having established the distinction between Davidsonian and Kimian statives in the previous section, let’s look at the criteria Maienborn establishes in order to define the class of Kimian statives. K-states are taken to be ontological objects. These objects, being non-linguistic entities, are taken to be part of the world, therefore they have properties on their own. Because of their abstract nature, K-states cannot be perceived. These states are abstract objects which are located in time, but not in location. In particular, they are used for cognitive operations which include but are not limited to linguistic mechanisms. Maienborn summarizes these properties in the following list. (45) Ontological properties of K-states: a. K-states, being abstract objects, are not accessible to direct perception and have no location in space. b. K-states, being abstract objects, are accessible to (higher) cognitive operations. c. K-states can be located in time. [Maienborn 2005b: (48)] The ontological properties of K-states are reflected in linguistic structure. Hence, Kstates, contrasting Davidsonian statives, cannot be the complement of perception verbs. Additionally, adverbial modification is severely restricted. Davidsonian states,in contrast, do not show these restrictions on adverbial modification. K-states may not be modified by manner adverbials; locative modifiers (in particular, event related locative modifiers, but see below) are interpreted differently than they are with Davidsonian



Chapter 2.  Theoretical considerations 

statives. Furthermore, K-states can be detected via anaphorical reference. In other words, anaphorical elements like this can refer to a preceding K-state. An overview of the linguistic properties of K-states corresponding to the ontological properties is listed in (46). (46) Linguistic diagnostics for K-states: a. K-state expressions cannot serve as infinitival complements of perception verbs and do not combine with locative modifiers. b. K-state expressions are accessible for anaphoric reference. c. K-state expressions combine with temporal modifiers. [Maienborn 2005b: (49)] In turn, I will discuss the linguistic diagnostics in more detail: Section 2.4.4 treats the infinitival complements of perception verbs, and Section 2.4.5.2 shows that it is not possible to combine Kimian states with locative modifiers.

2.4.2 A filter that groups together D- and K-states Both Kimian and Davidsonian statives must not be continued with this happened .... Originally, this test was invented by Davidson 1967 in order to show that action sentences contain a hidden event variable, since they may be continued with this happened .... Davidson argued that the anaphoric element this refers to the hidden event argument. It was assumed then that verbs which fail this test lack the Davidsonian argument. Hence, they are not part of action sentences but part of stative ones. Maienborn 2003 shows that both K- and D-states fail this test. Neither of them allow for the continuation with this happened .... (47) Eva besaß ein Haus. *Das geschah während... statives Eva owned a house. This happened while... [Maienborn 2005b: (13a)] (48) Eva stand am Fenster. *Das geschah während... D-state verbs Eva stood at.the window. This happened while... [Maienborn 2005b: (12a)] Nevertheless, anaphoric reference to the Kimian state is possible. The Kimian state in (49) is picked up by the following das (this). (49) Carolin wog zu viel. Das endete erst mit der Pubertät. Carolin weighed too much. This ended not-until with the puberty. [Maienborn 2007b: (11)] Thus, it is not anaphoric reference that is excluded, but reference to an event argument. The eventive character, stressed by the following verb happened, is only present in



The Structure of Stative Verbs

action sentences. Anaphoric reference, on the other hand, is possible for action sentences and for statives.

2.4.3 Syntactic tests Maienborn 2003 only goes briefly through two syntactic tests for Kimian statives. For her, these tests function only as a first characterization, whereas the tests based on adverbial modification (see below) provide the major evidence for the Kimian state. First, K-states cannot undergo event nominalization. This morphosyntactic process creates a noun that refers to an iteration of events denoted by the respective verb. Event nominalization can either be carried out with the help of the suffix -erei or with the circumfix Ge-...-e. Action verbs, as in (50), allow this process freely. In (50), the derived nominals Schreierei and Geschreie (screaming) denote an iteration of screaming-events. See (51) for the unavailability with Kimian statives; (52) shows that this is not dependent on the type of compound. (50) schreien – Schreierei – Geschreie scream – screaming – screaming (51) * Romy-Schneider-Geähnle Romy-Schneider-resemblance [Maienborn 2003: p. 61 (38a), my translation] (52) * Geähnle resemblance Second, do-periphrasis is not available for stative verbs. Certain variants of German, including Bavarian dialects, allow for an optional insertion of tun (do).6 This tun is only present with activity verbs such as lesen (read), illustrated in (53). Kimian statives do not allow for tun-insertion at all as in (54), whereas Davidsonian statives are marginal in this construction; see (55). Note that not all variants of German allow for do-periphrasis. (53) Da Poidi duad grod a Heftl lesn. Viennese The Poldi does right now a booklet read. ‘Poldi is reading a booklet right now.’

6. The insertion of tun is one of the genuine focus constructions in Bavarian/Viennese. (1) Lesn duad di Iami imma nua in da Nochd. Viennese Reading does the Irmi always only in the night. ‘It is reading that Irmi does only at night.’ However, the tun-insertion which is used as a test for stativity is not related to the focus construction.



Chapter 2.  Theoretical considerations

(54) * Catherine tut diese Oper gut kennen. Catherine does this opera well know. [Maienborn 2003: p. 63 (42a), my translation] (55) * Renate tut verlegen in der Ecke stehen. Renate does bashfully in the corner stand. [Maienborn 2003: p. 63 (41a), my translation] Maienborn 2003 uses both event nominalization and do-periphrasis as a heuristic in order to determine the class of Kimian statives. She does not use these tests as an argument for the existence of Kimian states.

2.4.4 Infinitival complements of perception verbs In contrast to all other verb classes, statives cannot serve as the complement of perception verbs. (56) * Ich sah meine Tante Romy Schneider ähneln. I saw my aunt Romy Schneider resemble. [Maienborn 2005b: (9c)] One problem that comes up when using this test is that psychological activities are not easy to perceive. For example, one cannot see someone think. Therefore, when trying to separate psychological states from psychological activities later on, this test cannot be employed.

2.4.5 Tests based on adverbial modification The major evidence for the difference between Kimian statives and verbs containing a Davidsonian argument comes from adverbial modification. Following the Davidsonian paradigm, Maienborn assumes that the event argument is the place where certain (namely, event-related) adverbs are connected to the linguistic structure. As adverbials are always optional elements, they have to be anchored within the linguistic structure at some point. Therefore, the presence of an adverbial hints at the existence of a Davidsonian argument. Kimian states, on the other hand, do not contain the Davidsonian argument. Hence, adverbial modification is severely restricted. 2.4.5.1 Manner adverbials Manner adverbials modify the way an event happens. Thus, these adverbs need the presence of a Davidsonian argument in order to be used feliciously. Sentences without an event argument such as Kimian state expressions do not allow for manner modification.





The Structure of Stative Verbs

(57) * Paul besitzt sparsam / spendabel viel Geld. Paul owns thriftily / generously much money. [Maienborn 2005b: (30b)] However, Davidsonian statives, bearing a full-blown event argument, do allow for manner modification freely. (58) Carol saß reglos / kerzengerade am Tisch. D-state verbs Carol sat motionless / straight.as.a.die at.the table. [Maienborn 2005b: (31b)] Similarly, comitatives may modify the internal structure of events but not of Kimian states. Hence, they are excluded with the latter, but are grammatical with Davidsonian statives. (59) * Maria ähnelt mit ihrer Tochter Romy Schneider. statives Maria resembles with her daughter Romy Schneider. [Maienborn 2005b: (30a)] (60) Paul schläft friedlich / mit seinem Teddy / ohne Schnuller. D-state verbs Paul sleeps calmly / with his teddy / without dummy. [Maienborn 2005b: (31a)] Certain manner adverbials such as mit Begeisterung (with enthusiasm) seem to occur together with Kimian states, but they are all subject to a shifted interpretation. Therefore, these examples do not provide counterevidence to the claim that Kimian states do not allow for manner modification. See Maienborn 2003 for a detailed discussion of these coerced readings. 2.4.5.2 Locative modifiers The interpretation of local modifiers functions as a further hint that there the Kimian state argument differs from the Davidsonian event argument. In order to find the relevant evidence, different local modifiers have to be set apart. Maienborn illustrates in (61) that there are at least three different local modifiers which must not be mixed up. (61)

weil in den Anden Schafe auf dem Marktplatz because in the Andes sheep on the market place an den Ohren markiert werden at the ears marked are [Maienborn 2005b: (16)]

First, in den Anden is a frame-setting adverbial. It does not specify the location where the event is happening, but it provides a general context in which the sentence is interpreted. Frame-setting adverbials, therefore, can be interpreted in several ways. With respect to (61), the frame-setting adverbial can serve as a general restriction, i.e., it can mean that people who live in the Andes do in general treat sheep this way. See Maienborn 2001 for a detailed discussion of the possible interpretations of frame-setting adverbials.



Chapter 2.  Theoretical considerations 

The other two local modifiers are eventuality-related. They specify the location where the event takes place. Crucally, in order to be located in space, an event must be there in the first place. Maienborn distinguishes between external and internal eventuality-related locative adverbials, which both modify the Davidsonian event argument. Example (61) contains an external locative modifier (auf dem Marktplatz) and an internal one (an den Ohren). As the eventuality-related locative modifiers are restricted to sentences which contain a Davidsonian argument, they are ungrammatical with Kimian statives, as shown in (62). (62) * Die Tomaten wiegen neben den Zwiebeln 1 Kg. statives The tomatoes weigh besides the onions 1 kg. [Maienborn 2005b: (26a)] Notice that locative modifiers are not excluded in general for Kimian statives. If they do cooccur, as in (63), the locative adverbials must be interpreted as frame-setting adverbials. (63) Eine Flasche Rotwein kostet im Restaurant 45 DM. One bottle red wine costs in.the restaurant 45 DM. [Maienborn 2001: 217 (60c)] Example the locative modifier im Restaurant of example (63) is understood as a framesetter. In other words, (63) means that red wine costs 45 DM when it is sold in restaurants. It cannot mean that the bottle I bought at the super marked for 20 DM suddenly costs 45 DM the moment I enter the restaurant, carrying the bottle with me. Thus, if locative modifiers do occur at all with Kimian statives, they are always interpreted as frame-setting adverbials. Davidsonian statives, containing an event argument, do allow for eventuality-related locative modification. The modifier bei Calais of example (64) clearly specifies where geographically the lying event took place. (64) Die spanische Armada lag bei Calais vor Anker. D-state verbs The Spanish Armada lay near Calais at anchor. [Maienborn 2005b: (27b)] 2.4.5.3 Degree readings of ein bisschen In general, the modifier ein bisschen (a little bit) is ambiguous in two ways. It can be interpreted either as a degree modifier or as modifying the temporal extension of an event. Both readings are possible with an eventive verb like schwitzen (sweat) as in (65). (65) Carol hat gestern ein bisschen geschwitzt. degree and eventive reading Carol has yesterday a little bit sweated. [Maienborn 2005b: (36)] The eventive interpretation of ein bisschen in (65) is that Carol was sweating for a short period of time. To illustrate an appropriate context for the time-span reading, imagine



The Structure of Stative Verbs

Carol is moving from one apartment to another, and yesterday she was sweating while she carried a box of books. The rest of the day she only wrapped old newspapers around her china, which didn’t cause her to sweat. The degree interpretation describes how much sweat there was, i.e., how wet Carol got. In the case of (65), her t-shirt might have stayed almost completely dry. Crucially, the time-span reading is only available for verbs that bear a Davidsonian argument, since ein bisschen specifies that the event didn’t last very long (see (66)). (66) Paul hat ein bisschen im Garten gesessen. eventive reading Paul has a little bit in.the garden sat. [Maienborn 2003: (37b)] In Kimian statives, there is no event argument, hence, a time-span reading is excluded. Kimian statives therefore either allow for the degree reading only, as in (67a), or they are completely ungrammatical, as (67b). (67) a. Carol ähnelte ein bisschen ihrer Großmutter. only degree reading Carol resembled a little bit her grandmother. b. * Nach ihrer 5. Heirat hieß Liz ein bisschen Burton. After her 5th marriage was-named Liz a little bit [Maienborn 2005b: (39ca)] Note that the degree reading is only possible for gradable predicates. Non-gradable predicates like the one in (67b) are excluded for the modification with ein bisschen.

2.4.6 Summary In the previous sections, several linguistic tests were described. These tests show that the underlying Kimian state argument differs from the Davidsonian event argument. This difference can be detected via several linguistic phenomena, in particular, adverbial modification is severely restricted for Kimian statives. In the next chapter, I will use these tests as evidence that a Kimian stative reading exists with different verbs. The list of criteria summarized below will function as a “checklist” that helps to identify Kimian statives in the following chapters. Syntactic tests: – Event nominalization – tun-insertion Infinitival complements of perception verbs Tests based on adverbial modification: – Manner adverbials – Locative modifiers – Degree readings (ein bisschen)



Chapter 2.  Theoretical considerations 

2.5 Other kinds of stative expressions Apart from stative verbs, which are the topic of this dissertation, three other main kinds of stative expressions exist: generic sentences, perfect constructions and the adjectival passive. All of them render a sentence stative. In other words, these mechanisms may stativize a sentence. Take, for example, a telic expression such as eat an apple. This verb phrase is clearly eventive when used in contexts such as Last sunday, Mary ate an apple. Such verb phrases may be interpreted stative, once they are part of a generic construction, or if they are used in the perfect, or in the adjectival passive. (68) a. Last Sunday Mary ate an apple. b. Maria ißt Äpfel. (generic) Maria eats apples. ‘Maria is eating apples.’ c. Maria hat einen Apfel gegessen. (perfect) Maria has an apple eaten. ‘Maria ate an apple.’ d. Der Apfel ist aufgegessen. (adjectival passive) The apple is up-eaten. ‘The apple is eaten.’ Lexical statives, on the other hand, are verbs that are stative in the first place, i.e., they are stative without a generic operator, and they are stative even if they do not appear in perfect tense. Those statives, which are not derived, are the topic of the current work. Nevertheless, let’s take a brief look on the other main mechanisms for stativizing a sentence.

2.5.1 Generic expressions Krifka et al. 1995 distinguish two main types of genericity: reference to a kind and general property. Example (69) illustrates the former case. (69) The potato was first cultivated in South America. [Krifka et al. 1995: 2 (1a)] In this example, the noun phrase the potato does not refer to a particular potato, but to the kind of potato. Hence, the sentence does not describe the particular event of cultivating the very first potato ever, but refers to the start of the cultivation of potatoes in general. Very often, reference to a kind is done with bare plurals, such as Potatoes were first cultivated in South America. The second kind of generic expressions, general properties or characterizing sentences are exemplified in (70). Characterizing sentences describe properties that peo-



The Structure of Stative Verbs

ple do generally or as a habit. Hence, so called habitual sentences are related to generic constructions as well. (70) John smokes a cigar after dinner. [Krifka et al. 1995: 3 (2a)] Example (70) describes a habit of John: he usually smokes a cigar after dinner. Therefore, it is not a single event of smoking that (70) refers to, but a whole series of such events.

2.5.2 Perfect as a stativizer Perfect sentences such as (71) show a similar behavior to stative ones. For example, Katz 2003 discusses the similarities between the two. (71) Irmi has eaten a cookie. Both statives and perfect sentences, e.g., have a present orientation. They are allowed to occur in their simple form, whereas other types of expressions must occur in the progressive, in order to be acceptable in a sentence bearing present tense. Still, the perfect applies to a verb that is already given, and it does not affect the lexicalsemantic structure of words. Therefore I do not discuss the perfect in what is to follow.

2.5.3 Adjectival passive Another construction that qualifies as stative is the adjectival passive. A substantial amount of work regarding the German Zustandspassiv (including, among others, Haiden 2005, Helbig and Kemptner 1973, Kratzer 2000, Maienborn 2007a, Rapp 1997, Rapp 1998) has dealt with the question if the verb form in sentences such as (72) is lexically or syntactically derived. (72) Das Bett ist zerwühlt. The bed is ruffled. ‘The bed is ruffled.’ As it may be, the adjectival passive is not a base form of a verb. As the scope of this book concentrates on the lexical-semantic structure of underived verb forms, I leave aside the question of the adjectival passive.

chapter 3

Stative/eventive ambiguities 3.1 Introduction This chapter discusses a number of verbs which display a stative/eventive reading, i.e., several types of verbs exist that allow for a stative reading as well as an eventive one. In the following sections, this ambiguity will be discussed in detail, whereby the elements that are responsible for the eventive reading will be identified. Thus, the questions that guide this chapter are: what kind of grammatical mechanism is responsible for the eventive / the stative reading? Does a systematic set of alternations exist and if so, why should that be the case? Verbs that display the stative/eventive ambiguity are verbs that undergo the instrumental alternation (e.g. surround, obstruct), object-experiencer verbs assigning accusative case (e.g. depress, frighten), dispositional verbs (help), threaten-type verbs and perception verbs (see, hear). This chapter is organized as follows. A section is devoted to each of the verbs that display the stative/eventive ambiguity. Within each section, I will first examine the verbs with the help of the tests described in Section 2.4. Next, the event structure will be identified. Crucially, these verbs express a kind of result state that is created. Different kinds of adverbials help to identify the presence of this state. Finally, I will suggest a lexicalsemantic structure for each of the verbs that display a stative/eventive ambiguity.

3.2 Ambiguity due to instrumental alternation 3.2.1 Introduction One group of verbs that are ambiguous between a stative and an eventive reading are verbs that undergo the instrumental alternation. When Kratzer 2000 discusses different kinds of adjectival passives, she comes across certain verbs which are ambiguous between a stative and an eventive reading. These include obstruct, surround, cover, support and illuminate. Kratzer points out that some psych verbs like depress and worry (object experiencer verbs) alternate between these readings as well (Kratzer 2000: 9). In the following section, I will concentrate on the former group; the latter one is discussed in a separate section. Kratzer 2000 characterizes these verbs as having a Davidsonian event argument, or – in case of the stative reading – a state argument, together with a target state



The Structure of Stative Verbs

argument (like an obstructed blood vessel). Kratzer points out that verbs like obstruct always denote a causal relationship. In case of the eventive variant, the verb expresses a causal relation between the event and the resultant state. The stative variant is quite similar, here the causal relation holds between two states. This difference is illustrated with animate, agentive subjects (73a) and non-agentive subjects (73b), respectively. (73) a. The army surrounds the village. b. The wall surrounds the village. Several other verbs display the same ambiguity, a non-exhaustive list is given in (74). (74) füllen (fill), schmücken (decorate), überdachen (roof), bedecken (cover), verbinden (connect), stören (disturb), versperren (lock), blockieren (block) All of these verbs undergo the instrumental alternation, i.e., they either have an agent bearing nominative case together with a possible instrumental phrase introduced by a PP, or there is only the instrument present, this time bearing nominative case. Most important, the agentive version is eventive, whereas the instrumental counterpart corresponds to the stative causation verbs described by Kratzer. The pattern of the instrumental alternation is illustrated in (75) and (76). (75) a. Die Irmi füllt die Vase mit Wasser. The Irmi fills the vase with water. ‘Irmi is filling the vase with water.’ b. Wasser füllt die Vase. Water fills the vase. ‘Water is filling the vase.’ (76) a. Der Poldi schmückt die Torte mit Kerzen. The Poldi decorates the cake with candles. ‘Poldi is decorating the cake with candles.’ b. Kerzen schmücken die Torte. Candles decorate the cake. ‘Candles are decorating the cake.’ Kratzer demonstrates that the instrument is independent from the agent. When discussing the case of The tissue obstructs the blood vessel, Kratzer points out that this state could have been created without there being an agent, i.e., the blood vessel could have been obstructed ever since the person’s birth. Still, even this case is part of the more general pattern of instrumental alternation, since the verb obstruct takes part in the instrumental alternation, even if that particular obstruction of the blood vessel is not caused by an agent. The instrument and the agent are independent; any possible relation between them is due to world knowledge.



Chapter 3.  Stative/eventive ambiguities 

A similar case regarding the stative/eventive ambiguity with respect to verbs with a causative flavor is reported by Torrego 1998 and Pesetsky and Torrego 2004 for Spanish. Although the cases described do not involve the instrumental alternation explicitly, the pattern that emerges nevertheless fits into the current discussion. As reported by Torrego, Spanish direct objects can be either marked with abstract accusative (i.e., bare accusative) or with marked accusative. If marked accusative is assigned, the object is licensed via the preposition a. In general, the marked accusative is used when the subject is agentive, i.e., if there is a dative preposition introducing the accusative object, the subject must denote an agent. It is not grammatical to use the marked accusative together with a non-agentive subject. This ambiguity is demonstrated with the verb tapar (block). In (77a), only the stative reading is available where the police blocked the view of the speakers. In this case, it is not neccessary that the police did an intentional action as to disturb the view. They might be doing something else but happen to hide the speakers. However, there is no action going on. In contrast, (77b) refers to the (intentional) action of the police which results in the view being blocked. Only in this latter, eventive case, the object is introduced via the preposition a. (77) a marks non-stativity: tapar a. La policía tapaba los oradores. ‘The police blocked the view of the speakers.’ [stative only] b. La policía tapaba a los oradores. ‘The police moved so as to block the view of the speakers.’ [non-stative only] [Pesetsky and Torrego 2004: (17ab)] However, the alternation between marked and structural accusative case is intertwined with some facts about definiteness. As Torrego 1998 discusses (Chapter 2.8), the case alternation only appears with indefinites. As soon as definites come into play, the marked accusative and therefore the active reading becomes obligatory. (78) a. b.

Escondieron *(a) su amigo. pro hid to their friend [Torrego 1998: 66 (59b)] Llamaron *(a) los bomberos. ‘They called the firemen’ [Torrego 1998: p. 66(60a)]

Similarly, partitives require the marked accusative and count as definites in Spanish. (79) Escondieron *(a) cinco de ellas. pro hid (to) five of them [Torrego 1998: 68 (64a)] In sum, the Spanish data support the claim that two different readings of verbs with a causative flavor exist: one where the agent is engaged in some kind of action causing



The Structure of Stative Verbs

some kind of effect, the other where only the state of affairs is described, involving neither an action of an agent nor an increasing development of the resultant state.

3.2.2 Tests for a Kimian stative reading Whereas the agentive variant is clearly eventive (see (80)), the instrumental counterpart has a Kimian stative reading, as indicated in the next few paragraphs. Passing the tests for a Davidsonian reading, the agentive sentence in (80) is compatible with an eventuality-related locative modifier, a manner adverbial and an instrumental phrase. (80) Die Irmi schmückt die Torte in der Küche phantasievoll mit Kerzen. The Irmi decorates the cake in the kitchen creatively with candles. ‘Irmi is decorating the cake creatively in the kitchen.’ 3.2.2.1 Manner adverbials Event-related manner adverbials are excluded with Kimian stative verbs. Nevertheless, verbs that undergo the instrumental alternation seem to allow them freely. Consider the examples in (81). (81) a. Die Irmi schmückt die Torte auf eine phantasievolle Weise. The Irmi decorates the cake in an imaginative way. ‘Irmi is decorating the cake in an imaginative way.’ b. Kerzen schmücken phantasievoll die Torte. Candles decorate imaginatively the cake. ‘Candles are decorating the cake imaginatively.’ c. Die Kerzen schmücken auf phantasievolle Weise die Torte. The candles decorate in imaginative way the cake. ‘The candles are decorating the cake in an imaginative way.’ d. * Die Kerzen schmücken auf bunte Weise die Torte. The candles decorate in colorful way the cake. ‘The candles are decorating the cake in a colorful way.’ e. Das Spitzendeckerl schmückt den Tisch auf elegante Weise. The lace cloth decorates the table in elegant way. ‘The lace cloth is decorating the table in an elegant way.’ It is not surprising that the agentive variant of the verb schmücken (decorate) in (81a) allows for event-related manner modification. The adverbial auf phantasievolle Weise (in an imaginative way) expresses that the way of decorating was imaginative, i.e., Irmi had some good ideas about the arrangement of the candles which – in the end – caused the cake to look beautiful.



Chapter 3.  Stative/eventive ambiguities 

However, example (81d) suggests that the set of possible manner adverbials is restricted. Although one can easily imagine a situation where a brown cake is decorated with colorful candles, the corresponding sentence is nevertheless unacceptable. Contrary to the expectation, the stative variants of verbs that undergo the instrumental alternation do not prohibit manner adverbials. Hence, (81c) and (81e) are not ruled out. This fact is further illustrated in (82b). (82) a. Die Glücksfee hat das Konto auf magische Weise gefüllt. The fortune fairy has the account in a magical way filled. ‘The fortune fairy filled the account in a magical way.’ b. Das Geld hat das Konto auf magische Weise gefüllt. The money has the account in a magical way filled. ‘Money filled the account in a magical way.’ However, I take these adverbials not to be event-related, but to specify the result of a previous action. In particular, the examples in (81c), (81e) and (82b) do not contain event-related manner adverbials, but adverbials thatspecify the resultant state. Therefore, the judgment of these data does not count as evidence that verbs that undergo the instrumental alternation describe an event that can happen in a certain way. Resultative adverbs, like in the ones in (83), have been discussed by Geuder 2000 and Maienborn 2003. Crucially, these adverbs do not specify the way an event happened, but they express the way the resultant state was. So, in (83), the way of a created resultant state is specified, i.e., the size of the opening and the amount of plants. (83) a. Ein Fenster war weit offen. A window was wide open. [Maienborn 2003: 97 (93a); my translation] b. Die Beete waren üppig bepflanzt. The flower beds were amply planted. [Maienborn 2003: 97 (93b); my translation] Most important, result adverbs are only acceptable if there was an action that resulted in the given state. Example (84) is odd because caves, unlike doors, never change their state of being open. (84) ?# Die Höhle war weit offen. The cave was wide open. [Maienborn 2003: 98 (96); my translation] Hence, I take (81c) and (81e) to be resultative adverbials much like those in (83). In particular, it is the decoration that is imaginative or elegant. Still, the question of the exact nature of how resultative adverbials are licensed with verbs that undergo the instrumental alternation remains open.



The Structure of Stative Verbs

3.2.2.2 Event-related locative modifiers When occurring with Kimian stative verbs, locative modifiers cannot be interpreted in a way such that they modify the event. The only possible interpretation is that they are understood as frame-setters, thus contrasting Davidsonian statives and eventive verbs, which allow for event-related modification. (85) a. * Auf dem Tisch haben Kerzen die Torte geschmückt. On the table have candles the cake decorated. ‘On the table, candles decorated the cake.’ b. * Wasser hat die Vase neben dem Computer gefüllt. Water has the vase next to the computer filled. ‘Water filled the vase next to the computer.’ Locative modifiers are not allowed. Note that (85a) and (85b) are grammatical if the locative modifier is interpreted as specifying the location of the accusative object, i.e., the cake can be understood to be on the table and the vase to be next to the computer. The locative modifiers cannot be interpreted in a way that the state of decorating or filling holds at a particular location. 3.2.2.3 Degree readings Next, consider the different readings a degree expression such as ein bisschen (a little) may get. While Davidsonian (eventive) verbs allow for both a degree and a time-span reading, Kimian stative verbs do not permit the latter. Hence, the ein bisschen-test indicates that verbs that are part of the instrumental alternation do have a true Kimian stative reading, as the time-span reading is not allowed for the examples given in (86). (86) a. Kerzen haben die Torte ein bisschen geschmückt. Candles have the cake a little decorated. ‘Candles decorated the cake a little.’ b. Wasser hat die Vase ein bisschen gefüllt. Water has the vase a little filled. ‘Water filled the vase a little.’ Regarding (86a), the modifier expresses the fact that the degree of decoration was little. Thus, the sentence is compatible in a situation where a pretty plain cake was decorated with some candles, enhancing the over all look just a little bit. However, it is not compatible with a situation where the candles decorated the cake only for a short period of time (e.g. because they were burnt down soon).



Chapter 3.  Stative/eventive ambiguities 

3.3.3 Stative causation? 3.2.3.1 Counterfactual analysis of causation After having established that verbs of causation can have a Kimian stative reading, I would now like to take a closer look at the notion of “stative causer”. This section first discusses the notion of “cause” as developed by David Lewis, and then goeson to describe how these insights are captured in Semantic Form. Causation, as analyzed by Lewis 1973 corresponds to a counterfactual relation between two propositions. In his terms, only the occurrence of an event c, expressed as O(c), and the occurrence of an event e, labeled O(e), can stand in a causal relation. The occurrence of an event such as O(c) is a proposition, i.e., it is either true or false that the event c has occurred. Lewis takes causation to be expressed directly by the English complementizer because, which relates two propositions by connecting two sentences explicitly. Verbs can express causation as well, but they do it in a more indirect way than complementizers. Crucially, only a proposition but not a noun or a person can stand in the causal relation, excluding an “assignment” of causation to a particular theta role. The only thing that causation allows, at least in the way it is defined by Lewis, is that two elements, i.e., two propositions, can stand in relation to each other. Of course, this relation can be labeled causation or causative relation, but that is only a metatheoretical term. Lewis 1973 captures causation in terms of counterfactuals. Roughly speaking, the idea embraces the following two conditions. The proposition expressing the result only holds if the causing one holds, and if the causing proposition does not occur, the resulting one doesn’t either. The two conditions can be rephrased in terms of events: a proposition “s1 causes s2 if and only if both s1 and s2 occur but s2 would not have occured if s1 had not occurred” (Wunderlich 1997: 35). Dowty 1979 illustrates the counterfactual analysis of causation with the help of the following example. (87)

a. b. c. d.

John opened the door. The door was not open just before John acted. The door was open just after John acted. The door would not have become open on that particular occasion if John had not acted and all else had remained the same. [Dowty 1979: 99f. (116)-(117)] Dowty points out that the cause relation may not only hold between two occurrences of events, but also between two occurrences of states. He illustrates this non-eventive causation in natural language with the sentence in (88), crediting Charles Fillmore. (88) Mary’s living nearby causes John to prefer this neighborhood. [Dowty 1979: 103 (122)] Most important, Lewis “does not assume a relation of temporal priority between cause and effect and can thus potentially deal with phenomena such as backwards causation



The Structure of Stative Verbs

and closed causal loops.” (Dowty 1979: 109). Therefore, the possibility that the cause and the effect hold simultaneously is allowed as well. Of course, there is much more to say about causation. On the one hand, there is a vast discussion on this topic within philosophy which I won’t go into here. On the other hand, many open questions regarding the causation relation between events (e.g., direct vs. indirect causation, the size of events that take part in causation etc.) cannot be addressed at this point either. The aim of the previous section is only to remind the reader of two things: first, only propositions but not individuals may stand in the causation relation. Second, the causation relation is not limited to (occurrences of) events, but includes relations among (occurrences of) states as well. 3.2.3.2 Causation and event structure: A note on Kaufmann and Wunderlich While the philosophical view on causation converges on the claim that causation is a relation between two propositions or two events, the linguistic perspective questions how language expresses these relations. What kind of mechansims are there to expressa causative relation? In general, there exist different possibilities of expressing such a relation. For example, one might use two clauses and a complementizer like because. Other causative constructions that appear across languages include the usage of an extra verb like make, or some special causative morphology (cf. Marantz 1984, Baker 1988 among many others). Of interest here are lexical causatives, i.e., verbs that describe a causative relation but do not employ explicit syntactic or morphological means. Wunderlich 1997 discusses two mechanisms which underlie causative verbs. In particular, Wunderlich focuses on lexical causatives such as empty the bottle and resultative constructions such as drink the bottle empty. It is the mechanism that he develops for lexical causatives which provides useful insight for stative causers. First, consider lexical causatives as in (89). (89) *Er leerte die Flasche. he emptied the bottle [Wunderlich 1997: 35 (8a)] Wunderlich argues that constructions as the one in (89) describe relations between individuals and propositions (Wunderlich 1997: 36). This contrasts with the standard philosophical view which requires causation to be a relation between two propositions. But, as Wunderlich points out, it is not possible to have a sentential subject denoting the causing proposition with deadjectival verbs like to empty. Therefore, Wunderlich concludes that the causing situation is not present in the underlying structure of lexical causative verbs. It is only a not further specified entity that is present in the Semantic Form of lexical causatives. Hence, the structure of to empty is as in (90). (90) to empty: cause(x, become(empty(y))) (s) [Wunderlich 1997: 36 (11)]



Chapter 3.  Stative/eventive ambiguities 

Cause, as characterized by Wunderlich 1997, “abstract[s] away from any properties of the causing situation except the existence of an agent or causer entity.” (Wunderlich 1997: 36). Thus, cause does not provide any constraints on the first argument. It is designed to accommodate all possible kinds of subjects. Wunderlich contrasts lexical causatives with resultative constructions as in (91). (91) Er trank die Flasche leer. he drank the bottle empty [Wunderlich 1997: 35 (8b)] According to Wunderlich 1997, resultative constructions are “extensions of verbs” (Wunderlich 1997: 36) that include a resultant state which is expressed by the addition of the conjunct become(P(y)) in the representation at SF. Hence, (91) is assigned the structure in (92). (92) drink empty: {drink(x,z) & become(empty(y))}(s) [Wunderlich 1997: 36 (12)] Following Kaufmann 1995a and Kaufmann 1995b, Wunderlich postulates two principles that govern the well-formedness of complex lexical entries such as (92). These two principles restrict the possible structures of complex predicates, i.e., of predicates that consist of two conjuncts. Simple predicates such as activity verbs are not discussed by Wunderlich, and are taken not to fall under those principles. First, the Coherence principle given in (93) requires the propositions (or the conjuncts) to be interpreted either causally or contemporaneously. (93) Coherence A lexical SF conjunction is contemporaneously or causally interpreted. [Wunderlich 1997: 36 (13)] Second, the Connexion principle in (94) governs the position of the become operator within the structure of causatives. According to (94), the transition must be in the second conjunct; it is not allowed to occur in the first conjunct. Therefore, a structure which contains a become operator in its first conjunct is ruled out by (94). (94) Connexion The second member in a lexical SF representation specifies inferences about the first member. [Wunderlich 1997: 42 (31)] Apart from the distinction between deadjectival verbs and resultative constructions, Wunderlich 1997 does not make any further claims about the representation of lexical causative verbs: it remains open whether the cause-operator or the structure consisting of two conjuncts is best for representing causative relations in general. For example, Wunderlich offers both the structures in (95b) and (95c) when discussing three-place verbs.



