The Working Memory of Argument–Verb Dependencies

Spatiotemporal Brain Dynamics during Sentence Processing

Impressum

Max Planck Institute for Human Cognitive and Brain Sciences, 2013

Diese Arbeit ist unter folgender Creative Commons-Lizenz lizenziert:

http://creativecommons.org/licenses/by-nc/3.0

Druck: Sächsisches Druck- und Verlagshaus Direct World, Dresden

© Titelbild: Foto Lars Meyer, 2013 (Detail aus dem Wandmosaik „Der Mensch bezwingt den
Kosmos“ von Fritz Eisel in Potsdam)

ISBN 978-3-941504-29-5
 T H E WO R K I N G M E M O RY
OF A R G U M E N T–V E R B D E P E N D E N C I E S
Spatiotemporal Brain Dynamics during Sentence Processing

Dissertation

zur Erlangung des akademischen Grades eines

Doctor philosophiae (Dr. phil.)

der Humanwissenschaftlichen Fakultät der Universität Potsdam

vorgelegt

von Lars Meyer, M.Sc.

Leipzig, 2012

Gutachterinnen

Prof. Dr. Angela D. Friederici

Prof. Dr. Isabell Wartenburger

Für Marlene Fock-Greulich in dankbarer Erinnerung
AC K N O W L E D G M E N T S

Amongst the fellow neuroscientists who deserve credit for the content, look, and feel of this thesis, I am

obliged to Angela D. Friederici and Isabell Wartenburger for acceptance and assessment of my work.

I cordially thank my supervisors: The climate of Angela D. Friederici’s unconditional supervision

meandered between research freedom and rigid discussion—both of which were highly stimulating. For

his advice and high analytical standards as well as our arguments and creative juggling of datasets I thank

my friend, partner in crime, and supervisor, Jonas Obleser.

For sharing data-screening experience, skeptical methodological and witty engineering skills,

deeply felt thanks to Alfred Anwander, Maren Grigutsch, Thomas Gunter, Molly Henry, Annette

Horstmann, Stefan J. Kiebel, Burkhard Maess, Michiru Makuuchi, and Helena Trompelt.

Data quality, research quality, text quality, graphics quality, and emotional quality—choose

those that fit you best—were ensured by Kerstin Flake, Andrea Gast-Sandmann, Sarah Gierhan, Björn

Herrmann, Mike Hove, Sarah Jessen, Iris N. Knierim, Ina Koch, Anke Kummer, Stephan Liebig, Claudia

Männel, Jutta Mueller, Helga Smallwood, Celia Sommer, Rosie Wallis, Annett Wiedemann, Simone

Wipper, and at least twelve anonymous reviewers. I am grateful to all of you.

CONTENTS

Preface viii

1 General Introduction 1

1.1 Argument–Verb Dependencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Working Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.3 Argument–Verb Dependencies and Working Memory . . . . . . . . . . . . . . . . . . . . . . 6

1.4 Open Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2 General Methodology 13

2.1 Magnetic Resonance Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2.1.1 Functional Magnetic Resonance Imaging . . . . . . . . . . . . . . . . . . . . . . . . . 13

2.1.2 Voxel-Based Morphometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.1.3 Diffusion-Weighted Imaging and Diffusion-Tensor Imaging . . . . . . . . . . . . . 17

2.2 Electroencephalography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

2.2.1 Event-Related Brain Potentials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

2.2.2 Time–Frequency Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

2.2.3 Source Localization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

2.3 Experimental Paradigm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

2.3.1 Linguistic Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

2.3.2 Psycholinguistic Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

2.3.3 Psychological Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

i
Contents

2.3.4 General Hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

3 Behavioral Study 31

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

3.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

3.2.1 Participants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

3.2.2 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

3.2.3 Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

3.3 Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

3.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

3.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

4 Magnetic-Resonance-Imaging Study 39

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

4.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

4.2.1 Participants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

4.2.2 Working-Memory Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

4.2.3 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

4.2.4 Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

4.2.5 Magnetic-Resonance-Imaging Data Acquisition . . . . . . . . . . . . . . . . . . . . . 43

4.2.6 Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

4.2.6.1 Behavioral Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

4.2.6.2 Functional-Magnetic-Resonance-Imaging Data . . . . . . . . . . . . . . . 45

4.2.6.3 Correlation Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

4.2.6.4 Voxel-Based Morphometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

4.2.6.5 Diffusion-Tensor Imaging and Fractional Anisotropy . . . . . . . . . . . 47

4.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

4.3.1 Behavioral Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

4.3.2 Functional-Magnetic-Resonance-Imaging Results . . . . . . . . . . . . . . . . . . . . 49

4.3.3 Voxel-Based-Morphometry Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

ii
 Contents

4.3.4 Diffusion-Tensor-Imaging and Fractional-Anisotropy Results . . . . . . . . . . . . 53

4.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

4.4.1 Behavioral Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

4.4.2 Functional-Magnetic-Resonance-Imaging Results . . . . . . . . . . . . . . . . . . . . 55

4.4.2.1 The Left Inferior Frontal Gyrus Activates for Reordering . . . . . . . . 56

4.4.2.2 The Left Temporo-Parietal Region Activates for Storage . . . . . . . . . 59

4.4.3 Individual Differences: Asymmetry of the Inferior Frontal Gyrus . . . . . . . . . 61

4.4.4 Diffusion-Tensor-Imaging and Fractional-Anisotropy Results . . . . . . . . . . . . 64

4.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

5 Patient Study 69

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

5.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

5.2.1 Participants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

5.2.2 Working-Memory Test Battery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

5.2.3 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

5.2.4 Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

5.2.5 Magnetic-Resonance-Imaging Data Acquisition . . . . . . . . . . . . . . . . . . . . . 73

5.2.6 Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

5.2.6.1 Working-Memory Test Battery . . . . . . . . . . . . . . . . . . . . . . . . . 74

5.2.6.2 Behavioral Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

5.2.6.3 Diffusion-Tensor Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

5.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

5.3.1 Working-Memory Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

5.3.2 Sentence-Processing Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

5.3.3 Diffusion-Tensor-Imaging Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

5.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

5.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

6 Electroencephalography Study 83

iii
Contents

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

6.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

6.2.1 Participants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

6.2.2 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

6.2.3 Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

6.2.4 Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

6.2.5 Source Localization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

6.2.6 Correlation Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

6.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

6.3.1 Behavioral Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

6.3.2 Time–Frequency Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

6.3.3 Source Localization of Alpha Activity: Results . . . . . . . . . . . . . . . . . . . . . 91

6.3.4 Results of the Correlation Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

6.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

6.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

7 Combined Analyses 101

7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

7.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

7.2.1 Participants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

7.2.2 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

7.2.3 Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

7.2.4 Data Acquisition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

7.2.5 Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

7.2.6 Combined Analyses in Sensor and Source Space . . . . . . . . . . . . . . . . . . . . . 107

7.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

7.3.1 Behavioral Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

7.3.2 Combined Analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

7.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

7.4.1 Behavioral Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

iv
 Contents

7.4.2 Combined Analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

7.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

8 General Discussion 119

8.1 Summary of Experimental Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119

8.2 The Inferior Frontal Gyrus: Reordering, Rehearsal, or Both? . . . . . . . . . . . . . . . . . 122

8.3 The Temporo-Parietal Region: Storage, Retrieval, or Both? . . . . . . . . . . . . . . . . . . . 125

8.4 The Dorsal Tract: Syntax, Working Memory, or Both? . . . . . . . . . . . . . . . . . . . . . 129

9 Towards a Testable Framework 135

9.1 Reordering: Beyond Sentence Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137

9.2 The Inferior Parietal Cortex: Beyond Working Memory . . . . . . . . . . . . . . . . . . . . 138

9.3 Disentangling Reordering and Rehearsal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139

9.4 Brain Oscillations and Fronto-Temporal Connectivity . . . . . . . . . . . . . . . . . . . . . . 140

References 143

List of Figures 185

List of Tables 189

Abbreviations 191

Appendix A Stimuli from Behavioral, fMRI and EEG Studies A-1

Appendix B Stimuli from Patient Study B-7

Appendix C Fillers from Behavioral Study C-13

Appendix D Fillers from Behavioral and MRI Studies D-19

v
PR E F AC E

Sentences are series of symbols arranged in a specific order. Some of these symbols are verbs, expressing

actions—others of these symbols are arguments, expressing participants that join in the actions. This

thesis investigates the spatiotemporal brain dynamics of two cognitive processes that enable the compre-

hension of the mutual dependencies between arguments and verbs. When an argument and its verb do

not occur in succession, the argument must be remembered during the time interval between argument

and verb. When the relative argument order differs from the order that the arguments combine in with

their verb, reordering of the remembered arguments must take place.

Argument storage and reordering are working-memory processes: Storage is remembering, coun-

teracting forgetting. Reordering is an executive process operating on stored items. Chapter 1 introduces

this conceptualization from linguistic, psycholinguistic, and cognitive-neuropsychological viewpoints,

as well as the open issues: Linguistics has a rich terminology for describing argument–verb dependencies,

psycholinguistics has established the psychological reality of some of these terms, and cognitive neu-

ropsychology can relate those terms that are psychologically real to the underlying cognitive processes

and the subserving functional neuroanatomy.

The goal of this thesis is to understand the spatiotemporal brain dynamics that subserve argument

storage and reordering during sentence processing. The goal is not to formulate a psycholinguistic theory

of argument–verb-dependency processing, since multiple frameworks are already available. Because

spatiotemporal brain dynamics can only be investigated exploiting high-resolution temporal and spatial

brain-data-analysis methods, this thesis applied various data acquisition and analysis methods to a

single experimental paradigm. Chapter 2 provides a critical overview of these methods, as well as the

to-be-tested experimental paradigm.

vii
Contents

The following experimental chapters are devoted to investigate the open issues in the spatiotem-

poral dynamics of argument storage and reordering: Chapter 3 asks about the behavioral relationship

between argument storage and reordering. Chapter 4 asks about the functional neuroanatomy of

argument storage and reordering. Chapter 5 asks about the consequences of selective damage to

this functional neuroanatomy. Chapter 6 asks about the brain-electric substrate of argument stor-

age. Chapter 7 asks about the spatiotemporal dynamics inside the functional-neuroanatomical brain

network established in the previous chapters. Chapter 8 summarizes the experimental results, ex-

tracts their three leitmotifs and relates them to cognitive-neuropsychological research from outside the

sentence-processing domain. Finally, Chapter 9 ends the thesis, proposing both a testable neurocognitive

framework of argument–verb-dependency processing and a number of related research opportunities.

viii
1
GENERAL INTRODUCTION

1.1 Argument–Verb Dependencies

In a sentence, arguments symbolize the who and whom, while a verb symbolizes an action jointly

involving the who and whom (Bühler, 1934; Frege, 1879). Because arguments and verbs jointly constitute

sentence meaning, they form mutual argument–verb dependencies, which are language’s core device of

symbolizing relations between entities. Hence, speakers’ mental-lexical knowledge about stereotypical

argument–verb combinations (argument–verb sets) reflects speakers’ experience with stereotypical

actions and stereotypically involved entities (Goldberg, 1992; Lapata, Keller, & Walde, 2001; Rumelhart,

1980; for an overview, see Tomasello & Merriman, 1995). This section provides an introduction to the

linguistic properties of argument–verb dependencies.

A verb’s mental-lexicon entry contains two sources of information about the type and number

of arguments that need to co-occur with the verb to form a coherent sentence (Chomsky, 1965, 1981):

The arguments’ formal syntactic properties (e.g., nominative case) are specified by the subcategorization

frame, whereas the arguments’ functional thematic roles (e.g., the who performing an action) are

specified by the θ-grid. The neuropsychological reality of subcategorization frames and θ-grids is

suggested by clinical evidence on their selective disruptability (Bastiaanse, Edwards, Mass, & Rispens,

2003); in addition, the neurophysiological sensitivity to subcategorization frames and θ-grids is suggested

by their neuroanatomical dissociability (Shetreet, Palti, Friedmann, & Hadar, 2007).