The Structure of Stative Verbs

(95) a. b. c.

Sie gab ihm ein Wörterbuch. she gave him dat a dictionary acc [Wunderlich 1997: 55 (59a)] λz λy λx λs cause(x, become poss(y,z))(s) [Wunderlich 1997: 55 (60a)] λz λy λx λs {act(x, ...) & become poss(y,z)}(s) [Wunderlich 1997: 56 (63a)]

Of course, one could argue that the structure involving of two conjuncts should only be used if sentential subjects are allowed, as the two conjuncts represent the two propositions which stand in the cause relation directly. This is clearly not the case with (95a). Sentential subjects are not possible with three-place verbs such as give. (96) *That Mary moved the book through the room, gave him the dictionary. Although this is clearly a disadvantage of the structure in (95c), Wunderlich does not exclude it. Moreover, Kaufmann 1995a states that a structure such as (97) is ruled out if the two conjuncts do not share an argument and are not causally related subsituations. (97) *[P(x) & Q(y)] [Kaufmann 1995a: 199 (2)] (98) Coherence of Situations The participants of a situation encoded by a verb must be meaningfully related. [Kaufmann 1995b: 86 (P1)] The principle of Coherence rules out possible verb structures like the one in (99), where a certain thing (say a bomb) explodes and another thing (say the sky) lights up at the same time. (99) §[explode(x) & light-up(y)] [Kaufmann 1995b: 85 (14a)] Kaufmann concludes that it is not sufficient for two conjuncts to occur at the same time to be able to form a verb structure. In order to form a representation that fulfills the requirements of grammar, the two conjuncts must be related in a “meaningful” way, i.e., “a (more or less) immediate “manipulation” of the participant of the resulting situation by the participant of the causing situation must be construable” (Kaufmann 1995b: 86). In conclusion, I will follow Wunderlich and use the cause-operator to represent causative relations within a single verb. Moreover, I take the structure consisting of two conjuncts represent resultative constructions, which do not require such a tight a relation between the arguments of the first conjunct and those of the second one.



Chapter 3.  Stative/eventive ambiguities 

3.2.4 The structure of stative verbs of causation 3.2.4.1 The stative reading According to the previous section, the lexical representation of a causative relation boils down to a predicate that contains a cause-operator. Consider, for example, (100). (100) a. Der Baum hat die Aussicht auf den See gestört. The tree has the view to the lake disturbed. ‘The tree blocks the view of the lake.’ b. Die Haare verstopfen den Abfluß. The hair obstruct the drain. ‘Hair obstruct the drain.’ (101) a. b.

λy λx λs cause(x, disturb(y))(s) (tree) (view) = = λs cause(tree, disturb(view))(s) λy λx λs cause(x, obstruct(y))(s) (hair) (drain) = = λs cause(hair, obstruct(drain))(s)

Note that it is possible to construct a paraphrase containing a sentential subject for these cases. Example (102) shows the sentential subject in its base position; in example (103) it is extraposed. (102) a. Daß ein Baum vor dem Fenster steht, behindert die Aussicht. That a tree in front of the window stands disturbs the view. ‘It is blocking the view that there is a tree in front of the window.’ b. Daß Haare im Abfluß sind, verstopft ihn. That hair in the drain are obstructs it. ‘It obstructs the drain that there are hair in it.’ (103) a.

Es behindert die Aussicht, daß ein Baum vor dem It disturbs the view that a tree in front of the Fenster steht. window stands. ‘It is blocking the view that there is a tree in front of the window.’

b. Es verstopft den Abfluß, daß Haare drinnen sind. It obstructs the drain that hair inside are. ‘It obstructs the drain that there are hair in it.’ The important point here is that stative causers do not contain an activity or a transition operator. The cause-operator alone does not give rise to an eventive interpretation. As far as the syntactic structure is concerned, I assume that the cause-operator is hosted within a V-head. The stative causative verb, then, projects only a VP layer, which expresses a resultant state. This state is subject to optional become-modification (see below).



The Structure of Stative Verbs

(104) a. Die Haare verstopfen den Abfluß. the hair obstruct the sink. ‘Hair obstruct the sink.’ b. VP wo DP V 6 wo die Haare V V cause wo DP V 6 verstopfen den Abfluß 3.2.4.2 Stative/eventive ambiguities As pointed out in Section 3.2, certain verbs that undergo the instrumental alternation are ambiguous between an eventive and a stative reading. While the agentive alternant (Sue is obstructing the traffic with her car) is of course eventive, the instrumental realization (The tissue obstructed the blood vessel) is stative. This was shown in Section 3.2.2, the structure is given in (105). (105) Stative: a. Das Gewebe verstopfte das Blutgefäß. The tissue obstructed the blood vessel. ‘Tissue obstructed the blood vessel.’ b. λy λx λs cause(x, obstruct(y)) (s) (blood vessel) (tissue) = = λs cause(tissue, obstruct(blood vessel))(s) The agentive reading contains a subject which is intentionally doing a certain action in order to bring about the resultant state. In the case of obstruction as in (107), the subject Irmi might park her truck in such a way that no other vehicle can pass. I take this kind of agent, which renders the sentence eventive, to be licensed by a do-operator. Do expresses the fact that a person is engaged in doing something. For example, Dowty 1979 defines it in the following way: a certain proposition “is under the unmediated control of the agent (individual denoted by α)” (Dowty 1979: 118). (106) ☐[do(α,ϕ) ↔ ϕ ∧ u.t.u.c.o.a. (ϕ)]] [Dowty 1979: 118 (152)] Several possibilities of how to best capture the causer vs. agent distinction have been made in the literature. For example, some recent analyses of the agentive/causative alternation conclude that little v comes in different types (e.g. Harley 1995, Folli and Harley 2005) depending on the kind of subject they license. According to this line of thought, there could be a stative little v for the trigger argument and an agentive little



Chapter 3.  Stative/eventive ambiguities 

v that licenses the animate agent. In recent minimalist terms, this would mean that there is a feature of the relevant type (or different sets of features) in the respective verbal projection (see also Kallulli 2006 for this line of thought). A slightly different solution has been offered in Arad 1998b. In this approach, there is no little v layer in case of a trigger argument, whereas an agent argument requires a full verbal projection including little v. In this way, the inventory of verbal heads reflects the difference between an agent and a trigger argument. As it may be, I do not prefer one particular solution. Regarding the question what kind of mechanisms allow for a stative/eventive ambiguity, the aforementioned approaches differ only with respect to the name of the label or the feature. For now, I leavethe question aside by which functional head or feature the subject is licensed, but see Chapter 7 for a detailed discussion. Nevertheless, I want to emphasize that an agentive subject requires a do-operator, whereas a bare causer argument does not. Hence, the nature of the distinction between agent and causer arguments is reflected in the presence or absence of the well-known do-operator (cf. Dowty 1979 and followers). For the sake of concreteness, I call the node hosting do little v. In sum, the do-operator requires two arguments: an agent and a proposition. The agentive reading of lexical-causative verbs incorporates the do-operator. (107) Agentive: a. Die Irmi verstopft die Straße mit ihrem Lastwagen. The Irmi obstructs the street with her truck. ‘Irmi is obstructing the street with her truck.’ b. λz λy λx λs do(x, cause(y, obstruct(z))) (s) (Irmi) (truck) (street) = = λs do(Irmi, cause(truck, obstruct(street))) (s) As far as the phrase-structure representation is concerned, I take the do-operator to occupy a functional projection which I label little v in order to distinguish it from the stative part of the verbal projection (big VP). (108) vP wo DP v 6 3 Irmi v VP do wo DP V 6 3 mit dem Lastwagen V V cause 3 DP V 6 verstopft die Straße



The Structure of Stative Verbs

Nevertheless, it is possible for the instrumental alternant (the one lacking an agent) to receive an eventive interpretation as well. This eventive reading arises if an adverbial expressing gradual change like nach und nach (bit by bit) is added. (109) Das Gewebe hat nach und nach das Blutgefäß verstopft. The tissue has bit by bit the blood vessel obstructed. ‘The tissue obstructed the blood vessel bit by bit.’ For example, (109) has an eventive causative reading. During the course of time, the obstruction of the blood vessel increases bit by bit. In contrast to other eventive verbs, the subject of this non-stative variant of instrumental verbs receivesa particular interpretation. Regarding (109), the subject is undergoing a change, i.e., the tissue is understood to be growing. As a result of its increasing size, the tissue obstructs the blood vessel more and more. In this way, the htissue is delimiting the event of increasing obstruction. The tissue, together with the blood vessel, “measure out” (in the sense of Tenny 1994) the event. In other words, there is a maximum amount of tissue such that the bloodvessel is obstructed completely. The reinterpretation of the subject can also be observed in (110): in the first sentence, the amount of leaves is increasing until there are so many leaves that the floor is completely covered. In the second sentence, the curtain is being moved such that the amount of the stage that is visible decreases. In both examples, the amount of covering or hiding is limited by the accusative object. Once every piece of the floor is covered, there can be no further covering. Once every piece of the stage is hidden, there can be no further hiding. (110) a. Die Blätter haben nach und nach den Boden bedeckt. The leaves have bit by bit the floor covered. ‘Leaves were covering the floor bit by bit.’ b. Der Vorhang hat nach und nach die Bühne verdeckt. The curtain has bit by bit the stage hidden. ‘The curtain was hiding the stage bit by bit.’ Note that the obligatory reinterpretation of the subject is special to verbs of causation that display the stative/eventive ambiguity. Other verbs that take an instrumental argument, such as wipe, do not allow for such a reinterpretation. While the subject in (111c) has the property of increasing gradually, the subject of the corresponding example in (111b) does not. In (111b), the broom is not undergoing any change during the wiping-event. (111) a. Der Poldi hat den Boden mit einem Staubsauger gereinigt. The Poldi has the floor with a hoover cleaned. ‘Poldi hoovered the floor with a vacuum cleaner.’



Chapter 3.  Stative/eventive ambiguities

b. Der Staubsauger hat den Boden nach und nach gereinigt. The hoover has the floor bit by bit cleaned. ‘The hoover was cleaning the floor bit by bit.’ c. Die Blätter haben nach und nach den Boden bedeckt. The leaves have bit by bit the floor covered. ‘The leaves were covering the floor bit by bit.’ Whereas in the construction in (111a) there is an activity of hoovering followed by a state of cleanliness, in the eventive example in (111c) the leaf-falling takes place while the floor gets covered. The obligatory reinterpretation of the subject in the eventive cases can now be accounted for. The adverbial nach und nach (bit by bit) indicates the presence of a become-operator, stating that the resultant state of being obstructed is achieved gradually. Hence, the change itself is modified by this adverbial. (112) incorporates the become-operator. (112) Eventive (no agent): a. Das Gewebe hat nach und nach das Blutgefäß verstopft. The tissue has bit by bit the blood vessel obstructed. ‘The tissue was obstructing the blood vessel bit by bit.’ b. λs become (cause(tissue, obstruct(blood vessel)))(s) (113)

VP wo V VP become wo DP V 6 3 das Gewebe V V cause 3 DP V 6 verstopft das Blutgefäß

3.2.5 Conclusion In conclusion, the discussion of the stative/eventive ambiguitiy regarding verbs that undergo the instrumental alternation has revealed that there are two possible mechanisms for rendering a sentence eventive: the insertion of a do-operator for agentive subjects,





The Structure of Stative Verbs

and the insertion of a become-operator which expresses the gradual onset of the resultant state. Thus, we have found evidence that these two operators occur independently: – The stative case lacks both the do and the become-operator. – The agentive case contains a do-operator which expresses that the agent is acting actively and intentionally. – The case of gradual onset contains a become but no do-operator. Most important, the subject is included in the resultant state and therefore undergoes a change as well.

3.3 Object-experiencer verbs assigning accusative 3.3.1 Introduction This section discusses the stative/eventive ambiguity of object-experiencer verbs that assign accusative case to their object. In addition to the previous section, these verbs provide evidence for the existence of a stative causer relation as well, since the stative reading involves a cause (the stimulus argument) and an effect (experienced by the object). Some German examples are given in (114). (114) deprimieren (depress), ärgern (annoy), erinnern (remind), amüsieren (amuse), erschrecken (frighten), beeindrucken (impress), anwidern (disgust) (115) Das Buch hat die Irmi beeindruckt. The book has the Irmi impressed. ‘The book impressed Irmi.’ First, I will give a brief overview of the treatment of object-experiencers in the literature. Second, I will present cases of the stative/eventive ambiguity that have been observed in different languages. After applying the tests for the detection of a Kimian stative reading, I conclude with a possible analysis of object-experiencer verbs in terms of Semantic Form.

3.3.2 Object-experiencer verbs in previous work Before turning to the tests for the existence of a Kimian stative reading, this section provides a short overview of the analyses of object-experiencer verbs that have been proposed up to now. As there exists a vast amount of research on these verbs, I will focus on the discussion of stative/eventive ambiguities within this verb class. It was first noted by Postal 1971 and Dowty 1991, among others, that object-experiencer verbs have peculiar properties when they are stative. In the following, I will present cases of this kind of ambiguity in Italian, Spanish and Finnish.



Chapter 3.  Stative/eventive ambiguities 

3.3.2.1 A note on terminology Before turning to the stative/eventive ambiguity that arises with object-experiencer verbs, I would like to make a short note on the terminology regarding psych verbs. Over the years there has been a vast discussion of the different kinds of psychologicalpredicates (e.g. Abraham 1995, Arad 1998a, Belletti and Rizzi 1988, Bennis 2004, Bouchard 1995, Croft 1993, Dowty 1991, Grimshaw 1990, Landau 2005, Pesetsky 1995, Reinhart 2001a and Wechsler 1995). I will present a rough summary regarding the stative/eventive ambiguity in the next section. The accounts that have been put forward in the literature converge on the claim that there are three different types of experiencer verbs: one subject-experiencer verb (SubjExp) and two different object-experiencer verbs (ObjExp). In German, these two types assign different cases to their experiencer object. On the one hand, verbs like those in (117), which are known as the preoccupare or worry-class, assign accusative case to their object, whereas the verbs in (118), which are referred to as piacere or appeal to-class, assign dative case. It is the former group that is ambiguous between a stative and an eventive reading. The object-experiencer verbs in (118) and the subjectexperiencer verbs in (116) allow for a stative interpretation only (see Chapter 4 for further discussion). (116) subject-experiencer verbs temere/fear-class a. Der Poldi fürchtet Unwetter. (German) b. John fears storms. c. Gianni teme questo. (Italian) ‘John fears this.’ [Belletti and Rizzi 1988: 291 (1); my translation] (117) object-experiencer verbs acc preoccupare/worry-class a. Die Nachricht beunruhigt den Poldi. (German) b. The message worries John. c. Questo preoccupa Gianni. (Italian) ‘This worries John.’ [Belletti and Rizzi 1988: 291 (2); my translation] (118) object-experiencer verbs dat piacere/appeal to-class a. Das Lied gefällt dem Poldi. (German) b. The song appeals to John. c. A Gianni piace questo. (Italian) ‘This appeals to John.’ [Belletti and Rizzi 1988: 291 (3a); my translation] 3.3.2.2 Stative and eventive object-experiencer verbs in Italian Arad 1998a and Arad 1998b discusses the different readings of object experiencer verbs regarding the stative/eventive distinction. First of all, Arad distinguishes three different readings of object experiencer verbs: the agentive reading, the eventive reading



The Structure of Stative Verbs

and the stative reading. In the agentive reading, there is an agent who “deliberately does something in order to bring about a mental state in the experiencer” (Arad 1998b: 181). (119) Nina frightened Laura deliberately / to make her go away. [Arad 1998b: 181 (2)] The eventive reading differs from the agentive reading in that the event is not brought about on purpose. Similar to the agentive reading, there is a change of state in the experiencer. See (120) for an illustration. (120) a. Nina frightened Laura unintentionally / accidentially. b. The explosion / the noise / the thunderstorm frightened Laura. [Arad 1998b: 182 (3)] The stative reading, as defined by Arad, “has neither an agent nor any change of mental state in the object” (Arad 1998b: 182). However, there is still some perception of the stimulus in the experiencer. The stative reading holds as long as the perception trigger is exposed to the experiencer. As soon as the two are separated, the described state ceases to exist.1 (121) a. b. c. d.

John / John’s haircut annoys Nina. John / John’s behavior / nuclear war frightened Nina. This problem concerned Nina. Blood sausage disgusts Nina. [Arad 1998b: 182 (4)]

This insight is illustrated in the diagram in (122), where the mental state holds as long as the perception of stimulus lasts. Of course, the perception is only possible if the trigger is present. (122)

stative: perception of stimulus mental state [Arad 1998b: 185 (7a)]

...................stop ...................stop

Arad goes on to show that the stative and the eventive readings of psych verbs can be seen overtly in Italian. Evidence that this distinction is reflected in the grammar comes from data regarding reflexivization through cliticization, from the causative construction and from restrictions of the extraction from the object. These criteria have been already discussed in Belletti and Rizzi 1988. However, the authors did not point out 1. Arad 1998b:185f. points out that some object-experiencer verbs allow only for the stative reading (e.g. concern and worry), others such as surprise permit only the eventive interpretation. Moreover, verbs like frighten may receive the agentive, the eventive and the stative reading. Thus, it is the idiosyncratic information of the verb that restricts the availability of the different interpretations of object-experiencer verbs.



Chapter 3.  Stative/eventive ambiguities 

that there is a systematical relationship between the stative psych verbs and their eventive counterparts. The reflexive clitic si in Italian can only be bound by an external argument. Therefore, passives, unaccusatives and raising verbs do not allow for the reflexive clitic to show up. As the stative reading of object experiencer verbs does not include an agent, it does not allow for the subject clitic si to appear. (123) a. ?? Gianni si spaventa. (on the reflexive reading) Gianni self frightens b. Gli studenti si spaventano prima degli esami per indursi a studiare di più. ‘The students frighten themselves before exams in order to urge themselves to study harder.’ [Arad 1998b: 188 (13)] (124) a. ?? Gianni si diverte. Gianni si amuses b. Gianni si diverte facendosi le bocacci allo specchio. ‘G. amuses himself by making funny faces in the mirror.’ [Arad 1998b: 188 (14)] Next, the causative construction in Italian provides further evidence that there is a grammatical distinction between eventive and stative psych verbs. The causative construction in Italian is not grammatical with derived subjects. Hence, Arad shows that agentive psych verbs, which include an underived subject, may take part in the causative construction. (125) Gli ho fatto spaventare il candidato per farlo lavorare di più. ‘I made him frighten the candidatei to make himi work harder.’ [Arad 1998b: 189 (16)] On the other hand, stative psych verbs contain a derived subject. This prevents them from taking part in the causative construction. (126) * Questo lo ha fatto preoccupare/commuovere/attrare ancora di più a Mario. ‘This made Mario worry / move / attract him even more.’ [Arad 1998b: 189 (15)] Finally, extraction from the object is not allowed with stative object experiencer verbs (c.f. Belletti and Rizzi 1988), but as Arad shows, as soon as there is an agent, extraction is possible. (127) * La ragazza di cui Gianni preoccupa il padre the girl of which Gianni worries the father [Arad 1998b: 190 (17a)]



The Structure of Stative Verbs

(128) La ragazza di cui Gianni ha divertito / impressionato / spaventato i genitori perchè gliela facessero sposare. ‘The girl whose parents G. amused / impressed / frightened so that they will allow him to marry her.’ [Arad 1998b: 190 (18)] In sum, the Italian data show that in the stative readings, there is no external argument, and that the object is non-canonical (as it does not allow for extraction). In the eventive readings, on the other hand, there is an external argument and a canonical object present. 3.3.2.3 Stative and eventive object-experiencer verbs in Spanish Spanish object-experiencer verbs exhibit a stative/eventive ambiguity as well. Arad 1998b shows that these verbs differ with respect to the case they assign to their object, depending on the interpretation of the subject. In cases where the subject performs an intentional act, such as (129a), the object bears accusative case, contrasting examples such as (129b), where the subject does not engage in any action. The latter cases assign dative to their object. (129) a. el niño la molestó. the boy her(acc) bothered (accusative pronoun) [Arad 1998b: 197 (30a)] b. el niño / la musica le molestó. the boy / the music her-dat bothered (dative pronoun) [Arad 1998b: 197 (29a)] Arad concludes that “the object case marking [...] in Spanish ObjExp verbs depends not on the change of state in the experiencer, but on the agentivity of the subject” (Arad 1998b: 198). I take these data to show that the eventivity is due to the action of the agent. In Spanish this is reflected morphologically. On the other hand, eventivity can be due to the gradual onset of the resultant state. In Spanish, this option is subsumed together with its purely stative counterpart under a single morphological marking (both require dative case on the object). In this way, Spanish contrasts Finnish, where this difference is reflected in the morphology overtly. 3.3.2.4 Stative and eventive object-experiencer verbs in Finnish Pylkkänen 2000 discusses various experiencer verbs in Finnish. She reports that Finnish object-experiencer verbs are derived from their subject-experiencer correspondents via the overt causative morpheme -tta.



Chapter 3.  Stative/eventive ambiguities 

(130)

Subject-experiencer: a. inhoa (find disgusting) b. sääli (pity) [adapted from Pylkkänen 2000: 419]

(131)

Object-experiencer: a. inho-tta (disgust) b. sääli-ttä (cause to pitty) [adapted from Pylkkänen 2000: 419]

This type of object-experiencer verb always assigns partitive case to its object, indicating that verbs like those in (131) cannot be telic. Consider (132) for an illustration. (132) a. Mikko inhoa-a hyttysi-ä. Mikko.nom findDisgusting-3sg mosquitos-par ‘Mikko finds mosquitos disgusting’ b. Hyttyset inho- tta- vat Mikko-a mosquitos.nom findDisguisting-caus-3pl Mikko-par ‘Mosquitos disgust Mikko’ [Pylkkänen 2000: 418 (1)] Both the subject-experiencer verb and its derived object-experiencer counterpart are stative. With the help of the inchoative suffix -stu, the corresponding eventive variants can be generated. Example (133) expresses that a process of Mikko getting angry was going on which terminated in the state of Mikko being angry. (133)

Uutiset viha-stu-tti-vat Mikko-a news.nom anger-inchoative-cause.past-3pl Mikko-par ‘The news made Mikko become angry’ [Pylkkänen 2000: 418 (2b)]

This group of eventive object-experiencer verbs assigns the common pattern of either accusative or partitive case to its object, depending on whether the object is definite or not. The fact that accusative is possible shows that the verbs including -stu are telic. Whereas subject-experiencers in Finnish refer to unbounded states (Pylkkänen associates them with individual-level predicates), the causative statives are more limited: the state of experience is always interpreted to hold simultaneously to the state ofperception. For example, (134) can only be understood in such a way that some mosquitoes are around Mikko while the sentence is uttered. (134)

Hyttyset inho- tta- vat Mikko-a. mosquitoes.nom findDisgusting-caus-3pl Mikko-par ‘Moskitoes disgust Mikko (now)’ [Pylkkänen 2000: 431 (30b)]



The Structure of Stative Verbs

Similarly, the episodic context in (135) can only be continued with the stative objectexperiencer verb. The subject-experiencer is unacceptable, because it can not refer to a clear-cut episode of disgust (at least in Finnish). (135) a. Menin eilen kalatorille, mutta en ostanut mitään. Kalaa käsiteltiin paljain käsin ja... ‘Yesterday I went to the fish market, but I didn’t buy anything. They handled the fish with bare hands and...’ b. ... se inho -tti minu-a. that.nom findDisgusting-caus.past I-par ‘... that disgusted me’ c. ... ?? minä inho- si- n sitä. I-nom findDisgusting-past-1sg that.par ‘... I found that disgusting’ [Pylkkänen 2000: 432 (22)] In sum, the stative object-experiencer verbs in Finnish require the trigger to be present. Moreover, the stative/eventive ambiguity regarding this group of verbs is reflected in Finnish morphology overtly. The eventive version is derived from the stative one via an inchoative suffix. Thus, the Finnish data support the view that object-experiencer verbs may either express a state of experience (a stative causation) or the gradual onset of such an effect in the experiencer. 3.3.2.5 Previous accounts of object-experiencer verbs In their seminal paper, Belletti and Rizzi 1988 posit an unaccusative analysis for object-experiencer verbs. In particular, the structure assumed for those that assign accusative case to their object is given in (136). (136) S wo NP VP ec 3 V’ NP 3 Gianni V NP preoccupa questo [Belletti and Rizzi 1988: 293 (6)] According to this analysis, both the experiencer and the trigger argument are basegenerated within VP. In other words, this type of verb does not have an external argument. The authors implement this idea with the “assignment” of an empty category (ec) to the position of the external argument. In the course of the derivation, the trigger argument questo (this) moves out of its base position in order to receive nominative



Chapter 3.  Stative/eventive ambiguities 

case. Hence, object-experiencer verbs assigning accusative case have an underlying unaccusative structure. The main argument for the unaccusative analysis comes from binding data. Whereas a non-experiencer verb does not allow for the bindee to occur before the binder, psych verbs permit this configuration. Note the difference between the psychverb example and the ungrammatical non-psych counterpart. (137) a. Questi pettegolezzi su di sé preoccupano Gianni più di ogni altra cosa. ‘These gossips about himself worry Gianni more than anything else.’ b. * Q  uesti pettegolezzi su di sé descrivono Gianni meglio di ogni biografia ufficiale. ‘These gossips about himself describe Gianni better than any official biography.’ [Belletti and Rizzi 1988: 312 (57)] In order to fulfill the binding requirements, the bindee must at some point in the derivation be c-commanded by the binder. Thus, it is the initial configuration of the arguments (D-structure at the time of Belletti and Rizzi) where the binder c-commands the bindee. Therefore, object-experiencer verbs have an unaccusative structure, whereas “common” verbs don’t. Pesetsky 1995 discusses the arguments of Belletti and Rizzi in detail, and incorporates the unaccusative analysis into his own one. Nevertheless, the two accounts differ in many respects. In particular, Pesetsky argues that the two kinds of object-experiencer verbs (those that assign dative versus those that assign accusative case to their object) do not share a common set of theta roles. Still, both Pesetsky and Belletti & Rizzi essentially agree on the unaccusative analysis. Arad 1998b does not follow this tradition and rejects the unaccusative analysis. Besides refuting the arguments put forward by Belletti and Rizzi 1988, Arad points out that the trigger argument, which is assumed to undergo movement from an initial object position, is functioning as the cause argument in object-experiencer verbs. Following the assumption that syntactic structure reflects event structure, causer arguments must occupy a higher position than their results. Therefore, Arad’s account has an advantage from a theoretical point of view. Arad argues that the structures of the stative and the eventive version of objectexperiencer verbs differ with respect to the functional category that licenses the subject. In case of stative verbs, the subject is licensed within the big VP projection.2

2. Arad takes a verb such as frighten to incorporate the noun fright, thus corresponding roughly to a construction such as to put fright on somebody. This is reflected in the decomposition of fright-en in the phrase marker.

 The Structure of Stative Verbs

(138) incorporated, stative VP 3 stimulus V 3 V VP 3 NP V Exp. 3 V NP

[Arad 1998b: 212 (58b)] In case of eventive verbs, little v licenses the subject. This head is responsible for the eventive reading and expresses the fact that the subject is an agent who is acting intentionally. (139) vP 3 NP v Nina 3 v VP 3 NP V Lily 2 V NP

‘Nina frightened Lily’ [Arad 1998b: 213 (60)] In sum, Arad shows that the event structure is reflected in the syntactic behavior of object-experiencer verbs. She claims that the peculiarities associated with verbs of the preoccupare-group are due to their stative reading. Assuming a tight correspondence between event structure and the syntactic structure of the verb phrase, the stative reading differs from the eventive one in terms of the complexity of v/VP.

3.3.3 Tests for a Kimian stative reading In this section, I will show that object-experiencer verbs possess a Kimian stative reading. This is done by applying the usual set of tests which discriminate between Davidsonian and Kimian statives. 3.3.3.1 Manner adverbials Manner adverbials are allowed for the examples containing a non-animate nominal subject. Consider (140b). The joke may be of such a kind, that it makes fun of Poldi in



Chapter 3.  Stative/eventive ambiguities 

an inappropriate way. For example, the joke could refer to Poldi’s dead family member or it could contain swear-words. Similarly, it could annoy Poldi in an elegant way, by making fun of him without referring to anything that makes Poldi feel uncomfortable. (140) a.

Das Grinsen ärgerte den Poldi auf eine elegante / The grinning annoyed the Poldi in an elegant / unpassende / zynische / raffinierte Weise. inappropriate / cynical / sophisticated way. ‘The grin annoyed Poldi in an elegant / inappropriate / cynical / sophisticated way.’

b.

Der Witz ärgerte den Poldi auf eine elegante / dreiste / The joke annoyed the Poldi in an elegant / impudent / unpassende / brutale / spöttische / zynische / raffinierte Weise. inappropriate / brutal / quizzical / cynical / sophisticated way. ‘The joke annoyed Poldi in an elegant / impudent / inappropriate / brutal / quizzical / cynical / sophisticated way.’

These adverbials clearly refer to the event that is denoted by the noun. In (141), the subject is not an event nominal, hence the manner adverbials are odd, if not even ungrammatical. (141) ?? Der Stau ärgerte den Poldi auf eine unpassende / zynische Weise. The traffic jam annoyed the Poldi in an inappropriate / cynical way. ‘The traffic jam annoyed Poldi in an inappropriate / cynical way.’ Sentential subjects do not allow for manner modification with object-experiencer verbs, as demonstrated in (142). No adverbial in the main clause can modify an event that is refered to by the subordinate clause. (142) ?*Daß die Irmi im Lotto gewonnen hatte, ärgerte den Poldi That the Irmi in lottery won had, annoyed the Poldi auf eine elegante / dreiste / unpassende / brutale / in an elegant / impudently / inappropriate / brutal / spöttische / zynische / raffinierte Weise. quizzical / cynical / sophisticated way. ‘It annoyed Poldi in an elegant / impudently / inappropriate / brutal / quizzical / cynical / sophisticated way that Irmi had won in the lottery.’ As soon as there is an animate subject, object-experiencer verbs can be interpreted in an eventive way. In (143), Irmi clearly does some action on purpose to annoy Poldi. (143)

Die Irmi ärgerte den Poldi vorsichtig / zärtlich / langsam / The Irmi annoyed the Poldi carefully / tenderly / slowly / lauthals / direkt / stumm. loudly / directly / dumbly. ‘Irmi annoyed Poldi carefully / tenderly / slowly / loudly / directly / dumbly.’



The Structure of Stative Verbs

To sum up, if manner adverbials can be used at all, they modify the trigger event denoted by the subject. However, this is only possible with nominal but not with sentential subjects. Thus, manner adverbials show that object-experiencer verbs with sentential subjects have Kimian stative readings. 3.3.3.2 Locative modifiers Object-experiencer verbs do not allow for locative modification at all. All kinds of subjects, even non-animate causers, are ungrammatical with locative adverbials, as in (144). (144) a. *Das Grinsen ärgerte die Irmi unter einem Baum. The grinning annoyed the Irmi under a tree. ‘The grin annoyed Irmi under a tree.’ b. *Der Witz ärgerte die Irmi unter einem Baum. The joke annoyed the Irmi under a tree. ‘The joke annoyed Irmi under a tree.’ c. *Daß der Poldi im Lotto gewonnen hatte, ärgerte That the Poldi in the lottery won had, annoyed die Irmi unter einem Baum. the Irmi under a tree. ‘It annoyed Irmi under a tree that Poldi had won in the lottery.’ In order to allow for a locative adverbial, object-experiencer verbs must take an animate subject. In (145), Irmi does something (for example, she might tickle Poldi). This action can take place at a particular location. (145) Die Irmi ärgerte den Poldi unter einem Baum. The Irmi annoyed the Poldi under a tree. ‘Irmi annoyed Poldi under a tree.’ In sum, event-related locative modification is not acceptable with object-experiencer verbs. If such adverbials are possible at all, they must be interpreted as frame-setting adverbials. This indicates that object-experiencer verbs do have Kimian stative readings. 3.3.3.3 Degree readings The third test is how degree phrases such as ein bisschen (a little) are interpreted. They receive a time-span reading only with eventive, but not with Kimian stative verbs. (146) a. Der Witz deprimierte die Irmi ein bisschen. The joke depressed the Irmi a little. ‘The joke depressed Irmi a little.’ b. Daß der Poldi abgesagt hatte, deprimierte die Irmi ein bisschen. That the Poldi not accepted had, depressed the Irmi a little. ‘It depressed Irmi a little that Poldi had not accepted the invitation.’