Psycholinguistic data support the significance of argument–verb dependencies for sentence pro-

cessing: Fodor, Garrett, and Bever’s (1968) early work suggests processing difficulty to increase with the

1
1. General Introduction

number of possible argument–verb sets, with verbs that allow for multiple possible sets showing higher

processing load than verbs whose mental-lexicon entry specifies only a single possible set. Later work

showed that the combination of an argument and a transitive verb that belong to a joint argument–verb

set pre-activates the syntactic characteristics of the second argument, and thus the verb’s subcategoriza-

tion frame (Clifton et al., 1984; for further discussion, see Trueswell & Kim, 1998). While Shapiro,

Zurif, and Grimshaw (1989), Shapiro, Brookins, Gordon, and Nagel (1991), and McElree and Dosher

(1993) suggested that argument pre-activation carries over to the θ-grid, McElree and Griffith’s (1995) re-

sults suggest that subcategorization and θ-grid information are pre-activated serially: subcategorization

violations are recognized quicker than θ-grid violations. Further studies extended these results (Garnsey,

Pearlmutter, Myers, & Lotocky, 1997; Holmes, Stowe, & Cupples, 1989; Trueswell & Kim, 1998),

also showing that argument–verb-set pre-activation is a two-way mechanism (cf. Friederici & Frisch,

2000): Verbs pre-activate their subcategorization frames and θ-grids, and to a lesser extent, arguments

pre-activate their verbs (Kamide, Altmann, & Haywood, 2000; Kamide, Scheepers, & Altmann, 2003;

Tsuzuki, Uchida, Yukihiro, Hisano, & Tsuzuki, 2004).

To infer sentence meaning, those phrases of a joint argument–verb set that constitute the who,

the whom, and the action the who and whom are involved in must be linked to one another. De-

scriptive work has long assumed that argument–verb linking occurs in a serial order idiosyncratic to

any particular language (Chomsky, 1955; Fodor, 1978)—for instance, subject–object–verb in German

(Bader & Bayer, 2006; Haider, 1993). For English, which follows subject–verb–object order, Clifton et

al. (1984) found evidence for this proposal: The authors observed that grammaticality judgment was

slowed down when transitive verbs were not followed by their object, but by a prepositional phrase,

suggesting that the missing argument at the post-verbal position had been predicted from the verb’s

subcategorization information in combination with the English subject–verb–object order (cf. Holmes,

1987; Shapiro, Gordon, Hack, & Killackey, 1993). Further work showed that the preference to interpret

a post-verbal noun phrase as the verb’s direct object is so strong as to erroneously hinder this noun

phrase’s interpretation as the subject of a subsequent sentence (Trueswell, Tanenhaus, & Kello, 1993).

Since a verb’s subcategorization frame and θ-grid are activated on encounter, the number and

type of required arguments becomes known with the verb; those arguments in the sentence whose

syntactic features fit the verb’s subcategorization frame can then be linked to the verb. There is evidence

2
 1.2. Working Memory

that this linking is triggered by an alignment of syntactic argument features (e.g., a suffix marking

nominative case) with the verb’s subcategorization frame and θ-grid: In an elegant study, Van Dyke and

McElree (2006) manipulated the number of noun phrases whose syntactic features matched the verb

subcategorization frame and θ-grid of a given verb, finding reading times at the verb to increase with

the number of syntactically matching noun phrases (cf. Fedorenko, Gibson, & Rohde, 2006; Gordon,

Hendrick, & Johnson, 2001).

Argument–verb linking may be straightforward when the arguments occur in the vicinity of their

subcategorizing verb and in the serial order that arguments combine in with their subcategorizing verbs

in that particular language. However, a direct argument–verb linking is not possible when arguments

occur at sentential positions remote from their subcategorizing verb (e.g., an argument is in front of a

sentence, whereas its verb is at the sentence ending), or when the relative order of arguments differs from

the serial order that arguments combine in with their subcategorizing verbs (e.g., the object is in front

of the subject). In the resulting argument–verb dependency, argument–verb linking can only occur

indirectly by storing the remote argument across the resulting argument–verb distance and re-activating

the remote argument at its verb. Section 1.2 will next discuss the underlying cognitive mechanisms

that enable these processes, while Section 1.3 will then relate these mechanisms more specifically to the

sentence-processing domain.

1.2 Working Memory

Working memory is the workbench of cognition, storing information while required for manipulation

by the brain’s executive processing systems (Baddeley, Eysenck, & Anderson, 2009; Rumelhart, Lindsay,

& Norman, 1972; Shah & Miyake, 1999). Across sensory and cognitive domains, at least a visual- and

a verbal-working-memory system are commonly differentiated, of which the latter is crucial to the

cognitive processes investigated in the current thesis: It is clear that verbal working memory is involved

in language processing (Baddeley & Hitch, 1974; Wingfield & Butterworth, 1984).

Current models largely agree that working memory is an active system involving executive

components (Oberauer, Süß, Wilhelm, & Wittman, 2003; cf. Miyake & Shah, 1999). In particular,

the contents of verbal working memory can constantly be refreshed by articulatory rehearsal, which

effectively lengthens the possible retention interval (Baddeley, 2012). There is imaging evidence on the

3
1. General Introduction

neuroanatomical dissociation of working-memory-storage and -rehearsal sub-systems (Awh, Smith, &

Jonides, 1995; Paulesu, Frith, & Frackowiak, 1993; for discussion, see Chapter 4); furthermore, work

on the elderly suggests that their reduced working-memory retention duration results from decreased

brain activity in those prefrontal regions that underlie rehearsal (Johnson, Mitchell, Raye, & Greene,

2004; Johnson, Reeder, Raye, & Mitchell, 2002). In addition to rehearsal, Baddeley’s (2012) model of

working memory proposes executive sub-systems concerned with information manipulation. While

there is some consensus that working memory does involve such systems (Miyake & Shah, 1999), there

is little consensus as to which processes this entails (Baddeley, 1996).

Its active nature distinguishes working memory from other systems of short-period information

retention: Echoic memory (Neisser, 1967) and short-term memory (Brown, 1958; Peterson & Peterson,

1959) have also been proposed as subservants of temporary information storage. While some authors

treat these systems as synonymous with working memory, the content of echoic and short-term memory

degenerates more rapidly than that retained in working memory. This rapid decay of the content stored

by echoic memory and short-term memory has been proposed to result from their proximity to sensory

input systems and their according lack of an information-refreshing sub-system (Baddeley & Larsen,

2007; Jones, Hughes, & Macken, 2007).

Information refreshing by rehearsal reduces the degeneration susceptibility of working memory,

but its retention duration is still limited by decay and interference (Miller, 1956): Early work on

forgetting suggests that decay induced by increased retention duration is an inherent property of

working memory (Anderson, Reder, & Lebiere, 1996; Peterson & Peterson, 1959). The neural reality of

decay is suggested by single-cell recordings in monkeys, which show neurons involved in retention to

reduce their firing rates along the retention interval (Fuster, 1999). In addition to decay, similarity-based

interference of synchronously stored items limits retention duration: Conrad and Hull (1964) find

that acoustic similarity of to-be-remembered letters decreases memory span (cf. Lewandowsky, Geiger,

& Oberauer, 2008; Oberauer & Lange, 2008; Potter, 1976; Shulman, 1970). It is, however, debated

whether decay or interference is a stronger limiter of retention duration, ever since McGeoch (1932)

argued that the task performed during retention limits retention duration over and above decay (cf.

Altmann & Schunn, 2002; Campoy, 2012; Conrad, 1967; Keppel & Underwood, 1962; Lewandowsky,

4
 1.2. Working Memory

Oberauer, & Brown, 2009). Given this open discussion, the experiments of the current thesis controlled

for interference as far as possible to isolate decay as a controlled experimental factor (see Section 2.3).

The susceptibility to degenerative factors might distinguish working memory from long-term

memory1 ; however, Anders Ericsson and Kintsch (1995) proposed a direct link between working mem-

ory and long-term memory by conceptualizing working memory as an attentional mechanism that

selectively activates long-term-memory content (Parkin, 2001; Unsworth & Engle, 2007; for discus-

sion, see Cowan, 2008). The major argument for this conceptualization is the finding that individuals’

working-memory performance can be predicted by their fields of expertise (Ericsson & Chase, 1982).

For instance, professional chess players have been found to access long-term memories when verbally

reporting on unfamiliar chess positions they were experimentally prompted to solve (de Groot, 1978), a

finding which has been replicated for other fields of expertise (for review, see Anders Ericsson & Kintsch,

1995). To still account for the fact that working-memory processes may recur to novel information,

Anders Ericsson and Kintsch (1995) propose a differentiation between short-term working memory

and long-term working memory, of which the former designates the temporary storage of novel infor-

mation, whereas the latter designates the temporary attentional selection of long-term-memory content.

This conceptualization is particularly interesting given a possible neuroanatomical overlap between

brain regions involved in working memory, attention, long-term memory, and the mental lexicon (cf.

Chapter 8; see also Cabeza, Ciaramelli, & Moscovitch, 2012a).

A final issue to be addressed before turning to the specific role of working memory in

argument–verb-dependency processing is the question of the fractionation of verbal working memory

into phonological, semantic, and syntactic sub-parts. Martin, Shelton, and Yaffee (1994) suggested that

a holistic view of verbal working memory may not capture the diversity of language processing: Martin

et al. (1994) provide clinical evidence that phonological and semantic aspects of verbal working memory

can be selectively impaired, backed up by functional-magnetic-resonance-imaging (fMRI) evidence from

healthy populations (Shivde & Thompson-Schill, 2004). Martin and Romani’s (1994) data additionally

illustrate the possibility of a selectively impaired ability to store syntactic information. While these re-

sults principally demonstrate a fractionation of verbal working memory, disagreement prevails: Caplan

1
Long-term memory may not be interference-proof during retrieval, and especially not during verbal retrieval (Cowan, 2008).

5
1. General Introduction

and Waters (1999) differentiate phonological and syntactic working memory, Fiebach, Friederici, Smith,

and Swinney (2007), Martin and He (2004), and Shivde and Anderson (2011) focus on the semantic

aspects, and Caspari, Parkinson, LaPointe, and Katz (1998) and Fedorenko, Gibson, and Rohde (2007)

propose an additional working-memory system shared by any symbolic process, both during language

processing and outside of the language-processing domain. Whether or not these fractionation proposals

are of relevance to argument–verb-dependency processing will be discussed in the following section,

which focuses on the role of verbal working memory in the processing of argument–verb dependencies.

1.3 Argument–Verb Dependencies and Working Memory

An important assumption of the current thesis is that working memory is the cognitive basis for

argument–verb-dependency processing. The difficulty of argument–verb-dependency processing is de-

termined by two cognitive processes: First, when an argument occurs at a sentential position remote

from its subcategorizing verb, temporary storage of the argument in working memory becomes nec-

essary, until the argument is retrieved for argument–verb linking at its subcategorizing verb. In this

case, a so-called gap remains where the subcategorizing verb necessitates its now-remote argument; the

argument itself has now become a so-called filler (Fodor, 1978; Frazier, Clifton, & Randall, 1983; Kluen-

der & Kutas, 1993)2 . Second, when multiple arguments of a joint argument–verb set occur in a relative

order other than the serial order idiosyncratic to any particular language, working memory needs to

recruit an executive mechanism of argument reordering (Carpenter & Just, 1989; Just & Carpenter,

1992; Kintsch & Van Dijk, 1978).

The psychological reality of argument storage was established by Wanner and Maratsos (1979),

who found that concurrent memory load decreases performance during argument–verb-dependency

processing (Gordon et al., 2001; Gordon, Hendrick, Johnson, & Lee, 2006; Gordon, Hendrick, & Levine,

2002; Van Dyke, 2007; cf. Frazier et al., 1983). Strong evidence for the neuropsychological reality of

argument storage was later provided by Kluender and Kutas’ (1993) seminal event-related-brain-potential

(ERP) study: Their paradigm investigated sentences containing a pronoun-argument–verb dependency,

and found a sustained negative effect spanning the argument–verb distance for these sentences when

2
The theoretical precursor of Fodor’s (1978) filler–gap dichotomy is Chomsky’s (1955) antecedent–trace dichotomy; for a
concise introduction, see Radford (1997).

6
 1.3. Argument–Verb Dependencies and Working Memory

compared to a baseline condition. Later studies corroborated these results (Phillips, Kazanina, & Abada,

2005; Ueno & Kluender, 2003), and extended the findings to interrogative-pronoun and full-noun-phrase

arguments in German (Felser, Clahsen, & Münte, 2003; Fiebach, Schlesewsky, & Friederici, 2001,

2002; Matzke, Mai, Nager, Rüsseler, & Münte, 2002)3 . It is, however, problematic that these prior

studies did not separate argument storage and reordering: Object-first and subject-first sentences at

long argument–verb distance were compared only, but never short and long argument–verb distances

directly—in spite of evidence for argument storage being common to any argument order. I will come

back to this critique when discussing the present experimental design (see Section 2.3).