Chapter 3.  Stative/eventive ambiguities 

Example (146a) cannot mean that that Irmi was annoyed by the joke for a short period of time. Hence, object-experiencer verbs do not allow for the running-time reading of modifiers such as ein bisschen. In conclusion, all three tests indicate that object-experiencer verbs have a Kimian stative reading.

3.3.4 Properties particular to this verb class Next, I take a look a some properties that are related to the target state of object-experiencer verbs. In the eventive variants, there is clearly a state of experience that is obtained, i.e., a kind of target or resultant state. Crucially, this state holds within the experiencer, contrasting verbs like helfen (help), to be discussed in the next chapter. While the target state is not expressed overtly with verbs of the helfen-type, this section questions whether the accusative marking of theexperiencer is an overt realization of the target state. 3.3.4.1 Temporal modification Temporal modification with object-experiencer verbs affects the target state of the emotion. In other words, temporal adverbials such as drei Wochen lang (for three weeks) determine the length of the experience. (147) a.

Der Witz, der auf dem Fest erzählt worden ist, ärgerte The joke that at the party told been is, annoyed die Irmi drei Wochen lang. the Irmi three weeks long. ‘The joke that was told at the party annoyed Irmi for three weeks.’

b. Die täglichen Witze ärgerten die Irmi die ersten drei Wochen lang. The daily jokes annoyed the Irmi the first three weeks long. ‘The daily jokes annoyed Irmi for the first three weeks.’ 3.3.4.2 Modifiers that express the start of the target state Object-experiencer verbs do not behave in a unified way when combined with timespan adverbials. Pesetsky 1995 already observed that these verbs vary with respect to how quick the experiencer feels the emotion. Emotions which appear instantaneous do not allow for the time-span adverbials. Such experiencer verbs include ärgern (annoy) as in (148). Emotions that may grow over time like deprimieren (depress), in contrast, are compatible with time-span adverbials as in (149). (148) a. *Der Witz ärgerte den Poldi innerhalb von fünf Minuten. The joke annoyed the Poldi within of five minutes. ‘The joke annoyed Poldi within five minutes.’



The Structure of Stative Verbs

b. *Daß die Irmi im Lotto gewonnen hatte, ärgerte That the Irmi in the lottery won has, annoyed den Poldi innerhalb von fünf Minuten. the Poldi within of five minutes. ‘It annoyed Poldi within five minutes that Irmi had won in the lottery.’ (149) a.

Die Umstände deprimierten die Irmi innerhalb von The conditions depressed the Irmi within of wenigen Monaten. few months. ‘The conditions depressed Irmi within a few months.’

b. ?Daß die Umstände so ungünstig waren, deprimierte die That the conditions so bad were, depressed the Irmi innerhalb weniger Monate. Irmi within few months. ‘It depressed Irmi within few months that the conditions were so bad.’ To sum up, time-span adverbials express how long it takes until the emotion starts to hold. This indicates that the structure of eventive object-experiencer verbs contains therefore a kind of “effect”, caused by the trigger or agent argument, that allows – at least with some verbs – for a gradual onset. 3.3.4.3 Modification of domains Compared to helfen-type verbs (help), object-experiencer verbs lack the evaluative component and hence the domain specification. Therefore, they do not allow for the modification of that domain, as illustrated in (150). So, (150a) cannot be interpreted in such a way that the joke made fun of Poldi’s health or profession. (150) a. *Der Witz ärgerte den Poldi gesundheitlich / finanziell The joke annoyed the Poldi sanitary / financially / beruflich. / professionally. ‘The joke annoyed Poldi sanitary / financially / professionally.’ b. *Daß der Poldi gestern ein Bier getrunken hat, That the Poldi yesterday a beer drunken has, ärgerte die Irmi gesundheitlich / finanziell / annoyed the Irmi sanitary / financially / beruflich / emotional. professionally / emotionally. ‘It annoyed Irmi sanitary / financially / professionally / emotionally that Poldi drank a beer yesterday.’



Chapter 3.  Stative/eventive ambiguities 

This indicates that there is no “hidden” state of effect with object-experiencer verbs. Rather, the effect holds within the experiencer itself. Note that experiencers of this verb class are marked accusative. Accusative case more often than not expresses that the noun is part of a resultant or target state.

3.3.5. The structure of object-experiencer verbs Following the discussion of stative causers in Section 3.2.3, I take a cause-operator to be present in all object-experiencer verbs. The structure of the stative reading, which is exemplified with a sentential subject in (151), contains only this operator. 3.3.5.1 The lexical-semantic structure The basic structure of an object-experiencer verb that assigns accusative case is shown in (151). This representation contains neither a do nor a become-operator. There is only a cause operator that relates two separate situations. (151) a. b.

Daß die Irmi im Lotto gewonnen hat, ärgert den Poldi. That the Irmi in the lottery won has annoyes the Poldi. ‘It annoyed Poldi that Irmi had won in the lottery.’ λy λx λs cause(x, annoyed(y))(s) (Irmi-wins-in-lottery) (Poldi) = λs cause(Irmi-wins-in-lottery, annoyed(Poldi))(s)

Those verbs that allow for a gradual onset of the psychical effect (e.g. as noted in Pesetsky1995) like deprimieren (depress), may contain a become-operator. This reading is illustrated in (149a) above. If this operator is present, the verb is no longer stative. (152) a. b.

Die Umstände deprimieren den Poldi. The circumstances depress the Poldi. ‘The circumstances depress Poldi.’ λs cause(circumstances, become (depressed(Poldi)))(s)

Finally, the presence of an agent requires a do-operator in the lexical-semantic structure of the verb. (153) a. b.

Die Irmi ärgert den Poldi. The Irmi annoys the Poldi. ‘Irmi is annoying Poldi.’ λy λx λs do(x, cause(x, become (annoyed(y)))) (s)(Irmi) (Poldi) = = λs do(Irmi, cause(Irmi, become (annoyed(Poldi)))) (s)

3.3.5.2 The syntactic structure Regarding the syntatic structure of object-experiencer verbs, I follow Arad 1998b and take the causer argument to occupy a high position within the structure. This allows for a direct correspondence between event structure and syntax. The experiencer, which carries accusative case in the eventive variant, is part of the resultant state, i.e.,



The Structure of Stative Verbs

it is part of the experiencing effect that is caused by the stimulus or the agent. The active version including the do-operator is depicted in (154). (154) The active reading, eventive a. Die Irmi ärgert den Poldi (mit einem Lied). The Irmi annoys the Poldi (with a song). ‘Irmi is annoying Poldi (with a song).’ b. vP wo DP v’ 5 3 die Irmi v VP do wo (PP) V 6 3 (mit einem Lied) V V cause 3 V V become 3 DP V 6 ärgert den Poldi Note that it is possible to add an instrumental phrase like mit einem Lied (with a song) to the active reading. This instrumental phrase functions as the core trigger of the stimulus. Regarding (154), the agent causes the presence of the instrument, in this case the presence of the song. The song, then, is the trigger of the experience in the object. In the stative case, the state of experience holds only as long as Irmi is singing the song, as pointed out by Arad 1998b. As soon as she stops singing, Poldi is no longer annoyed. (155) The stative reading (nominal subject) a. Das Lied ärgert den Poldi. The song annoys the Poldi. ‘The song annoys Poldi.’ b. VP wo DP V 6 3 das Lied V V cause 3 DP V 6 ärgert den Poldi



Chapter 3.  Stative/eventive ambiguities 

The same configuration arises with a sentential subject. (156) The stative reading (sentential subject) a. Daß die Irmi im Lotto gewonnen hat, ärgert den Poldi. That the Irmi in the lottery won has, annoys the Poldi. ‘It annoys Poldi that Irmi has won in the lottery.’ b. VP wo CP V 6 3 daß die Irmi im V V Lotto gewonnen cause 3 hat DP V 6 ärgert den Poldi Finally, the eventive reading incorporates the become-operator. (157) The eventive reading (non-agentive) a. Die Umstände deprimierten die Irmi (innerhalb von wenigen The conditions depressed the Irmi (within of few Monaten). months). ‘The conditions depressed Irmi (within a few months).’ b. VP wo DP V 6 3 Die Umstände V V cause 3 V V become 3 DP V 6 deprimierten die Irmi



The Structure of Stative Verbs

3.3.6 Conclusion As with verbs that undergo the instrumental alternation in Section 3.2, the stative/ eventive ambiguities that arise with object-experiencer verbs are due to the same mechanisms: the do and the become-operator. – The stative case lacks both the do and the become-operator. – The agentive case contains a do-operator which expresses that the agent is acting actively and intentionally. – The case of gradual onset contains a become but no do-operator. – The experiencer object is included in the state of experience. This is indicated by accusative marking in the eventive case.

3.4 The threaten-class 3.4.1 Introduction A further case of the stative/eventive ambiguity is exemplified by the verbs of the threaten class. Johnson 1985, Prinzhorn 1990 and Arad 1998b show that verbs like threaten or require are ambiguous between a stative and an eventive reading. As illustrated in (158), the eventive reading contains an active agent who is doing something. (158) Nina threatens to kiss Paul. [Arad 1998b: 155 (21b)] In the stative reading, there is no agent who is acting intentional. (159) a. The rock threatens to fall. b. Paul’s bag threatens to break open. [Arad 1998b: 115 (22ab)] This ambiguity can be further clarified by looking at the German example and its two distinct interpretations, given by Prinzhorn 1990. (160) weil Hans dem Kind die Sandburg zu zerstören drohte because Hans the child the sand-castle to destroy threatened [Prinzhorn 1990: 211 (43)] (161) a. Hans drohte, daß er dem Kind die Sandburg zerstören würde Hans threatened that he the child the sand-castle destroy would b. Es drohte, daß Hans dem Kind die Sandburg zerstören würde It threatened, that Hans the child the sand-castle destroy would [Prinzhorn 1990: 211 (44)]



Chapter 3.  Stative/eventive ambiguities 

The interpretation in (161a) corresponds to the active variant of drohen (threaten). Here, Hans does a particular act of threatening the child, e.g. he is shouting. The nonactive interpretation, given in (161b), expresses that incidentially Hans is threatening to destroy the sand-castle. For example, he might be stumbling and almost fall into it. Prinzhorn captures the difference between the two variants of drohen along the following lines. He assumes that there are two different lexical entries for drohen. The first one has a control structure, where both the matrix and the embedded verb assign a full theta-role to their subject. This corresponds to the eventive variant, where the subject engages in an intentional act of threatening, say via a certain speech act. The stative reading, in contrast, contains verb-complex formation, i.e., the matrix and the embedded verb together form a complex verb. In this case, Prinzhorn takes the matrix verb to assign no real theta role to its subject.3 Therefore, stative drohen may occur together with an expletive, as in (161b). Only if verb-complex formation has taken place, raising is possible, thus explaining the stative reading of structures such as (160). In (162), the corresponding trees are given, the varying theta-roles printed in bold face. (162) CP CP 3 3 weil IP weil IP 3 3 NP I’ NP I’ | 2 | 2 Hans VP I e VP I 3 3 CP V CP V 2 | 2 | C IP droh- C IP droh 3 3 NP I’ NP I’ | 2 | 2 PRO VP I Hans VP I 3 3 NP V’ NP V’ | 3 | 3 dem Kind NP V dem Kind NP V | | | | die zu die zu Sandburg zerstören Sandburg zerstören [Prinzhorn 1990: 212 (45); my emphasis]

3.

Following Zubizarreta 1982, the author terms this an adjunct theta role.



The Structure of Stative Verbs

Moreover, Prinzhorn 1990 shows that verb-complex formation prohibits the extraposition of the infinitival complement. In other words, only the active reading allows for the complement clause to be extraposed. (163) a. *weil der Orkan t drohte, [die Häuser zu zerstören]t because the thunderstorm t threatened the house to destroy [Prinzhorn 1990: 201 (5c); my annotation] b. weil der Otto t drohte, [die Häuser zu zerstören]t because the Otto t threatened the house to destroy [Martin Prinzhorn, p.c.] Maya Arad independently comes to a similar conclusion. Following Johnson 1985, Arad takes the stative reading to be due to reanalysis, i.e., “the matrix verb and its complement form a complex predicate” (Arad 1998b: 116). The possible complement of threaten on the reanalysis reading is restricted to unaccusatives or verbs with a causer. Unergatives and agentive transitives, on the other hand, may not take part in the reanalysis reading. Thus, an agentive complement such as dance in (164c) can only have the control reading. (164)

a. The ice-cream threatens to melt. (unaccusative) b. The sun threatens to melt the ice-cream. (causer) c. *Paul threatens to dance. [Arad 1998b: 117 (25ab), (26a)]

The difference between the control and the reanalysis reading is, according to Arad, reflected in the possible complements of threaten. They are listed in (165). (165) a. threaten [v, CP] (control) b. threaten [TP] (reanalysis) [Arad 1998b: 131 (48)] Arad 1998b concludes that in the reanalysis version of threaten there is no little v head present in the complement verb. According to her view, little v is responsible for licensing the agent or the external argument. Therefore, the stative or reanalysis version of threaten requires its complement to consist only of a layered VP structure. As shown in (166a), there is no little v-projection present for the embedded verb melt, since it does not allow for an agentive reading. The sun, in this case, is not understood as an intentional agent. Rather, it serves as a stative causer which is licensed as the specifier of the whole complex threaten to melt.



Chapter 3.  Stative/eventive ambiguities 

(166) a. vP 3 the sun v 3 v V 3 V TP threaten 3 T VP to 3 V NP melt the ice b. The sun threatened to melt the ice. [Arad 1998b: 133 (52)] In contrast, the control reading of threaten requires the complement to contain a little v-projection, since the embedded verb has an agent argument. This is illustrated in (167a), where Mary is the agent of the embedded verb paint. It is the lower little vprojection which prohibits reanalysis. (167) a. VP 3 V TP threaten 3 T vP to 3 Mary v 3 v VP 3 V NP paint the wall b. Mary threatens to paint the wall. [Arad 1998b: 133 (53); my emphasis] This way of capturing the different versions of threaten is rephrased in Arad 1999 within the Distributed Morphology framework, where the author assumes that all verbal projections consist of a category-neutral root and a vP layer. Apart from the commonly known little v there exists a different type of little v which hosts stative causers. In this way, the distinction between agentive and stative variants of the threaten-class is implemented in the different variants of little v.4 4. The different variants of little v are also discussed in Folli and Harley 2005. However, these authors do not talk about stative v.



The Structure of Stative Verbs

Arad 1998b reports that threaten is incompatible with an expletive in both English (see 168) and French. This fact indicates that threaten does not have the structure of a raising verb such as seem, which occurs together with an expletive freely. Hence, it is the reanalysis mechanism that is underlying the stative variant of threaten and not a raising construction. (168) *It threatens that the book will fall off the shelf. [Arad 1998b: 116] The same pattern emerges for German versprechen (promise), which cannot occur with an expletive either. (169) a. This young boy promises to become a good musician. [Ruwet 1991; cited after Arad 1998b: 115 (22c)] b. Der Bub verspricht ein guter Musiker zu werden. The boy promises a good musician to get. ‘The boy promises to become a good musician.’ c. *Es verspricht der Bub ein guter Musiker zu werden. It promises the boy a good musician to get. ‘There promises the boy to become a good musician.’ d. *Es verspricht, daß der Bub ein guter Musiker wird. It promises that the boy a good musician gets. ‘There / It promises that the boy will become a good musician.’ Nevertheless, as Prinzhorn 1990 has shown, the stative variant of drohen is perfectly acceptable with an expletive subject in German. It may occur as a raising construction (see (170a)) or with an expletive and a finite clause (see (170b)). (170) a. Die Kekse drohten schlecht zu werden. The cookies threatened rotten to get. ‘The cookies threatened to go bad.’ b. Es drohte, daß die Kekse schlecht würden. It threatened that the cookies rotten would get. ‘It was likely that the cookies would go bad.’ In sum, verb-complex formation seems to be underlying the stative variant of threaten. Moreover, it serves as a necessary condition for raising and the occurrence of an expletive. As we have seen, there are more cases of verb-complex formation (English, French, certain German cases) than there are of raising constructions. Raising and expletives are therefore only allowed if there is verb-complex formation as well. Nevertheless, I leave the question of the exact syntactic mechanism of allowing or disallowing the expletive for future research.



Chapter 3.  Stative/eventive ambiguities 

3.4.2 Tests for a Kimian stative reading In the next section, I will go through the tests for the existence of a Kimian stative reading of verbs of the threaten-class. 3.4.2.1 Manner adverbials Manner modification, at a first glance, is hard to construct with verbs of the threatenclass. The most prominent reading a manner adverbial can get is one where it modifies the embedded verb. For example, the answer to the question in (171a) could be “by hanging”. Likewise, a possible answer to the question in (171b) is “slowly”, meaning the breaking will be slowly. As far as (171c) is concerned, the question itself is odd, precluding any possible answer. (171) a. Auf welche weise droht die Irmi sich umzubringen? In what way threatens the Irmi herself to kill? ‘In what way is Irmi threatening to kill herself?’ b. Auf welche Weise droht die Tasche aufzubrechen? In what way threatens the bag to break open? ‘In what way is the bag threatening to break open?’ c. ??Auf welche Weise droht das Weinen die Nachtruhe zu stören? In what way threatens the crying the sleep to disturb? ‘In what way is the crying threatening to disturb the sleep?’ If the way of threatening is modified, the question in (171a) could be answered with hysterisch (hysterical), meaning that Irmi was screaming hysterically that she was going to commit suicide. Regarding the question in (171b), it is hard to construct a manner adverbial, since all possible adverbs that come to mind refer to the way of expressing the speech act of threatening. Needless to say, a non-animate subject like a bag does not allow for the speech act reading. Consider (172). (172) a. Der Ballon drohte mit einem lauten Knall zu zerplatzen. The balloon threatened with a loud bang to burst. ‘The balloon threatened to burst with a loud bang.’ b. ?Das Haus drohte gefährlich in Flammen aufzugehen. The house threatened dangerously in flames up to go. ‘The house threatened to dangerously burst into in flames.’ Example (172a) shows that the instrumental phrase can only belong to the embedded verb, i.e., it is the burst that happens with a loud bang, not the threatening. The manner adverbial in (172b) is even ungrammatical. Although it might be a dangerous situation when a house almost burns, neither the threatening nor the burning can be modified by gefährlich. I conclude that there is no manner modification possible for the stative variants of threaten-type verbs.



The Structure of Stative Verbs

3.4.2.2 Locative modifiers Locative modifiers, if possible at all, again hint towards the presence of a Davidsonian state/event reading. In predicates which have only Kimian state reading, eventualityrelated locative modifiers are not possible. (173) a. Die Irmi drohte im Badezimmer die Sandburg zu zerstören. The Irmi threatened in the bathroom the sand-castle to destroy. ‘In the bathroom, Irmi threatened to destroy the sand-castle.’ b. *Die Welle drohte im Meer die Sandburg zu zerstören. The wave threatened in the sea the sand-castle to destroy. ‘In the sea, the wave threatened to destroy the sand-castle.’ c. *Das Weinen drohte im Bett die Nachtruhe zu stören. The crying threatened in the bed the sleep to destroy. ‘In bed, the crying threatened to disrupt the sleep.’ The agentive version of drohen (threaten) is able to license an eventuality-related locative modifier. In (173a), the act of threatening happens in the bathroom, i.e., while Irmi is standing in the bathroom, she says that she is going to jump into the sandcastle later that day. The stative variants in (173b) and (173c) cannot be modified by event-related locative adverbials. For example, (173b) cannot mean that a certain wave is going to destroy the sand-castle soon, and that this state of threatening is located in the sea. This lack of event-related locative modifiers with verbs of the threaten-class suggests that stative variants of drohen have a Kimian stative reading. 3.4.2.3 Degree readings The interpretation of degree modifiers such as ein bisschen (a little) serves as another test for the existence of a Kimian stative reading. While these modifiers are usually ambiguous between a time-span reading and a degree reading, Kimian statives do not allow for the latter. (174) a. Die Irmi drohte ein bisschen die Sandburg zu zerstören. The Irmi threatened a little the sand-castle to destroy. ‘Irmi threatened to destroy the sand-castle a little.’ b. Die Welle drohte ein bisschen die Sandburg zu zerstören. The wave threatened a little the sand-castle to destroy. ‘The wave threatened to destroy the sand-castle a little.’ Both examples in (174) are two-way ambiguous. First, the degree modifier ein bisschen may refer to the degree of the threatening, i.e., there was little danger that the sand-castle got destroyed. Second, it may refer to the degree of the damage, i.e., there was a threatening (not specified to which degree) that the sand-castle got destroyed only a little bit (e.g. only its outer wall will be damaged).



Chapter 3.  Stative/eventive ambiguities 

The time-span reading of the degree modifier, indicating the presence of a Davidsonian event argument, can be can be associated with the active version, i.e., Irmi was only threatening for a short period of time. Example (174b) and its correspondent with an infinitive subject in (175) do not allow for the time-span reading. (175) Das Weinen drohte ein bisschen die Nachtruhe zu stören. The crying threatened a little the sleep to distrub. ‘The crying threatened to disrupt the sleep a little.’ Likewise, the stative version together with a sentential complement prohibits the presence of the degree modifier. (176) *Die Welle drohte ein bisschen, daß sie die Sandburg zerstören würde. The wave threatened a little that she the sand-castle destroy would. ‘The wave threatened a little that it would destroy the sand-castle.’ 3.4.2.4 Summary In sum, verbs of the threaten-class may have a Kimian stative reading. All three tests (manner modification, event-related locative adverbials, and the interpretation of degree modifiers) indicate that the Kimian stative reading exists.

3.4.3 The modal component of threaten In this section, I will discuss the underlying modal component of threaten-type verbs. First, I will show that the stative reading has a modal flavor which is preserved in the eventive reading. Second, I will take a brief look at the type of modality involved. 3.4.3.1 The DO/MOD alternation In the eventive variant of threaten, there is an activity and a proposition as its complement. This proposition can be expressed with the help of an infinitive including subject control, as in (177). In addition to that, the same meaning can be conveyed with a finite clause, as in (178). Avoiding the distractions of a control structure, it is particularly apparent in the latter case that the embedded clause refers to a proposition which does not have a fixed truth value, but which is likely or possible to occur. (177) Die Irmii drohte, proi die Kekse aufzuessen. The Irmi threatened pro the cookies up-to-eat. ‘Irmi threatened to finish the cookies.’ (178) Die Irmi drohte, daß sie die Kekse aufessen würde. The Irmi threatened that she the cookies up-eat would. ‘Irmi threatened that she would finish the cookies.’ Thus, the proposition that Irmi would eat the cookies is evaluated with respect to a modal operator which is part of the matrix verb drohen (threaten). In other words, it is



The Structure of Stative Verbs

neither true or false that Irmi ate the cookies. Rather, this proposition is likely to occur. In the stative case, this likelihood is even more transparent. Consider (179). (179) Es drohte, daß die Kekse schlecht würden. It threatened that the cookies rotten got. ‘It threatened that the cookies would go bad.’ In order to utter a sentence like (179) feliciously, the fact that the cookies get rotten must be quite likely to happen soon. Hence, it is the verb drohen that modalizes this proposition. 3.4.3.2 The modality of drohen A further question to which I cannot go into here in detail is how drohen (threaten) relates to other modal verbs. In particular, German modal verbs always select bare infinitives, but drohen subcategorizes for a larger constituent involving the infinitival marker zu.5 Thus, German modal verbs are sometimes called “modal auxiliaries”. Drohen, in contrast, is more like a full verb than an auxiliary. After having established in the previous section that drohen (threaten) contains a modal operator, the question arises what kind of modality it expresses. In order to do this, let’s look at the hierarchy of modal expressions by Kratzer 1981. Table 3.1 lists four types of modality. Both of the modal verbs can and must occur in raising constructions as well as in control contexts. The raising variant, often called the epistemic reading, is taken to be the “higher” modality, i.e., the modal operator scopes over the whole proposition. The control variant, known as the root reading, expresses a particular possibility or obligation of the subject. Hence, it has a lower scope than the epistemic modal. The distinction between root and epistemic modality is illustrated by Stechow 2004 with the help of an example like (180). (180) Poldi muß zu Hause sein. Poldi must at home be. [adopted from von Stechow 2004] The root modality reading says that according to some law (e.g. according to the rules of Poldi’s parents), it is required that Poldi is at home. The epistemic modality reading, in contrast, expresses that according to what we (the speaker/hearer) know, it has to be the case that Poldi is at home. Table 3.1  Types of modality epistemic necessity ⇒ epistemic possibility root necessity root possibility

5.

must, raising can, raising must, control can, control

Thanks to Martin Prinzhorn, p.c., for pointing this out.



Chapter 3.  Stative/eventive ambiguities 

Regarding drohen, the proposition in question may or may not take place (e.g. the cookies may or may not get rotten). Therefore, the kind of modality we have here is a possibility. Moreover, it is epistemic modality, because the whole proposition is modalized. Root modality, on the other hand, would relate the subject of the sentence to a particular obligation (in the case of root necessity). In the stative drohen case, it is not the subject that is modalized, but the whole proposition. The modality of drohen can be analyzed in the spirit of Kratzer 1981. According to her, modals take a conversational background and an ordering source. In the case of drohen, the conversational background consists of the situations that may occur or that are likely to occur. The ordering source provides an ordering between these situations. A formal definition of human possibility (“it is likely that”) is given by Stechow 2004: (181) Modal component of “drohen”: λHλOλpλs [¬∀u[u∈H(s)] ¬∃v[v∈H(s)] (v and therefore open to Semantic Incorporation. The meaning of (336) is given in (343). (343) ∃y[cost(book, y) ∧ euro(y) ∧ |y| = 10 ∧ earn(I, z) ∧ ιz.(z = y)] Earn(I, z) refers to the relative clause which I have earned, and ιz.(z = y) introduces the uniqueness constraint required by the demonstrative. Note that the DPs that can be semantically incorporated correspond to those DPs that qualify as having weak Case in the sense of Hoop 1996. Very roughly, the idea behind the distinction between weak and strong Case is that DPs with weak readings



Chapter 4.  Non-ambiguous statives 

receive weak Case and stay within the VP. DPs that have strong readings receive strong Case and move out of the VP. For example, the partitive/accusative distinction that arises with Finnish telic verbs is an instance of the weak/strong distinction. Measure phrases, containing only weak determiners, therefore, are compatible with the group of DPs that receive weak case. 4.3.4.2 The syntactic structure of measure verbs A possible structure that is compatible with the lexical semantics of measure verbs is given in (344b). (344) a. Das Buch kostet 10 Euro. The book costs 10 euro. ‘The book costs ten euros.’ b.

VP wo DP V’ 6 3 das Buch NP V 6 kostet 10 Euro

4.3.5 Conclusion In this section, the structure of measure verbs was discussed. It turned out that measure verbs, always expressing a property of their subject, never allow for a systematic stative/eventive ambiguity. Hence, there is no way of creating an eventive reading out of a measure verb. The two core mechanisms that are responsible for an eventive reading, the do and the become-operator, may not be inserted into the structure of measure verbs. In addition, these verbs do not express a causal relation, hence, there is no cause-operator present. Therefore, the structure of measure verbs is simple: it consists only of a single-layered verbal projection. Moreover, the complement of measure verbs displays an interesting property. The measure phrase only allows for weak determiners; measure phrases that contain a strong determiner are unacceptable. This fact was accounted for by assuming that measure verbs contain a degree argument that must be specified. In this way, measure phrases specify the property that is predicated of the subject, and they do not take part in a measure event.

 The Structure of Stative Verbs

4.4 PP-complement verbs Several verbs selecting for a PP-complement are stative only. Some examples are listed under (345). (345) schmecken nach (taste of), bestehen aus (consist of), grenzen an (border with), vorkommen in (appear in), beruhen auf / fußen auf / basieren auf (be based on), folgen aus (follow from), hervorgehen aus (come from), stinken nach (smell of) (346) Das Grundstück grenzt an den Fluß. The property borders to the river. ‘The property borders to the river.’

4.4.1 Remark A further example of this group is stinken, which is often thought of as a one-place verb. However, I would like to argue that the underlying form comes with a PP complement nach as in (347a). The frequently used one-place variant, exemplified in (347b), has a generic object that means something like bad smell. Moreover, there is also a dative variant of stinken, which has an idiomatic meaning; see (347c). It belongs to the group of dative experiencer verbs. (347) a. Der Fußballer stinkt nach Schweiß. The soccer player smells of sweat. ‘The soccer player smells of sweat.’ b. Der Fußballer stinkt. The soccer player smells. ‘The soccer player is smelly.’ c. Das stinkt mir. (Idiom) This nom stinks me dat ‘To me, this sucks.’

4.4.2 Evidence for a Kimian stative reading 4.4.2.1 Manner adverbials Event-oriented manner adverbials are not possible together with PP-complement verbs. Neither the question in (348) nor the modifiers in (349) are acceptable. (348) a. * Auf welche Weise grenzt das Grundstück an einen Fluß? In what way borders the property to a river? ‘In what way does the property border to a river?’



Chapter 4.  Non-ambiguous statives 

b. * Auf welche Weise stinken die Socken nach Käse? In what way smell the socks of cheese? ‘In what way do the socks smell of cheese?’ (349) * Das Grundstück grenzt auf geradlinige / legale / The property borders in straight / legal / baubedingte Weise an einen Fluß. due to the way of building way to a river. ‘The property borders in a straight way / legally / by the way how it was built to a river.’ Since manner adverbials are not compatible with these verbs, this test shows that verbs which select for a PP-complement have a Kimian stative reading. They do not allow for a Davidsonian stative nor for an eventive reading. 4.4.2.2 Locative modifiers Event-related locative modifiers are not licensed by the PP-complement verbs, pointing towards the lack of an eventive reading. (350) a. * Das Grundstück grenzt im Salzkammergut an einen Fluß. The property borders in the Salzkammergut to a river. ‘The property borders to a river in the Salzkammergut.’ b. Die Socken stinken neben den Schuhen nach Käse. The socks smell next to the shoes of cheese. ‘The socks smell of cheese next to the shoes.’ Both examples in (350) can be interpreted. However, in this case the locative aderbials are interpreted as frame-setters. For example, the property can be so large that it is located in different regions. Only in the region Salzkammergut, it borders with a river. Similarly, neben den Schuhen can be understood as a condition: whenever the socks are next to the shoes, they smell of cheese (because both the socks and the shoes were used for sports). In sum, these locative modifiers indicate that the (Davidsonian) eventive reading is excluded for verbs with a PP-complement. 4.4.2.3 Degree readings Next, consider the test of the interpretation of ein bisschen. As shown in (351), this modifier can only receive a degree reading. It cannot be interpreted as a time-span, again hinting towards the fact that verbs with PP-complements allowfor a Kimian stative reading only. (351) a. Das Grundstück hat ein bisschen an einen Fluß gegrenzt. The property has a little to a river bordered. ‘The property bordered to a river a little.’

 The Structure of Stative Verbs

b. Die Socken haben ein bisschen nach Käse gestunken. The socks have a little of cheese smelled. ‘The socks smelled of cheese a little.’ 4.4.2.4 Complement of perception verbs In order to serve as a complement of a perception verb, a predicate not only has to contain a Davidsonian event variable, but it also has to have some visible property. Verbs of cognitive activity like think or grief are difficult in this respect, but the stative PP-complement constructions denote properties that are possible to perceive. (352) a. ?? Die Irmi sieht das Grundstück an einen Fluß grenzen. The Irmi sees the property with a river border. ‘Irmi sees the property bordering to a river.’ b. ? Die Irmi sieht das Werk aus drei Bänden bestehen. The Irmi sees the book of three volumes consist. ‘Irmi sees the book consisting of three volumes.’ Note that the correspondents with sentential complements improve remarkably in grammatical status. (353) a. Die Irmi sieht, daß das Grundstück an einen Fluß grenzt. The Irmi sees that the property to a river borders. ‘Irmi sees that the property borders to a river.’ b. Die Irmi sieht, daß das Werk aus drei Bänden besteht. The Irmi sees that the book of three volumes consists. ‘Irmi sees that the book consists of three volumes.’ The perception verb test shows that there is no Davidsonian eventive reading with PPcomplement verbs listed in (15).

4.4.3 The structure of verbs with a PP-complement 4.4.3.1 The lexical-semantic structure Similar to measure verbs, the PP-complement is interpreted as a specification of the property that is expressed by the verb. The analysis proposed for measure verbs in Section 4.3.4 cannot be carried over to verbs that select for a PP-complement, since the DP can be definite as in (354), contrasting the complements of measure verbs. (354) Das Grundstück grenzt an den Fluß / an die Donau. The land borders to the river / to the Danube. ‘The property borders to a river / to the Danube.’