In addition to argument storage, filler–gap dependencies also necessitate argument retrieval at

their subcategorizing verbs (Lewis, 1996; Lewis & Vasishth, 2005). While Corbett and Chang’s (1983)

initial off-line results suggested that pronouns trigger retrieval of their antecedent nouns, Tanenhaus,

Carlson, and Seidenberg (1985) on-line data suggested that an argument gap leads to argument retrieval:

Tanenhaus et al. (1985) presented words that rhymed with a remote filler at its gap, finding that these

words were primed. Nicol and Swinney (1989) report that Swinney, Ford, Frauenfelder, and Bresnan

(1988) adapted Tanenhaus et al.’s (1985) cross-modal priming technique, finding that also words that

are semantically related to a remote filler are primed at the remote filler’s gap. Similar experiments by

Bever and McElree (1988) and Love and Swinney (1996) concluded that argument retrieval does not

only occur for pronoun reference and interrogative phrases, but for proper-noun phrases as well (for

review, see Nicol, Fodor, & Swinney, 1994; Nicol & Swinney, 1989)4 .

So, we can reasonably deduce that argument–verb-dependency processing involves storage and

retrieval. However, reports are equivocal as to whether these recur to phonological, semantic, or syntactic

representations: The cross-modal priming studies discussed above mention fillers to elicit rhyme priming

(Tanenhaus et al., 1985), prosodic priming (Nicholas Nagel, Shapiro, & Nawy, 1994), and semantic

priming (Bever & McElree, 1988; Nicol & Swinney, 1989; Swinney et al., 1988) at subcategorizing verbs

3
These effects are similar to sustained negative ERPs observed for verbal-working-memory-storage tasks outside of the
sentence-processing domain (Johnson, 1995; Ruchkin et al., 1992). This points to a degree of domain-generality of the
verbal-working-memory system involved in argument–verb-dependency processing—although Ruchkin et al.’s (1999) data
suggest that words elicit increased sustained negative ERP effects relative to non-words.
4
For brevity’s sake, this section confines to the simplistic notion of argument retrieval at subcategorizing verbs. There is,
however, evidence that arguments can be retrieved before the actual encounter of their subcategorizing verb; for discussion,
see Section 6.4, Section 7.4.2, and Chapter 8.

7
1. General Introduction

(cf. Clahsen & Featherston, 1999; Featherston, 2001). Furthermore, self-paced-reading (SPR) studies

suggest that syntactically interfering noun phrases increase both argument storage and retrieval efforts

(Van Dyke & McElree, 2006). The data thus suggest that arguments are stored in a rich verbal code.

As any working-memory content, arguments decay over time (cf. Section 1.2). This led to

the proposal that argument–verb distance be a major determinant of sentence-processing difficulty

(Frazier, 1987; Frazier et al., 1983; Yngwe, 1960): The longer the argument–verb distance, the

longer the argument-storage interval prior to retrieval, and the more the argument will degener-

ate. For the same reason, long argument–verb distances may exacerbate argument retrieval at the

subcategorizing verb. Cross-linguistic experiments found that dependency length indeed increases

argument–verb-linking efforts (Babyonyshev & Gibson, 1999; Cowper, 1976; Gibson, 2000; Grodner

& Gibson, 2005)—rendering dependency length a reliable tool to query argument storage and retrieval

during argument–verb-dependency processing5 .

So far, this section considered the interrelation between decay and argument–verb distance.

However, when an argument–verb set’s arguments do not occur in their idiosyncratic relative order,

the re-establishment of their preferred relative order is an additional challenge to working memory (cf.

Section 1.1)—necessitating the postulation of an executive mechanism of argument reordering in addition

to argument storage. In support of this postulate, Osterhout and Swinney’s (1993) cross-modal-priming

study found that only the appropriate subject from a transitive object-first argument order primes

semantically related words at the empty subject gap of a subsequent empty-subject sentence (cf. Nicol &

Swinney, 1989). This implies that fillers are retrieved only at their appropriate gap. An overwhelming

body of cross-linguistic behavioral evidence shows object-first as compared to subject-first argument

orders to increase processing load (Hyönä & Hujanen, 1997; MacWhinney & Pléh, 1988; Mazuka, Itoh,

& Kondo, 2002; Miyamoto & Takahashi, 2002), most profoundly at subcategorizing verbs (Frauenfelder,

Segui, & Mehler, 1980; Grodner & Gibson, 2005). In a nutshell, a mismatch between occurrence order

and gap order necessitates argument reordering.

5
As an interesting linguistic aside, some authors have proposed that working-memory decay may be the evolutionary reason
for the cross-linguistic tendency of languages to keep dependents—such as arguments—close to their heads—such as verbs (cf.
Temperley, 2007).

8
 1.4. Open Issues

Previous neuroimaging data on the proposed distinction between argument storage and reordering

are not unequivocal. While studies on argument reordering from languages other than English agree on

an involvement of inferior frontal brain regions (Ben-Shachar, Palti, & Grodzinsky, 2004; Friederici,

Fiebach, Schlesewsky, Bornkessel, & von Cramon, 2006; Kim et al., 2009), Rogalsky and Hickok

(2010) insist on a reductionist perspective equating reordering- and storage-related brain activations as

both reflecting rehearsal. The picture is similarly complex for argument storage: While some studies

suggest that inferior frontal cortex be involved in argument storage distinctly from argument reordering

(Fiebach, Schlesewsky, Lohmann, von Cramon, & Friederici, 2005; Makuuchi, Bahlmann, Anwander,

& Friederici, 2009; Santi & Grodzinsky, 2007), these studies contrasted different kinds of syntactic

dependencies. They also disagree with work from the sentence-processing domain that found inferior

parietal regions involved in argument storage (Novais-Santos et al., 2007). While Novais-Santos et al.’s

(2007) result converges on work from outside the sentence-processing domain (for review, see Owen,

McMillan, Laird, & Bullmore, 2005; Smith & Jonides, 1998; Wager, Keller, Lacey, & Jonides, 2005),

their experimental paradigm confounded dependency length and phrasal size of the stored argument,

leaving its result ambiguous. Clearly, the imaging literature is in need of better-controlled work.

This chapter has laid out the linguistic and psycholinguistic foundations of the argument–verb

dichotomy, the underlying cognitive processing mechanisms, and their specific manifestations during

filler–gap-dependency processing. When it comes to the neural reality of dependency-length-induced

storage and retrieval load and order-mismatch-induced reordering, there are bodies of evidence available.

Since these are under debate, they will be introduced in Section 1.4, alongside the open issues in the

research on the spatiotemporal dynamics of argument–verb-dependency processing in the human brain.

1.4 Open Issues

The previous sections reported on attempts to segregate argument storage and retrieval (depending on

working-memory decay) from argument reordering (depending on a mismatch between argument occur-

rence order and gap order). Throughout the thesis, this will guide the following hypothetical scenario:

Argument–verb-dependency processing involves the storage of arguments across the argument–verb

distance; at their subcategorizing verb, these arguments are retrieved, and in case of a relative argument

order that deviates from the idiosyncratic order, argument reordering re-establishes this order.

9
1. General Introduction

While these hypotheses have firm bases, some authors do not assume active executive mech-

anisms, treating argument–verb-dependency processing as fully emerging from decay, interference,

and syntactic-feature matching (Lewis, Vasishth, & Van Dyke, 2006), or simply rehearsal (Rogalsky &

Hickok, 2010). These positions may result from the fact that English rarely allows for argument-order

variations, and possible cases such as relative clauses or embedded questions confound argument–verb

distance and argument order. While there is evidence for deriving the processing difficulty associated

with increasing argument–verb-dependency length from increased decay-exposure of the filler (Fra-

zier, 1987; Gibson, 2000), there is also clear evidence that argument-order variations decrease response

accuracy—even if argument–verb-dependency length is controlled for (Obleser, Meyer, & Friederici,

2011). Because it is unclear whether argument storage and retrieval on the one side and argument

reordering on the other side exhibit an independent or interactive relationship, Chapter 3 reports on a

behavioral SPR and rating experiment that fully crossed the two factors.

While behavioral results can complement psycholinguistic theory, they do not suffice to conclude

on the neural interdependence of argument storage on one side and argument reordering on the other:

Behavioral interactions between experimental variables may result from brain-functional interactions,

but they may also derive from brain-structural properties that selectively constrain the output of

principally independent brain functions: If reordering and storage were performed by independent

regions, their anatomical connection could selectively facilitate the processing of sentences that tax both

reordering and storage. Because previous neuroimaging data do not speak to this possibility, Chapter 4

reports on an fMRI study on the paradigm used in Chapter 3, supplemented by an analysis of the

involved regions’ connectivity and anatomical substrates. To elucidate on the related question of the

functional-neuroanatomical isomorphism between the storage–reordering and the storage–rehearsal

dichotomies, Chapter 4 also investigated whether activations of the argument–verb-dependency network

correlate with working-memory abilities.

Because fMRI evidence can in principle only work out the necessary, but not the sufficient brain

parts involved in a cognitive process (for discussion, see Caplan, 2009), it needs backup by clinical

data from patients with selective damage to—preferentially—one of those brain parts. This in mind,

Chapter 5 presents a case study on a patient whose selective neuroanatomical damage gives reason to

assume a selective deficit in those processes characterized in Chapter 3 and localized in Chapter 4.

10
 1.4. Open Issues

The manifold previous research on the brain-electric equivalents of the working-memory pro-

cesses involved in argument–verb-dependency processing does not speak to the differentiation between

argument-storage and -retrieval processes and argument-reordering mechanisms. Furthermore, all pre-

vious electroencephalography (EEG) studies used ERPs, which present a valuable, yet somewhat remote

characterization of brain-electric activity (as discussed in Section 2.2.1 and Section 2.2.2). Thus, Chap-

ter 6 focuses on argument storage independently of argument order, approaching the paradigm from

the previous experiments with a combination of time-frequency analysis (TFR) and source localization.

Chapter 7 asks how argument retrieval and reordering progress during argument–verb linking

at subcategorizing verbs. It combines data from Chapter 4 and Chapter 6 into two methods of joint

data analysis to exploit both spatial and temporal specificity. While there is a firm EEG literature on

argument storage and diverse behavioral indications on argument retrieval at subcategorizing verbs,

the spatiotemporal dynamics of these processes inside the underlying neuroanatomy are far from

understood. When argument retrieval, as suggested by previous results, is common to subjects and

objects, what is its functional relation to argument reordering? What is the spatiotemporal dimension

of the neuroanatomical network established in Chapter 4 and Chapter 5?

In sum, the following chapters will approach the following questions: What is the behavioral rela-

tionship between argument storage and reordering (Chapter 3)? What is the functional neuroanatomy

of argument storage and reordering, and what its connectivity? Are there individual differences in the

functional neuroanatomy? How does the functional neuroanatomy subserving argument storage and

reordering relate to the functional neuroanatomy underlying storage and rehearsal in working memory

(Chapter 4)? Does a selective disruption of the involved regions’ connectivity support the neuroanatomi-

cal dissociation of argument storage and reordering (Chapter 5)? What is the brain-electric counterpart

of argument storage, regardless of the argument order (Chapter 6)? What are the spatiotemporal dynam-

ics inside the functional-neuroanatomical brain network once argument storage has successfully been

performed, and arguments need retrieval and reordering for final argument–verb linking (Chapter 7)?

Since these questions have been approached with a number of acquisition and analysis techniques whose

understanding is viable to understanding the experimental findings, Chapter 2 will provide the reader

with a critical overview of these methods, followed by an exhaustive description of the experimental

paradigm used in all experiments of the current thesis.

11
2
GENERAL METHODOLOGY

The goal of the current thesis is to understand the spatiotemporal brain dynamics that subserve argument

storage and reordering. To this aim, spatially (i.e., magnetic resonance imaging (MRI); Section 2.1) and

temporally (i.e., EEG; Section 2.2) fine-grained data acquisition and analysis methods were used, each

of which has virtues and downsides. To exploit the former while circumnavigating the latter, detailed

knowledge about the technical details of these analyses is important. Thus, this chapter will not so much

focus on the basic principles of signal generation, but discuss data analysis per se6 .

2.1 Magnetic Resonance Imaging

2.1.1 Functional Magnetic Resonance Imaging

Assessing in-vivo the functional brain network underlying particular cognitive processes is made possible

by fMRI, which can track neuronal activity on the whole-brain level by virtue of being sensitive to

indirect measures of blood oxygenation. The physiological basis of this method is that neuronal activity

during a cognitive task is followed by increased blood flow towards those neurons active in the task,

since the neuronal metabolism consumes, inter alia, oxygen (Huettel et al., 2004). Blood contains

hemoglobin, which binds oxygen and unloads it at consuming neurons. Thus, a brain region active

in a cognitive task contains an increased amount of deoxygenated blood after activity. The magnetic

moment of deoxygenated hemoglobin locally raises the energy level of the proton spins in surrounding

6
For an introduction to the basic physical mechanisms behind MRI, I refer the reader to the exhaustive description by Huettel,
Song, and McCarthy (2004); for an introduction to the electrophysiological basis of EEG signal generation, overviews can
be found in the collection by Niedermeyer and Lopes da Silva (1993).

13
2. General Methodology

hydrogen molecules (Pauling, 1935). This shortens the time for the re-establishment of the original

energy level of these spins (i.e., T2*), which leads to a delayed larger post-excitation MR spin echo in

brain regions that engaged in a cognitive process during MR excitation (Huettel et al., 2004; Thulborn,

Waterton, Matthews, & Radda, 1982), called the blood-oxygen-level-dependent (BOLD) signal. The

correlation between neural activity and the BOLD signal was verified by Logothetis et al. (2001), who

showed that BOLD correlates with local field potentials during intracranial recordings in monkeys.

Computationally, fMRI quantifies the BOLD response during an experimentally administered cognitive

task inside the cells of a three-dimensional spatial image grid of variable spacing. Depending on the

acquisition duration, an fMRI experiment will acquire a large number of such images.

To reliably quantify the BOLD response, spatial correction of the acquired image volumes is nec-

essary: Humans inside an MR scanner exhibit undesired movement, further amplified by physiological

factors such as respiration and heartbeat. In effect, the values in an given sub-volume—or voxel—along

the experimental time line do not correspond to a single brain voxel. To re-establish spatial correspon-

dence, all images of the time line are re-aligned to the first image acquired (Friston, Frith, Frackowiak,

& Turner, 1995). In addition to this spatial correction, temporal preprocessing steps are needed: Since

whole-brain fMRI involves a temporally-staged slice-by-slice acquisition of two-dimensional voxel grids,

there is a temporal offset between image slices (Sladky et al., 2011). Slice-timing correction compensates

for this by first estimating an individual voxel’s activation time course and then shifting its phase

(Henson, Burgess, & Frith, 1999). Further preprocessing involves correcting for magnetic-field inhomo-

geneities ( Jezzard & Balaban, 1995) and filtering of slow global signal changes (Della-Maggiore, Chau,

Peres-Neto, & McIntosh, 2002). Typically, preprocessing ends with normalization of the individual data

to a standard brain template, as well as spatial smoothing.

Statistical analysis of fMRI data requires accounting for an underlying property of the BOLD

signal: BOLD is a sluggish measure of brain activity, since oxygenated blood reaches a consuming

region only with a temporal delay (roughly 5 s; cf. Buxton, Wong, & Frank, 1998)7 . As a result,

BOLD magnitude depends on factors such as epoch length and stimulus length, which is compensated

for by convolving the raw data with an a-priori hemodynamic-response function (HRF) that mimics

7
This delay differs inter-individually and inter-regionally; see Schacter, Buckner, Koutstaal, Dale, and Rosen (1997) and
Aguirre, Zarahn, and D’Esposito (1998) for discussion.

14
 2.1. Magnetic Resonance Imaging

the prototypical BOLD time course (Glover, 1999). After convolution, an average BOLD signal is

calculated for the experimental condition of interest and statistically compared to either a global baseline

or another experimental condition. The resulting contrasts are then submitted to across-participants

general-linear-model (GLM) statistics.

When using fMRI, it is important to keep in mind its methodological limitation: Due to the

sluggishness of the BOLD response, fMRI data acquired with whole-brain coverage and at reasonable

spatial resolution do not have a high temporal resolution. Thus, meaningful temporal conclusions (i.e.,

in the order of seconds) can only be drawn for stimuli of sufficient length—such as sentences (Friederici,

Fiebach, et al., 2006) or extended delay periods (Ravizza, Hazeltine, Ruiz, & Zhu, 2011). Alternatively,

a temporal dimension can be added to the data by recording data from the same paradigm used during

fMRI data acquisition with a temporally-fine-grained acquisition modality (e.g., EEG), preferably in

parallel to exclude any test–retest effects. An example for a combined analyses is given in Chapter 7 of

the current thesis, in which EEG data were used to deduce the activation time course of a BOLD effect,

and the activation magnitude of a BOLD effect was used to constrain an EEG topography.

2.1.2 Voxel-Based Morphometry

To perform in-vivo structural whole-brain analyses, voxel-based morphometry (VBM) allows for

across-participant voxel-wise analyses of local microstructural brain-tissue properties (Wright et al.,

1995). Its main challenge is to establish inter-individual voxel-wise spatial correspondence. While the

inter-individual variability of anatomical features such as gyrification or thickness of the gray-matter

layer is the main problem in alignment (Mechelli, Price, Friston, & Ashburner, 2005), it forms the basis

of statistical analysis (cf. Ashburner & Friston, 2001): Paradoxically, efficient spatial alignment needs to

enforce the global leveling-out of variability while locally keeping it.

To achieve spatial alignment, each voxel’s gray value is first automatically assigned a probabil-

ity value of belonging to one of multiple a-priori-defined tissue classes (Ashburner & Friston, 2005).

For a typical anatomical brain scan, the original representation of light gray shades (fatty tissue, e.g.

white matter) and dark gray shades (watery tissue, e.g. gray matter) is mapped onto p-values (exemplary

tissue-probability-map-(TPM)-segmented gray-matter images are shown in panel A of Figure 2.1). While

previous algorithms refined segmentation by applying this procedure recursively (Good et al., 2001), the

15
2. General Methodology

current diffeomorphic-anatomical-registration-using-exponentiated-lie-algebra (DARTEL) approach in-

volves an interactive progression of segmentation and registration (Ashburner, 2007). After segmentation

(typically into white and gray matter, cerebrospinal fluid, and bones8 ), inter-individual alignment of the

segments of interest (see Section 4.2.6.4) takes place. The approach used in the current thesis recursively

warps the segmented individual tissue volumes of a given class onto their mean (Ashburner, 2007) to

generate individual deformation fields and a constantly-self-refining template (an exemplary template,

based on 24 participants from the current thesis, is shown in panel B of Figure 2.1)9 .

Finally, a process termed modulation is applied to the segmented and co-registered tissue volumes.

Modulation compensates for the fact that non-linear registration compresses or inflates individual voxels

to individual degrees: If, for instance, a given brain region is spatially extended in one participant but

spatially confined in another participant, warping them to their common mean compresses the region

in one participant while inflating it in another. To reduce this confound, each voxel’s initial p-value

is multiplied by its relative volume prior and after warping (Good et al., 2001), effectively darkening

compressed voxels and lightening up inflated ones. The modulated volumes are then smoothed and

optionally warped to a standard space (see panel C in Figure 2.1 and Section 4.2.6.4).

A B C

L R

Figure 2.1: Segmentation, template-generation, and co-registration steps in the VBM pipeline: (A)
individual gray-matter segments of 9 exemplary participants from the current investigation; (B) av-
erage group template after six rounds of iterative averaging; (C) final co-registered, smoothed, and
Montreal-Neurological-Institute-(MNI)-normalized gray-matter segments of the same participants.

8
Further tissue classes can be introduced, such as a lesion class for automated lesion mapping (Seghier, Ramlackhansingh,
Crinion, Leff, & Price, 2008).
9
Because of its superior quality (Klein et al., 2009; Takao, Abe, & Ohtomo, 2010; Yassa et al., 2010), DARTEL is used in
functional imaging as well (cf. Wilson et al., 2010).

16
 2.1. Magnetic Resonance Imaging

An important procedure that should be applied prior to statistical assessment relates to the

non-uniform smoothness of segmented and co-registered structural brain images: Regions that exhibit a

large spatial inter-participant correspondence before co-registration will keep a comparably low degree

of spatial smoothness after co-registration as compared to regions with low spatial inter-participant

correspondence—et vice versa. As a result, the occurrence probability of significant statistical tests

increases in smooth regions, invalidating cluster-size statistics. To overcome this pitfall, Worsley, Ander-

mann, Koulis, MacDonald, and Evans (1999) and Hayasaka, Phan, Liberzon, Worsley, and Nichols (2004)

introduced a so-called non-stationarity correction, correcting observed clusters for the across-sample

spatial smoothness of the underlying region.

While VBM is sophisticated, it is also limited: For once, computationally economical non-linear

warping of a brain-tissue volume needs to constrain its spatial resolution (typically to 1.5-mm isotropic

voxels). Furthermore, local inter-individual gyrification differences may result in registration errors

(Bookstein, 2001). It should also be noted that the tissue-probability values assessed statistically do not

exhibit an unambiguous neuroanatomical correlate, but rather reflect a mixture of neuron size, arboriza-

tion of dendrites or axons, gray-matter thickness, and increased gyrification (Mechelli et al., 2005). While

these concerns limit the neuroanatomical precision and meaningfulness of VBM results (Ridgway et al.,

2008), they do certainly not imply arbitrariness of VBM-based microstructural structure-to-function

mapping on a neuropsychologically adequate level—especially so if structural measures are interpreted

in concert with their functional correlates (see Section 4.2.6.4).

2.1.3 Diffusion-Weighted Imaging and Diffusion-Tensor Imaging

While functional neuroimaging can investigate the building blocks of the functional neuroanatomy

of language processing and VBM analyses are capable of assessing the micro-structural properties

of the underlying brain tissue, the concerted interplay of these neuroanatomical substrates is only

possible given structural connections between the concerned regions—which can be assessed using

diffusion-weighted imaging (DWI). This technique tracks the local diffusion of water molecules, that

is, proton-movement-related properties of brain tissue (Le Bihan et al., 1986; Mori, Wakana, Nagae-

Poetscher, & Van Zijl, 2006).

17
2. General Methodology

The physical basis of DWI is the phenomenon that brain-tissue macrostructure constrains the

natural, random, three-dimensional diffusion of water molecules (Brown, 1827): In an unconstrained

medium, water molecules move freely and randomly due to internal thermodynamic excitation (Einstein,

1905), leading to so-called isotropic diffusion. However, external forces from surrounding particles—

such as the myelin sheath of neuronal axons, inter alia—constrain this free random motion, effectively

increasing diffusion directionality along white-matter fibers and decreasing diffusion directionality

perpendicular to white-matter fibers, leading to so-called anisotropic diffusion (Mori et al., 2006).

To detect anisotropic diffusion, DWI applies symmetric spin-echo MR gradients along multiple

spatial directions (at least six for a three-dimensional brain voxel). The MR signal is maximally attenuated

(i.e., the MR echo is minimized) for the gradient applied in the main diffusion direction of the water

molecules in a given voxel (Basser, Mattiello, & Le Bihan, 1994; Basser & Pierpaoli, 1996; Reese, Heid,

Weisskoff, & Wedeen, 2003). The spin echo from all acquired diffusion directions in a given voxel

forms a matrix, allowing for the voxel-wise reconstruction of the diffusion tensors to enable both the

calculation of characteristic diffusion parameters and whole-brain structural-connectivity analyses.

The diffusion parameter most frequently analyzed in white-matter studies is the fractional

anisotropy (FA), which parametrizes the shape of the diffusion tensor by dividing the variance of

the spin-echo matrix eigenvalues by the overall magnitude of the tensor matrix. Basser and Pierpaoli

(1996) showed that the FA effectively approaches zero in isotropic media (e.g., a sphere such as a drop of

cerebrospinal fluid) and one in anisotropic media (e.g., a tube such as an axon)10 . According to Basser

et al. (1994), this computation ensures a more reliable estimation of the main diffusion direction: A

diffusion-weighted imaging protocol that can be applied in an ethical scan time—and thus the matrix

of spin echoes—does not provide an exhaustive picture of diffusion directionality. The inclusion of all

acquired gradient directions in the computation of FA compensates for this problem.

Based on the voxel-wise calculation of diffusivity, tractography can be performed, which allows

for the in-vivo assessment of short- and long-range connectivity in the brain (cf. Figure 2.2, panel C). To

this end, algorithms for tracking the contiguity of the diffusion tensor across adjacent voxels have been

10
Further information about the underlying white matter can be derived by calculating the axial (first eigenvalue of the
diffusion tensor; i.e., axonal direction, cf. Galantucci et al., 2011; see also panel B in Figure 2.2), radial (mean of the two
eigenvalues; i.e., axonal diameter, cf. Phillips et al., 2010), and mean diffusivities (mean of three first eigenvalues; i.e., average
per-voxel diffusion irrespective of directionality, cf. Papoutsi, Stamatakis, Griffiths, Marslen-Wilson, & Tyler, 2011).