Chapter 4.  Non-ambiguous statives 

Note that definite DPs are only compatible with locative verbs in this context. Nonlocative PP-complement verbs disallow the definite determiner as expected, even when the noun refers to (an indefinite number of) atoms such as boys: (355) Der Chor besteht aus (*den) Buben. The choir consists of (*the) boys. ‘The choir consists of (*the) boys.’ The definite DP in (354) has a mereological structure that is relevant for locative predicates, allowing for the PP-complement to get the right type for Semantic Incorporation. The locative extension of a river (and even the Danube) consists of various subparts that are also locations of this river. As it is unspecified in (354) which part of the river is part of the complex predicate border-to, we get an indefinite amount of river parts. This indefinite amount boils down to a non-existential reading of the indefinite which may take place in Semantic Incorporation. Following the account of Mador-Haim and Winter 2007 for incorporated PP-locatives, the meaning of a PP-complement verb can be derived as follows. First, objects are mapped to their eigenspaces, i.e., to the space they occupy. Then the preposition is applied. (356) P(loc(x)) Thereafter the result space is mapped to the objects contained in that space. (357) loc-1 (P(loc(x))) [Mador-Haim and Winter 2007: 8 (15)] The meaning of a river makes direct reference to its part-whole structure (≤). While y in (358) refers the Danube as such, singled out by the iota-operator, some unspecified part x is available for further processing. Thisis the unspecified, non-existential part that allows Semantic Incorporation. (358) λxιy(x ≤ y) ∧ danube(y) The structure of the PP to the Danube is shown in (359b), x being an indefinite of the type < e, t >. (359) a. to the Danube b. λx.loc-1 (to(loc(ιy[(x ≤ y) ∧ danube(y)]))) With the lexical-semantic structure of a verb that selects a PP-complement as in (360) and the meaning of the PP in (359b), the meaning of the whole sentence results in the structre in (361). (360) border to: λP∃y[border (x, y) ∧ P(y)] (361) ∃y[border (land, y) ∧ [loc-1 (to(loc(ιz[(y ≤ z) ∧ danube(z)])))]]

 The Structure of Stative Verbs

4.4.3.2 The syntactic structure The lexical-semantic representation which was developed above, is compatible with the following structure of the verb phrase: (362) a. Das Grundstück grenzt an einen Fluß. The property borders to a river. ‘The property borders to a river.’ b.

VP wo DP V’ 6 3 das Grundstück PP V 6 grenzt an einen fluß

4.4.4 Conclusion To sum up, this section has discussed verbs that select a PP-complement. It was shown that all of these verbs receive a Kimian stative interpretation only. There is no grammatical means to create an eventive reading from verbs that require a PP-complement. Thus, like the verbs discussed above (measure verbs, dative-experiencer/possessor verbs and subject-experiencer verbs), the structure of PP-complement verbs does not contain any of the aspectual operators. In addition, the structure does not express a causal relation and does therefore not consist of a relation between two states or two events. Thus, the underlying structure of PP-complement verbs is simple, expressing a property that is predicated of the subject. Moreover, the PP-complement was analyzed similar to the measure phrase in the previous section: there is an implicit argument within the lexical-semantic structure of the verb that must be specified by the prepositional phrase. Whereas this implicit argument is a degree argument that is restricted to weak DPs with measure verbs, the PPargument further specifies the property that is predicated of the subject. The restriction that the PP may only contain weak DPs holds not only with measure phrases but also with non-locative PP-complements. In the case of locative PP-complements, a non-specific subpart of the DP is selected due to its mereological structure, resulting in an indefinite that may be semantically incorporated. In sum, the PP-complement as well as the measure phrases are licensed via a different mechanism, namely Semantic Incorporation, compared to run-of-the-mill arguments.



Chapter 4.  Non-ambiguous statives 

4.5 Conclusion In this chapter, I have discussed several different types of verbs that express a Kimian stative reading only. These verbs do not share a single underlying structure, but have different lexical-semantic representations. – Experiencer verbs: both subject and dative-object experiencer (the piacere-group) verbs allow for a stative interpretation only. – Subject-experiencer/possessor verbs contain no aspectual operators, their lexical-semantic structure is therefore simple. However, they have an external argument which is clearly not an agent and therefore not licensed by the dooperator. – Dative-experiencer/possessor verbs again do not contain any of the aspectual operators. Thus, their lexical-semantic structure is simple. – Verbs that have a “semantically incorporated”-like argument are predicates that hold of the subject. Their object-like argument is licensed via a variable that is existentially bound within the lexical-semantic structure of the verb. This variable may be specified via Semantic Incorporation (i.e., it is a predicate and not an entity that the verb subcategorizes for). This mechanism accounts for the fact that the objects are either restricted with respect to the possible determiners (measure phrases) or must be licensed via prepositions. – Measure verbs contain a degree argument that is specified via the measure phrase. – Verbs with a PP complement contain a further specification of the predicate that is licensed via the same mechanism as the measure phrase. It is not possible to insert a become-operator into the lexical-semantic structure of these verbs. Nevertheless, it is possible to coerce these verbs into a reading that expresses a gradual onset, as this is possible with any expression via an operation that takes place outside of the verb itself (e.g., it is possible to refer to the beginning of an action like run, to the start of a state like be sick etc.). Moreover, none of the verbs expresses a causal relation, i.e., the verbs that do not display a stative/eventive ambiguity never express a relation between two situations or two events. Therefore, the operator cause is not present in their lexicalsemantic structure.

chapter 5

Verbs of position Davidsonian statives, which seem to form a distinct class that lies between Kimian statives and eventive verbs, consist of several subgroups of verbs. This chapter focuses on verbs of position, which select for an obligatory locative argument. There are two major classes of verbs of position that display a distinct behavior in terms of event structure: stative verbs of position (Section 5.1) and verbs of body posture (Section 5.2). After discussing the status of the eventuality that they refer to I will suggest a structure for each of them in turn.

5.1 Stative verbs of position Stative verbs of position like the German examples in (363) pose an interesting question regarding their event structure status: do they refer to an eventive eventuality, to a Kimian state, or is it necessary to introduce a third type of eventuality, the Davidsonian stative, into the ontology? (363) hängen (hang), sitzen (sit), stehen (stand), lehnen (lean), liegen (lie) According to Levin and Rappaport Hovav 1995, verbs of simple position as in (364) are non-agentive and have an obligatory locative phrase. Levin 1993 and Levin and Rappaport Hovav 1995 argue that these verbs are in fact verbs of existence, much like exist and remain.1 1. Other verbs that probably belong to the class of verbs of existence include those listed in (1) and (2).

(1) stimmen (be correct), zählen (count), dafürstehen (stand for), genügen (suffice), währen (last)

(2) a. Die Antwort stimmt. The answer is right. ‘The answer is correct.’ b. Nur die Liebe zählt. Only the love counts. ‘Only love counts.’ c. Ein kleines Präsent genügt. A small present suffices. ‘A small present suffices.’

 The Structure of Stative Verbs

(364) The papers lay on the desk. [Levin and Rappaport Hovav 1995: 127] The authors assume that verbs of existence have an unaccusative structure with two internal arguments, as they do not form causative transitive counterparts. ( 365) a. A star appeared in the sky. b. * The darkness appeared a star in the sky. [Levin and Rappaport Hovav 1995: 122 (99a)] (366) a. A solution exists. b. * The mathematician existed a solution. [Levin and Rappaport Hovav 1995: 122 (99c)] The apparent counterparts that exist within verbs of spatial configuration (such as transitive and intransitive hang) are not derived from one another. Levin and Rappaport Hovav conclude that these apparent pairs only “happen to involve the same constant and therefore may have the same name” (Levin and Rappaport Hovav 1995: 132). Verbs of existence, then, do not have an unergative structure like the verbs of internal causation. Their structure is given in (367). (367) [x be at z / hang] [Levin and Rappaport Hovav 1995: 132 (119a)] The next section takes a look at the tests for a Kimian stative reading in order to determine the event structure of stative verbs of position.

5.1.1 Tests for a Kimian stative reading 5.1.1.1 Manner adverbials Although many of the stative verbs of position seem to allow manner adverbials, a closer look reveals that these are not adverbials that specify the way in which the event happened. The first group of adverbials resembling manner adverbials describe in which direction the object is positioned. (368) a. Der Pflock steckt gerade / schief in der Erde. The picket is inserted straight / bent in the ground. ‘The picket is straight / bent in the ground.’ b. Das Bild hängt gerade / schief an der Wand. The painting hangs straight / askew on the wall. ‘The painting hangs straight / askew on the wall.’ For example, it is the main axis of the picket that is straight in (368a), and it is the frame of the painting in (368b) that is not parallel to the floor. Note that all of these examples can be paraphrased by a copular construction, which never refers to a Davidsonian eventuality.



Chapter 5.  Verbs of position 

(369) a. Der Pflock ist gerade in der Erde. The picket is straight in the ground. ‘The picket is straight in the ground.’ b. Das Bild ist schief an der Wand. The painting is askew on the wall. ‘The painting is askew on the wall.’ The fact that these adverbials are compatible with copular constructions, supports the idea that they are predicated of the subject. The second group of adverbials that might be manner adverbials again refer to a property of the subject, not of the event. This time it is not the orientation but the current condition of the subject that is specified. (370) a. Das Buch liegt aufgeblättert / zugeschlagen auf dem Tisch. The book lies opened / closed on the table. ‘The book lies opened / closed on the table.’ b. Die Decke liegt zerknittert / gefaltet auf dem Bett. The blanket lies crumpled / folded on the bed. ‘The blanket lies crumpled / folded on the bed.’ c.

Der Mantel hängt schlampig / auf unordentliche Weise The coat hangs sloppily / in messily way über dem Sessel. over the chair. ‘The coat is sloppily / messily hanging over the chair.’

Third, some adverbials that look like manner adverbials specify the way the situation is experienced by the hearer. (371) Der Ziegel ragt alarmierend / gefährlich über die Kante. The brick looms alarming / dangerously over the edge. ‘The brick is alarmingly / dangerously looming over the edge.’ The situation described in (371) is dangerous for the people standing below that particular brick. The examples in (372) show that manner adverbials are ungrammatical when combined with stative verbs of position. (372) a. * Das Buch liegt auf unordentliche Weise auf dem Tisch. The book lies in sloppily way on the table. ‘The book is sloppily lying on the table.’

 The Structure of Stative Verbs

b. * Der Pflock steckt auf verbotene / erlaubte Weise in The picket is inserted in prohibited / allowed way in der Erde. the ground. ‘The picket is sticking in the ground in a prohibited / allowed way.’ In sum, we do not find manner adverbials that go with stative verbs of position. 5.1.1.2 Locative modifiers Stative verbs of position are more often than not combined with a locative PP. What is the status of this PP? If it is an adverbial, the verb does not select for it. If it is an argument, it should not be optional. Crucially to the current discussion: if it is an event-related locative adverbial, the eventuality the verb refers to must be a Davidsonian one. From a syntactic point of view, the main property of adverbials is that they are never obligatory. However, the distinction between locative arguments and locative adverbials in the case of stative verbs of position is not straightforward to see, since the locative PP may be omitted in certain contexts. Maienborn 1996 argues that it is possible to omit locative adjuncts in general, whereas locative arguments can only be dropped under certain conditions: it is possible to omit locative arguments if the mode of position is more prominent than the specification of the location. In other words, in order to omit a locative argument the mode of position must be focussed and the locative component defocussed. Maienborn illustrates this idea with the help of the following example: the sentence in (373) is only acceptable if Julius is a toddler who is about to learn how to stand, the manner of position therefore being focussed. (373) Julius steht. Julius stands. [Maienborn 1996: 62, my translation] In contrast, the activity Julius schläft (Julius is sleeping) does not require a comparable context. Maienborn argues that the process of defocussing boils down to the fact that the locative argument is existentially bound. Only under this condition it is allowed for the argument to remain silent. The locative PP is therefore obligatory and bears argument status. Moreover, Maienborn 2003 points out that there is a clear parallelism between stative verbs of position and locative copular constructions as in (374a) and (374b). (374) a. Heidi liegt auf dem Sofa. Heidi lies on the sofa. b. Heidi ist liegenderweise auf dem Sofa. Heidi is lying-way on the sofa. [Maienborn 2003: 86 (60), my translation]



Chapter 5.  Verbs of position 

Copular constructions, referring to a Kimian state, may contain a locative PP as in (374b). Since copular sentences never refer to a Davidsonian event, the PP cannot locate that event. So what the PP does is locate the subject of the copular clause. This parallelism sugests that the locative PPs that come with stative verbs of position, therefore, do not count as evidence that these verbs are eventive. It is important, therefore, to carefully separate two modes of localization: event localization, which may be referred to by event-related locative adverbials, and object localization which is carried out by locative arguments. Czinglar 1997 substantiates the claim that locative PPs denote object localization. She points out that the three German verbs sein (to be) and geben (to give) differ with respect to a single property: The verb geben (give) as in (375) expresses the existence of its subjects, the locative PP being optional, whereas the copula sein as in (376) denotes the localization of the subject. (375)

Es gibt Elefanten (in Afrika). It gives sg elefants (in Africa). ‘There are elefants (in Africa).’ [Czinglar 1997: 48 (23a); my gloss, my translation]

(376)

Es sind Elefanten *(in Afrika). It are pl elefants (in Africa). ‘There are elefants (in Africa).’ [Czinglar 1997: 48 (23c), my gloss, my translation]

Although the constructions look similar on the first glance, the locative PP is obligatory in (376), contrasting (375). To sum up, in contrast to verbs of body posture, stative verbs of position merely express the localization of an object, not of a situation. The locative PP has the status of an argument and not of an adjunct. I conclude that the locative PP in stative locative verbs does not express event-related locative modification (since there is no event, as corroborated by the tests on manner modification), but expresses object localization. In other words, the PP refersto the location of the subject, not to the location of the event expressed by the verb of position.

5.1.2 The structure of stative verbs of position Contrasting verbs of body posture, stative verbs of position do not express an action that is deliberately carried out by the agent. The discussion in Section 5.1.1 shows that the eventuality stative verbs of position refer to is a Kimian state. Stative verbs of position, therefore, express that an object is located at a particular position in space. Building on ideas put forward in Bierwisch 1988, Wunderlich and Kaufmann 1990 and

 The Structure of Stative Verbs

Wunderlich and Herweg 1991, among others, Maienborn suggests the following structure for verbs of position in general: (377) (λP[–Dir]) λo λs [s inst [pos-mod(o) & P(o) ] ] [Maienborn 1996: 56 (16)] The lexical-semantic structure in (377) captures the fact that the locative PP is an argument. Its licensing position is within the lexical entry of the verb, it predicates of the relevant object, not of the situation. The predicate pos-mod specifies the mode of position, e.g. if the object is aligned relative to its horizontal or its vertical axis. Note that pos-mod does not refer to any activity or change-of-state, nor does it specify the manner of a possible locative event. Due to he absence of all aspectual operators the eventuality is a Kimian state. The brackets around λP capture the fact that the locative argument is optional if the context requires that the manner of position is focused. Again pushing on the similarity between locative copular constructions and stative verbs of position, I take those constructions to have a similar syntactic structure. The structure for the locative copular construction is given in (378). (378) a. Das Buch ist am Tisch. The book is on the table. ‘The book is on the table.’ b.

predP wo DP pred’ 6 3 Das Buch pred0 PP ist 6 am Tisch

Apart from the label of the predicative/verbal head, the structure for stative locative verbs is the same: (379) a. Das Buch liegt am Tisch. The book lies on the table. ‘The book is lying on the table.’ b.

VP wo DP V’ 6 3 Das Buch V0 PP liegt 6 am Tisch



Chapter 5.  Verbs of position 

The locative PP is the sister of the lowest verbal head v0, contrasting verbs of body posture. The predicative head pred0 introduces the situation argument that is later bound at infl. I do not attempt to tackle the question where morphological insertion of the copula sein takes place – either at pred0 or at i0 – both answers are compatible with the theory spread out here.

5.2 Verbs of body posture Next, consider verbs of body posture as in (380). (380) hocken (crouch), kauern (squat), knien (kneel), lümmeln (loll), thronen (to be enthroned) According to Levin and Rappaport Hovav 1995, the verbs in (380) can be classified asagentive verbs which express that the subject is maintaining a position. The subject of these verbs is intentionally remaining at a given position, although she is not executing any particular action. The authors report that these verbs have an optional locative phrase. (381) Yvonne stood alone (in the hallway) for six hours. [Levin and Rappaport Hovav 1995: 127] Kaufmann 1995a, who suggests that verbs of body posture specify the orientation of the object being localized, i.e., it is the location of the object and not the situation that is relevant. Kaufmann does not examine whether there is a difference between verbs of body posture and stative verbs of position. In particular, the author does not assume that there is an agentive or an eventive component within the structure of verbs of body posture. To illustrate, consider Kaufmann’s analysis of hocken (crouch): (382) a. hocken: λP λx [hock(x) & P(x)] b. Int (hock(x)) = ∃y (supportƒ (d-os(y), us(x)) & & angezogen(beine(x))) [Kaufmann 1995a: 103:(11ab)] The interpretation (Int) of the predicate hock is given in (382b): it involves the support relation supportƒ specifying that the upper surface of the object y (d-os(y)) supports the bottom side of the located object x (us(x)). In addition to that, verbs of body posture contain a specification of the posture of relevant body parts. The predicate angezogen(beine(x)) requires x’s legs (beine) to be bent (angezogen).

5.2.1 Tests for an eventive reading In order to determine the exact status of the eventuality, I will consider manner modification and the status of the locative PP in turn.

 The Structure of Stative Verbs

5.2.1.1 Manner adverbials To start with, the examples in (383) show that event-related manner adverbials combine with verbs of body posture. (383) a.

Der Frosch hockt verzweifelt / mit angezogenen The frog crouches desparately / with bent Beinen im Tümpel. legs in the pond. ‘The frog is desperately crouching in the pond with bent legs.’

b.

Das Küken hockt bewegungslos / auf reglose Weise The chick crouches motionless / in motionless way im Gras. in.the grass. ‘The chick is motionlessly / in a motionless way crouching in the grass.’

c.

Die Prinzessin thront majestätisch / auf anmutige The princess is enthroned majestetically / in graceful Weise auf der Erbse. way on the pea. ‘The princess is majestetically / in a graceful way sitting enthroned on the pea.’

While the examples in (383) contain verbs that have a posture reading only, other verbs like stehen (stand) and liegen (lie) are ambiguous between a posture reading and a stative-verb-of-position reading. On their posture reading, they are compatible with a manner adverbial as well: (384)

Yvonne steht sechs Stunden lang auf anmutige / unruhige / Yvonne stands six hours long in graceful / moving / bewegungslose Weise (am Gang). motionless way (in.the hallway). ‘Yvonne is gracefully / moving / motionlessly standing in the hallway for six hours.’ [modelled after Levin and Rappaport Hovav 1995: 19]

Maienborn 2003 and Maienborn 2005b, when discussing Davidsonian statives, points out that verbs like sitzen (sit) in (385) allows for manner modification, Maienborn concludes that all verbs of position refer to a Davidsonian eventuality. (385)

Carol saß reglos / kerzengerade am Tisch. Carol sat motionless / straight.as.a.die at.the table. D-state verbs [Maienborn 2005b: (31b)]

However, I want to stress that verbs of position are only eventive on their posture reading. Hence, sitzen in (385) is used as a posture verb, not as a stative verb of position. In sum, the examples above show that verbs of body posture allow for a manner modification.



Chapter 5.  Verbs of position 

5.2.1.2 Locative modifiers Regarding the status of the locative PP that comes with verbs of body posture, the question arises whether these PPs are event-related locative modifiers. If these verbs refer to a Davidsonian eventuality, event-related locative modifiers are expected to be acceptable. If, on the other hand, the locative PPs turn out to have argument status, they cannot serve as a test for the status of the eventuality. To start with, lets determine the status of the PP with respect to the argument/adjunct distinction. If no further further context is given, the examples in (386) are odd. (386) a. ?? Der Poldi hockt. The Poldi crouches. ‘Poldi is crouching.’ b. ?? Die Irmi kauert. The Irmi squats. ‘Irmi is squatting.’ c. ?? Die Irmi thront. The Irmi is enthroned. ‘Irmi is enthroned.’ Note that I could not find an example containing a German posture verb without a locative PP on a google search. In order to be fully acceptable in an out-of-the-blue context, there has to be a locative PP: (387) a. Der Poldi hockt am Boden. The Poldi crouches on the floor. ‘Poldi is crouching on the floor.’ b. Die Irmi kauert in einer Ecke. The Irmi squats in a corner. ‘Irmi is squatting in a corner.’ c. Die Irmi thront auf dem Sessel. The Irmi is enthroned on the chair. ‘Irmi is sitting enthroned on the chair.’ In order for the sentences in (386) to be fully acceptable, the discourse context has to provide a reason for contrasting the current posture with an alternative one (see also the discussion in Section 5.1.1.2). (388) a. A: Der Poldi sitzt am Boden, mitten im Dreck! A: The Poldi sits on the floor, in the middle of the dirt! ‘A: Poldi is sitting on the floor, in the dirt!’ b. B: Nein, er hockt, seine Hose bleibt sauber. B: No, he crouches, his pants stay clean. ‘B: No, he is crouching, so his pants will stay clean.’

 The Structure of Stative Verbs

Thus, the locative PP that comes with verbs of body posture does not have adjunct status. Rather, it has the status of a facultative argument that can be omitted only if a reasonable contrast can be inferred from the context. In particular, the locative PP is not an event-related locative adverbial. Therefore, it cannot help to determine the exact ontological status of the eventuality expressed by posture verbs.

5.2.2 The structure of posture verbs When an animal or a human deliberately remains at a location and holds its body in a particular position, then this is similar to more prototypical activities such as jumping or clapping that involve the movement of the body. In other words, from the eventstructure perspective, prototypical activity verbs and verbs of body posture share their lexical-semantic structure: the aspectual operator involved in both verb classes is do, allowing for intentionality to be attributed to thesubject. The lexical-semantic structure of posture verbs is therefore as follows. (389) a. b.

Der Poldi kniet in der Ecke. The Poldi kneels in the corner. ‘Poldi is kneeling in the corner.’ λx λs [do(x, kneel(x)) & loc(x, corner)] (s) (Poldi) = = λs [do(Poldi, kneel(Poldi)) & loc(Poldi, corner)] (s)

Posture verbs consist of the structure of stative verbs of position plus the do-operator. The stative part, then, expresses that the subject is located at a certain position. The do-operator adds to this that the subject is under the active control of remaining at that particular position. Thus, the general lexical-semantic structure of posture verbs is as shown in (390). (390) (λP)λx λs [do(x, pos-mod(x)) & P(o)] (s) (x) (P) Note that mod-pos refers to the mode of position, i.e., the orientation of the object that is located. With respect to the verb knien (kneel) as in (389a), mod-pos would refer to something like “on one’s knees”. As with stative verbs of position, mod-pos is not an aspectual operator, hence it does not render a verb eventive. The syntactic structure of posture verbs consists of the one proposed for stative verbs of position combined with a do-head. The locative PP originates as a sister of v0 in an argument position. (391) a. Der Poldi hockt am Boden. The Poldi crouches on the floor. ‘Poldi is crouching on the floor.’



Chapter 5.  Verbs of position 

b.

doP wo DP do’ 6 3 Der Poldi do0 VP | V’ 3 V0 PP liegt 6 am Boden

5.3 PP-argument versus PP-incorporation Stative verbs of position and verbs of body posture resemble the group of verbs I have termed PP-complement verbs in Section 4.4, since all of them select for a PP-argument. However, the PP-argument is licensed via two different mechanisms. Whereas the PP in PP-complement verbs incorporates and forms a complex predicate with the basic verbal predicate, the PP in stative verbs of position and in verbs of body posture fills the argument slot directly. For convenience, I repeat the lexical-semantic structures below. (392) a. Stative verbs of position b. (λP[–Dir]) λx λs [s inst [pos-mod(x) & P(x) ] ] (393) a. Verbs of body posture b. (λP) λx λs [do(x, pos-mod(x)) & P(o)] (s) (x) (P) (394) a. PP-complement verbs b. λ P ∃ y [border(x, y) ∧ P(y)] Remember from Section 4.3 and 4.4 that definites are not compatible with PP-complement verbs. Verbs of body posture and stative verbs of position, in contrast, are perfectly acceptable if the PP contains a definite noun phrase. (395) Das Buch liegt auf der roten Mappe. The book lies on the red folder. ‘The book is lying on top of the red folder.’ The complement of stative verbs of position and of posture verbs is therefore an entity of type < e >, whereas the complement of PP-complement verbs is an indefinite of type < e, t > that is licensed via Semantic Incorporation.

 The Structure of Stative Verbs

5.4 Evidence from Tibeto-Burman Evidence for the distinction between verbs of body posture and stative verbs of position comes from Chantyal, a Tibeto-Burman language. In Chantyal, the distinction between these two verb classes can be seen overtly. Following the work of Noonan and Grunow-Hårsta 2002, this section presents a rough summary of the relevant data. Posture verbs in Chantyal can be expressed by simplex verbs or by complex expressions. The latter consist of an orientational word and one of the three light verbs la(do), ta- (become) or pәri- (happen). The first two light verbs correspond nicely to two of the aspectual operators: do and become. The third one, pәri- (happen) refers to a resultant state: “it contributes the sense that the state of affairs thus described is the result of an event [...] mentioned [...] earlier” (Noonan and Grunow-Hårsta 2002: 84). There are three main posture verbs in Chantyal which correspond roughly to the English verbs stand, sit and lie. To start with, the translation of lie can only be a complex expression, a simplex verb does not exist. All three light verbs combine with the orientational word: (396) a. terso la- ‘maintain a horizontal position’ (literally: ‘horizontal/flat do’) b. terso ta- ‘assume a horizontal position’ (literally: ‘horizontal/flat become’) c. terso pәri- ‘come to be in a horizontal position’ (literally: ‘horizontal/ flat happen’) [Noonan and Grunow-Hårsta 2002: 81 (4)] Only Example (396a) is of interest with respect to the analysis of stative locative verbs. Although this verb takes animate as well as inanimate subjects, the authors stress that an inanimate “subject is presented as a volitional actor” (p. 84). This fact provides evidence for the idea that verbs of body posture contain the do-operator that expresses the fact that the subjects deliberately hold their body in a particular posture. The second verb, the English stand, can be expressed by a simplex verb or by a complex expression in Chantyal. The simplex verbs yep- (maintain a standing position) and yes- (assume a standing position), as well as the complex expression formed with la- (do) are incompatible with inanimate subjects. This substantiates the claim that the light verb la- (do) refers to an active, intentional eventuality. Third, the English verb sit can only be translated into the simplex verb ci-, which allows for animate subjects only. In sum, complex expressions involving la- (do) in Chantyal require the subject to be actively involved in the eventuality. In this way, these data substantiate the claim that verbs of body posture always contain the aspectual operator do. Note that the expression of manner is only possible with complex expressions. The authors claim that there is a “lack of simplex verbs encoding posture + manner [...;] manner is most frequently encoded by expressive vocabulary, often with la- ‘do’ functioning as an anterior coverb” (p. 86f.).



Chapter 5.  Verbs of position 

In Chantyal, the location of inanimate objects is typically expressed with the help of copular constructions: (397)

pBara-ye tawko-ri tBim mu-õ mountain-gen edge-loc house be-impf ‘The house was/stood on the edge of a cliff.’ [Noonan and Grunow-Hårsta 2002: 86 (20)]

Stative verbs of position, therefore, do not exist in Chantyal. What seem to be stative locative verbs form two classes: – animate, volitional posture verbs involving la- (do) and an orientational word – locative expressions involving a copular verb In this way, the data from Chantyal provide empirical evidence for the distinction between posture verbs and stative verbs of position.

5.5 Conclusion This chapter focused on verbs of position. Maienborn 2003 argued that verbs of position belong to the group of Davidsonian statives that differ from Kimian statives with respect to the nature of their eventuality argument. Davidsonian statives, although they do not describe an action of any sort, still fail the tests for Kimian stativity. For this reason, Maienborn assumed that Davidsonian statives express a third type of eventuality. A closer look on the nature of verbs of position revealed that this class of verbs must be split into two: stative verbs of position and posture verbs. Section 5.1 showed that the former refer to a Kimian state, contrary to the assumption in Maienborn 2003. Section 5.2 revealed that the latter are eventive, although nothing seems to happen in these verbs. Nevertheless, their lexical-semantic structure contains the do-operator, rendering posture verbs eventive. Consequently, there is no reason to assume a third type of eventuality, as far as verbs of position are concerned. The next chapter questions if this type of eventuality is necessary for verbs of internal causation.

chapter 6

Verbs of internal causation This chapter investigates the nature of verbs of internal causation. Despite verbs of position, the behavior of these verbs pose the second reason for assuming a stative Davidsonian eventuality. In what is to follow, I will show that the lexical-semantic structure of verbs of internal causation consists of the same building-blocks as the other eventive verbs addressed so far: a basic predicate and some of the aspetual operators. Therefore, verbs of internal causation refer to an eventive Davidsonian eventuality, rendering the stative Davidsonian eventuality redundant.

6.1 Non-agentive verbs of internal causation Non-agentive verbs of internal causation such as verbs of emission denote an “action” that is invoked by the subject without any intentional component. For example, a piece of gold may shimmer in the sun because its surface reflects the light. It is easy to see that the subject of such verbs does not contribute intentionally to what is happening, as the piece of gold does not have the necessary prerequisites for intentionality – in fact, it has no cognitive capacity at all. On the other hand, it is not straightforward to acknowledge that verbs of internal causation refer to an “action” or to “something that is happening”. In the case of the shimmering gold, there is nothing really going on, except from the colors that the gold exhibits. Extending the work of Perlmutter 1978, Levin and Rappaport Hovav 1995 distinguish four types of verbs of emission, according to what kind of substance is emitted. (398) a. Sound: burble, buzz, clang, crackle, hoot, hum, jingle, moan, ring, roar, whir, whistle,... b. Light: flash, flicker, gleam, glitter, shimmer, shine, sparkle, twinkle, ... c. Smell: reek, smell, stink d. Substance: bubble, gush, ooze, puff, spew, spout, squirt, ... [Levin and Rappaport Hovav 1995: 91 (19)] Following Levin and Rappaport Hovav 1995 (p. 90f.), verbs of (involuntary) emotional reactions such as blush and tremble, verbs of emission and activity verbs constitute the group of verbs of internal causation. Since activity verbs are clearly eventive, only non-agentive verbs of internal causation are relevant to the discussion of Davidsonian statives. Davidsonian statives (cf. Maienborn 2007b) contain an event argument but

 The Structure of Stative Verbs

differ form run-of-the-mill eventive verbs like activities in that they cannot be followed by it happened. (399) Die Schuhe glänzten. *Das geschah während ... The shoes gleamed. This happened while... [Maienborn 2005b:(12c)] In contrast to Kimian statives, it is possible to combine verbs of internal causation with event-related locative modifiers and with manner adverbials. (400) Die Perlen glänzen in ihrem Haar. The pearls gleam in her hair. [Maienborn 2007b:(5c)] (401) Die Perlen glänzen matt / rötlich / feucht. The pearls gleam dully / reddishly / moistly. [Maienborn 2007b:(6c)] In addition, modification with ein bisschen allows for the time-span interpretation with verbs of emission: (402)

Diese Hose glänzt nach dem Bügeln ein bisschen (aber das These trousers gleams after the ironing a little (but this geht schnell vorbei). goes quick away). [Maienborn 2003: 101 (103f), my translation]

Finally, verbs of emission may function as the complement of perception verbs, thus qualifying them again as Davidsonian statives. (403) Die Konquistadoren sahen überall Gold glänzen. The conquistadors saw everywhere gold gleam. [Maienborn 2003: 66 (7c), my translation] Now that it is clear that non-agentive verbs of internal causation belong to the group of Davidsonian statives, the question arises what their lexical-semantic structure is. To start with, Levin and Rappaport Hovav 1995 claim that verbs of emission and verbs of internal causation have an unergative structure. These verbs are characterized by the fact that some internal property of the argument causes the eventuality, contrasting verbs of external causation. This distinction is due to the fact that verbs of external causation show causative transitives, whereas verbs of internal causation do not. (404) a. The jewels glittered/sparkled. b. * The queen glittered/sparkled the jewels. [Levin and Rappaport Hovav 1995:p 92 (20)]



Chapter 6.  Verbs of internal causation 

( 405) a. The stream burbled/roared. b. * The rocks burbled/roared the stream. [Levin and Rappaport Hovav 1995:p 92 (21)] Furthermore, the fact that verbs of emission cannot have an external cause is illustrated in (406). The use of the reflexive stresses that some external cause is equal to the undergoing object, which cannot be the case with verbs of internalcausation. (406) a. * The diamond glowed by itself. b. * Jane trembled by herself. [Levin and Rappaport Hovav 1995:p 93 (23)] In sum, Levin and Rappaport Hovav 1995 assume the representation in (407) to be the underlying structure of verbs of internal causation. In addition, the authors takte those verbs to have an unergative structure, blocking the derivation of a causative transitive couterpart. (407) [x predicate] [Levin and Rappaport Hovav 1995: 94 (27b)] Verbs of emission or verbs of inner causation have been discussed in the following works.