18
 2.1. Magnetic Resonance Imaging

devised. The approach used in the current thesis assesses this contiguity by considering the deflection

angle between diffusion tensors from adjacent voxels (Lazar et al., 2003)—if a smooth deflection of the

incoming vector by the target eigenvector is given, a connection will be established; if a deflection angle

is too steep, tracking will stop at the voxel. This procedure leads to a realistically-smooth curvature of

the reconstructed fiber tracts, avoiding erroneously-steep tracking-continuation angles. When applied

to all brain voxels, a tractogram (as shown in panel C of Figure 2.2) results, based on which selective

deterministic (i.e., voxel-to-voxel, see Chapter 4) or explorative probabilistic (i.e., all tracts departing

from a given voxel, see Chapter 5) fiber tracking can be performed.

A B C
b=2 b=9 b = 17

b = 24 b = 31 b = 38

b = 46 b = 53 b = 60
L R L R

Figure 2.2: Overview of DWI analysis steps: (A) axial slices through diffusion-weighted data for 9
exemplary diffusion directions in a single participant; (B) reconstructed diffusion-tensor orientations
for a single axial slice of the same participant; (C) exemplary three-dimensional result of the whole-brain
fiber-tracking procedure.

When tract selection is constrained by functional-imaging or source-localization results, statisti-

cal analyses on the diffusion parameters are a powerful technique for establishing white-matter

structure-to-function correlations—especially when combined with sophisticated non-linear

white-matter-alignment algorithms (see Chapter 4). However, since fiber tracking necessitates

high-level reconstruction techniques based on the calculated diffusion parameters, it involves the

propagation of DWI-inherent degrees of freedom: At a typical resolution of DWI data, the FA of a

given voxel does not necessarily provide the main diffusion direction of a single white-matter fiber tract.

Rather, the voxel may contain fibers that are crossing, merging or fanning out, leading to an ambiguous

FA value, potentially introducing streamlines of arbitrary angles into any diffusion-tensor-imaging

19
2. General Methodology

(DTI) reconstruction (Lazar et al., 2003). While an increase of the spatial resolution of DWI-data

acquisition would circumvent this problem, it would also boost acquisition time above the duration

practical and ethical for a human participant. As a compromise, parallel enhancement of acquisition

resolution and speed (e.g., Jian & Vemuri, 2007) in combination with more sophisticated reconstruction

methods that can extract multiple diffusion tensors from a single voxel (e.g. Descoteaux, Deriche,

Knösche, & Anwander, 2009) seems a promising future direction. Having considered the acquisition

and analysis methods for functional and structural MRI, the following section will provide a critical

discussion of the EEG data analysis methods employed in the current thesis.

2.2 Electroencephalography

As noted in Section 2.1.1, MRI exhibits a high spatial resolution, but lacks temporal resolution. When

questions about the timing of a cognitive process are in the focus of research, EEG is a classical tool:

Depending on the recording hardware, EEG can reflect electrophysiological activity on the millisecond

level and below (Rugg & Coles, 1995). The EEG records ongoing scalp-level voltage fluctuations that

neurophysiologically originate from ion flow through dendritic membranes of post-synaptic neurons

(mostly cortical pyramidal cells; cf. Allison, Wood, & McCarthy, 1986 and Speckmann & Elger, 1993).

When a sufficient number of neurons is oriented in parallel, a synchronous discharge of their

assembly will yield an electric current between their apical dendrites (negative polarity) and cell bodies

(positive polarity), leading to a measurable voltage change perpendicular to the cortical sheath (Regan,

1989). Electrodes positioned at the scalp can record such voltage changes against a remote reference

electrode, which is positioned at a site relatively unaffected by the ongoing electrophysiological activity.

The raw signal from the scalp electrodes is amplified on-line (typically at 60–100 dB; Kamp, Pfurtscheller,

Edlinger, & Silva, 1993) to increase its magnitude over the noise exhibited by the analog recording

equipment and to meet the dynamic range and recording resolution of the digital recording equipment.

2.2.1 Event-Related Brain Potentials

Since electrophysiological brain activity is the substrate of any cognitive process, external stimulation

modulates the EEG. While the modulatory effect from a single stimulation (i.e., ±1–20 µV) is only a

small fraction of the ongoing brain activity (as large as ±100 µV), repeated stimulation with members

20
 2.2. Electroencephalography

of a pre-defined stimulus class and subsequent averaging across those epochs of the EEG that follow

stimulation can unhinge the event-related part—the ERP—from the background EEG. When the ERPs

to a set of stimulus classes are compared statistically, their difference and variance can guide inferences on

the magnitude of the involvement of a neurophysiological process in the processing of these stimuli; this

neurophysiological process is then presumed to underlie the cognitive process linked to the difference

in stimulus classes (Kutas & Van Petten, 1994; Kutas, Van Petten, & Kluender, 2005).

To obtain a reliable ERP, preprocessing of the raw EEG is necessary. The raw EEG contains arti-

facts (i.e., voltage differences induced by electric currents stemming from internal muscular activity or

external gear; for an overview, see Blume, Kaibara, & Young, 1995), which outsize the ERP in magnitude

and thus may drive spurious ERP effects if not removed from the EEG. Various automatic techniques are

available for artifact identification and removal, such as filtering (e.g., for steady-wave artifacts such as the

mains frequency and its harmonics; cf. Edgar, Stewart, & Miller, 2005), amplitude-based identification

(e.g., for muscular activity; cf. Oostenveld, Fries, Maris, & Schoffelen, 2011) or independent-component

analysis (ICA; Bell & Sejnowski, 1995; Delorme, Sejnowski, & Makeig, 2007).

While ERPs are a relatively low-cost and easy-to-use method to study the neurophysiological appa-

ratus, ERP analysis makes strong assumptions, some of which have been criticized. First, the assumption

of latency invariance (which legitimizes across-trial averaging) may only hold for ERP components that

link to physical stimulus properties—but not for those which reflect higher-level cognitive processing,

for which jitter of single-trial ERPs increases (cf. Donchin & Heffley, 1978). Different circumventions

have been devised, such as an artificial establishment of across-trial temporal correspondence (Woody,

1967); more recently, ICA techniques have been used to avoid averaging altogether and enable the

analysis of single-trial ERPs (Jung et al., 2001; Turi et al., 2012). As a second questionable assumption,

the classical, two-dimensional characterization of ERPs by latency and amplitude potentially neglects

the multi-dimensionality of the neurophysiological bases of the EEG. Traditional ERP analysis assumes

ongoing oscillatory activity—which constitutes a major part of the raw EEG—as random, normally

distributed noise, whereas the ERP is assumed to be independent and invariantly time-locked to the

stimulus. Since it is known that oscillations are functionally meaningful (cf. Berger, 1929; see also

Section 2.2.2), these simplifications potentially underestimate stimulus-linked, but latency-jittered oscil-

latory influences on the ERP, such as band-specific amplitude increases or phase resetting (Başar, 1998;

21
2. General Methodology

Buzsáki, 2006). It is unclear whether oscillatory factors underlie the ERP altogether (for review, see

Klimesch, Sauseng, Hanslmayr, Gruber, & Freunberger, 2007) or constitute an independent marker of

electrophysiological brain activity (Fell et al., 2004; Fuentemilla, Marco-Pallarés, & Grau, 2006; Mäki-

nen, Tiitinen, & May, 2005). In either case, their potential condition-specific effect is not accounted for

in the ERP and may be a lurking variable in statistical analysis.

2.2.2 Time–Frequency Analysis

While ERPs are a valuable tool for analyzing stimulus-dependent electrophysiological brain activity, they

are restricted to capturing time-locked parameters of the EEG; also, their statistical analysis is vulnerable

to condition-specific oscillatory effects (Makeig, Debener, Onton, & Delorme, 2004). Time–frequency

analysis quantifies these oscillatory contributions to the EEG; on the scale that is accessible to scalp-level

EEG, neural oscillations reflect the continuous periodic fluctuation of the membrane potentials of large

groups (> 1000; cf. Nunez & Srinivasan, 1981) of cortical or subcortical neurons that are synchronized

by synaptic interaction (for review, see Bremer, 1958 and Wang, 2010)11 .

Neural oscillations have been early found to be cognitively meaningful (cf. Berger, 1929), playing a

major role in local neuronal processing and long-distance communication between neuronal ensembles.

Potentially, local processing involves high-frequency oscillations, whereas long-distance communication

involves lower frequency bands (Lisman & Idiart, 1995; Sarnthein, Petsche, Rappelsberger, Shaw, & von

Stein, 1998): Groups of neurons may locally increase the amplitude of their fluctuations or lock their

phase with external stimuli, other groups of neurons may lock their fluctuations with distant neuronal

groups to enable joint processing.

The basic assumption underlying TFR is that oscillatory activity can be exhaustively expressed

by a set of sinusoids (Buzsáki, 2006), defined by their amplitude and phase for a given frequency

window. Decomposition of the EEG signal into such a set of amplitude–phase duplets (see panel (B) in

Figure 2.3) for a given frequency window is classically performed by the Fourier transform (Fourier,

1822). While the output of this method is restricted to either the time or the frequency domain (since it

assumes an infinite stationary signal), the later introduced Gabor transform—also known as short-time

11
The full oscillatory spectrum is commonly divided into δ (0.5–4 Hz), θ (4–8 Hz), α (8–12 Hz), β (12–30 Hz), and γ (>
30 Hz) bands. The arbitrariness of this nomenclature has historical reasons.

22
 2.2. Electroencephalography

Fourier transform (Allen, 1977)—enables decomposition for pre-defined time windows (which assumes

stationarity only for each consecutive time window). While the Gabor transform limits the size of the

analysis time window12 , the more recently introduced wavelet transform (Grossmann & Morlet, 1984;

Lachaux, Rodriguez, Martinerie, & Varela, 1999; see also Chapter 6) circumvents this restriction by

convolving the EEG across time–frequency windows with short sinusoidal periods of rising and falling

amplitude (Samar, Bopardikar, Rao, & Swartz, 1999; see panel (A) in Figure 2.3)13 .

A B C
Shift

Real part
frequency

frequency
High

High

Phase
Imaginary part
Magnitude

Frequency
waveform

Power
EEG

Real part
frequency
Low
frequency

Phase
Low

Shift Imaginary part Time

Magnitude

Figure 2.3: Overview of TFR steps in a wavelet analysis; (A) wavelet convolution of the preprocessed
EEG and derivation of transformation coefficients for each time window (small-scale wavelet for
high frequencies at the top, large-scale wavelet for low frequencies at the bottom); (B) calculation
of phase and magnitude of the frequency components based on the transformation coefficients; (C)
resulting time–frequency representation of the original signal, color shading indicating power inside a
time–frequency bin (cf. Samar et al., 1999 and Roach & Mathalon, 2008).

After transformation of the EEG signal, the desired amplitude and phase parameters for a given

frequency and time window correspond to the real and imaginary parts of the complex output of the

short-time Fourier or wavelet transform (Roach & Mathalon, 2008); the squared Euclidean distance

between the real and imaginary parts corresponds to the power inside a time–frequency window of the

EEG spectrum (see panel (B) in Figure 2.3). Any of these parameters for a given experimental condition

can then be compared to their counterpart in a baseline interval or another experimental condition to

quantify stimulation-dependent changes.

12
This is due to Heisenberg’s (1927) uncertainty principle, which states that two parameters of a given observation cannot
be measured simultaneously (cf. Quiroga, 1998); in other words: The Gabor transform per definitionem cannot exactly
determine both amplitude and phase for small time windows.
13
The scalability of these wavelets links their time and frequency domains and implies that time windows decrease with
the analysis frequency. While this permits an asymptotically short analysis time window for high frequencies and an
asymptotically long analysis time window for low frequencies, it also limits the frequency resolution (for discussion, see
Başar, Demiralp, Schürmann, Başar-Eroglu, & Ademoglu, 1999 and Samar et al., 1999).

23
2. General Methodology

2.2.3 Source Localization

To infer the neuroanatomical sources of the results of ERP or TFR, source localization seeks to find

those spatially-localized currents inside a model brain volume that best explain a measured scalp-level

ERP or time–frequency topography (Ramírez, Wipf, & Baillet, 2010). The electrophysiological basis

for source localization is that active neuronal compounds can be well expressed by directional currents

in space (de Munck, Van Dijk, & Spekreijse, 1988; cf. Section 2.2.1), whose magnitude fluctuations over

time drive the EEG topography.

The first step to source localization is to model how hypothetical dipole activity at pre-chosen

sites inside a model brain volume propagates towards the scalp-level EEG electrodes, which is called

the forward model or lead-field matrix (for review, see Grech et al., 2008). It contains micro-level infor-

mation about the local current density of individual neurons as well as macro-level information about

the propagation of electric fields inside the brain and the different head tissues. While early analyses

had used single (Frank, 1952) or multiple concentric (Berg & Scherg, 1994) spheres to model the head’s

macro structure, more recent boundary-element models (BEMs) use a finer-grained three-dimensional

representation of the brain, skull, and scalp surfaces (Besl & McKay, 1992; He et al., 1987; cf. Sec-

tion 6.2.4 and Section 7.2.5). Following the classification of Grech et al. (2008), parametric (a.k.a. dipole

models) and non-parametric (a.k.a. distributed-source models) forward models exist: Parametric models

distribute a small number of dipoles across the model brain volume, whose locations and orientations

are subject to solution. Non-parametric models distribute a large number of dipoles at known locations

across the model brain volume, whose orientations are subject to solution.

In a second step, the forward model is inverted: While the lead-field matrix derives a hypo-

thetical scalp topography based on hypothetical neural generators, inversion enters a measured ERP

or time–frequency topography into the forward model to solve it for the neural generators. Both

classes assess the ability of the dipole arrangement (location, orientation, amplitude) to generate the

to-be-localized data by quantifying the divergence between generated and measured data.

Nevertheless, a drawback of source localization is that it suffers from a spatial version of the

inverse problem (Ambarzumian, 1929; Helmholtz, 1881): In principle, a scalp-level EEG topography at

a given time point can result from an infinite number and infinite arrangements of cortical generators

(for review, see Grech et al., 2008). While this problem is mathematically ill-posed when only the EEG

24
 2.3. Experimental Paradigm

topography is provided for solution, it can be alleviated by incorporating prior knowledge (Ramírez

et al., 2010): The model brain volume used in the forward computations can be further constrained

by incorporating individual structural-MRI data. Second, individual EEG electrode positions can be

determined during recording and linked exactly to the model brain volume. Thirdly, a-priori knowledge

from prior or parallel fMRI studies can be used to constrain dipole positions (cf. Section 7.2.5).

After having critically reviewed the MRI and EEG acquisition and analysis techniques used in the

current thesis, the following section will introduce the experimental paradigm used in all of the current

experiments. Because this thesis interfaces linguistic syntax, experimental psycholinguistics of sentence

processing, and the cognitive neuropsychology of working memory, this will concern the linguistic,

psycholinguistic, and psychological considerations involved in the design of the current paradigm.

2.3 Experimental Paradigm

To address the research questions of the current thesis (cf. Chapter 1), German sentences were

constructed that allow for the orthogonal manipulation of one factor each solely affecting

argument-reordering and argument-storage and -retrieval demands. The 2 × 2 factorial design ac-

cordingly crossed the factors argument reordering (subject-first versus object-first argument order) and

argument storage and retrieval (short versus long argument–verb distance), as shown in Figure 2.4.

Subject-first Shortdistance
Short distance

A Nach einer Saison in der Bundesliga hat der Trainer den

/after(DAT) a(AGR)-SG.F season(AGR) in(DAT) the(AGR)-SG.F german.soccer.league(AGR) have(PST.SG) the-NOM.SG.M coach(AGR)
Stürmer gewürdigt
the-ACC.SG.M striker(AGR) PTCP-honor-PTCP/

B Der Trainer hat nach einer

/the-NOM.SG.M coach(AGR) have(PST.SG) after(DAT) a(AGR)-SG.F
Saison in der Bundesliga den Stürmer gewürdigt
season(AGR) in(DAT) the(AGR)-SG.F german.soccer.league(AGR) the-ACC.SG.M striker(AGR) PTCP-honor-PTCP/

Long distance

Object-first Shortdistance
Short distance

C Nach einer Saison in der Bundesliga hat den Stürmer der

/after(DAT) a(AGR)-SG.F season(AGR) in(DAT) the(AGR)-SG.F german.soccer.league(AGR) have(PST.SG) the-ACC.SG.M striker(AGR)
Trainer gewürdigt
the-NOM.SG.M coach(AGR) PTCP-honor-PTCP/

D Den Stürmer hat

/the-ACC.SG.M striker(AGR)
nach einer Saison in der Bundesliga der Trainer gewürdigt
have(PST.SG) after(DAT) a(AGR)-SG.F season(AGR) in(DAT) the(AGR)-SG.F german.soccer.league(AGR) the-NOM.SG.M coach(AGR) PTCP-honor-PTCP/

Long distance

Figure 2.4: Overview of experimental materials; upper panel shows subject-first orders in the short
(A) and long (B) distance variants, lower panel shows object-first orders in the short (C) and long (D)
distance variants; interlinear glosses are provided below each example; subjects are marked in bold blue,
objects are marked in bold red, main verbs are marked in bold black; arrows illustrate the argument–verb
distance; (A) to (D) translate After a season in the German soccer league, the coach honored the striker.

25
2. General Methodology

In the first condition (A), subject and object are adjacent to the main verb—the argument order

is subject-first, and the argument–verb distance is short. In the second condition (B), the argument

order is still subject-first, while the subject is at the sentence beginning now. This increases storage

demands by lengthening the retention interval for the critical information of the subject noun phrase.

The third (C) and fourth (D) conditions involve the same two distance variants, but in object-first

argument order: The object is in front of the subject. The manipulations maximized the contrast along

an argument-reordering and an argument-storage-and-retrieval dimension, operationalizing argument

reordering as an executive operation on the arguments stored in working memory (cf. Wingfield &

Butterworth, 1984). The argument-reordering factor manipulated the relative argument order, while

not interfering with the argument-storage-and-retrieval factor.

2.3.1 Linguistic Considerations

To maximize the short–long contrast (see Section 2.3.3), past-perfect tense was used in the experimental

materials. In German, this tense necessitates the participle at the sentence-final position, enabling the

resolution of the sentences’ argument–verb dependencies only at the end of the sentence: Because the

sentence-final participle is the main verb in German past-perfect tense, its lexical entry specifies the

number and type of arguments and θ-roles (Binder, Duffy, & Rayner, 2001; Comrie, 1993).

German past-perfect tense also necessitates an auxiliary verb in sentence-second position, which

provides information on the gender and number of the subject argument across conditions (Klein, 2000),

thus enabling partial establishment of the subject–verb dependency across experimental conditions.

However, the auxiliary marks neither the number nor the relative order of arguments—that is, the

verb’s subcategorization frame and θ-grid is not provided.

To further prevent an unbalanced premature determination of the verb’s subcategorization frame

and the relative argument order, the object in the subject-first sentences and the subject in the object-first

sentences were positioned at the identical, verb-adjacent position across experimental conditions. Hence,

across conditions, both a subject and an object would await integration in working memory at the

verb-adjacent position. At all sentential positions prior to the verb, both subject-first and object-first

sentences did not reveal the total number of arguments (e.g., the sentences theoretically still allowed for

additional arguments of a ditransitive verb to occur).

26
 2.3. Experimental Paradigm

2.3.2 Psycholinguistic Considerations

In addition to controlling for formal linguistic features (see Section 2.3.1), the design of the current

paradigm aimed at controlling for a number of previously-reported processing phenomena which, if

unaccounted for, could—possibly condition-specifically—confound the experimental manipulations.

First, the absolute position of the main verb in our stimuli is kept constant across conditions to

counter an effect that has been termed sentence-final speed-up (Demberg & Keller, 2008; Ferreira &

Henderson, 1993). Sentence-final speed-up describes the phenomenon that processing is cumulatively

facilitated towards the end of sentences (Bastiaansen & Hagoort, 2006), potentially as an effect of the

cumulative decrease in the number of possible sentence continuations (Kutas, Lindamood, & Hillyard,

1984; Kutas & Van Petten, 1994; Levy, 2008). Since this decreased processing difficulty has differential

effects at different sentential positions, a variation of the main verb’s position across experimental

conditions can drive spurious effects, which would have been a particular problem when processing

effects at the main verb are the object of study (e.g., the analyses described in Chapter 7).

To counter the contrary processing phenomenon of sentence-final wrap-up (Friederici & Frisch,

2000; Friedman, Simson, Ritter, & Rapin, 1975; Osterhout & Holcomb, 1993), we added an identical

conjunct clause to all sentences of each four-sentence set (e.g., “und die Entwicklung bestätigt.”, translat-

ing to and stated the development. for the examples in Figure 2.4). Sentence-final wrap-up means that

processing difficulty increases in sentence-final regions, potentially due to working-memory saturation

(King & Just, 1991) or argument–verb integration. Since the working-memory resources are limited (see

Section 2.3.3), storing additional material gets increasingly difficult at the sentence ending (cf. Meyer,

2009). In sum, the conjunct clause separated the main verb (i.e., the critical integration point in a

verb-final sentence) from the region in which wrap-up effects become visible.

As an additional global control measure to counteract the processing influence of individual

lexical frequencies on the dependent variables (Allen, Badecker, & Osterhout, 2003; Van Petten, 1993;

Van Petten, Kutas, Kluender, Mitchiner, & McIsaac, 1991), the words at the corresponding sentential

positions across the 48 four-condition stimulus sets were selected according to a lemma-frequency

and syllable-count matching using the CELEX database (Baayen, Gulikers, & Piepenbrock, 1995).

Specifically, each verb’s subject and object were balanced in syllable count and lemma frequency to avoid

systematic confounding of the argument-reordering manipulation: If objects were of consistently higher

27
2. General Methodology

lexical frequency than subjects across the stimulus set, any effects of increased argument-reordering

demands might cancel out due to facilitated lexical access. In a similar vein, semantic coherence of

each individual subject–object–verb set was ensured by a sentential-neighborhood analysis using the

Projekt-Deutscher-Wortschatz database (Biemann, Bordag, Quasthoff, & Wolff, 2004): cross-linguistic

behavioral work has shown that arguments and verbs semantically pre-activate remaining members

of their argument–verb set (i.e., arguments and verbs) that they frequently co-occur with (Boland,

1993; Boland, Tanenhaus, Garnsey, & Carlson, 1995; Kamide et al., 2000, 2003; Marslen-Wilson, 1973;

Pappert, Schließer, & Pechmann, 2008; Trueswell & Kim, 1998; Tsuzuki et al., 2004); if uncontrolled for,

unbalanced typicality of the two arguments might have resulted in asymmetrical pre-activation effects.

2.3.3 Psychological Considerations

To experimentally query working-memory storage and retrieval, the current experiments manipu-

lated argument–verb distance, that is, the length of an argument–verb dependency, which is a classical

approach ever since it was recognized that long-distance dependencies impede comprehension (Behaghel,

1923; Cowper, 1976; Frazier, 1987; Gibson, 2000; Yngwe, 1960).

From the working-memory perspective of the current thesis, decay of the stored argument is

the reason for the increasing processing difficulty associated with long argument–verb dependencies

(cf. Baddeley, 2012; Miyake & Shah, 1999; Peterson & Peterson, 1959). With respect to our short–long

dimension, this entails that either the subject or the object argument in a verb-final sentence will

decay relatively more in a long as compared to a short argument–verb dependency. To maximize the

contrast between the short and long conditions, the arguments in the two short conditions were kept

immediately adjacent to the sentence-final main verb, minimizing the storage interval and according

retrieval demands. In contrast, the two long conditions maximized the argument–verb distance by

putting either the subject or the object argument immediately at the sentence beginning. Thus, the

storage interval in the short conditions crosses a single phrase (i.e., the object in the short subject-first

sentence and the subject in the short object-first sentence), whereas the storage interval in the long

conditions crosses four phrases. The use of the past-perfect tense (see Section 2.3.1) ensured that storage

demands be identical for the two short and the two long conditions, respectively.

28
 2.3. Experimental Paradigm

2.3.4 General Hypotheses

The general hypotheses for the current experimental paradigm are the following: The manipulation of

argument order (subject- versus object-first sentences) is assumed to increase reordering demands. This

should increase behavioral (e.g., reaction time) and functional-neuroanatomical (e.g., BOLD) measures

of processing difficulty for object- as compared to subject-first sentences. The experimental manipulation

of argument–verb distance (short versus long argument–verb-dependency sentences) is assumed to

increase both storage and retrieval demands and lead to according increases in the dependent measures

when long argument–verb-dependency sentences are compared to short argument–verb-dependency

sentences. For more detailed hypotheses with respect to the specific data acquisition and analysis

methods, the reader is referred to the introductions of the following experimental chapters.