6.1.1 The instrument role However, it is not straightforward to assign a thematic role to the subject of verbs of emission, which refer to an internal cause of an eventuality. Although these verbs share certain properties with activity verbs, the subject of verbs of emission clearly lacks intentionality. This idea has been fleshed out by Haiden 2005, who argues that the instrument cluster of theta features is assigned to subjects that “are interpreted as non-intentional emitters of force” (Haiden 2005: 73). Following Reinhart 2000 / Reinhart 2001b, Haiden takes theta roles to be composed of two theta features /±c/ (for “cause”) and /±m/ (for “mental involvement”). The instrument role corresponds to the feature cluster [+c-m], i.e., it acts as a cause to the eventuality, but it does not bear any mental involvement. Haiden discusses verbs of sound emission and verbs of light emission. First, some of these verbs permit an agentive reading, where the emitting source appears as the instrument in addition to the agentive subject. (408) a.

Hans leuchtet (absichtlich/gern) mit seiner Taschenlampe. Hans lights on-purpose/happily with his torch ‘Hans lights the way with his torch (on purpose/happily)’. [Haiden 2005: p. 73 (107a)]

 The Structure of Stative Verbs

b.

Hans klappert (absichtlich/gern) mit seinen Zähnen. Hans rattles on-purpose/happily with his teeth ‘Hans rattles his teeth (on purpose/happily).’ [Haiden 2005: p. 74 (108a)]

Of course, it is possible for these two verbs to have the emitting source as the subject. (409) a.

Die Lampe leuchtet. the torch lights ‘The torch is on.’ [Haiden 2005: p. 74 (109a)]

b.

Seine Zähne klapperten. his teeth rattled rather loudly ‘His teeth were rattling.’ [Haiden 2005: p. 74 (108b)]

Still, it is not the case that all verbs of emission systematically undergo the instrumental alternation, contrasting the group of verbs that were discussed in Section 3.1. For some verbs of emission it can be shown that the emitting source bears the instrument role, as these verbs allow for an agentive counterpart that requires an instrument role. In German, quite a few verbs of light emission and almost all verbs of sound emission have these agentive uses. (410) Verbs of light emission: strahlen (beam), blitzen (flash), blinken (blink) (411) Verbs of sound emission: quieken (squeak), quietschen (squeal), piepsen (beep), läuten (ring), dröhnen (drone), summen (buzz), surren (whir), brummen (buzz), knattern (crackle), knistern (crackle), rascheln (rustle), klappern (rattle), tuten (toot), pfeifen (whistle) 6.1.1.1 Verbs of sound emission With verbs of sound emission, the agenitve reading is possible if the agent uses a device to create the sound which must not have any intentionality itself. So, one can create a sound with a rubber duck, as in (412a), but not with a real duck, as in (413a). (412) a. Bert quietschte mit der Gummiente. Bert squealed with the rubber duck. ‘Bert squealed with the rubber duck.’ b. Die Gummiente quietschte. The rubber duck squealed. ‘The rubber duck squealed.’ (413) a. * Bert quietschte mit der Ente, die auf der Wiese saß. Bert squealed with the duck which on the meadow sat. ‘Bert squealed with the duck that was sitting on the meadow.’



Chapter 6.  Verbs of internal causation 

b. * Die Ente, die auf der Wiese saß, quietschte. The duck that on the meadow sat squealed. ‘The duck that was sitting on the meadow squealed.’ (414) a. Ernie dröhnte mit dem Preßlufthammer. Ernie droned with the jackhammer. ‘Ernie droned with the jackhammer.’ b. Der Preßlufthammer dröhnte. The jackhammer droned. ‘The jackhammer droned.’ (415) a. Bert knatterte mit seiner kleinen Dampflokomotive. Bert crackled with his small steam locomotive. ‘Bert crackled with his small steam locomotive.’ b. Die kleine Dampflokomotive knatterte. The little steam locomotive crackled. ‘The little steam locomotive crackled.’ (416) a. Ernie raschelte mit der Zeitung. Ernie rustled with the newspaper. ‘Ernie rustled with the newspaper.’ b. Die Zeitung raschelte. The newspaper rustled. ‘The newspaper rustled.’ The systematic alternations from (412) to (416) show that the subjects in the (b) examples bear the instrument role. This fact substantiates the claim that the subjects of verbs of sound emission are instruments. However, not all verbs of sound emission undergo the instrumental alternation. Consider the following examples which prohibit the agentive counterpart. (417) a. * Der Poldi knisterte mit dem Feuer. The Poldi crackled with the fire. ‘Poldi crackled with the fire.’ b. Das Feuer knisterte. The fire crackled. ‘The fire crackled.’ (418) a. * Die Irmi zwitscherte mit den Vögeln. The Irmi tweeted with the birds. ‘Irmi tweeted with the birds.’ b. Die Vögel zwitscherten. The birds tweeted. ‘The birds tweeted.’

 The Structure of Stative Verbs

This restriction is due to the fact that certain objects such as birds or fire cannot be controlled by humans. It is not possible to act upon a fire such that it emits a crackling sound. Likewise, it is impossible to force a bird to tweet. Consider the next example. The sentences in (419) show that the acceptability of the agentive alternant is dependent on the specific lexical item used: even though one cannot elicit thunders at will, it is possible to do so when using a sound machine, e.g. as it is used to create sound effects in television shows. (419) a. Der Sturm toste. The storm roared. ‘The storm roared.’ b. * Die Irmi toste mit dem Sturm. The Irmi roared with the storm. ‘Irmi roared with the storm.’ c. Die Irmi toste mit der Gewittermaschine. The Irmi roared with the storm machine. ‘Irmi roared with the storm machine.’ Also, (417a) is acceptable if the instrument noun is substituted. This fact backs up the claim that the verb knistern (crackle) undergoes the instrumental alternation, limitations being due to the world knowledge about the nouns involved. (420) a. Der Poldi knisterte mit dem Papier. The Poldi crackled with the paper. ‘Poldi crackled with the paper.’ b. Das Papier knisterte. The paper crackled. ‘The paper crackled.’ The idea that the subject of verbs of emission is an instrument can be further substantiated with the help of the following examples. Nowadays, cell phones are capable of producing all kinds of sounds, and humans can elicit all of them by simply pushing a button on the dial of the phone. So even sounds that could not be generated on purpose before the technical revolution are now compatible with agentive readings: (421) Die Irmi piepste / zwitscherte / zischte mit ihrem Handy. The Irmi beeped / tweeted / whooshed with her cell phone. ‘Irmi beeped / tweeted / whooshed with her cell phone.’ In sum, verbs of sound emission undergo the instrumental alternation systematically. The subject of the non-agentive variant, therefore, bears the instrument role. Those examples that do not allow for the agentive version do not count as counterexamples, since their failure to alternate is due to world knowledge regarding the noun that realizes the instrument role.



Chapter 6.  Verbs of internal causation 

6.1.1.2 Verbs of light emission Verbs of light emission undergo the instrumental alternation as well. Like verbs of sound emission, verbs of light emission show restrictions on this alternation. While the former group of verbs allows the agentive variant rather freely, the latter seems to restrict the agentive variant to a small number of verbs. In other words, there are only few verbs of light emission that permit the agentive variant. Assuming that all verbs of emission have the same underlying grammatical structure, the restriction with respect to verbs of light emission must come from world knowledge about the instrument noun: there are more objects in the world that allow for a controlled sound emission than there are objects that do so for light emission. Objects such as flashlights or turn lights that allow for controlled light emission are acceptable in the agentive variant. (422) a. Die Irmi blitzte mit dem Fotoapparat. The Irmi flashed with the camera. ‘Irmi flashed with the camera.’ b. Der Fotoapparat blitzte. The camera flashed. ‘The camera flashed.’ (423) a. Der Poldi blinkte mit dem rechten Blinker. The Poldi blinked with the right turn light. ‘Poldi blinked with the red turn light.’ b. Der rechte Blinker blinkte. The right turn light blinked. ‘The right turn light blinked.’ However, the majority of verbs of light emission do not allow for the agentive alternant, as the typical subjects of these verbs do not allow for their light emission to be brought about by humans. Hence, this restriction is due to world knowledge. (424) a. ?? Das Auto schimmerte mit der Kühlerhaube im Mondlicht. The car gleamed with the rooster in the moonlight. ‘The car gleamed with the rooster in the moonlight.’ b. Die Kühlerhaube schimmerte im Mondlicht. The rooster gleamed in the moonlight. ‘The rooster gleamed in the moonlight.’ (425) a. * Die Irmi schimmerte mit dem Metall im Mondlicht. The Irmi gleamed with the metal in the moonlight. ‘Irmi gleamed with the metal in the moonlight.’ b. Das Metall schimmerte im Mondlicht. The metal gleamed in the moonlight. ‘The metal gleamed in the moonlight.’

 The Structure of Stative Verbs

(426) a. * Die Irmi glitzerte mit den Augen. The Irmi glittered with the eyes. ‘Irmi glittered with her eyes.’ b. Die Augen glitzerten. The eyes glittered. ‘The eyes glittered.’ Consider (425). The verb schimmern (gleam) does not allow for an agentive alternant, since neither the rooster of a car nor a piece of metal can be forced to gleam by a human. The only way a verb like gleam could be used agentively would be to use a lamp as the instrument: (427) a. Die Irmi schimmerte mit der Lampe durch das Fenster. The Irmi gleamed with the lamp through the window. ‘Irmi gleamed through the window with the lamp.’ b. Die Lampe schimmerte durch das Fenster. The lamp gleamed through the window. ‘The lamp gleamed through the window.’ The sentence in (427a) refers to the following state of affairs: Irmi is deliberately holding down the switch of her lamp; she is doing something in order to keep the lamp glowing. The light of the lamp can then be seen through the window. The restriction on verbs of light emission can be captured as follows: A lamp, being an electrical device, can be forced to emit light, but a piece of metal cannot be controlled in such a way. Verbs of light emission that take a non-electrical object (which most of them do), are therefore not allowed in an agentive context. There is no grammatical constraint that prohibits the agentive variant. The sentences in (422a), (423a) and (427a) provide some of the rare examples. 6.1.1.3 Verbs of smell and substance emission Verbs of smell emission and verbs of substance emission are almost completely unacceptable in the agentive reading. Assuming that all verbs of emission, regardless of their specific source, have the same grammatical structure, this lack must be attributed to world knowledge. There are simply no objects that can be forced to emit a smell or a substance. (428) a. * Der Poldi duftete mit einem Parfum. The Poldi smelled with a perfume. ‘Poldi smelled with a perfume.’ b. * Der Poldi duftete mit einem Raumspray. The Poldi smelled with an air freshener ‘Poldi smelled with an air freshener.’



Chapter 6.  Verbs of internal causation 

c. Der Raumspray duftete. The air freshener smelled. ‘The air freshener smelled.’ (429) a. * Die Irmi blubberte mit dem Wasser. The Irmi bubbled with the water. ‘Irmi bubbled with the water.’ b. Das Wasser blubberte. The water bubbled. ‘The water bubbled.’ (430) a. ?? Die Hitze spritzte mit dem Öl. The heat spattered with the oil. ‘The heat spattered with the oil.’ b. Das Öl spritzte. The oil spattered. ‘The oil spattered.’ If one wants to express that a substance is emitted on purpose, one has to use a different verb in German. For example, parfümieren (perfume) expresses the idea that a substance (of perfume) is applied to the object. This verb structure corresponds like bestreichen (spread) which express that a substance is applied to the surface of an object. However, this verb cannot be used as a simple verb of smell emission: (431) a. Der Poldi parfümierte das Zimmer mit einem Raumspray. The Poldi perfumed the room with an air freshener. ‘Poldi perfuned the room with air freshener.’ b. * Der Raumspray parfümierte das Zimmer. The air freshener perfumed the room. ‘The air freshener perfumed the room.’ This pattern is replicated with verbs of the spray/load-type. They disallow the substance to show up in subject position as well. (432) a. Die Irmi bestreicht die Semmel mit Butter. The Irmi be-spreads the roll with butter. ‘Irmi spreads butter on the roll.’ b. * Die Butter bestreicht die Semmel. The butter spreads the roll. ‘Butter spreads the roll.’ Verbs of substance emission may be used in the agentive variant if a locative is added.

 The Structure of Stative Verbs

(433) a. Die Hitze spritzte das Öl aus der Pfanne. The heat spattered the oil out the pan. ‘The heat spattered the oil out of the pan.’ b. Die Irmi spritzte das Wasser an die Wand. The Irmi spattered the water onto the wall. ‘Irmi spattered the water onto the wall.’ 6.1.1.4 Conclusion In conclusion, this section has substantiated the idea of Haiden 2005 that the subject of verbs of emission bears the instrument role. Although only verbs of sound emission show the systematic instrumental alternation, I take all verbs of emission to consist of the same underlying grammatical structure. Those verbs that block the agentive variant do so since the instrument cannot be forced to emit. Hence, it is a matter of world knowledge and not a matter of grammatical knowledge that determines which verbs and which instruments are acceptable in the agentive variant. As it was shown with verbs of light emission, it is not the case that the substance of light can be emitted on purpose. Some ways of emission and some sources can, others can not. Therefore, it is not the case that the substance of emission determines whether the agentive variant is possible or not; i.e., there is no grammatical feature like /±light/, /±sound/ or /±smell/ (comparable to /±animate/) that determines the lexical-semantic status of a verb class. The discussion in the previous paragraphs has shown that there are clear differences between the kind of substance emitted. Whereas verbs of sound emission alternate freely, the possibility decreases until no agentive variant is allowed at all, as in verbs of smell and substance emission.

6.1.2 The structure of verbs of emission The study in Maienborn 2003 has revealed that verbs of emission pass all the tests that indicate that a verb contains a Davidsonian eventuality. The discussion above has shown that the subjects of verbs of emission bear the instrument role. I therefore conclude that verbs of emission are eventive and share their event structure with activity verbs. The lexical-semantic structure of verbs of emission contains the do-operator which renders the verb eventive. Die Gummiente quietschte. (434) a. b.

Die Gummiente quietschte. The rubber duck squealed. ‘The rubber duck squealed.’ do(rubber duck, (squeal(rubber duck)))

Within the syntactic structure of verbs of emission there is only a single argument licensed as the subject / external argument in the specifier of do.



Chapter 6.  Verbs of internal causation 

(435) doP wo DP do’ 6 3 die Gummiente do0 VP | V quietscht At this point, activity verbs and verbs of emission are treated alike. But how do they differ from each other? Both of them are acceptable with a single argument, and both types of verbs express some spontaneous activity that originates in or is initiated by its subject. The intuition about the difference is that activity verbs involve a deliberate action, i.e., the subject must intend the action that is carried out. In contrast, verbs of emission are perfectly compatible with a non-animate subject, i.e., with a subject that is not able to act intentionally. The feature theory of theta structure developed by Reinhart 2000 and Haiden 2005 provides elegant means to draw this distinction: agents are intentional subjects, hence, their feature structure is specified positively for cause and positively for mental involvement (or intentionality). Instruments, in contrast, are only positive for cause but negative for mental involvement. In other words, instruments are the same as agents apart from a single difference: intentionality (or mental involvement). This is exactly the distinction that comes into play when agentive verbs and verbs of emission are compared. (436) a. agent: [+c+m] b. instrument: [+c-m] The lexical-semantic structure of verbs of emission, therefore, consists only of the dooperator. Therefore, the structure is the same for activity verbs and for verbs of emission: (437) a. b.

Die Gummiente quietschte. The rubber duck squealed. ‘The rubber duck squealed.’ do(rubber duck[+c–m], (squeal (rubber duck)))

(438) a. b.

Die Irmi hüpfte. The Irmi jumped. ‘Irmi jumped.’ do(Irmi[+c+m], jump (Irmi))

Verbs of emission, then, have the same status with respect to the structure of eventualities as activity verbs have. Both refer to an eventive eventuality, which is due to the presence of the do-operator in their lexical-semantic structure. It has been tempting to analyze verbs of emission as statives, since some of their prototypical members do not refer to a straightforward “action”. For example, the expression “pearls gleaming in

 The Structure of Stative Verbs

the hair” seems to describe a static situation. Still, it was observed by Maienborn 2007b that verbs of emission belong to the group of Davidsonian statives. It was acknowledged that, in a way, these verbs are like “real” eventive verbs, although the intuitive idea of eventiveness as “something happens” is not met by most of them. This idea is fleshed out here by arguing that Davidsonian verbs may contain aspectual operators, in particular, certain Davidsonian verbs contain the do-operator.

6.1.3 Conclusion In this chapter verbs of emission were analyzed. It turned out that their lexical-semantic structure contains the do-operator. This analysis captures the fact that verbs of emission are eventive in that they pass the tests for the presence of a Davidsonian eventuality. In addition, the nature of the subject was examined: it bears the instrument role. In the next chapter, I am going to flesh out the details between instruments and agents in combination with a general theory of argument licensing.

chapter 7

Event structure and theta features The result from the analysis of verbs of emission leads to a broader conclusion regarding the relationship between theta features and event structure. The pattern that emerges is that certain combinations of theta features are compatible with specific aspectual operators. The comparison between activity verbs and verbs of emission has revealed that the do-operator is compatible with both the agent ([+c+m]) and the instrument ([+c-m]) role.

7.1 Features within the vP The idea that causers and actors compete for the subject position is by no means new. In this section I will review previous accounts on the feature structure of intentionality, activity and causation that have been put forward in the syntactic theory apart from the model proposed by Reinhart 2000. Thereafter a semantic account that implements some of Dowty’s aspectual operators in Minimalist terms will be reviewed.

7.1.1 Features in v To start with, Kallulli 2004, Kallulli 2006, Kallulli 2007, working in a purely syntactic framework, argues that the subject is licensed via feature checking. According to her analysis, little v may contain several features which license different types of subjects. In case of subjects that carry out an agentive activity, little v contains two features: [+intent] and [+act]. In contrast, subjects that are involved in a non-agentive activity require a little v bearing only the [+act] feature (Kallulli 2006:(29) and (30)). This difference can be illustrated with the help of the following example. The sentence in (439) is ambiguous between the intentional and the non-intentional reading. (439) Rosa screamed. [Kallulli 2006: 288 (25)] The intentional reading corresponds to the usual one: Rosa deliberately creates sounds, she is aware of screaming and willing to carry it out. The non-intentional reading, on the other hand, can be paraphrased as follows: “Rosa in (25) is an actor but not an agent if she does not intend her screaming activity (for instance, if she has taken drugs that make her scream)” (Kallulli 2006: 288, emphasis original). This analysis implies a general treatment of verbs of sound emission. Whereas in the active reading there are both

 The Structure of Stative Verbs

the [+intent] and the [+act] features present, in the non-agentive reading there appears only the [+act] feature, the [+intent] feature being absent. What has been characterized as instrument subjects in Section 6.1.2, which carry the features [+c-m] and require the aspectual operator do in the lexical-semantic structure of the verb, is captured here by a single feature [+act]. Note that Kallulli does not claim that non-intentional subjects of activity verbs are instruments, nor that there is a generalization from her structures (29) and (30) to the class of verbs of emission. According to Kallulli, the difference between activity verbs and causative verbs (or roots, in Kallulli’s terms) is reflected in the feature structure that the root projects into little v. Only activity verbs are able to project [+act], and only causative roots project [+cause]. This is the lexical and syntactic difference between build (activity) and break (causative). Kallulli discusses the difference between her feature system and that of Reinhart explicitly in Kallulli 2006: 287. Following Rivero and Savchenko 2004, she notes that the system of theta features argued for in Reinhart 2000 is not sufficient, since it does not capture the unintentional causer role. Although it is possible that a verb assigns only [+c] to one of its arguments, leaving the /±m/ feature unspecified, the cluster [+c] does not refer to unintentional causers, as the feature cluster [+c] is only an abbreviation for either [+c-m] (the instrument) or [+c+m] (the agent). Haiden 2005 questions this perspective. He argues that verbs of consumption assign different feature bundles to their subjects. For example, the German verb essen (eat) requires an animate, human subject. Hence, the verb selects for the fully specified bundle [+c+m]. The verb fressen (eat with nonhuman subject), on the other hand, is compatible with various kinds of subjects, including people, animals and machines (Haiden 2005: 62). Haiden concludes that fressen assigns the underspecified cluster [+c] to its subject, allowing for various instantiations of /±m/, such as agents ([+c+m]), instruments ([+c-m]) and bare causers [+c]. In particular, he contends that the interpretation of [+c] is to a considerable extent driven by world knowledge. In sum, the analysis of verbs of emission boils down to the fact that these unintentional causers can be analyzed as instruments. In addition, neither Reinhart’s nor Kallulli’s system are able to explain the systematic stative/eventive ambuguity that has been observed in Chapter 3. In order to capture the full range of data, a combination of aspectual operators and theta features leads to a simple that serves the requirement of explanatory adequacy to a greater amount.

7.1.2 Flavors of v The idea that little v comes in different versions has been put forward by Harley 1995. Although the author does not use features to implement the different types of v, her account is in principle compatible with Kallulli’s. Apart from the general discussion of this type of analysis in Section 2.3.2.5, the proposal of Foli and Harley 2005 is of interest to the discussion regarding agentive and instrumental subjects. The authors discuss several “flavors” of little v, attempting to capture the behavior of consumption verbs.



Chapter 7.  Event structure and theta features 

The main problem that the authors address is that consumption verbs that do not imply a resultant state may only have agentive subjects, as exemplified in (440). In contrast, consumption verbs that take a small clause as their complement are compatible with non-animate subjects as well. This fact is illustrated in (440a). Folli and Harley term these non-animate subjects “causers”. (440) a. The sea ate away the beach. b. *The sea ate the beach. [Folli and Harley 2005: (2a)] (441) The groom ate the wedding cake. [Foli and Harley 2005: (1b)] The authors argue that animate subjects are licensed by a special type of little v that is labeled vDO. It is this type of little v that licenses the animate, agentive subject in (441). In contrast, the subjects in (440) are licensed by vCAUS which does not impose the animacy restriction. However, there has to be a resultant state expressed explictly in the structure of the verb, in order for non-animate noun to function as the subject of a verb of consumption. “Causers” are characterized by having no intentional component in contrast to agents. This definition is comparable to the idea of Kallulli, who claims that the [+intentional] feature is missing in unintentional agents. It is also compatible with the claim proposed here that unintentional agents are instruments which are made up of the feature cluster [+c-m]. Furthermore, Folli and Harley define “causers” as initiating a change-of-state (Foli and Harley 2005: 16). Note that by including the notion of change-of-state, Folli and Harley rule out the idea that stative causers exist, since a change-of-state is always eventive. In this respect, their notion of “causer” suffers from a serious weakness. Apart from verbs of consumption, the authors analyze Romance causative constructions in Foli and Harley 2007. Again, the inventory of flavors of v is vDO, vCAUSE, vBECOME and vBE, of which only the first two license external arguments. In particular, the authors argue that avere (have) contains vBE and does not have an external subjects (Foli and Harley 2007: 19). Note that Section 4.1 challenges this claim by showing that even stative verbs with a very simple lexical-semantic structure (for instance subject experiencer/possessor verbs) have external arguments. Folli and Harley’s analysis of Romance causatives leads up to the same structures that were assumed in Foli and Harley 2005: vDO always assigns the agent theta role and is indefferent to its complement, wheras vCAUSE requires a small caluse complement, licensing both the causer and the agent theta role.

7.1.3 Dowty’s calculus from a minimalist perspective The idea that verbs should be decomposed into aspectual operators has been argued for by Arnim von Stechow (e.g. von Stechow 1995, von Stechow 1996). More recently,

 The Structure of Stative Verbs

the mechanism of aspectual decomposition has been rephrased in more minimalist terms (von Stechow 2007). The verb carries uninterpretable features that come with its lexical entry. These features must be checked against the aspectual heads that carry their interpretable counterparts. In other words, the decompositional analysis is implemented in terms of features that are part of the lexical entry. In syntax, these features must be checked in the same way as tense features on the verb (e.g. [u-past]) must be checked against T (or the relevant interpretable features on T). The two aspectual heads discussed in von Stechow 2007 are cause and become; do is not dealt with. Contrasting the various syntactic accounts, von Stechow offers a detailed semantic analysis of the various features. To illustrate, consider the inchoative verb aufwachen ‘wake up’ and the causative verb aufwecken ’awaken’. These two verbs share the same basic predicate, but differ with respect to their lexical features. (442) [[ aufwachte ]] = λw. λx. λt. x is awake in w at t. [after: von Stechow 2007: (7-2),(7-3)] The inchoative verb comes with an uninterpretable become feature [u-become], whereas the causative variant carries the uninterpretable cause feature [u-cause]. (443) a. The inchoative aufwachte ‘woke up’: type e(it) Features: [u-become], ... [[ aufwachte ]] b. The causative aufweckte ‘waked up’: type e(it) Features: [u-cause], ... [[ aufwachte ]] [after: von Stechow 2007: (7-2),(7-3)] The decompositional analysis put forward by von Stechow does not address the question of argument realization. In other words, it is not implied that the covert aspectual heads check some of their features against features of the nominal arguments of the verb. It is merely the verb that checks its uninterpretable features against their interpretable counterparts within the functional structure of the sentence. Being not concerned with intentional agents, von Stechow does not make explicit whether a different bundle of features is needed regarding animate (e.g. John) and inanimate (e.g. the alarm clock) subjects with causative verbs such as (443b).

7.2 Theta checking In the next section, I will outline an alternative account on argument realization and the architecture of the verb phrase. A verb consists of the aspectual operators do, cause and become (I disagree hereby with Jackendoff 1990 who uses several locative predicates such as go or path). The



Chapter 7.  Event structure and theta features 

structures which are made up of these operators specify the types of eventualities that can be expressed in natural language. In other words, there exists a linguistic restriction on the possible structure of eventualities in human language. Of course it is possible that humans can comprehend and process events (and causation, in particular) in ways extending the liminations of language. Each layout of the aspectual structure is compatible with certain argument realization patterns. The feature clusters of arguments (in terms of /±c/, /±m/) are checked against the aspectual operators, which are equivalent to functional heads. Checking functions along the same lines as it does in the domain of Case, person and so on (cf. Chomsky 1995, Chomsky 2001 and subsequent work). This approach to argument realization and eventuality structure generalizes Checking Theory from the A and A-bar domain to the domain of lexical structure. In this way, the general mechanisms that underlie the Minimalist Program are extended to an additional level of grammar. This section proceeds as follows. First, I will explore the possible configurations of the aspectual operators from a type-theoretic perspective. It will turn out that not all theoretically possible combinations are found in natural language. There exists an unexpected restriction that leads to a new insight about the architecture of grammar. Second, several example verb structures and their argument realization patterns are examined, leading to a systematic compatibility relation between theta features and aspectual heads detailed in Section 7.2.3. Finally, I will outline the notion of the situation argument that is implied within the current theory of verb structure.

7.2.1 Aspectual layout What is traditionally termed verb phrase consists only of the aspectual operators. They determine the overall layout of the eventuality that is expressed, and they license the arguments of the verb. It has become clear from the preceeding discussion that the only available operators are do, cause and become, in addition to the basic lexical predicate predicate. This section investigates what possible orders exist among these three operators. These possible orders are the only eventuality structures that occur in natural language. Starting from a type-theoretic perspective (von Stechow 2007 (and preceeding work)), I will examine the type-theoretic status and the possible complements of each aspecutal operator. Second, I will check which orders correspond to which verb classes in natural language. It will turn out that certain realization patterns of the aspectual operators are not ruled out for type-theoretic reasons, but are still not found in natural language. I will propose a theory that rules out those combinations in the third step. Before turning to the aspectual heads, consider the basic predicate predicate. This predicate forms the heart of a verb. In contrast to non-verbal predication, basic verbal predication amounts not only to the application of a predicate to an argument, it also introduces the situation argument. Verbal and only verbal predication boils down to property exemplification at a particular time. Whereas non-verbal predication

 The Structure of Stative Verbs

does not need any further machinery, verbal predication is enabled via the “aspectual” operator predicate. The head pred0 has the same function as the copula in copular constructions: it allows a property (a lexical, referential predicate) to be exemplified at the individual (the argument) at a particular pred0 is therefore the source of the situation argument. Note that this idea is compatible with the theory put forward in Baker 2003. Baker argues that the defining property of the category “verb” is that verbs have a specifier. In contrast to this syntactic perspective, I take the defining property of verbs to be the obligatory presence of predicate. For example, in (444), the property of being obstructed is exemplified at the individual street. (444) Agentive: a. Die Irmi verstopft die Straße mit ihrem Lastwagen. The Irmi obstructs the street with her truck. ‘Irmi is obstructing the street with her truck.’ b. λs do(Irmi, cause(truck, become(obstruct(street)))) (s) The fundamental part of the verb in (444) is the resultant state obstruct (street) that the street is obstructed. As more and more aspectual operators are added to the basic predicate, the structure of the verb becomes increasingly complex, such that additional arguments can be licensed. The tree in (446) provides a detailed view of the structure of the predicate phrase: (445)

predP wo (Argument) pred’ the street wo pred0 referential “noun” obstructed

The predicative head pred0 is merged with lexical predicate (obstructed, in the case at hand). pred0 is a functional head that takes this lexical predicate and adds the situation argument. In the next step, the argument (in our case street), which is of type e (referring to an individual), is added. The whole phrase then is a function that maps situation arguments to truth values. In the case at hand, this phrase would mean something like the street is obstructed. Pred varies with respect to the number of arguments it may take. With verbs that undergo the instrumental alternation as in (444), pred selects for a single argument. In contrast, subject experiencer verbs (as discussed in Section 4.1) and stative verbs of perception (see the analysis in Section 3.6) as well as object-experiencer verbs that assign dative case contain two arguments that are licensed within predP. In their stative



Chapter 7.  Event structure and theta features 

reading, the lexical-semantic structure of these verbs contains neither of the aspectual operators. In this case, pred looks as follows: (446) predP with two arguments: Poldi loves Irmi. predP wo (Argument) pred’ Poldi wo (Argument) pred’ Irmi wo pred0 referential “noun” love Note that it is not a property of the verbal predicate as such that it is able two license two arguments, since there exist adjectives which are able to do so as well. Consider (447): (447) stolz auf (proud of), zufrieden mit (satisfied with) In addition, predP also occurs with zero arguments. For example, activity verbs require a single argument, the agent, which is licensed by the do-operator. PredP, therefore, does not license any arguments in this verb class, it merely hosts the lexical information that distinguishes e.g. jump from hop. This line of thought has been put forward by Hale and Keyser 1993: 54, who aim to derive argument structure patterns with the help of structural and categorial information within the lexical syntax of verbs (see the discussion in Section 2.3.1.1). They argue that activity verbs consist of a single verbal projection that selects a noun as its complement. Other verb classes have a more complex structure, including several layers of V and/or PP-complements. Hale and Keyser’s analysis of activity verbs fits nicely the idea that predP does not license any arguments, but contains solely the specific lexical information (which Kayne 2008 takes to be nominal) in this case. The difference between a copular construction as in (448b) and the corresponding stative verb as in (448a) is therefore the bundling of morphological realization. In copular constructions, the lexical predicate and the copula are two separate morphological entities. In verbal constructions, pred0, which introduces the situation argument, and the lexical predicate form a single morphological entity. (448) a. this cheese smells b. this cheese is smelly [Pustet 2003: 91, cited from Maienborn 2007c: (4)] The predicative head pred0 and the copula, therefore, have the same meaning: a property exemplification at a particular time (a Kimian state). The meaning according to Maienborn 2005b is give in (449).