29
3
B E H AV I O R A L S T U D Y

3.1 Introduction

In order to understand a sentence it is crucial to determine who is doing what to whom. This necessarily

requires to link the doing—the verb—to the who, what, and whom—the arguments, e.g. subject and

objects. To succeed, sentence processing has to both store arguments in working memory until the

verb is reached (Aoshima, Phillips, & Weinberg, 2004; Clahsen & Featherston, 1999; Nakano, Felser,

& Clahsen, 2002; Nicol et al., 1994; Nicol & Swinney, 1989) and reorder the arguments in case of

object-first orders to re-establish the language’s idiosyncratic argument order. While argument storage

and reordering are intimately intertwined, we do not know exactly how.

There are proposals that inherent properties of working memory such as decay are sufficient

to explain both argument storage and argument reordering (Nakatani & Gibson, 2008; Peterson &

Peterson, 1959; Shulman, 1970; Vasishth & Lewis, 2006). These theories are based on behavioral

paradigms that tap working memory by manipulating the argument–verb distance, thereby lengthening

the argument-storage interval. Some of these paradigms have yielded increasing processing difficulty

with increasing argument–verb distance (in English: Frazier, 1987; Gibson, 2000; Grodner & Gibson,

2005), whereas others found decreasing processing difficulty (in English: Jaeger, Fedorenko, & Gibson,

2005; in German: Konieczny, 2000; Konieczny & Döring, 2003; in Hindi: Vasishth, 2003). This contrast

is usually explained by differences in speakers’ working-memory experience and language-specific

working-memory demands. These interpretations are interesting, but the behavioral findings have not

been substantiated by recurrence to concrete neurocognitive substrates.

31
3. Behavioral Study

Experimental results from languages that allow for argument-order variations question these

approaches: In German (Friederici, Fiebach, et al., 2006), Chinese (Hsiao & Gibson, 2003), Finnish

(Hyönä & Hujanen, 1997), Hungarian (MacWhinney & Pléh, 1988), Korean (Mazuka et al., 2002), and

Japanese (Miyamoto & Takahashi, 2002), object-first sentences were found to increase processing load,

even when controlling for argument–verb distance. Furthermore, there is evidence that retrieval of

arguments from working memory is insensitive to argument order (Nicol, 1993; Osterhout & Swinney,

1993). Working-memory-inherence approaches also disregard the strong evidence for the neural reality

of argument reordering during sentence processing: Cross-linguistic neuroimaging work from German

(Friederici, Fiebach, et al., 2006; Grewe et al., 2005), Hebrew (Ben-Shachar et al., 2004), and Japanese (Kim

et al., 2009; Kinno, Kawamura, Shioda, & Sakai, 2008) has found that activation of the left inferior frontal

gyrus (IFG) increases with argument-reordering demands. Although these languages use morphological

markings to distinguish arguments, these markings may produce ambiguous structures, which can only

be processed on the basis of argument-order information.

To elucidate on the conceptual dialectics of argument storage and reordering during sentence

processing, we crossed an argument-order manipulation with an argument–verb distance manipulation

in a behavioral study. To specifically pinpoint the locus of the assumed processing-difficulty sources, a

classical SPR (i.e., on-line) experiment in combination with a rating (i.e., off-line) study was performed.

The hypothesis is that both increasing argument-storage demands (as induced by increasing argument–

verb distance) and increasing argument-reordering demands (as induced by changing subject-first to

object-first order) linearly increase processing difficulty, independently of the respective second factor.

3.2 Methods

3.2.1 Participants

Forty participants took part (mean age 23.4 years, standard deviation (SD) 2.8 years; 20 females; all

native speakers of German). All of them were right-handed as assessed by an abridged version of

the Edinburgh Inventory (Oldfield, 1971), reported no neurological deficits, and had normal or

corrected-to-normal vision. They never participated in another study using these stimuli or exper-

imental paradigm and were naïve to the purpose of the study. They were paid A
C3.50 for participation

of half an hour.

32
 3.3. Data Analysis

3.2.2 Materials

The full 192-stimuli set (see Section 2.3) was divided into four lists, and each subset was interleaved and

pseudo-randomized with 72 fillers from previous experiments using MATLAB® (The MathWorks, Inc.,

Natick, MA, USA) scripts, giving a total of 120 stimuli per list and participant.

3.2.3 Procedure

Participants were seated in a dimly lit room 70 cm in front of a Sony Trinitron® Multiscan E200

cathode-ray tube (CRT) video graphics array (VGA) monitor with a refresh rate of 75 Hz (Sony Corpo-

ration, Tokyo, Japan), on which the sentences were presented in a proportional, sans-serif font (size 20

px, black letters, white background) using Presentation® (Neurobehavioral Systems, Inc, Albany, CA,

USA). After a fixation cross for 1150 ms, sentences were presented phrase by phrase; participants were

instructed to press the space bar of a computer keyboard to view each successive phrase once they had

understood the previous one. Reading time at each phrasal position was the dependent measure. After

another fixation cross of 1150 ms, a well-formedness-rating task on a 1–6 scale (1 = very good; 6 = very

bad) was presented, which participants were instructed to solve as intuitively as possible. Half of the

fillers additionally introduced a yes–no comprehension question to maintain participants’ attention.

3.3 Data Analysis

Data analysis was performed using SPSS® (SPSS Inc., Chicago, IL, USA). To check whether participants

had attended to the stimuli, mean percentages of correct responses to the comprehension questions were

compared to the overall mean. None of the participants’ mean RTs were further than 3.29 SDs away from

the mean, so none was excluded from statistical analysis. A participant-wise regression analysis was run

to determine multivariate outliers (n = 196) from a Chi-Square statistic on each response’s Mahalanobis

distance (Tabachnick, Fidell, & Osterlind, 2001). To keep the normal distribution spread as wide as

possible, only outliers with p < 0.0005 were excluded. The remaining responses were logarithmized

to correct for the typically F-shaped RT distribution and fulfill the requirements of the Mixed-Model

analysis (Baayen, 2008). Finally, Mixed-Model analyses were run on the RTs, including a participant

regressor as random factor. On the ratings, a multivariate Analysis of Variance (ANOVA) was performed.

33
3. Behavioral Study

3.4 Results

On the ratings, a main effect of argument order was found (F(1,39) = 78.3, p < 0.001), as well as an inter-

action between argument order and argument–verb distance (F(1,39) = 11.9, p < 0.002)—see Figure 3.1.

Clearly the well-formedness of object-first argument orders were rated worse than the well-formedness

of subject-first argument orders; still, object-first argument orders at long argument–verb distance were

rated significantly better than at short argument–verb distance, as were subject-first argument orders at

short argument–verb distance as opposed to long argument–verb distance.

2
Rating

3
Short
Long

Subject-first
Object-first 6

Figure 3.1: Average ratings (n=40); note that the y-axis is inverted (i.e., 1 = very good, 6 = very bad);
error bars represent standard error of the mean (SEM); blue bars mark subject-first sentences, red bars
mark object-first sentences, solid color marks short argument–verb-dependency sentences, striped color
marks long argument–verb-dependency sentences.

The overall Mixed-Model analysis on the RTs revealed main effects of argument order and senten-

tial position (F(1,17064) = 54.3, p < 0.001 and F(1,17064) = 194.5, p < 0.001), as well as interactions be-

tween argument order and argument–verb distance (F(1,17064) = 13.0, p < 0.001), argument order and

position (F(1,17064) = 2.0, p < 0.05), argument–verb distance and argument order (F(1,17064) = 148.1,

p < 0.001), all mirrored in a three-way interaction between argument order, argument–verb distance

and position (F(1,17064) = 7.4, p < 0.001). The position-wise results of the Mixed-Model analysis are

provided in Table 3.1 and Figure 3.2, showing a clear main effect of argument order at the short

argument–verb distance at positions 4 to 8, while such an effect was only present at positions 3 and 4

for the long argument–verb distance.

34
 3.4. Results

Table 3.1: Significant position-wise RT effects.

Argument order Argument–verb distance Interaction

Position
F P F P F P

1 32.72 < 0.001

2 854.84 < 0.001 11.51 < 0.005
3 586.38 < 0.001 70.04 < 0.005
4 6.06 < 0.05 95.20 < 0.001 28.27 < 0.005
5 12.89 < 0.001 10.96 < 0.005 21.70 < 0.005
6 11.73 < 0.005 3.98 < 0.05
7 25.80 < 0.001 4.56 < 0.05
8 10.92 < 0.005
9

A 2.9
Argument Verb
Subject-first
Object-first
Reaction time (log 10 ms)

2.8

2.7

0
Nach einer in der der Trainer den Stürmer
hat gewürdigt und die Entwicklung bestätigt
Saison Bundesliga den Stürmer der Trainer

B 3 Subject-first
Argument Verb
Object-first
Reaction time (log 10 ms)

2.9

2.8

2.7

0
Der Trainer nach einer in der den Stürmer
hat gewürdigt und die Entwicklung bestätigt
Den Stürmer Saison Bundesliga der Trainer

Figure 3.2: Position-wise reading times (log 10 ms; n = 40); (A) shows the short
argument–verb-distance conditions (subject-first in blue and object-first in red), (B) shows the
analogue for the long argument–verb-distance conditions; error bars represent SEM; the positions of
the first argument and the verb are marked in gray.

35
3. Behavioral Study

3.5 Discussion

Our combination of an on-line and off-line study set out to give evidence to the conceptual and neu-

rocognitive bisection of argument storage and argument reordering. To test whether these processes

are functionally distinct, we combined an argument–verb-distance manipulation (short versus long

argument–verb distances) with an argument-order manipulation (subject-first versus object-first sen-

tences). As opposed to the original hypothesis of an independent linear influence of the two factors

on behavioral performance, the results of the current study showed an interaction of argument–verb

distance and argument order, both in the SPR and rating study: Increasing argument–verb distance

seems to increase processing difficulty only for subject-first argument orders; for object-first argument

orders, a long argument–verb distance seems to facilitate processing. We will now discuss possible

explanations for this unexpected finding.

The current data fit results from English, Finnish, and Japanese that show objects in object-first

argument orders to facilitate verb processing (Hyönä & Hujanen, 1997; Kamide & Mitchell, 1999;

Nakatani & Gibson, 2008, 2010; Yamashita, 1997). As an example study from Japanese, Yamashita

(1997) found that object-first argument orders—which collaterally increase the argument–verb distance in

Japanese—do not increase processing load at sentence-final main verbs. As to English data, Nakatani and

Gibson (2010) find that long-distance dependencies facilitate processing only for object-first sentences.

Together, these results speak against proposals that long-distance dependencies in general facilitate pro-

cessing over their short-distance counterparts: While Jaeger et al. (2005) found a processing facilitation of

increasing argument–verb distance, their design did not disentangle argument order and argument–verb

distance, which also holds true for work on German (Konieczny, 2000; Konieczny & Döring, 2003).

A possible explanation for these previous as well as the current order-dependent distance effects

is that while argument order and argument–verb distance interact on a behavioral level, they may

be processed by separate brain regions. If, for example, one brain region were responsible for the

processing of argument order and a second brain region were responsible for argument storage, a direct

neuroanatomical link between these two regions might selectively facilitate the processing of sentences

that tax both argument storage and argument-reordering processes. This proposal would predict a

behavioral interaction—in the absence of interactive effects on the neural level. Furthermore, this

proposal would predict a neuroanatomical link between those brain regions that are active for argument

36
 3.5. Discussion

reordering and those brain regions that are active for argument storage. Such a picture would also

accommodate existing sentence-processing theories that postulate working-memory-inherent factors

as sole processing determinants (Gibson, 2000; Gibson & Thomas, 1999; Lewis & Vasishth, 2005;

Nakatani & Gibson, 2008): Working memory may be necessary for argument storage, but not sufficient

for argument reordering.