 The Structure of Stative Verbs

(449) be / sein / ser...: λP λx λz [z ≈[P(x)]] [Maienborn 2005b: (53)] The relation ≈ conveys the idea that a property exemplification at a particular time introduces a referential situation argument (a discourse referent in drt terms) into the lexicalsemantic representation. This argument refers to a Kimian state. The purpose of pred0 is therefore to introduce this situation argument into the representation of a verb. Now that we have a clear picture about predP, let’s consider the aspectual operators. To start with, become expresses the transition from a predicate not holding of its argument to a predicate holding of its argument (as defined by Dowty 1979). The type, therefore, is , < i, t >> if one takes into consideration that the predicate still needs a tense argument at this point in the derivation, as argued for by von Stechow 2007. In the current theory, the type < i > does not refer to a mere temporal argument, but to the situation argument referring to the Kimian state that is introduced by the predicative head pred0 . Although similar, it is not enough to think of < i > as mere tense; rather, it is a property exemplification at a time. Similar to bare tense, this situation argument is then existentially bound within the infl projection. In sum, become, maps an untensed predicate onto another one of the same type. (450) become : type (it)(it) [[become]] = λw. λPit.λt:¬P(beg(t)).P(end(t)) [von Stechow 2007: 6(7-5), further feature omitted] The phrase structure for the become-head is as follows: (451)



becomeP wo become0, predP

Second, cause is defined by von Stechow as taking an individual (the causer) and a proposition (the causee). He notes that this operator “can be further analysed in the style of Lewis and Dowty” (von Stechow 2007: 6, fn.3). However, the Lewis-style counterfactual analysis of causation requires that two propositions, two situations or two events stand in the causal relation. In its basic counterfactual definition, the causeoperator excludes a relation between an individual and a proposition, requiring that the two objects be of the same type. Contrary to this philosophical definition of causation, the linguistic version of cause, the one that comes into play when a verbal structure is created, relates an individual (of type < e >) with a proposition (that may need a tense argument in the course of the derivation, hence of type < i, t >). Cause is defined as follows: (452) cause: type (s(it))(e(it)) [[cause]] = λw.λPs(it).λt.x causes in w at t [λs’.P(s’)(end(t))] [von Stechow 2007: 6 (7-6), further feature omitted]



Chapter 7.  Event structure and theta features 

According to von Stechow, the cause-head is combined with its complement via intensional functional application. This is necessary, since cause sees not the truth value but the proposition in its complement. Extensional functional application does not provide means to take a proposition as a complement. The additional type s that is part of the type < s, < i,t >>, < e, < i,t >> in the structure depicted in (453) refers to the type of the world argument that is needed in intensional functional application. In particular, this world argument is not part of the situation argument that is introuduced by pred0, since not only predicates but also individuals can have intensional meanings. For example, the intensional meaning of Joe is a function from worlds to individuals. Given a world, this function then returns that individual that is characterzied as Joe in the given world. (453)





causeP wo Irmi cause’ wo cause0, predP

Note that a structure like the one in (453) requires the subject of cause to refer to an individual. This excludes sentential subjects, as they never refer to an individual of type < e > but rather to an intensionalized proposition of type . The question how to accommodate sentential subjects in spite of the obvious type mismatch is addressed in Section 7.2.3. The third operator, do, expresses that an agent has a proposition under her immediate control (cf. Dowty 1979 and the discussion in Section 3.2.4.2). The do-operator must therefore relate an individual (of type< e >) with a proposition, in the case at hand a proposition that still needs its temporal/situational argument to be bound (type ). The type of the do-head is therefore equivalent to the type of cause as put forward by von Stechow 2007: < s, >,. To illustrate, a typed tree structure for the do-projection looks as follows: (454)





doP wo Irmi do’ wo do0 , predP

As cause and do work along the same lines, I take do to require intensional functional application as well. Now that I have clarified the type-theoretic properties of the aspectual operators, the next question to address is what possible combinations of the aspectual operators occur in natural language. The become-operator requires its complement to be of type < i,t >. In principle, this is compatible with predP, causeP and doP, all of them being of type . Become0 takes predP as its complement in change-of-state verbs such as einschlafen

 The Structure of Stative Verbs

(to fall asleep). These verbs simply express the transition from a state in which the predicate is not holding to a state in which it is. (455) a. fall asleep b. λx λs [become(asleep(x))] (s) Become0 takes causeP as its complement in verbs of the instrumental alternation where the instrument increases in size (see also the discussion in Section 3.2.4.2) (456) a. b.

Das Gewebe hat nach und nach das Blutgefäß verstopft. The tissue has bit by bit the blood vessel obstructed. ‘The tissue was obstructing the blood vessel bit by bit.’ λs become (cause(tissue, obstruct(blood vessel)))(s) (repeated from (112))

In examples like (456), the instrument noun Gewebe (tissue) is the subject argument of cause, but occupies a position below become. This accounts for the fact that the amount of tissue is increaing bit by bit until the blood vessel is completely obstructed. Therefore, become0 has scope over cause0. However, become cannot take a structure as its complement that contains the dooperator, although the type of the do-phrase is the same as the type of the causephrase, namely . Therefore, this order is not excluded because of a type mismatch. I will return to the question what prohibits this disribution below. Turning to do, we find that this operator embeds all three aspectual heads: the basic predicate predP, a basic predicate that is headed by become, as well as causeP. All three of these formations are of type , thus it is expected that each of them can be the complement of do. This prediction is borne out. Consider the following verb types: in activity verbs, the basic predicate is the complement of do. (457) a. sing b. λx λs [do(x,sing)] (s) Verbs of consumption consist of do0 that embeds becomeP. Note that there is only a single sub-eventuality expressed in verbs of consumption, i.e., the process of eating and the process of being eaten must always be simulataneous, Since the structure in (458b) does not include cause, there cannot be two temporally distinct sub-eventualities. (458) a. eat b. λy λx λs [do(x, become(eaten(y)))] (s) Verbs that undergo the instrumental alternation, object-experiencer verbs that assign accusative case, and verbs of emission are realizations of a structure that consists of do and cause. An example is given in (459), but see Sections 3.2, 3.3 and 6.1 for a detailed discussion. (459) a. annoy (active reading) b. λy λx λs do(x, cause(x, become (annoyed(y)))) (s) (repeated from (153))



Chapter 7.  Event structure and theta features 

Third, cause requires its complement to be of type as well. Again, this fact allows in principle all aspectual heads to occupy the complement position. The combination of cause0 and predP is found in stative verbs that undergo the instrumental alternation, as well as in the stative variants of object-experiencer verbs assigning accusative case. For example: (460) a. decorate (stative reading) b. λy λx λs cause(x, decorate(y))(s) The combination of cause0 and becomeP is found in non-stative, no-agentive variants of verbs that undergo the instrumental alternation. These are similar to (461), but the eventive reading is usually forced by adverbials that require a gradual onset of the resultant predicate. See also Section 3.2.4.2 for a more detailed discussion. (461) a. obstruct (non-active, incremental reading) b. λy λx λs cause(x, become(obstruct(y)))(s) There is no corresponding verb class for the combination of cause0 and doP. As before, the do-operator cannot function as the complement of an aspectual head. Note that it is commonly assumed in the literature that cause takes do as its complement. Usually this is found when causative verbs are discussed, e.g., causative constructions such as feed (“to make someone eat”) or causative light verbs that take a full verb as its complement as in Japanese or Korean, or causative constructions in languages that mark the causative component overtly in morphology. The discussion in this work does not treat such constructions. Rather, I take the structure of main lexical verbs to consist of the aspectual operators. Of course it is possible that such a main verb functions as the complement of a causative light verb. To wrap up: although it is possible from a type-theoretic perspective that all aspectual operators take each of them as their complement, we never find that doP occurs in this position. In other words, doP must always be the highest aspectual layer within a verb structure. This view is compatible with the perspective of Kayne 2008, who argues that nouns but not verbs are open-class elements. The only lexical category that allows for new members to be added freely is therefore noun. In particular, verbs are closed-class elements. Following the tradition of Hale and Keyser 1993, Kayne argues that there are only a finite number of possible verb structures. What seems to render verbs an open class is the referential part that is included in verbs. Kayne puts forward the idea that this referential part is nominal, hence belonging to the open class. For example, whisper, shout and yell seem to be completely different words on the first glance. However, they differ only with respect to their nominal (referential) part that determines the particular way of speaking, while sharing the eventuality structure. Hence, what is a verb and what belongs to the closed class is the eventuality structure: there exist only finite possible structures. I want to strengthen this claim: only those structures that are made up of the aspectual operators, and therefore only those verbs, exist.

 The Structure of Stative Verbs

7.2.2 Example verb structures After inspecting the aspectual layout of the verb, let’s now turn to the relationship between aspectual heads and thematic features in detail. To start with, the subject of activity verbs like laugh carries two features which are relevant to the current discussion: [+c] and [+m]. This feature structure expresses that the subject of an activity verb is an intentional agent. The aspectual operator within this verb structure is do, which is able to check the [+c] feature of the subject. [+m] is interpretable on the noun phrase – it encodes intentionality. (462) a. Irmi laughed. b. doP wo DPAgent do’ [+c+m] wo Irmi do0 predP [+c] | pred’ 3 pred0 laugh Verbs of emission come in two forms: the subject is either an agent or the instrument that is emitting. In the agentive variant, the subject is licensed in the same way as in activity verbs. Thus, the mechanisms regarding the do-projection are the same as in (462b). The noun referring to the emitting source is specified as an instrument, bearing the feature cluster [+c-m]. This instrument is licensed via the cause-head that checks the [+c] feature. Again, [-m] is not checked since it is interpretable on the noun. (463) a. Bert squealed with his rubber duck. b. doP wo DPAgent do’ [+c+m] wo Bert do0 causeP [+c] wo dpInstr cause’ [+c-m] wo rubber duck cause0 predP [+c] | pred’ 3 pred0 squeal



Chapter 7.  Event structure and theta features 

In the non-agentive variant, the instrument and the do-operator remain. This is a legal structure, sine the resulting layout of the tree is not excluded by any grammatical means. It is the same as in activity verbs. In addition, it is possible for the instrument to be licensed by the do-head, since it needs to check the [+c] feature. do carries this feature. (464) a. The rubber duck squealed. b. doP wo dpInstr do’ [+c-m] 3 rubber duck do0 predP [+c] | pred’ 3 pred0 squeal Verbs of consumption are similar to verbs that undergo the instrumental alternation. However, the “instrument” in verbs of consumption (e.g. to eat with a spoon) does not bear the instrument role. What appears to be an instrument is an adverbial that modifies the way the consumption event happens, pretty much like a manner adverbial modifies the way an event takes place. Therefore, only the agentive subject and the object (the thing that is consumed) are licensed via checking by the aspectual operators. As before, the agentive subject is licensed by the do-head via checking of the [+c] feature, and the object is licensed by pred-head, either via the [-c] (or the absence of [+c]) feature. (465) a. Irmi ate an apple. b. doP wo dpAgent do’ [+c+m] wo Irmi do0 becomeP [+c] wo become0 predP 3 dpObj pred’ [-c-m] 3 apple pred0 eaten [-c] Verbs that undergo the instrumental alternation are the most complex verbs that exist, since their aspectual structure involves all three operators. These verbs select for three arguments. In the agentive form, the subject refers to an intentional agent, which is licensed by the mechanisms that were observed for the subject of activity verbs, the agentive subject of verbs of emission and the subject of consumption verbs: the [+c] feature

 The Structure of Stative Verbs

of the noun is checked against the [+c] feature of the do-head, the [+m] feature is interpetable on the noun. The instrument is licensed in the same way as it is in agentive verbs of emission. The noun carries the feature bundle [+c-m]. The [+c] feature is checked against its correspondent on the cause-head, and [-m] stays on the noun where it is interpretable. Finally, the object is specified as [-c-m]. It is licensed by the pred-head via checking of [-c], [-m] is again interpretable on the noun and must not be checked. (466) a. Irmi obstructed the street with her truck. b. doP wo DPAgent do’ [+c+m] wo Irmi do0 causeP [+c] wo dpInstr cause’ [+c-m] wo truck cause0 becomeP [+c-m] wo truck become0 predP 3 dpObj pred’ [-c-m] 3 street pred0 obstr. [-c] In the non-agentive variant of verbs that undergo the instrumental alternation, the structure lacks the do-operator and the agentive subject. Metaphorically speaking, the tree is the same as in (466b), but truncated above the cause-head. The remaining instrumental argument is promoted to subject status and receives nominative case in the further derivation. (467) a. The truck obstructed the street. b. causeP wo DPInstr cause’ [+c-m] wo truck cause0 becomeP [+c] wo become0 predP wo dpObj pred’ [-c-m] 3 street pred0 obstructed [-c]



Chapter 7.  Event structure and theta features 

7.2.3 Checking and verb phases There are three types of features that must be considered when dealing with the mechanisms of argument realization: the causative feature and the feature expressing mental involvement. However, it is not clear what happens to these features in the course of the derivation. In what follows, I will argue that these features, like all other features that play a role in syntax, must either be interpretable at some point, or must be checked against a suitable counterpart. In other words, lexical-syntactic features are subject to the same mechanisms as other features like [agree], [epp], [wh] or [Case]. Lexicalsyntactic features and core syntactic features, therefore, are processed by the same syntactic machinery, following the spirit of the Minimalist Program. Lexical features are checked against their counterparts in functional heads, which I take to be the aspectual operators. In this way, the grammatical system links aspectual operators (i.e. verbal event structures) with their arguments (i.e. argument realization patterns). The causative feature comes in two forms, [+c] and [-c]. It is uninterpretable on noun phrases, hence it must be checked in order to be invisible in the rest of the derivation. The results from Section 7.2.2 show that arguments carrying the [+c]-feature are licensed by either the do-operator (e.g. in activity verbs, verbs of emission, consumption verbs) or by the basic predicate. Both do and cause, therefore, check the [+c]-feature. The negatively specified cause feature [-c], on the other hand, can only be checked by the basic predicate as evidenced in Section 7.2.2. At this point, a simplification of the system is feasible.1 Instead of having a two valued feature system [+c] and [-c], it is possible to encode the same distinction simply by the presence or absence of [+c]. In other words, the argument that was said to carry [-c] is not licensed via this feature. Rather, it is the positively specifed [+c] feature that is checked against the two aspectual heads do and cause. In contrast, the patient or Table 7.1  Arguments licensed by do / cause Verb type activity verb verb of emission (agentive) verb of emission (non-agentive) verb of consumption instrumental alternation (agentive) verb of emission (agentive) instrumental alternation (non-agentive)

1.

licensing head do do do do do cause cause

Thanks to Edwin Williams for pointing this out.

argument intentional agent [+c+m] intentional agent [+c+m] instrument [+c-m] intentional agent [+c+m] intentional agent [+c+m] instrument [+c-m] instrument [+c-m]

example laugh squeal squeal eat obstruct squeal obstruct

 The Structure of Stative Verbs

the experiencer argument is distinguished from agents and instruments by the lack of the [+c]-feature (instead of having a negative feature [-c]). The resulting system is simpler: there exists a relation between the feature [+c] and the aspectual heads do and cause. Arguments lacking this feature are licensed by pred as an elsewhere condition. The cause features function as ordering devices. With the help of these features it is possible to distinguish two otherwise alike nouns in terms of their causer and causee role: one of the nouns is associated with the causing eventuality, whereas the other takes part in the caused eventuality. In this sense, the cause feature is semantically uninterpretable on a noun. It merely functions as a syntactic ordering device. The cause features contrast the feature that specifies mental involvement. The latter is semantically interpretable and, although visible at the lexical-syntactic level, not of a purely syntactic nature. Second, the feature expressing mental involvement ([±m]), which is again either speicified positively or negatively, is not checked against a head and eliminated. On the contrary, the feature is interpretable on the noun phrase. This indicates that mental involvement is an interpretable category comparable to number or gender on the noun. Both do and cause as well as the basic predicate are compatible with [+m] and [-m]. The overview in Section 7.2.2 and data in Table 7.1 summarize the patterns for do and cause. In particular, the do-operator is compatible with both [+m] (as in activity verbs) and [-m] (as in verbs of emission). The compatibility of predicate with the[±m] feature is summarized in Table 7.2 below. The relation between cause and [+m] is not included in Table 7.1. Example (468) shows that this is a legitimate checking relation. (468) Hänsel und Gretel verstopfen mit der Hexe den Ofen. Hänsel and Gretel obstruct with the witch the oven. ‘Hänsel and Gretel are obstructing the oven with the witch.’ Verbs that undergo the instrumental alternation like verstopfen (obstruct) can take an agent (Hänsel und Gretel) and an instrument (the witch). Regarding (468), Hänsel and Gretel could as well have obstructed the oven with stones (which are of course [-m]). This suggests that the specification of [m] is irrelevant for instruments: both [+m] (the Table 7.2  Arguments licensed by predicate Verb type verb of consumption instrumental alternation (agentive and non-agentive) object-experiencer ACC (agentve and non-agentive)

licensing head

argument

example

pred pred

patient [-c-m] patient [-c-m]

eat obstruct

pred

experiencer [-c+m]

annoy



Chapter 7.  Event structure and theta features 

witch) and [-m] (stones) are compatible with the cause-head. In other words, the [±m]-feature is invisible to cause. Additional evidence for the claim that [±m] is interpretable comes from studies on the interpretation of metaphor. Prinzhorn 2005 argues that metaphorical use is the systematic reinterpretation of the lexical features [±c] and [±m]. The examples in (469) illustrate this idea. (469) a. Robert is a bulldozer. b. The fog comes on little cat feet. [Carston 2002: (3), (6a)] Prinzhorn argues that the matching between the human bearing the cluster [+c+m] and the bulldozer carrying [+c-m] in (469a) runs into a feature mismatch. In order to interpret the sentence, the non-animate word must be shifted from [-m] to [+m]. In this way, some properties of the bulldozer are transferred to the animate subject. For instance, Robert is understood to act like a bulldozer in that he does not pay respect to people. Prinzhorn purports that people who suffer from autism have severe difficulties in shifting [-m] to [+m]: they cannot comprehend examples like those in (469). Hence, there exists a specific breakdown of the interpretation of the [m]-feature. This finding substantiates the idea that [m] is an interpretable feature on the noun. Moreover, the distinction between the purely grammatical, uninterpretable feature [c] and the interpretable [m] is confirmed by the fact that on a noun [c] but not [m] needs a verb to be interpretable at all. Take a noun without any further context like child, cat, book. It is always clear whether this noun refers to an animate/intentional entity or not. In contrast, one cannot tell from the noun alone if it is specified as [+c] or [-c]. In order to do so, one must take into consideration the verb that “assigns the theta role” to the word. For instance, book is specified as [+c] in (470a), but as [-c] in (470b). (470) a. The book scares Irmi. b. Irmi destroyed the book. Does the system proposed so far over-generate? An anonymous reviewer points out that the system predicts that verbs that undergo the instrumental alternation, in their non-agentive variant, might receive an episodic reading. For example, decorate contains the do-operator which is in principle compatible with the [+c]-feature of the instrument noun candles, incorrectly predicting that the example in (471) should allow for an eventive reading. (471) Candles decorated the cake. If a verb lice decorate that undergoes the instrumental alternation contains the dooperator, it contains the cause-operator as well. Hence, we only find the two eventuality structures in (472) with verbs that undergo the instrumental alternation.

 The Structure of Stative Verbs

(472) a. agentive: do(cause (decorated (cake))) b. non-agentive: cause (decorated (cake)) Note that the eventuality structure containing merely do and the basic predicate exists (e.g. in activity verbs and verbs of emission), but is not available for verbs that undergo the instrumental alternation. Taking the structures in (472) to be the only ones available for verbs that undergo the instrumental alternation (modulo, of course, the optional insertion of become which is irrelevant to the point here), only (472a) receives an eventive reading, (472b), lacking both do and become, being stative. The stative variant in (472b) may take both animate and inanimate nouns as its subject. Since [+m] is not interpretable at cause, this distinction does not have any consequences. Apart from the run-of-themill instrumental subject candles, consider the example in (473). (473) Ein Mädchen schmückte die Torte. A girl decorated the cake. ‘A girl decorated the cake.’ Think of a bachelor party: on top of a huge cake there sits a girl. In this context, the girl is decorating the cake pretty much like candles do. It is possible, therefore, that this example receives a stative reading. This example indicates that a subject containing [+m] does not automatically trigger an episodic reading. Returning to (472b), this structure always requires two arguments, one that is licensed by do, and one that is licensed by cause. Imagine the instrument subject ([+c-m]) being licensed by the do head. In this case, the agent ([+c+m]) must be licensed by cause. Since only do is sensitive to the presence of [+m], in this case, there is no [+m], the instrument not bearing a positive feature value. As cause is insensitive to the specification of [+m], both arguments are only characterized by their [+c] feature. Therefore, they are indistinguishable from the point of view of grammar. For this reason I take it to be the case that an argument bearing [+m] must be licensed by do if both do and cause are present. Therefore, it is not possible for an instrument like candles to be licensed by do in verbs that undergo the instrumental alternation. This leaves us with the question why it is not possible for verbs that undergo the instrumental alternation to have a structure consisting solely of do and a predicate, as in (474). (474) do(decorated(cake)) Of course, it is exactly this structure that licenses an instrumental subject in verbs of emission. I take it to be that there exist all possible combinations. In particular, there exists an agentive variant of verbs of emission, and an agentive variant of verbs that undergo the instrumental alternation, sharing their event structure. From this structure it is possible to drop the cause head in verbs of emission, and it is possible to drop the do-head in non-agentive verbs that undergo the instrumental alternation. At this point, I cannot say more than that.



Chapter 7.  Event structure and theta features 

After having clarified the core difference between the two lexical features, I will now turn to the two aspectual heads do and cause, focusing on the following questions: why must do not be dominated by another aspectual head? How is it possible that sentential subjects are licensed by the cause operator, given the type-mismatch (cause requiring an individual as its argument)? As discussed in Section 7.2.1, do and cause have the sametype-theoretic status, and both license arguments that bear the [+m] or the [-m]-feature. Yet, they differ with respect to whether they license sentential arguments (only cause does), and whether they have to be the topmost aspectual operator (only do has to be). So, what is the exact difference between do and cause? The answers to these questions address directly the nature of the verb phrase as a phase. A clear picture emerges from a phase-theoretic perspective on syntactic derivation (cf. Chomsky 2001 and subsequent work). Although do and cause work along the same lines with respect to their type-theoretic argument selection properties, they differ when it comes to their phase-theoretic status. The hypothesis is that only do0 but not cause0 marks the end of the phase. In other words, as soon as the aspectual head do0 is inserted, the verbal phase is closed and transferred to spell out or to a higher level of representation (depending on the exact version of Minimalism). In order to substantiate this claim, I will employ two arguments: first, since do marks the end of the phase, it must always occur as the topmost aspectual head within a verb. Second, the distribution of sentential subjects underpins the difference between do and cause. Again, this distinction can be captured in phase-theoretic terms. In Section 7.2.1, the general aspectual layout of possible verb structures was examined. One major result was that do must be the highest aspectual head within event structure. It is not possible that cause or become take do as their complement. This fact indicates that do has a special status within the aspectual (or functional) strucutre of the verb: it is always the topmost functional head, i.e., it closes the phase. Cause, in contrast, may also be the highest aspectual head, but only if there is no do-head present. In these cases, cause does not necessarily mark the end of the phase. When it comes to licensing of sentential subjects, the following pattern emerges: while cause may freely license sentences, do cannot do so. For instance, stative verbs that involve a sentential subject like verbs that undergo the instrumental alternation or stative object-experiencer verbs that assign accusative case are grammatical with sentential subjects. The latter is given in (475). Both the stative and the eventive (gradual) reading are grammatical, compare the lexical-semantic representations in (475b) and (475c). (475) a. b. c.

Daß die Irmi im Lotto gewonnen hat, ärgert den That the Irmi in the lottery won has annoys the Poldi (nach und nach). Poldi (bit by bit). ‘It is nnoying Poldi (bit by bit) that Irmi has won in the lottery.’ λs cause(Irmi-wins-in-lottery, annoyed(Poldi))(s) λs cause(Irmi-wins-in-lottery, become(annoyed(Poldi)))(s)

 The Structure of Stative Verbs

Note that it is not the case that cause is always able to license sentential arguments. Although it can do so when it is the highest aspectual operator as in (475), this possibility ceases as soon as do is present, e.g. in verbs containing an agent (licensed by do) and an instrument (licensed by cause). Hence, instruments can never be realized as sentences, they always have to be nominal. Consider (476). (476) * Die Irmi hat (mit), daß sie im Lotto gewonnen hat, The Irmi has (with), that she in the lottery won has, den Poldi geärgert. the Poldi annoyed. ‘Irmi annoyed Poldi (with the fact) that she had won in the lottery.’ However, if a pronoun is added to the preposition mit (with) that refers to the instrumental clause, the sentence becomes grammatical. (477)

Die Irmi hat damit, daß sie im Lotto gewonnen The Irmi has that-with, that she in the lottery won hat, den Poldi geärgert. has, the Poldi annoyed. ‘Irmi annoyed Poldi with it that she had won in the lottery.’

It is possible that sentential instruments are excluded on Case-theoretic grounds. For example, sentences such as (476) could be illicit since the preposition needs a DP to assign its Case to. Moreover, cause can occur as the complement of become. In this case, cause again cannot license sentential arguments. For example, the non-agentive, eventive reading of verbs that undergo the instrumental alternation like cover forces that the subject increases in its size. Consider (478), repeated from (110). (478) a. Die Blätter haben nach und nach den Boden bedeckt. The leaves have bit by bit the floor covered. ‘Leaves were covering the floor bit by bit.’ b. Der Vorhang hat nach und nach die Bühne verdeckt. The curtain has bit by bit the stage hidden. ‘The curtain was covering the stage bit by bit.’ Examples like (478) are rendered ungrammatical when they contain a sentential subject: (479) *Daß es heruntergefallen ist, hat nach und nach den Boden bedeckt. That it downfallen is, has bit by bit the floor covered. ‘It was covering the floor bit by bit that it fell.’ Contrasting cause, the operator do never licenses sentential subjects. Consider the activity verb in (480a), the verb of consumption in (481a), and the verb of emission in (482a).



Chapter 7.  Event structure and theta features 

(480) a. *Daß er lacht, lacht am längsten. That he laughs, laughs the longest. ‘That he is laughing, is laughing the longest.’ b. Der Poldi lacht. The Poldi laughs. ‘Poldi is laughing.’ (481) a. *Daß sie kaut, ißt einen Apfel. That she chews eats an apple. ‘That she is chewing, is eating an apple.’ b. Die Irmi ißt einen Apfel. The Irmi eats an apple. ‘Irmi is eating an apple.’ (482) a. * Daß es brennt, leuchtet. That it burns, glows. ‘That it is burning is glowing.’ b. Das Feuer leuchtet. The fire glows. ‘The fire is glowing.’ In sum, there is a clear distinction between do and cause regarding the licensing of sentential subjects. Only if cause is the highest aspectual head, CP-arguments are grammatical. Note that only subordinated clauses that are introduced by a complementizer exhibit this pattern. Relative clauses, in contrast, are perfectly acceptable in all of the above constructions. In particular, relative clauses can be licensed by do and by cause and they can act as an argument that is increasing in its size. (483) a. Wer zuletzt lacht, lacht am längsten. Who at last laughs, laughs the longest. ‘The one who is laughing at the end will be laughing for the longest time.’ b. Was der Irmi passiert ist, ärgert den Poldi. What nom the Irmi dat happened is, annoys the Poldi. ‘The thing that happened to Irmi is annoying Poldi.’ c. Was heruntergefallen ist, hat nach und nach den Boden bedeckt. What downfallen is, has bit by bit the floor covered. ‘The thing that had fallen was covering the floor bit by bit.’ Relative clauses denote predicates that act as modifiers of a noun (cf. Heim and Kratzer 1998, Chapter 5). In the cases at hand it is likely that there is a phonetically null noun within the matrix clause, the relative clause acting as a modifier to this noun. Clauses that are introduced by a complementizer, in contrast, do not refer to an entity; rather, they express a proposition that has a truth value. This is a fundamental difference

 The Structure of Stative Verbs

between these two types of subordinated clauses. Therefore, the fact that relative clauses are always licit does not weaken the claim that only cause in the top position may license sentential subjects. The arguments that are licensed via the aspectual heads directly must be nominal for two main reasons: recall from Section 7.2.1 that the type of the operators requires them to refer to an entity. Only nouns (or, rather, DPs) are able to denote an entity. Complement clauses always refer to a proposition and never to an entity. First, the arguments must contain a bundle of thematic features that are checked against the aspectual head; for example, the [+m]-feature of an agent noun is checked against the aspectual do-head. A sentence introduced by a complementizer does not have such features, as the complementizer closes off the complete clause, rendering any unchecked features within it inaccessible from outside. In particular, the features of the nouns within the complement clause are already checked and therefore invisible to the verb of the matrix clause. Relative clauses, in contrast, always modify a noun in the matrix clause, even if this noun is phonetically silent. Relative clauses, therefore, refer to an entity and, therefore, are possible in both the agent and the instrument positions. Second, in Representation Theory (Williams 2003), a general mechanism for embedding is propsed: the Level Embedding Conjecture (LEC) (Williams 2003: 63f.) requires that “an item can be embedded exactly at the level at which it is defined, and no other” (p. 64). In other words, the embedding as well as the embedded structure must be at the same level of representation when the embedding occurs. For instance, a verb that is at Theta Structure embeds a noun (that is as well at Theta Structure). A verb that is at Case Structure (it has tense features, or it projects a TP) embeds a noun at Case Structure (e.g. a noun that bears accusative case). Hence, a VP cannot be merged with a complete sentence until the verb phrase has been lifted to a higher level of representation. In particular, that-clause embedding occurs at Surface Structure (Williams 2003: 73). Therefore, the embedding of sentential subjects must occur at a later stage in the derivation. While the do-head indicates the end of the verb phase, the causehead, if it occurs in top position, does not do so. Hence, in the latter case the subject position of the verb stays open at least until Surface Structure, whereas in the former case, it must be filled at Theta Structure. Putting it all together, the system under development here works as follows: there are two aspectual heads, do and cause, that are able to license arguments within their specifier position. The two heads differ with respect to two properties: do, if present, has to be the highest head (therefore indicating the end of the verbal phase and transferring the verb into the next level of representation). In contrast, there is no such requirement on cause. This head can appear as the highest head and as the complement of do. If it occupies the highest position, it does not indicate the end of the verbal phase, leaving its features open to be checked by a sentential subject at a later stage of the derivation. The feature [±m] is only checked or, rather, interpreted at the do-head. So only do but not cause is sensitive to this distinction, the specification of [m] not being interpreted at the cause-head.



Chapter 7.  Event structure and theta features 

7.2.4 The situation argument This section discusses the nature of the situation argument. The theory of verb structure put forward here makes specific claims about the situation argument. In essence, a verb contains a situation argument that refers to a Kimian state. With the help of the aspectual operators, this situation argument can be turned into an eventive eventuality. Hence, the notion of Davidsonian event argument is superfluous. It is an epiphenomenon. Regarding the treatment of the situation argument at the level of lexical-semantic representation within Semantic Form as it is employed here, Wunderlich 1997 argues that the situation argument is not a temporal index alone; rather, it consists at least of a temporal component and a component handling possible worlds, since a situation may be true only with respect to some worlds. According to Wunderlich, both nouns and verbs carry referential arguments. These arguments must be combined with contextual information in order to be able to refer to a particular entity. This combination functions as a constraint on the referential argument; it is carried out by functional categories: determiners in the case of nouns, and aspect, tense, mode, among others, in the case of verbs. In other words, it is always the same type of situation argument that is combined with functional categories that gives rise to different types of eventualities. Extending this claim to the situation arguments of verbs, I take it that there is no need for a fundamental distinction between stative and eventive arguments. It is possible that the latter are construed out of the former by certain functional heads. I will diverge from Wunderlich’s system in the following respects: first, Wunderlich claims that the various kinds of adverbials are only compatible with different sorts of the situation index. Impossible combinations of adverbs and verb types are therefore ruled out on the basis of a simple type mismatch. According to the theory presented here, adverbials must be anchored on one of the aspectual heads. It is therefore the absence of either do, cause or become that renders certain adverbials ungrammatical. Second, Wunderlich does not give any reason why natural language allows the cause relation to hold between an individual and a proposition (or situation), in contrast to the classical philosophical theory. The theory proposed here tries to derivethis property in terms of more basic mechanisms of natural language (see Section 7.2.3). Third, Wunderlich (Wunderlich 1997:(19)) proposes that the situation argument is introduced into syntactic structure at the infl node. This contrasts the view of Maienborn 2003 who claims that the situation argument is only existentially bound, but not introduced at this position. a particular point of time. In her book, Maienborn argues that copular constructions always refer to a Kimian state. She argues that every verb contains a referential argument. For example, a verb like sleep can be represented as in (484), where e indicates the referential argument (which is in this case presumably a Davidsonian event argument), and the theme argument is introduced in a neoDavidsonian way.

 The Structure of Stative Verbs

(484) schlafen λx λe [sleep(e), theme(e,x)] [Maienborn 2005b: (51b)] This referential argument is then existentially bound at Infl. More precisely, Maienborn contends that Infl introduces a discourse referent for the situation argument. In other wordds, the situation argument is fully determined at Infl, much like a noun is determined by its determiner. The semantic structure is given in (485), in DRT style (in (a)) and in predicate logic (in (b)). (485) a. b.

Infl: λP [s | P(s)] [Maienborn 2005b: (51c)] Infl: λP ∃s [P(s)] [Maienborn 2005b: fn. 23]

The application of infl to a verb existentially binds the situation argument, as can be seen explicitely in (485b), where P is then the predicate referred to by the verb. For the current theory this implies that as soon as infl binds the situation argument, it is no longer possible to add aspectual operators. The Kimian situation, which is included in all types of eventualities, is a property exemplification at a particular time. While also some nouns may have temporal indices, the situational index of verbs is different. An event noun can be modified by temporal adverbs: (486) a. b.

yesterday’s explosion die gestr-ige Entscheidung the yesterday-aff decision ‘yesterday’s decision’

However, verbs but not nouns are sensitive to the possible world to which it is evaluated. Moreover, it is the situation argument of verbs that may refer to an event: the event is created out of the situation argument and the aspectual operators. 7.2.4.1 The elimination of the Davidsonian event argument In order to show that a reduction to a situation argument and aspectual operators is possible, I will go through the main arguments for having this variable, showing for each of them how the same task can be modeled by the situation argument and the aspectual operators alone. First, Davidson 1967 and many authors after him have argued that the Davidsonian event argument is necessary in the structure of verbs, since adverbials are anchored at the event argument. In particular manner adverbials and local adverbials seem to need this entity in a semantic representation. Statives, which do not allow for these adverbials, were then said to lack the Davidsonian argument. Von Stechow (personal communication) points out that adverbials are modifiers that are anchored at the aspectual heads. This allows for a fine grained distinction: certain adverbials select for a specific type of head. For example, agent-oriented manner adverbials are only compatible with



Chapter 7.  Event structure and theta features 

the do-head. They include a presupposition that the agent is about to do the action. This is at least one step more explanatory than simply postulating that the event argument comes in different types and that adverbials are compatible with one or the other type. Second, according to Neo-Davidsonian theories of argument selection (cf. Parsons 1990), thematic roles are solely introduced via thematic predicates that anchor an argument to the event. This type of theory does not restrict the number and type of arguments, nor does it predict any systematic relation between the event structure that is expressed and the type of arguments that are selected. The third main argument for the existence of the Davidsonian argument is that action sentences can be continued with it happened, where the pronoun it is said to refer to the event (cf. Davidson 1967). Maienborn 2005b shows that this fact is not restricted to action sentences, i.e., the pronoun does not refer to the (eventive) Davidsonian event argument. She gives examples of sentences expressing a Kimian state which may be referred to by a following pronoun. (487) Carol ist wütend. Das wird bald vorbei sein. Carol is angry. This will soon over be. [Maienborn 2005b: (43a)] Fourth, it is the case that only verbs referring to an (eventive) Davidsonian eventuality may be the complements of perception verbs. Kimian statives may not occur in this position. It is tempting to analyze this constraint in a way that perception verbs require the event argument to be present in their complement. However, I put forward the idea that it is not a selection restriction that goes for the event argument, but that only actions and changes are perceptible. Hence, at least one of the aspectual operators must be included in the complement of perception verbs. Finally, there is the question why there is only a single eventuality expressed by a verb; if verbs contain more than one sub-eventuality, why is it always the case that they are sub-eventualities of a single eventuality? Why is it impossible for a verb to refer to two eventualities, for example, a causing and a causee eventuality? This is the case since a single verb contains only a single lexical predicate that anchors the situation argument. The other components of the eventuality which may be captured as subeventualities are always the aspectual operators. They never introduce a situation into the structure of the verb. This can only be done by the basic lexical verbal predicate.

7.3 Conclusion This chapter proposed an architecture for the event structure of verbs and the linking of arguments. The event structure of a verb is derived from its components, the aspectual heads do, cause and become, as well as the basic predicate predicate which introduces the situation argument that is characteristic for verbs. Although in principle

 The Structure of Stative Verbs

we expect to find any possible combination of these operators, we find that there exists a substantial restriction: the do head must never be subordinated. Agent and instrument arguments are licensed by the two heads do and cause; both of them check the [+c] feature on the argument. This feature is a purely syntactic one, its mere purpose is to distinguish causers from causees. The second feature, [±m] is only interpreted on do. In other words, do has a specification for [m], and cause has no [m]. Arguments that bear the [-c] feature are licensed via predicate directly. All other complements of verbs are not licensed via feature checking. Rather, they are anchored to the predicate via different means, e.g. Semantic Incorporation in the case of measure verbs, or via cipient predication for dative aruments. The difference between do and cause is furthermore replicated when sentential subjects are considered. Only cause but not do is able to license them. This correlation hints towards the fact that do with its [m]-feature closes the phase. All of its features within its domain are sent to the interpretative component of language as soon as this head enters the derivation. In contrast, cause, which is defined by the lack of [m], may stay open until a later phase is ended, thereby being able to license sentential subjects. In this way, the system proposed here extends the spirit of the Minimalist Program to the lexical domain. In order to build a verb, the well-known ingredients “functional” projections (the aspectual heads), uninterpretable features ([+c], [-c]), an interpretable feature ([m]) and phases are employed.

chapter 8

Conclusion This book closes with an outlook on two classes of verbs that have not been studied within the framework proposed so far. After that a final conclusion is drawn.

8.1 Further verb classes 8.1.1 Modals It is not reasonable to analyze modal verbs with respect to their event structure status, since they differ greatly with respect to lexical verbs. Modal verbs and lexical verbs come from two different domains of grammar: whereas lexical verbs select for arguments to form a proposition (i.e., to become fully saturated), modal verbs are operators that act on these propositions. In other words, modals, in contrast to lexical verbs, are propositional operators (as argued for by Lewis 1973, Kratzer 1981, Kratzer 1991, von Stechow 2004 and e.g. Butler 2004). According to the standard semantic view, modal verbs are quantifiers over possible worlds which are interpreted with respect to two entities: a modal base or a conversational background that specifies what the rules are and an ordering source that determines the order of the possible worlds. Therefore, I will leave the question open whether it does make any sense at all to ask if modals refer to a Kimian or to a Davidsonian eventuality. Likewise, I do not want to make any claims about whether it is reasonable to investigate if some of the aspectual operators are present in the structure of modals.

8.1.2 Sensation predicates Very briefly, this section examines the status of sensation predicates as in (488). These verbs are no longer productive in German, therefore I will leave the detailed analysis of this verb class to future research. (488) frieren (freeze), dürsten (thirst), hungern (hunger), frösteln (shiver), ekeln (nauseate) Sensation predicates, in contrast to subject-experiencer verbs like love, may occur with only a single argument. This argument bears nominative case as in (489a), but for most of the verbs there exists a variant where the single argument carries accusative, as in

 The Structure of Stative Verbs

(489b). The latter cases are marked and have an outdated flavour, but are still used in more formal registers. (489) a. Der Poldi friert. The Poldi nom freezes. ‘Poldi is freezing.’ b. Den Franzi friert (es). The Franzi acc freezes (it). ‘Franzi is freezing.’ c. Den Sepp ekelt es. The Sepp acc nauseates it. ‘It nauseates Sepp.’ (480) a. Es dürstet den Poldi nach Wissen. It thirsts the Poldi acc after knowledge. ‘Poldi is thirsty for knowledge.’ b. Der Poldi dürstet nach Wissen. The Poldi nom thirsts after knowledge. ‘Poldi is thirsty for knowledge.’ Some verbs like the one in (491) do not allow for the subject to carry accusative case. Only the nominative, active variant is possible. (491) * Den Poldi hungert. The Poldi acc hungers. ‘Poldi is hungering.’ In turn, I will show that sensation predicates do not have a Kimian stative reading. As far as the diagnostics for Kimian stative readings are concerned, these verbs pass none of them. 8.1.2.1 Manner adverbials As illustrated in (492), all kinds of manner adverbials are acceptable with sensation predicates. This contrasts strongly the behavior of Kimian stative verbs, which do not allow for event-related manner modification. (492) a. Die Irmi hungert auf eine ungesunde Weise. The Irmi nom hungers in a unhealthy way. ‘Irmi is hungering in an unhealthy way.’ b. Die Irmi friert zähneklappernd. The Irmi nom freezes teeth-rattling. ‘Irmi is freezing and rattling her teeth.’ c. Den Poldi fröstelt. The Poldi acc shivers. ‘Poldi is shivering.’



Chapter 8.  Conclusion 

Although the cases where the experiencer bears nominative are more natural and permit adverbials more easily, there exist even cases with manner adverbials for those verbs that select a non-nominative argument, as in (493). (493) a.

sie fröstelt unwillkürlich She nom/acc shivers involuntarily. ‘She is shivering involuntarily.’ [from: http://www.literaturwerkstatt.at/texte2/2texte00/muel-s3.html; my translation]

b.

Wie kuschelig fröstelt’s einen, wenn [...] How cosy shivers-it oneACC if [...] ‘How cosy is one shivering if ...’ [from: http://www.zeit.de/2003/04/KJ-Luchs2002; my gloss, my translation]

These data strongly suggest that there is a Davidsonian eventuality involved. True Kimian statives do not allow for this kind of modification. 8.1.2.2 Locative modifiers Event-related locative modifiers are only allowed if there is an event going on. With sensation predicates, these types of locative modifiers are perfectly acceptable, as shown in (494). Example (494a) indicates that even inner and outer event-related locative modifiers are possible with sensation predicates, whereas (494c) shows that these adverbials are also grammatical together with non-nominative assigning verbs. (494) a.

Im Winter friert der Poldi auf dem Schulweg an In winter freezes the Poldi nom on the way-to-school on den Füßen. the feet. ‘In winter Poldi’s feet are freezing on his way to school.’

b. Die Irmi hungert vor dem Parlament. The Irmi nom hungers in front of the parliament. ‘Irmi is on hunger strike on front of the parliament.’ c.

da fröstelt es manchen im Gemüt there shivers it someone acc in the soul ‘It shivers in someone’s soul.’ [from: http://r.bluewin.ch/?id=58704; my translation]

Again, this test shows that sensation predicates cannot be interpreted as Kimian statives.

8.1.2 Degree readings The degree adverbial ein bisschen (a little) allows for a temporal interpretation only, if there is an event expressed by the verb. In Kimian stative cases, its interpretation is

 The Structure of Stative Verbs

restricted to degree readings. With sensation predicates, this degree adverbial allows for a time-span problem easily. (495) a. Die Irmi hungert ein bisschen. The Irmi nom hungers a little. ‘Irmi is hungering a little.’ b. Den Poldi fröstelt ein bisschen. The Poldi acc shivers a little. ‘Poldi is shivering a little.’ In sum, sensation predicates permit a temporal interpretation of the degree adverbial ein bisschen (a little), indicating that these verbs have a Davidsonian (eventive) reading. 8.1.2.4 Conclusion In the previous section, I took a brief look on sensation predicates. Although these verbs are at the first glance very similar to subject-experiencer verbs, their aspectual behavior is entirely different. Whereas subject-experiencer verbs were shown to express a Kimian state, sensation predicates fail all the test that predicates which allow for a Kimian stative reading pass. Moreover, sensation predicates do not behave in a uniform way. Those predicates that assign nominative case to their subject behave similar to activity verbs like run. Both allow for the experience to be controlled intentionally, e.g. as in Die Irmi hungert absichtlich (Irmi hungers intentionally; she is on hunger-strike). The verbs that assign accusative or dative to their subject do not pass the diagnostics for the presence of a Kimian state so easily. This may be due to the fact that these forms are already obsolete. Nevertheless, some rare cases of event-related modification were found. As it may be, I leave the precise status and their underlying lexical-semantic representation for future research.

8.2 Conclusion Finally, I would like to recapitulate the goal of this book and summarize its main results. In addition, I will address what kind of implications my results have on the current picture of grammar. I will conclude with some open questions that are left for future research. The core results regarding the stative/eventive distinction are summarized in the following list: – Several different types of verbs allow for a Kimian stative reading. It is not the case that statives form a uniform class of verbs. In particular, stative verbs are not the building blocks for verbs displaying a more complex event structure. – A Kimian stative reading arises if both the do and the become-operator are absent.



Chapter 8.  Conclusion 

– If a verb contains the cause-operator within its lexical-semantic structure, it is able to receive a stative reading if both the causer and the causee sub-event are stative. If cause is present, the verb always displays a systematic stative/eventive ambiguity. – Verbs that do not allow for the eventive reading contain one eventuality only. The first goal of this thesis was to sketch a picture on how stative verbs are seen in current linguistic theory. When reviewing the assumptions that have been made in the literature it turned out that the work regarding argument structure contains a lot of implicit assumptions on the nature of stative verbs. The literature suggested that stative verbs should have one or all of the following properties: first, it has been assumed that stative predicates are the smallest and simplest building-blocks of event structure. However, it remained unclear whether this assumption only holds of the building-blocks of event structure or if it may be carried over to stative verbs as well. In particular, it has not been investigated whether or in what way stative verbs are different from pure stative predicates or from the buildingblocks of event structure. Next, it turned out that there is a persistent assumption that stative verbs have a less complex structure than eventive verbs do. Moreover, several authors assume (e.g. Hale and Keyser 2002) that stative verbs correspond directly to copular constructions. As attractive as this may be, this generalization does not hold, since there are certain stative verbs that are more complex than copular constructions. Related to the assumption that stative verbs are like copular constructions is the view that stative verbs contain only a single structural argument that is the holder of the state, as e.g. put forward by Ramchand 2008. Starting from the distinction between Kimian and Davidsonian statives (cf. Maienborn 2003), which provides a good working definition of “stative verb”, this thesis has shown that the common views about stative verbs have to be questioned. Whereas stative predicates in the lexical-semantic structure of a verb might be the smallest building-blocks of event structure, stative verbs by no means are. In particular, there exists a variety of different stative verbs, all having their distinct lexical-semantic representation. Some of them may involve the cause-operator, thus containing two events/states. In particular, there is no common property that all stative verbs share. In contrast, there is a property that all eventive verbs have in common: they contain a do and/or a become-operator. Thus, in order to display an eventive reading, a verb must either contain an intentional activity by an agent, or it must express a change of state. Of course, it is possible for a verb to express both of them. Therefore, it is rather the eventive reading that is “simple”. It is “simple” in the sense that there are only two grammatical mechanisms that are responsible for generating this meaning. Chapter 3 discussed the systematic stative/eventive ambiguity that arises with certain verbs. In order to do this, the first step was to show that all of those verbs indeed do allow for a stative reading (a Kimian stative reading in the sense of Maienborn

 The Structure of Stative Verbs

2003). The first class displaying this property are verbs that undergo the instrumental alternation such as obstruct. Verbs like obstruct may either have an agentive reading, which is, needless to say, eventive. In this reading, an agent is performing an action intentionally which results in a certain state brought about. In contrast, the stative reading expresses only this resultant state. However, it is not necessary that an action by an agent led to this state, as pointed out by Kratzer 2000. Despite expressing only a state like in the blood vessel is obstructed, these verbs may also express the gradual onset of this resultant state. The reason for the stative/eventive ambiguity is that the verbs contain a cause-operator that relates two situations or two events to one another (the causer and the causee). The stative interpretation arises when both sub-eventualities are stative, hence the absence of the do (usually present in the causing event) and the become (usually expressing the cange of state to the resultant event) operator. In addition to verbs that undergo the instrumental alternation, object-experiencer verbs assigning accusative case like depress and perception verbs (hear vs. listen) are subject to the same ambiguity. Section 3.5 discussed the behavior of dispositional verbs (e.g. help). Following Engelberg 2005, it was shown that these verbs display the systematic stative/eventive ambiguity as well, again due to the fact that two states/events are related within the verb meaning. Moreover, it was shown the dative assignment of dispositional verbs is not specified in the lexical entry of the verb. In other words, the dative assignment of these verbs is not an inherent property. On the contrary, it it turned out that the dative assignment of dispositional verbs is related to the dative assignment of zu-comparatives in German, and to double objects (cf. Brandt 2003, Brandt 2005b). Thus, the dative is licensed via a semantic mechanism that relates the degree of the comparative or the degree of an effect (e.g. a helping-effect) to the personal scale of an affected person. Finally, verbs like threaten display the stative/evenitve ambiguity as well. However, verbs of the threaten-class does not express a causal realtion. The stative reading arises because threaten is understood as a modal operator that scopes over its entire complement clause. The eventive reading, on the other hand, is due to the presence of the do-operator, i.e., in the eventive reading an agent is doing an act of threatening (usually by performing the relevant speech act). Chapter 4 dealt with stative verbs that do not alternate between a stative and an eventive reading in a systematic way. These verbs show a Kimian stative reading only. Again, contrary to the assumptions in the literature, the verbs that allow for a stative reading only do not form a single uniform class. First, there are subject-experiencer and dative-experiencer/possessor verbs which share the property of having an experiencer theta role. On the other hand, measure verbs and verbs that selecta PP-complement express a single property, i.e., they do not relate two autonomous participants of an eventuality. All of the verbs share the property that they express only a single eventuality. Moreover, when discussing subject-experiencer verbs, it turned out that the relation between the presence of an external argument and the presence of an agent does not



Chapter 8.  Conclusion 

hold. Subject-experiencer verbs like love provide the prime example of a non-agentive, stative verb that contains an external argument. Furthermore, verbs that express a single property of their subject, i.e., measure verbs and verbs that select for a PP-complement, have in common that their argument is anchored not via the genuine argument licensing mechanism, but via Semantic Incorporation. Evidence for this structure comes from the fact that measure phrases do not allow for strong determiners. In contrast to genuine arguments, the degree variable in measure verbs (and the corresponding one one for the PP-complement) are existentially bound within the lexical-semantic structure. Chapter 5 explored the nature of verbs of position, which were judged as Davidsonian statives by Maienborn 2003, thus constituting a separate class of stative verbs. It turned out that verbs of position consist of two sub-classes: stative verbs of position and verbs of body posture, the former referring to a Kimian eventuality, the latter denoteing a Davidsonian event. Whereas stative verbs of position simply express the location of their subject, verbs of body posture are in fact activity verbs.Therefore, the class of Davidsonian statives turned out to be superfluous, at least for verbs of position. Chapter 6 examined the next candidates for Davidsonian statives: verbs of internal causation. A closer look revealed that they pass all the test for the presence of an eventive eventuality and that their subjects bear the instrument role. Their lexical-semantic structure is therefore similar to the one of activity verbs. However, activity verbs and verbs of internal causation differ with respect to the intentionality of their subject. Chapter 7 offered a unified account on eventuality structure and argument licensing. The aspectual operators do and cause as well as the basic predicate form the skeleton of a lexical-semantic verb structure, become maybe optionally inserted. predicate introduces the situation argument. This is the distinct feature of the lexical category verb. When building this skeleton, the restriction that do, if present, must be the topmost head has to be obeyed.This fact is due to the idea that do but not cause fully close the (first) verb phase. In the next step, the arguments, bearing the uninterpretable feautures [+c] or [-c] and the interpretable feature [m] are checked against the aspectual heads. Do but not cause is able to check [m], hence animacy/intentionality can only be interpreted at the do-head, thereby excluding sentential subjects. The feature [-c] is checked by predicate. All other arguments are not licensed via the checking of lexical features but are anchored into the lexical-semantic structure via other mechanisms such as Semantic Incorporation or cipient predication. What do the results of the exploration of the nature of stative verbs tell us about the nature of the human faculty of language? What do they say about the architecture of grammar? The investigation of the behavior of Kimian statives suggests strongly that event structure is reflected in grammar. In particular, it seems as if human grammar is designed to create discrete entities within the stative-eventive continuum. There is a distinct class of

 The Structure of Stative Verbs

Kimian stative verbs, and another one for eventive predicates. It is a fundamental property of grammar that there are no gradual transitions between these categories. It turned out that the transition from a Kimian stative reading to an eventive reading is due to two grammatical operators. Thus, it is their presence or their absence that is responsible for the interpretation, not leaving any room for a gradual transition. Finally, note that the difference between Kimian states and Davidsonian events is a purely grammatical one and therefore not reflected in the real world: what should the difference be between someone experiencing love and someone experiencing hunger? In what way does the situation in which a picture is hanging on the wall differ from the one in which a cup is filled with water? It is grammar that categorizes along the lines of different situation types.

References Abraham, W. 1995. Deutsche Syntax im Sprachenvergleich. Grundlegung einer typologischen Syntax des Deutschen. Tübingen: Gunter Narr. Adger, D. 1996. Aspect, agreement and measure phrases in scottish gaelic. In The Syntax of the Celtic Languages, R. Borsley & I. Roberts (eds), 200–222. Cambridge: CUP. Adger, D.& Ramchand, G. 2005. Merge and move: Wh-dependencies revisited. Linguistic Inquiry 36(2): 161–193. Arad, M. 1998a. Psych-notes. UCL Working Papers in Linguistics 10. Arad, M. 1998b. VP-Structure and the Syntax-Lexicon Interface. MIT Working Papers in Linguistics. Arad, M. 1999. On "little v". MIT Working Papers in Linguistics (Papers on Morphology and Syntax, Cycle One) 33: 1–25. Bach, E. 1986. The algebra of events. Linguistics and Philosophy 9(1): 5–16. Baker, M. 1988. Incorporation: A Theory of Grammatical Function Changing. Chicago IL: University of Chicago Press. Baker, M. 1997. Thematic roles and syntactic structure. In Elements of Grammar: Handbook of Generative Syntax, L. Haegeman (ed.), 73–137. Dordrecht: Kluwer. Baker, M. 2003. Lexical Cateogories. Verbs, Nouns and Adjectives. Cambridge: CUP. Belletti, A. & Rizzi, L. 1988. Psych-verbs and theta-theory. Natural Language and Linguistic Theory 6(3): 291–352. Bennis, H. 2004. Unergative adjectives and psych verbs. In The Unaccusativity Puzzle, A. Alexiadou, E. Anagnostopoulou & M. Everaert (eds), 84–113. Oxford: OUP. Bierwisch, M. 1982. Formal and lexical semantics. Linguistische Berichte 80: 3–17. Bierwisch, M. 1987. Semantik der Graduierung. In Grammatische und konzeptuelle Aspekte von Dimensionsadjektiven, M. Bierwisch & E. Lang (eds), 91–286. Berlin: Akademie-Verlag. Bierwisch, M. 1988. On the grammar of local prepositions. In Syntax, Semantik und Lexikon, M. Bierwisch, W. Motsch & I. Zimmermann (eds), 1–65. Berlin: Akademie-Verlag. Bierwisch, M. & Lang, E. 1987. Etwas länger – viel tiefer – immer weiter: Epilog zum Dimensionsadjektiveprojekt. In Grammatische und konzeptuelle Aspekte von Dimensionsadjektiven, M. Bierwisch & E. Lang (eds), 649–699. Berlin: Akademie-Verlag. Borer, H. 1994. The projection of arguments. In University of Massachusetts Occasional Papers 17: Functional Projections, E. Benedicto & J. Runner (eds). Amherst MA: GLSA. Borer, H. 2005. Structuring Sense. Oxford: OUP. Bouchard, D. 1995. The Semantics of Syntax: A Minimalist Approach to Grammar. Chicago IL: University of Chicago Press. Brandt, P. 2003. Cipient Predication: Unifying Double Object, Dative Experiencer and Existential/Presentational Constructions. PhD dissertation, Utrecht University. (LOT Dissertation Series 74). Brandt, P. 2005a. Conditioning cipient datives. Ms, Universität Frankfurt/Main. Brandt, P. 2005b. (No) binding in (German) double object constructions. Generative Grammatik des Südens.

 The Structure of Stative Verbs Bresnan, J. 2001. Lexical-Functional Grammar. Oxford: Blackwell. Butler, J. 2004. Phase Structure, Phrase Structure, and Quantifcation. PhD dissertation, University of York. Carlson, G. 1977. Reference to Kinds in English. PhD dissertation, University of Massachusetts, Amherst. Carston, R. 2002. Metaphor, ad hoc concepts and word meaning – more questions than answers. UCL Working Papers in Linguistics 14: 83–105. Chomsky, N. 1981. Lectures on Government and Binding: The Pisa Lectures. Dordrecht: Foris. Chomsky, N. 1995. The Minimalist Program. Cambridge MA: The MIT Press. Chomsky, N. 2001. Derivation by phase. In Ken Hale. A Life in Language, M. Kenstowicz (ed.), 1–52. Cambridge MA: The MIT Press. Croft, W. 1993. Case marking and the semantics of mental verbs. In Semantics and the Lexicon, J. Pustejovsky (ed.), 255–272. Dordrecht: Kluwer. Czinglar, C. 1997. Bemerkungen zur Existenzbehauptung und Ortsbestimmung im Deutschen und einer Alemannischen Variante. Groninger Arbeiten zur Germanistischen Linguistik 41: 39–59. Davidson, D. 1967. The Logic of Decision and Action. Chapter: The logical form of action sentences, 81–95. Pittsburgh PA: Pittsburgh University Press. de Hoop, H. 1996. Case Configuration and NP Interpretation. New York NY: Garland. Diesing, M. 1992. Indefinites. Cambridge MA: The MIT Press. Dowty, D.R. 1979. Word Meaning and Montague Grammar. The Semantics of Verbs and Times in Generative Semantics and in Montague’s PTQ. Dordrecht: Reidel. Dowty, D.R. 1991. Thematic proto-roles and argument selection. Language 67(3): 547–619. Elbourne, P. 2007. Demonstratives as individual concepts. Ms, Queen Mary, University of London. Engelberg, S. 2000. Verben, Ereignisse und das Lexikon. Tübingen: Niemeyer. Engelberg, S. 2005. Stativity, supervenience, and sentential subjects. In Event Arguments in Syntax, Semantics, and Discourse, C. Maienborn & A. Wöllstein (eds), 45–68. Tübingen: Niemeyer. Fillmore, C.J. 1968. The case for case. In Universals in Linguistic Theory, E. Bach & R. T. Harms (eds), 1–88. New York NY: Holt, Rinehard. Folli, R. & Harley, H. 2005. Consuming results in Italian & English: Flavours of v. In Aspectual Inquiries, P. Kempchinsky & R. Slabakova (eds), 95–120. Dordrecht: Springer. Folli, R. & Harley, H. 2007. Causation, obligation, and argument structure: On the nature of little v. Linguistic Inquiry 38: 197–238. Frey, W.& Pittner, K. 1998. Zur Positionierung der Adverbiale im Deutschen Mittelfeld. Linguistische Berichte 176: 489–534. Geuder, W. 2000. Oriented Adverbs. Issues in the Lexical Semantics of Event Adverbs. PhD dissertation, Universität Tübingen. Goldberg, A. 1995. Constructions: A Construction Grammar Approach to Argument Structure. Chicago IL: University of Chicago Press. Grimshaw, J. 1990. Argument Structure. Cambridge MA: The MIT Press. Gruber, J. S. 1976. Lexical Structures in Syntax and Semantics. Amsterdam: North-Holland. Haiden, M. 2005. Theta Theory. Berlin: Mouton de Gruyter. Hale, K. 1995. Have and be: Linguistic diversity in the expression of simple relations. Presidential address, LSA Annual Meeting. Hale, K. & Keyser, S.J. 1993. On argument structure and the lexical expression of syntactic relations. In The view from building 20, K. Hale & S.J. Keyser (eds), 53–109. Cambridge MA: The MIT Press.



References  Hale, K. & Keyser, S.J. 2002. Prolegomenon to a Theory of Argument Structure. Cambridge MA: The MIT Press. Halle, M. & Marantz, A. 1993. Distributed morphology. In The View from Building 20, K. Hale & S.J. Keyser (eds), 111–176. Cambridge MA: The MIT Press. Hallman, P. 2004. NP-interpretation and the structure of predicates. Language 80(4): 707–747. Harley, H. 1995. Subjects, Events and Licensing. PhD dissertation, MIT. Heim, I. & Kratzer, A. 1998. Semantics in Generative Grammar. Oxford: Blackwell. Helbig, G. & Kemptner, F. 1973. Das Zustandspasssiv. Zur Theorie und Praxis des Deutschunterrichts für Ausländer. Leipzig: VEB Verlag Enzyklopädie. Härtl, H. 2001. Cause und Change: Thematische Rollen und Ereignisstrukturen in Konzeptualisierung und Grammatikalisierung. Berlin: Akademie-Verlag. Isenberg, H. 1968. Das direkte Objekt im Spanischen. Berlin: Akademie-Verlag. Jackendoff, R. 1990. Semantic Structures. Cambridge MA: The MIT Press. Johnson, K. 1985. A Case for Movement. PhD dissertation, MIT. Kallulli, D. 2004. The syntactic visibility of intentionality: Evidence from dyadic unaccusatives. In Proceedings of WECOL 2004. Kallulli, D. 2006. Unaccusatives with dative causers and experiencers: A unified account. In Datives and Similar Cases, D. Hole, A. Meinunger & W. Abraham (eds.), 271–301. Amsterdam: John Benjamins. Kallulli, D. 2007. Rethinking the passive/anticausative distinction. Linguistic Inquiry 38(4): 770–780. Katz, G. 1995. Stativity, Genericity, and Temporal Reference. PhD dissertation, University of Rochester. Katz, G. 2003. On the stativity of the English perfect. In Perfect Explorations, M. Rathert, A. Alexiadou & A. von Stechow (eds), 205–234. Berlin: Mouton de Gruyter. Kaufmann, I. 1995a. Konzeptuelle Grundlagen semantischer Dekompositionsstrukturen. Die Kombinatorik lokaler Verben und prädikativer Komplemente. Tübingen: Niemeyer. Kaufmann, I. 1995b. What is an (im)possible verb? Restrictions on semantic form and their consequences for argument structure. Folia Linguistica 29(1–2): 67–103. Kayne, R. 2008. Antisymmetry and the lexicon. Ms, New York University. Keenan, E. 1996. The semantics of determiners. In The Handbook of Contemporary Semantic Theory, S. Lappin (ed.), 41–63. Oxford: Blackwell. Kim, J. 1969. Events and their descriptions: Some considerations. In Essays in the Honor of Carl G. Hempel, N. Rescher (Ed.), 198–215. Dordrecht: Reidel. Kim, J. 1976. Events as property exemplifications. In Proceedings of the Winnipeg Conference on Human Action, M. Brand & D. Walton (eds), 159–177. Dordrecht: Reidel. Kiparsky, P. 1998. Partitive case and aspect. In The Projection of Arguments: Lexical and Compositional Factors, M. Butt & W. Geuder (eds), 265–307. Stanford CA: CSLI. Kiparsky, P. 2005. Gradience and telicity: The semantics of structural case in Finnish. Workshop zu Ehren von Manfred Bierwisch, Berlin-Brandenburgische Akademie der Wissenschaften. Koopman, H. & Sportiche, D. 1988. Subjects. Ms, UCLA. Kratzer, A. 1981. The notional category of modality. In Words, Worlds, and Contexts, H. Eikmeyer & H. Rieser (eds), 38–74. Berlin: de Gruyter. Kratzer, A. 1991. Modality. In Semantik: Ein internationales Handbuch zeitgenössischer Forschung, A. von Stechow & D. Wunderlich (eds), 639–650. Berlin: de Gruyter. Kratzer, A. 1995. Stage-level and individual-level predicates. In The Generic Book, N.G. Carlson & J. Pelletier (Eds.). Chicago IL: Chicago University Press.

 The Structure of Stative Verbs Kratzer, A. 1996. Severing the external argument from its verb. In Phrase Structure and the Lexicon, J. Rooryck & L. Zaring (eds), 109–138. Dordrecht: Kluwer. Kratzer, A. 2000. Building statives. Berkeley Linguistic Society 26. Kratzer, A. 2002. Telicity and the meaning of objective case. Ms, University of Massachusetts at Amherst. Krifka, M. 1989. Nominalreferenz und Zeitkonstitution: Zur Semantik von Massentermen, Pluraltermen und Aktionsarten. München: Wilhelm Fink. Krifka, M., Pelletier, F.J., Carlson, G.N., ter Meulen, A., Link, G. & Chierchia, G. 1995. Genericity: An Introduction. In The Generic Book, N. G. Carlson & J. Pelletier (eds), 1–124. Chicago IL: Chicago University Press. Landau, I. 2005. The locative syntax of experiencers. Ms, Ben Gurion University. Larson, R. 1988. On the double object construction. Linguistic Inquiry 19(3): 335–391. Levin, B. 1993. English Verb Classes and Alternations: A Preliminary Investigation. Chicago IL: University of Chicago Press. Levin, B. 1999. Objecthood: An event structure perspective. CLS 35. Levin, B. & Rappaport Hovav, M. 1995. Unaccusativity. Cambridge MA: The MIT Press. Levin, B. & Rappaport Hovav, M. 2005. Argument Realization. Cambridge: CUP. Lewis, D. 1973. Counterfactuals. Cambridge MA: Harvard University Press. Mador-Haim, S. & Winter, Y. 2007. Non-existential indefinites and semantic incorporation of PP-complements. In Proceedings of SALT 17, M. Gibson & T. Friedman (eds). Maienborn, C. 1996. Situation und Lokation. Die Bedeutung lokaler Adjunkte von Verbalprojektionen. Tübingen: Stauffenburg. Maienborn, C. 2001. On the position and interpretation of locative modifiers. Natural Language Semantics 9: 191–240. Maienborn, C. 2003. Die logische Form von Kopula-Sätzen. Berlin: Akademie-Verlag. Maienborn, C. 2005a. A discourse-based account of Spanish ser/estar. Linguistics 43(1): 155–180. Maienborn, C. 2005b. On the limits of the Davidsonian approach: The case of copula sentences. Theoretical Linguistics 31(3): 275–316. Maienborn, C. 2007a. Das Zustandspassiv: Grammatische Einordnung – Bildungsbeschränkungen – Interpretationsspielraum. Zeitschrift für Germanistische Linguistik 35(1): 83–114. Maienborn, C. 2007b. On Davidsonian and Kimian states. In Existence: Semantics and Syntax, I. Comorovski & K. von Heusinger (eds), 107–130. Dordrecht: Springer. Maienborn, C. 2007c. Review of Regina Pustet. Copulas. Universals in the Categorization of the Lexicon. Studies in Language 31(1): 232–244. Marantz, A. 1984. On the Nature of Grammatical Relations. Cambridge MA: The MIT Press. Marantz, A. 1997. No escape from syntax: Don’t try morphological analysis in the privacy of your own lexicon. UPenn Working Papers in Linguistics 4(2): 201–225. Marantz, A. 2001. Words. Ms, MIT. McClure, W. 1995. Aspect and direct objects in Japanese. In WECOL 94, 164–180. Milsark, G. 1974. Existential Sentences in English. PhD dissertation, MIT. Möller, M. 2004. Psychische Wirkungsverben des Deutschen. Diplomarbeit, Humboldt Universität zu Berlin. Noonan, M. & Grunow-Härsta, K. 2002. Posture verbs in two Tibeto-Burman languages of Nepal. In The Linguistics of Sitting, Standing, and Lying, J. Newman (ed.), 79–102. Amsterdam: John Benjamins. Parsons, T. 1990. Events in the Semantics of English. Cambridge MA: The MIT Press.



References  Partee, B. 1987. Noun phrase interpretation and type-shifting principles. In Studies in Discourse Representation Theory and The Theory of Generalized Quantifiers, J. Groenendijk, D. de Jongh & M. Stokhof (eds), 115–143. Dordrecht: Foris. Perlmutter, D. 1978. Impersonal passives and the unaccusative hypothesis. Berkeley Linguistics Society 4: 157–189. Pesetsky, D. 1995. Zero Syntax: Experiences and Cascades. Cambridge MA: The MIT Press. Pesetsky, D. & Torrego, E. 2004. Tense, Case, and the nature of syntactic categories. In The Syntax of Time, J. Gueron & J. Lecarme (eds), 495–538. Cambridge MA: The MIT Press. Postal, P. 1971. Cross-Over Phenomena. New York NY: Holt, Rinehart, and Winston. Prinzhorn, M. 1990. Head movement and scrambling domains. In Scrambling and Barriers, G. Grewendorf & W. Sternefeld (eds), 199–215. Amsterdam: John Benjamins. Prinzhorn, M. 2005. Metaphern in der Grammatik. Presentation at Metapher in Alltag, Philosophie und Wissenschaft, Wien. Pustejovsky, J. 1991. The syntax of event structure. Cognition 41: 47–81. Pustet, R. 2003. Copulas. Universals in the Categorization of the Lexicon. Oxford: OUP. Pylkkänen, L. 2000. On stativity and causation. In Events as Grammatical Objects, C. Tenny & J. Pustejovsky (eds), 417–444. Stanford CA: CSLI. Ramchand, G. 2008. Verb Meaning and the Lexicon. A First Phase Syntax. Cambridge: CUP. Rapp, I. 1997. Partizipien und semantische Struktur. Zu passivischen Konstruktionen mit dem 3. Status. Tübingen: Stauffenburg. Rapp, I. 1998. Zustand? Passiv? – Überlegungen zum sogenannten “Zustandspassiv”. Zeitschrift für Sprachwissenschaft 15(2): 231–265. Rappaport Hovav, M. & Levin, B. 1998. Building verb meanings. In The Projection of Arguments:Lexical and Compositional Factors, M. Butt & W. Geuder (eds), 97–134. Stanford CA: CSLI. Reinhart, T. 2000. The theta system: Syntactic realization of verbal concepts. Ms, Tel Aviv University/Utrecht University. Reinhart, T. 2001a. Experiencing derivations. SALT 11. Reinhart, T. 2001b. A synopsis of the theta system. Ms, Tel Aviv University/Utrecht University. Ritter, E. & Rosen, S. 1998. Delimiting events in syntax. In The Projection of Arguments:Lexical and Compositional Factors, M. Butt & W. Geuder (eds), 135–164. Stanford CA: CSLI. Rivero, M.-L. & Savchenko, U. 2004. Russian anticausatives with oblique subjects. In Formal Approaches to Slavic Linguistics 13, The South Carolina Meeting, S. Franks, F. Gladney & M. Tasseva-Kurktchieva (eds), 276–288. Ann Arbor MI: Michigan Slavic Publications. Rizzi, L. 1990. Relativized Minimality. Cambridge MA: The MIT Press. Rogers, A. 1971. Three kinds of physical perception verbs. CLS 7: 206–222. Ruprecht, A. 2005. Grundfragen der Sprachwissenschaft. Nach Prof. Wolfgang U. Dressler vidiertes Skriptum. Wien: WUV-Universitätsverlag. Ruwet, N. 1991. Syntax and Human Experience. Chicago IL: University of Chicago Press. Smith, C.S. 1997. The Parameter of Aspect, 2nd edn. Dordrecht: Kluwer. Tenny, C.L. 1994. Aspectual Roles and the Syntax-Semantics Interface. Dordrecht: Kluwer. Torrego, E. 1998. The Dependencies of Objects. Cambridge MA: The MIT Press. Travis, L. 1994. Event phrase and a theory of functional categories. Canadian Linguistics Association. Toronto Working Papers in Linguistics, 559–570. van Geenhoven, V. 1998a. On the argument structure of some noun incorporating verbs in West Greenlandic. In The Projection of Arguments: Lexical and Compositional Factors, M. Butt and W. Geuder (Eds.), 225–263. Stanford CA: CSLI.

 The Structure of Stative Verbs van Geenhoven, V. 1998b. Semantic Incorporation and Indefinite Descriptions: Semantic and Syntactic Aspects of West Greenlandic Noun Incorporation. Stanford CA: CSLI. van Hout, A. 1998. Event Semantics of Verb Frame Alternations. New York NY: Garland. van Valin Jr., R.D. 1990. Semantic patterns of split-intransitivity. Language 66(2): 221–260. Vendler, Z. 1957. Verbs and times. The Philosophiocal Review 66(2): 143–160. von Stechow, A. 1995. Lexical decomposition in syntax. In Lexical Knowledge in the Organization of Language, U. Egli, P.E. Pause, C. Schwarze, A. von Stechow & G. Wienold (eds), 81–118. Amsterdam: John Benjamins. von Stechow, A. 1996. The different readings of Wieder “again”: A structural account. Journal of Semantics 13(2): 87–138. von Stechow, A. 2004. Modalität. Vorlesung, Universität Wien. von Stechow, A. 2007. Syntactic and lexical causativization: BECOME and CAUSE again. Presentation at ConSOLE XV, Brussels. Wechsler, S. 1995. The Semantic Basis of Argument Structure. Stanford CA: CSLI. Williams, E. 1981. Argument structure and morphology. The Linguistic Review 1: 81–114. Williams, E. 2003. Representation Theory. Cambridge Ma: The MIT Press. Wunderlich, D. 1997. Cause and the structure of verbs. Linguistic Inquiry 28(1): 27–68. Wunderlich, D. & Herweg, M. 1991. Lokale und Direktionale. In Semantik: Ein internationalesHandbuch der zeitgenössischen Forschung, A. von Stechow & D. Wunderlich (eds), 758–785. Berlin: de Gruyter. Wunderlich, D. & Kaufmann, I. 1990. Lokale Verben und Präpositionen-semantische und konzeptuelle Aspekte. In Sprache und Wissen. Studien zur kognitiven Linguistik, S. Felix, S. Kanngießer & G. Rickkeit (eds), 222–252. Opladen: Westdeutscher Verlag. Zimmermann, T.E. 1993. On the proper treatment of opacity in certain verbs. Natural Language Semantics 1(2): 149–179. Zubizarreta, M.L. (1982). On the Relationship of the Lexicon to Syntax. PhD dissertation, MIT.

Author index A Adger  9 Arad  49, 53–60, 68–72 B Bach  4 Baker  178 Belletti  53–54, 58–59, 124, 130 Bierwisch  8, 133, 151 Borer  10, 23–25 Brandt  95–100 Butler  9–10 C Chomsky  177, 191 Czinglar  151 D Davidson  4–5, 29, 196–197 de Hoop  138–139 Diesing  19 Dowty  4, 43–44, 48–49, 100–101, 180–181 E Elbourne  138 Engelberg  80–84, 86–89, 93–96 F Folli  22, 48–49, 175 Frey  115–116 G Grunow-Härsta  158–159 H Haiden  123–124, 163–164, 171, 174

Hale  5–6, 11–13, 25, 179 Hallman  19–21 Harley  16, 22–23, 25, 48, 174–175 Härtl  120–121 I Isenberg  113 J Jackendoff  176 K Kallulli  173–175 Katz  36 Kaufmann  44–46, 153 Kayne  183 Keyser  5–6, 11–13, 25, 179 Kim  6–8 Kiparsky  110–112 Kratzer  7, 16–18, 21–22, 37–38, 76–77, 109, 112

P Parsons  5 Pesetsky  39, 59, 63, 113 Pittner  115–116 Prinzhorn  68–70, 189 Pustejovsky  6 Pustet  179 Pylkkänen  56–58, 112 R Ramchand  9, 25–27 Rapp  117–119, 121 Rappaport  10, 14–15, 27, 131, 147–148, 153–154, 161–163 Reinhart  163, 171, 174 Rizzi  53–54, 58–59, 124, 130, 134–136 Ruprecht  131 S Smith  101

L Levin  10–11, 14–15, 27, 131, 147–148, 153–154, 161–163

T Tenny  50 Torrego  39, 113–114

M Mador-Haim  143 Maienborn  7–8, 28–34, 41, 150, 152, 154, 161–162, 179–180, 195–197 Marantz  16, 21 Milsark  134 Möller  118

V van Geenhoven  136–137 Vendler  3 von Stechow  76–77, 175–177, 180–181

N Noonan  158–159

W Williams  194 Wunderlich  8, 43–46, 78, 195 

Subject index A accomplishment  3, 15, 110 accomplishment verb  95, 98 achievement  3, 15, 101, 110 activity  3, 14, 24, 30, 45, 47, 51, 75, 80, 89, 100, 131, 134, 150, 152, 156, 161, 163, 170, 171, 173 f., 179, 182, 184, 187, 190 adjectival passive  35–37 adjective  12, 16, 22, 24, 84, 136, 179 adverbial manner adverbial  5, 28, 31 f., 40 f., 60–62, 73, 82–84, 92, 95, 102 f., 113–116, 124–126, 131 f., 140 f., 148–150, 154, 162, 185, 196, 200 f. locative adverbial  33, 62, 74 f., 85–87, 103, 126 f., 150 f. agent  3, 5, 11, 21, 38–40, 45, 48 f., 54, 65 f., 68, 70 f., 99, 122, 136, 145, 164, 173, 175, 181, 184–188, 190, 192 194, 197 f. agentivity  56 agree  9 argument introduction  9 aspect  11, 21, 78 atelic  24 B Bavarian  30 C case assignment  13, 80, 110 f. causation  43–47, 173, 180 external  163 internal  161–172 stative  38, 43–47, 52, 58, 65, 70 f., 107, 175 causer  25, 45, 48 f., 59, 65, 174, 188 stative see stative causation

change of state  6, 12–15, 54–56, 152, 175 complex predicate  143, 157 conceptual structure, CS  8 control  69–71, 75 f. copula  151, 153, 178 copular construction  7 f., 13, 27, 149–152, 159, 179, 195 counterfactual  43, 180 culmination  4 D Davidsonian argument  17, 24, 31, 33, 196 f. Davidsonian stative (verb)  6–8, 28–30, 32–33, 42, 154, 161 f. dative dative case  27, 53, 56, 80, 94 f., 97, 100, 124, 178 dative causer  23 dative experiencer  124–131 dative object  95, 120, 124 dative possessor  17, 129 f. decomposition  11, 25, 176 drohen  69, 72, 74–80 dynamic  4 E event argument  5–7, 9, 21 f., 24 f., 29, 31–34, 37, 75, 82, 132, 161, 195–197 event semantics  4, 14 event structure  3, 6, 10 f., 14, 24–28, 44–48, 59 f., 65, 88–91, 121, 123, 130, 147, 170, 187, 197 f. experiencer verb see object-experiencer verb (ACC), object-experiencer verb (DAT), subject-experiencer verb F feature [+act] feature  173 f. [+c] feature  174, 184–190

head feature  23 [+intent] feature  173 f. lexical feature  176, 187–191 [+m] feature  123, 163, 171, 174 f., 184–198 [telic] feature  112 tense feature  176, 194 theta feature  123, 163, 174–177 feature valuation  9 Finnish  56–58, 110–112, 139 functional application  181 functional head  10, 16, 21–23, 49, 177 f., 187, 191, 195 G grammatical feature  170 H helfen  63 f., 80–107 I ILP see also individual level predicate  7 imperfective  101 individual level predicate see also ILP  7 f., 16–21, 57 inference  45, 103 Italian  53–56 K K-state  29 Kimian state  7 f., 29, 34, 180, 197 L lexical category  13, 16, 183 little v  6, 9, 11, 16, 20–26, 48 f., 60, 70 f., 79, 109, 124, 173–175 logical form (LF)  8 M measure phrase  13, 131–139, minimalism  9 f., 49, 175–177, 187 Minimalist Program  177, 187, 198 modal operator  75–80

 The Structure of Stative Verbs modal verb  76, 78, 199 modality  76 f. N Neo-Davidsonian view  5, 197 nominalization  30–34, 121 f. O ontology  28 f. P partitive case  57, 110–112 partitive construction  39, 133 passive  10, 13, 22, 55, 121–124 perfect  35 f. perfect form  118 perfect tense  35 phase  191, 198 possible world  77, 195 f. Post-Davidsonian view  5 f., 25–27 PP-complement  28, 140- 144 predication  177 f. presupposition  96, 197 Principle of Full Interpretation  17 process  4, 6, 25 f., 117, 182 process-oriented adverbial  115 progressive  4, 36, 113 property exemplification  177– 180, 196 proposition  43–49, 75–78, 120, 180 f., 193–195 R Representation Theory  194

resultant state  25, 38, 40 f., 45, 47 f., 51 f., 56, 63, 65, 83, 92, 95, 98–101, 118, 158, 175–178 Russian  101, 106 S Semantic Form  8–10, 43 f., 77, 122, 195 sensation predicate  199–202 sentential subject  44–47, 65, 80, 82, 88, 191, 194 simple predicate  11, 45, 105 situation argument  153, 177–181, 195–197 small clause  175 Spanish  8, 39, 56, 113 f. stage level predicate (SLP)  7 f., 16–21 starting phase  117–121 stativity  13, 24 sub-event  5 f., 11 f., 182, 197 subinterval property  3 T target state  37, 63–65, 89–94 telicity  24 tense argument  180 theta role  11, 25, 43, 59, 6, 124, 135 f., 163, 176 Tibeto-Burman  158 f. U unaccusative  16 f., 22 f., 55, 58 f., 70, 130, 148 V verb activity verb

see activity of body posture  153–159 of cognitive activity  142 of consuption  174 f., 182–192 dispositional verb  28, 80–100 of external causation  162 of light emission  163–168 of location  143, 158 f. measure verb  27, 104, 131–139, 142 object experiencer verb (ACC)  28, 52–68, 81, 94, 182, 188, 191 object experiencer verb (DAT)  27, 53, 124–13, 178 perception verb  27, 29, 31, 100–107, 112, 119, 129, 142, 162, 197 PP-complement verb  140– 144 of possession  109–124, 124–131, 175 of smell emission  168 f. of sound emission  163–166 subject experiencer verb  19, 27 f., 53, 56–58, 109–124 verb phase  187–194 verbal phase see verb phase voice  21 f. vP  22, 25, 71, 98 VP  5, 11, 17 f., 21 f., 24, 26, 47, 59, 70, 98, 100, 106, 194

Linguistik Aktuell/Linguistics Today A complete list of titles in this series can be found on the publishers’ website, www.benjamins.com 146 Gelderen, Elly van (ed.): Cyclical Change. viii, 319 pp. + index. Expected July 2009 145 Westergaard, Marit: The Acquisition of Word Order. Micro-cues, information structure, and economy. xii, 242 pp. + index. Expected July 2009 144 Putnam, Michael T. (ed.): Towards a Derivational Syntax. Survive-minimalism. x, 264 pp. + index. Expected August 2009 143 Rothmayr, Antonia: The Structure of Stative Verbs. 2009. xv, 216 pp. 142 Nunes, Jairo (ed.): Minimalist Essays on Brazilian Portuguese Syntax. 2009. vi, 243 pp. 141 Alexiadou, Artemis, Jorge Hankamer, Thomas McFadden, Justin Nuger and Florian Schäfer (eds.): Advances in Comparative Germanic Syntax. 2009. xv, 395 pp. 140 Roehrs, Dorian: Demonstratives and Definite Articles as Nominal Auxiliaries. 2009. xii, 196 pp. 139 Hicks, Glyn: The Derivation of Anaphoric Relations. 2009. xii, 309 pp. 138 Siddiqi, Daniel: Syntax within the Word. Economy, allomorphy, and argument selection in Distributed Morphology. 2009. xii, 138 pp. 137 Pfau, Roland: Grammar as Processor. A Distributed Morphology account of spontaneous speech errors. 2009. xiii, 372 pp. 136 Kandybowicz, Jason: The Grammar of Repetition. Nupe grammar at the syntax–phonology interface. 2008. xiii, 168 pp. 135 Lewis, William D., Simin Karimi, Heidi Harley and Scott O. Farrar (eds.): Time and Again. Theoretical perspectives on formal linguistics. In honor of D. Terence Langendoen. 2009. xiv, 265 pp. 134 Armon-Lotem, Sharon, Gabi Danon and Susan Rothstein (eds.): Current Issues in Generative Hebrew Linguistics. 2008. vii, 393 pp. 133 MacDonald, Jonathan E.: The Syntactic Nature of Inner Aspect. A minimalist perspective. 2008. xv, 241 pp. 132 Biberauer, Theresa (ed.): The Limits of Syntactic Variation. 2008. vii, 521 pp. 131 De Cat, Cécile and Katherine Demuth (eds.): The Bantu–Romance Connection. A comparative investigation of verbal agreement, DPs, and information structure. 2008. xix, 355 pp. 130 Kallulli, Dalina and Liliane Tasmowski (eds.): Clitic Doubling in the Balkan Languages. 2008. ix, 442 pp. 129 Sturgeon, Anne: The Left Periphery. The interaction of syntax, pragmatics and prosody in Czech. 2008. xi, 143 pp. 128 Taleghani, Azita H.: Modality, Aspect and Negation in Persian. 2008. ix, 183 pp. 127 Durrleman-Tame, Stephanie: The Syntax of Jamaican Creole. A cartographic perspective. 2008. xii, 190 pp. 126 Schäfer, Florian: The Syntax of (Anti-)Causatives. External arguments in change-of-state contexts. 2008. xi, 324 pp. 125 Rothstein, Björn: The Perfect Time Span. On the present perfect in German, Swedish and English. 2008. xi, 171 pp. 124 Ihsane, Tabea: The Layered DP. Form and meaning of French indefinites. 2008. ix, 260 pp. 123 Stoyanova, Marina: Unique Focus. Languages without multiple wh-questions. 2008. xi, 184 pp. 122 Oosterhof, Albert: The Semantics of Generics in Dutch and Related Languages. 2008. xviii, 286 pp. 121 Tungseth, Mai Ellin: Verbal Prepositions and Argument Structure. Path, place and possession in Norwegian. 2008. ix, 187 pp. 120 Asbury, Anna, Jakub Dotlačil, Berit Gehrke and Rick Nouwen (eds.): Syntax and Semantics of Spatial P. 2008. vi, 416 pp. 119 Fortuny, Jordi: The Emergence of Order in Syntax. 2008. viii, 211 pp. 118 Jäger, Agnes: History of German Negation. 2008. ix, 350 pp. 117 Haugen, Jason D.: Morphology at the Interfaces. Reduplication and Noun Incorporation in Uto-Aztecan. 2008. xv, 257 pp. 116 Endo, Yoshio: Locality and Information Structure. A cartographic approach to Japanese. 2007. x, 235 pp. 115 Putnam, Michael T.: Scrambling and the Survive Principle. 2007. x, 216 pp.

114 Lee-Schoenfeld, Vera: Beyond Coherence. The syntax of opacity in German. 2007. viii, 206 pp. 113 Eythórsson, Thórhallur (ed.): Grammatical Change and Linguistic Theory. The Rosendal papers. 2008. vi, 441 pp. 112 Axel, Katrin: Studies on Old High German Syntax. Left sentence periphery, verb placement and verbsecond. 2007. xii, 364 pp. 111 Eguren, Luis and Olga Fernández Soriano (eds.): Coreference, Modality, and Focus. Studies on the syntax–semantics interface. 2007. xii, 239 pp. 110 Rothstein, Susan (ed.): Theoretical and Crosslinguistic Approaches to the Semantics of Aspect. 2008. viii, 453 pp. 109 Chocano, Gema: Narrow Syntax and Phonological Form. Scrambling in the Germanic languages. 2007. x, 333 pp. 108 Reuland, Eric, Tanmoy Bhattacharya and Giorgos Spathas (eds.): Argument Structure. 2007. xviii, 243 pp. 107 Corver, Norbert and Jairo Nunes (eds.): The Copy Theory of Movement. 2007. vi, 388 pp. 106 Dehé, Nicole and Yordanka Kavalova (eds.): Parentheticals. 2007. xii, 314 pp. 105 Haumann, Dagmar: Adverb Licensing and Clause Structure in English. 2007. ix, 438 pp. 104 Jeong, Youngmi: Applicatives. Structure and interpretation from a minimalist perspective. 2007. vii, 144 pp. 103 Wurff, Wim van der (ed.): Imperative Clauses in Generative Grammar. Studies in honour of Frits Beukema. 2007. viii, 352 pp. 102 Bayer, Josef, Tanmoy Bhattacharya and M.T. Hany Babu (eds.): Linguistic Theory and South Asian Languages. Essays in honour of K. A. Jayaseelan. 2007. x, 282 pp. 101 Karimi, Simin, Vida Samiian and Wendy K. Wilkins (eds.): Phrasal and Clausal Architecture. Syntactic derivation and interpretation. In honor of Joseph E. Emonds. 2007. vi, 424 pp. 100 Schwabe, Kerstin and Susanne Winkler (eds.): On Information Structure, Meaning and Form. Generalizations across languages. 2007. vii, 570 pp. 99 Martínez-Gil, Fernando and Sonia Colina (eds.): Optimality-Theoretic Studies in Spanish Phonology. 2007. viii, 564 pp. 98 Pires, Acrisio: The Minimalist Syntax of Defective Domains. Gerunds and infinitives. 2006. xiv, 188 pp. 97 Hartmann, Jutta M. and László Molnárfi (eds.): Comparative Studies in Germanic Syntax. From Afrikaans to Zurich German. 2006. vi, 332 pp. 96 Lyngfelt, Benjamin and Torgrim Solstad (eds.): Demoting the Agent. Passive, middle and other voice phenomena. 2006. x, 333 pp. 95 Vogeleer, Svetlana and Liliane Tasmowski (eds.): Non-definiteness and Plurality. 2006. vi, 358 pp. 94 Arche, María J.: Individuals in Time. Tense, aspect and the individual/stage distinction. 2006. xiv, 281 pp. 93 Progovac, Ljiljana, Kate Paesani, Eugenia Casielles and Ellen Barton (eds.): The Syntax of Nonsententials. Multidisciplinary perspectives. 2006. x, 372 pp. 92 Boeckx, Cedric (ed.): Agreement Systems. 2006. ix, 346 pp. 91 Boeckx, Cedric (ed.): Minimalist Essays. 2006. xvi, 399 pp. 90 Dalmi, Gréte: The Role of Agreement in Non-Finite Predication. 2005. xvi, 222 pp. 89 Velde, John R. te: Deriving Coordinate Symmetries. A phase-based approach integrating Select, Merge, Copy and Match. 2006. x, 385 pp. 88 Mohr, Sabine: Clausal Architecture and Subject Positions. Impersonal constructions in the Germanic languages. 2005. viii, 207 pp. 87 Julien, Marit: Nominal Phrases from a Scandinavian Perspective. 2005. xvi, 348 pp. 86 Costa, João and Maria Cristina Figueiredo Silva (eds.): Studies on Agreement. 2006. vi, 285 pp. 85 Mikkelsen, Line: Copular Clauses. Specification, predication and equation. 2005. viii, 210 pp. 84 Pafel, Jürgen: Quantifier Scope in German. 2006. xvi, 312 pp. 83 Schweikert, Walter: The Order of Prepositional Phrases in the Structure of the Clause. 2005. xii, 338 pp. 82 Quinn, Heidi: The Distribution of Pronoun Case Forms in English. 2005. xii, 409 pp. 81 FuSS, Eric: The Rise of Agreement. A formal approach to the syntax and grammaticalization of verbal inflection. 2005. xii, 336 pp.

80 Burkhardt, Petra: The Syntax–Discourse Interface. Representing and interpreting dependency. 2005. xii, 259 pp. 79 Schmid, Tanja: Infinitival Syntax. Infinitivus Pro Participio as a repair strategy. 2005. xiv, 251 pp. 78 Dikken, Marcel den and Christina Tortora (eds.): The Function of Function Words and Functional Categories. 2005. vii, 292 pp. 77 Öztürk, Balkız: Case, Referentiality and Phrase Structure. 2005. x, 268 pp. 76 Stavrou, Melita and Arhonto Terzi (eds.): Advances in Greek Generative Syntax. In honor of Dimitra Theophanopoulou-Kontou. 2005. viii, 366 pp. 75 Di Sciullo, Anna Maria (ed.): UG and External Systems. Language, brain and computation. 2005. xviii, 398 pp. 74 Heggie, Lorie and Francisco Ordóñez (eds.): Clitic and Affix Combinations. Theoretical perspectives. 2005. viii, 390 pp. 73 Carnie, Andrew, Heidi Harley and Sheila Ann Dooley (eds.): Verb First. On the syntax of verbinitial languages. 2005. xiv, 434 pp. 72 FuSS, Eric and Carola Trips (eds.): Diachronic Clues to Synchronic Grammar. 2004. viii, 228 pp. 71 Gelderen, Elly van: Grammaticalization as Economy. 2004. xvi, 320 pp. 70 Austin, Jennifer R., Stefan Engelberg and Gisa Rauh (eds.): Adverbials. The interplay between meaning, context, and syntactic structure. 2004. x, 346 pp. 69 Kiss, Katalin É. and Henk van Riemsdijk (eds.): Verb Clusters. A study of Hungarian, German and Dutch. 2004. vi, 514 pp. 68 Breul, Carsten: Focus Structure in Generative Grammar. An integrated syntactic, semantic and intonational approach. 2004. x, 432 pp. 67 Mišeska Tomić, Olga (ed.): Balkan Syntax and Semantics. 2004. xvi, 499 pp. 66 Grohmann, Kleanthes K.: Prolific Domains. On the Anti-Locality of movement dependencies. 2003. xvi, 372 pp. 65 Manninen, Satu Helena: Small Phrase Layers. A study of Finnish Manner Adverbials. 2003. xii, 275 pp. 64 Boeckx, Cedric and Kleanthes K. Grohmann (eds.): Multiple Wh-Fronting. 2003. x, 292 pp. 63 Boeckx, Cedric: Islands and Chains. Resumption as stranding. 2003. xii, 224 pp. 62 Carnie, Andrew, Heidi Harley and MaryAnn Willie (eds.): Formal Approaches to Function in Grammar. In honor of Eloise Jelinek. 2003. xii, 378 pp. 61 Schwabe, Kerstin and Susanne Winkler (eds.): The Interfaces. Deriving and interpreting omitted structures. 2003. vi, 403 pp. 60 Trips, Carola: From OV to VO in Early Middle English. 2002. xiv, 359 pp. 59 Dehé, Nicole: Particle Verbs in English. Syntax, information structure and intonation. 2002. xii, 305 pp. 58 Di Sciullo, Anna Maria (ed.): Asymmetry in Grammar. Volume 2: Morphology, phonology, acquisition. 2003. vi, 309 pp. 57 Di Sciullo, Anna Maria (ed.): Asymmetry in Grammar. Volume 1: Syntax and semantics. 2003. vi, 405 pp. 56 Coene, Martine and Yves D’hulst (eds.): From NP to DP. Volume 2: The expression of possession in noun phrases. 2003. x, 295 pp. 55 Coene, Martine and Yves D’hulst (eds.): From NP to DP. Volume 1: The syntax and semantics of noun phrases. 2003. vi, 362 pp. 54 Baptista, Marlyse: The Syntax of Cape Verdean Creole. The Sotavento varieties. 2003. xxii, 294 pp.  (incl. CD-rom). 53 Zwart, Jan-Wouter and Werner Abraham (eds.): Studies in Comparative Germanic Syntax. Proceedings from the 15th Workshop on Comparative Germanic Syntax (Groningen, May 26–27, 2000). 2002. xiv, 407 pp. 52 Simon, Horst J. and Heike Wiese (eds.): Pronouns – Grammar and Representation. 2002. xii, 294 pp. 51 Gerlach, Birgit: Clitics between Syntax and Lexicon. 2002. xii, 282 pp. 50 Steinbach, Markus: Middle Voice. A comparative study in the syntax-semantics interface of German. 2002. xii, 340 pp. 49 Alexiadou, Artemis (ed.): Theoretical Approaches to Universals. 2002. viii, 319 pp. 48 Alexiadou, Artemis, Elena Anagnostopoulou, Sjef Barbiers and Hans-Martin Gärtner (eds.): Dimensions of Movement. From features to remnants. 2002. vi, 345 pp.

47 Barbiers, Sjef, Frits Beukema and Wim van der Wurff (eds.): Modality and its Interaction with the Verbal System. 2002. x, 290 pp. 46 Panagiotidis, E. Phoevos: Pronouns, Clitics and Empty Nouns. ‘Pronominality’ and licensing in syntax. 2002. x, 214 pp. 45 Abraham, Werner and Jan-Wouter Zwart (eds.): Issues in Formal German(ic) Typology. 2002. xviii, 336 pp. 44 Taylan, Eser Erguvanlı (ed.): The Verb in Turkish. 2002. xviii, 267 pp. 43 Featherston, Sam: Empty Categories in Sentence Processing. 2001. xvi, 279 pp. 42 Alexiadou, Artemis: Functional Structure in Nominals. Nominalization and ergativity. 2001. x, 233 pp. 41 Zeller, Jochen: Particle Verbs and Local Domains. 2001. xii, 325 pp. 40 Hoeksema, Jack, Hotze Rullmann, Víctor Sánchez-Valencia and Ton van der Wouden (eds.): Perspectives on Negation and Polarity Items. 2001. xii, 368 pp. 39 Gelderen, Elly van: A History of English Reflexive Pronouns. Person, Self, and Interpretability. 2000. xiv, 279 pp. 38 Meinunger, André: Syntactic Aspects of Topic and Comment. 2000. xii, 247 pp. 37 Lutz, Uli, Gereon Müller and Arnim von Stechow (eds.): Wh-Scope Marking. 2000. vi, 483 pp. 36 Gerlach, Birgit and Janet Grijzenhout (eds.): Clitics in Phonology, Morphology and Syntax. 2001. xii, 441 pp. 35 Hróarsdóttir, Thorbjörg: Word Order Change in Icelandic. From OV to VO. 2001. xiv, 385 pp. 34 Reuland, Eric (ed.): Arguments and Case. Explaining Burzio’s Generalization. 2000. xii, 255 pp. 33 Puskás, Genoveva: Word Order in Hungarian. The syntax of Ā-positions. 2000. xvi, 398 pp. 32 Alexiadou, Artemis, Paul Law, André Meinunger and Chris Wilder (eds.): The Syntax of Relative Clauses. 2000. vi, 397 pp. 31 Svenonius, Peter (ed.): The Derivation of VO and OV. 2000. vi, 372 pp. 30 Beukema, Frits and Marcel den Dikken (eds.): Clitic Phenomena in European Languages. 2000. x, 324 pp. 29 Miyamoto, Tadao: The Light Verb Construction in Japanese. The role of the verbal noun. 2000. xiv, 232 pp. 28 Hermans, Ben and Marc van Oostendorp (eds.): The Derivational Residue in Phonological Optimality Theory. 2000. viii, 322 pp. 27 Růžička, Rudolf: Control in Grammar and Pragmatics. A cross-linguistic study. 1999. x, 206 pp. 26 Ackema, Peter: Issues in Morphosyntax. 1999. viii, 310 pp. 25 Felser, Claudia: Verbal Complement Clauses. A minimalist study of direct perception constructions. 1999. xiv, 278 pp. 24 Rebuschi, Georges and Laurice Tuller (eds.): The Grammar of Focus. 1999. vi, 366 pp. 23 Giannakidou, Anastasia: Polarity Sensitivity as (Non)Veridical Dependency. 1998. xvi, 282 pp. 22 Alexiadou, Artemis and Chris Wilder (eds.): Possessors, Predicates and Movement in the Determiner Phrase. 1998. vi, 388 pp. 21 Klein, Henny: Adverbs of Degree in Dutch and Related Languages. 1998. x, 232 pp. 20 Laenzlinger, Christopher: Comparative Studies in Word Order Variation. Adverbs, pronouns, and clause structure in Romance and Germanic. 1998. x, 371 pp. 19 Josefsson, Gunlög: Minimal Words in a Minimal Syntax. Word formation in Swedish. 1998. ix, 199 pp. 18 Alexiadou, Artemis: Adverb Placement. A case study in antisymmetric syntax. 1997. x, 256 pp. 17 Beermann, Dorothee, David LeBlanc and Henk van Riemsdijk (eds.): Rightward Movement. 1997. vi, 410 pp. 16 Liu, Feng-hsi: Scope and Specificity. 1997. viii, 187 pp. 15 Rohrbacher, Bernhard Wolfgang: Morphology-Driven Syntax. A theory of V to I raising and prodrop. 1999. viii, 296 pp. 14 Anagnostopoulou, Elena, Henk van Riemsdijk and Frans Zwarts (eds.): Materials on Left Dislocation. 1997. viii, 349 pp. 13 Alexiadou, Artemis and T. Alan Hall (eds.): Studies on Universal Grammar and Typological Variation. 1997. viii, 252 pp. 12 Abraham, Werner, Samuel David Epstein, Höskuldur Thráinsson and Jan-Wouter Zwart (eds.): Minimal Ideas. Syntactic studies in the minimalist framework. 1996. xii, 364 pp.