A second possibility is that argument order and argument–verb distance interact both on

the behavioral and neural level, as predicted by anticipation-driven or co-occurrence-frequency-based

sentence-processing theories (Konieczny, 2000; Konieczny & Döring, 2003; Levy, 2008). Under

such accounts, the occurrence of an object early in a sentence either opens a time window for

semantic-anticipation mechanisms (Konieczny & Döring, 2003) or reduces the number of possible

sentence continuations (Levy, 2008). Crucially, such a mechanism would be stronger for objects than

subjects (cf. Bader & Bayer, 2006; Bader & Lasser, 1994; Friederici, Wang, Herrmann, Maess, & Oertel,

2000; Schlesewsky & Bornkessel, 2004): While virtually any verb subcategorizes a subject, not any verb

subcategorizes an object (cf. Du Bois, 1987). Hence, the anticipation of a verb based on argument infor-

mation might be easier in object-first sentences. In line with this interpretation, a body of cross-linguistic

behavioral work has shown that arguments semantically pre-activate verbs they frequently co-occur

with (Boland, 1993; Boland et al., 1995; Kamide et al., 2000, 2003; Marslen-Wilson, 1973; Pappert et al.,

2008; Trueswell & Kim, 1998; Tsuzuki et al., 2004). On the neural level, this approach would suggest an

interaction of argument storage and argument reordering, most possibly involving brain regions that

are known to be sensitive to associative mechanisms in the mental lexicon.

Given these two interpretations for the behavioral results, the current experiment needed to

be repeated using a sophisticated method that can uncover the neural factors that underlie behavioral

patterns. Chapter 4 describes this experiment which used fMRI to tackle the neural substrates of the

interactive behavioral effect.

37
4
M A G N E T I C -R E S O N A N C E - I M A G I N G S T U D Y

4.1 Introduction

As outlined in Chapter 1 and Section 2.3, arguments must be linked to their subcategorizing verb to

determine who is doing what to whom from a sentence. This involves argument storage and reordering.

Both storage and reordering are conceptually independent of the subvocal-rehearsal component of

verbal working memory, evidence for which has been introduced in Section 1.2 and will be expanded

in the following sections. Rehearsal refreshes stored information, constantly across the conditions

of the current paradigm; reordering is an executive mechanism, operating on the representations

stored and rehearsed by verbal working memory. While there are firm behavioral and ERP indications

on the neuropsychological reality of both storage and reordering during argument–verb-dependency

processing (Clahsen & Featherston, 1999; Felser et al., 2003; Fiebach et al., 2001; Kluender & Kutas,

1993; Nakano et al., 2002; Nicol et al., 1994; Nicol & Swinney, 1989; Phillips et al., 2005; Ueno &

Kluender, 2003), their respective neural substrates are still a matter of intensive discussion (for review,

see Rogalsky, Matchin, & Hickok, 2008). Furthermore, argument storage and reordering were found

to interact behaviorally, questioning their neural independence. This in mind, an fMRI experiment

was performed (Section 4.2.6.2 and Section 4.3.2). To also link the functional results to the underlying

neuroanatomy more directly, additional diffusion-magnetic-resonance-imaging (dMRI; Section 4.2.6.5

and Section 4.3.4) and VBM (Section 4.2.6.4 and Section 4.3.3) analyses were performed14 .

14
Modified versions of the fMRI and dMRI sections of this chapter have been published previously (Meyer, Obleser, Anwander,
& Friederici, 2012).

39
4. Magnetic-Resonance-Imaging Study

With respect to storage, an imaging study from the sentence-processing domain reports activity

of left inferior parietal cortex to increase with the retention interval for disambiguating information in

ambiguous sentences, that is, storage demands (Novais-Santos et al., 2007). In addition, meta-analyses

and imaging studies from outside the sentence-processing domain suggest left posterior brain regions as

candidates for a storage substrate (Owen et al., 2005; Smith & Jonides, 1998; Wager et al., 2005). Paulesu

et al.’s (1993) seminal study required monolingual English-speaking participants to read, store, and

rehearse either English or Korean letters, of which only the English letters were hypothesized to activate

a phonological code and subsequent storage. This study found the left supramarginal gyrus (SMG) to be

active during the storage of English letters, a finding which was replicated by later studies (Awh et al.,

1996, 1995; D’Esposito, Postle, Ballard, & Lease, 1999; Jonides et al., 1998; Petrides, Alivisatos, Meyer, &

Evans, 1993). However, these studies contrast with other imaging work investigating aspects of working

memory during sentence processing that reports Brodmann area (BA) 45, sometimes extending to the

inferior frontal sulcus, in the left prefrontal cortex to play a role, using paradigms comparing different

syntactic dependencies (Fiebach et al., 2005; Makuuchi et al., 2009; Santi & Grodzinsky, 2007, 2010).

With respect to reordering during sentence processing, cross-linguistic imaging research has found

Broca’s area in the inferior frontal cortex to activate in languages as diverse as Hebrew, German, and

Japanese (Ben-Shachar, Hendler, Kahn, Ben-Bashat, & Grodzinsky, 2003; Bornkessel, Zysset, Friederici,

von Cramon, & Schlesewsky, 2005; Friederici, Fiebach, et al., 2006; Kim et al., 2002). While these studies

controlled for storage demands, they did not explicitly separate reordering and storage. One study

directly contrasted word order and distance between the verb and its arguments, hinting at a certain

independence of reordering and storage within the inferior frontal cortex, but restricted to specific

types of syntactic dependencies (Fiebach et al., 2005). Another study neuroanatomically separated the

processing of syntactic hierarchies in Broca’s area from working memory in the left inferior frontal

sulcus—but this study did not specifically vary reordering demands (Makuuchi et al., 2009).

Some authors have claimed that the role of Broca’s area in sentence processing does not lie in

reordering, but rather in subvocal rehearsal of information stored in working memory (Rogalsky &

Hickok, 2010; Rogalsky et al., 2008). Although Broca’s area was active during subvocal rehearsal in

Paulesu et al.’s (1993) and subsequent studies from outside the sentence-processing domain (Awh et al.,

1996, 1995; Petrides et al., 1993), no previous imaging results have disentangled reordering and rehearsal

40
 4.1. Introduction

during sentence processing. Although Rogalsky et al. (2008) report a decline in sentence-processing

performance under conditions of articulatory rehearsal, they found a control condition (finger tap-

ping during a sentence-processing task) to also selectively decrease sentence-processing performance,

merely suggesting that the presence of a secondary task can affect sentence processing behaviorally.

Furthermore, the claim that activation in Broca’s area during sentence processing stems from subvocal

rehearsal is questioned by clinical evidence that reordering during sentence processing can be impaired

independently from active subvocal rehearsal (Caplan & Waters, 1999; Waters & Caplan, 1996). Addi-

tionally, there is imaging evidence that subvocal rehearsal during sentence processing does not further

increase brain activation in Broca’s area (Caplan, Alpert, Waters, & Olivieri, 2000). In sum, while both

rehearsal of syntactic information and reordering of arguments may rely on Broca’s area, a conceptual

and neuroanatomical identity between reordering and rehearsal is doubtful.

Based on the assumption that the minimal neural representation of subject, object, and verb

involves at least a storage component which stores arguments across the argument–verb distance and a

reordering component which reorders these arguments to avoid a who–whom confusion (cf. Chapter 1),

the following was predicted: Argument reordering is supported by Broca’s area, and the storage of

relevant features of an argument over the argument–verb distance is supported by temporo-parietal (TP)

regions. This hypothesis is tested in the present study, which fully crossed reordering (i.e., argument

order) and storage (i.e., argument–verb distance) in an fMRI investigation. Importantly, if working

memory supports sentence processing by storage rather than subvocal rehearsal, TP-region rather

than prefrontal brain activation should be obtained for increased argument–verb distance. Moreover,

if TP regions subserve storage and Broca’s area subserves reordering, these regions should exhibit a

direct fiber connection, and the local microstructural properties of the fiber bundle—as quantified by

dMRI—should be related to the fMRI activations. To evaluate this, an analysis of dMRI data using

deterministic tractography was conducted and followed by a voxel-based statistic on the characteristic

diffusion parameters. Finally, I reasoned that interindividual variance of the functional activations

(see Section 4.3.2) could be partially predictable from the local microstructural properties of the

gray matter underlying these brain activations. This is especially relevant in the light of previous

discussions on the lateralization of prefrontal brain regions involved in sentence processing (for review,

see Bookheimer, 2002). To elucidate on the structure-to-function relation with respect to sentence

41
4. Magnetic-Resonance-Imaging Study

processing, I performed a correlation analysis between the functional brain activations gathered from

the fMRI data and microstructural properties of the underlying gray matter as gathered from an

additional VBM analysis. For methodological reasons (see Section 4.2.6.4), this analysis was confined to

the functional effects of the reordering factor.

4.2 Methods

4.2.1 Participants

Twenty-four university students (mean age 27.1 years, SD 3.2 years, 12 females, native German speakers)

took part in the experiment. They were matched for their reading span being in the range between

3 and 5 (mean 3.89, SD 0.77) according to an abridged version of the reading-span test (Daneman

& Carpenter, 1980). Participants were right-handed as assessed by an abridged version of the Ed-

inburgh Inventory (Oldfield, 1971), had no reported neurological or hearing deficits, and normal or

corrected-to-normal vision. Participants were paid A

C14 for participating. Written informed consent was

obtained. All procedures received ethical approval by the local ethics committee (University of Leipzig).

4.2.2 Working-Memory Test

Participants’ verbal-working-memory abilities for sequences of items were tested in the two digit-span

sub-tests from the German version of the Wechsler test (Jacobs, 1887; Tewes, 1994); mean forward

digit span was 9.54 (SD 1.69), mean backward digit span was 8.33 (SD = 2.10).

4.2.3 Materials

All sentences of the stimulus set described in Section 2.3 were recorded in a soundproof chamber by

a trained female German speaker with a Sennheiser® MKH 40 condenser microphone and a Roland®

CD-2 digital sound recorder. The recordings were cut and normalized in Praat (Boersma & Weenink,

2001) according to the root-mean-square amplitude of all files. To avoid onset and offset artifacts, a

cosine fade-in and -out sequence of 5 ms was attached. For each participant, an individual pseudo-

randomized list of 216 stimuli was generated using MATLAB® (The MathWorks, Inc., Natick, MA,

USA). A list of 144 stimulus sentences (36 per condition), 36 filler sentences and 36 null events was

drawn in a counterbalanced way from the entire stimulus pool. To maintain participants’ attention,

42
 4.2. Methods

24 trials introduced a who-did-what-to-whom yes–no comprehension question to be answered within

a limited time (e.g., Hat der Trainer den Stürmer geehrt? / Did the coach honor the center forward?).

Comprehension questions appeared in 16.7 % of trials in each of the four conditions at unpredictable

positions in the stimulus set. The proportion of yes–correct and no–incorrect questions was balanced.

4.2.4 Procedure

Stimuli were presented using Presentation® (Neurobehavioural Systems, Inc., Albany, CA, USA). Audi-

tory stimuli were presented using air-conduction headphones (Resonance Technology, Inc., Northridge,

CA, USA). Visual stimuli were presented on a Sanyo PLC-XP50L liquid-crystal display (LCD) extended

graphics array (XGA) mirror-projection system with a refresh rate of 100 Hz (Sanyo Electric Co., Ltd.,

Moriguchi, Japan), mounted onto the headcoil. A sans-serif font in black letters against a gray back-

ground (size 20 px) was used. A trial started with a fixation cross that stayed on screen for the whole

trial. After a random jitter of either 0, 500, 1000 or 1500 ms, an auditory stimulus started (mean length

4.9 s, SD 0.36 s). To keep the number of acquired volumes constant across trials, a trial always lasted

for 8 s, interpolating a silent period and an on-screen fixation cross between stimulus and trial end. In

the case of a comprehension question, such a sequence was followed by a fixation cross of a random

jitter and a visual comprehension question (16.7 % of all trials). The question remained on screen

for 1500 ms and had to be answered by the participant as quickly as possible during this time period.

Subsequent visual feedback was given for 1000 ms by a green happy or red sad emoticon. Again, in order

to keep the duration of the comprehension probes constant, silence and an on-screen fixation cross

were interpolated, such that each comprehension probe would last 4 s. Participants were instructed to

carefully listen to the sentences and to answer the comprehension questions via button press with either

their left or right hand, one hand corresponding to yes and the other to no. Button assignment was

counterbalanced across participants. Since participants were not aware of whether a task trial would

follow the auditory stimulus, all trials were included in the analysis.

4.2.5 Magnetic-Resonance-Imaging Data Acquisition

Functional-, structural-, and diffusion-MR images were acquired with a 3-T Siemens TIM TRIO

scanner (Siemens Healthcare, Erlangen, Germany) at the Max Planck Institute for Human Cog-

43
4. Magnetic-Resonance-Imaging Study

nitive and Brain Sciences in Leipzig, Germany. Functional data were acquired with a 12-channel

headcoil and a T2*-weighted gradient-echo echo planar imaging (EPI) sequence (data matrix 64 × 64,

repetition time (TR) = 2.0 s, continuous scanning, echo time (TE) = 30 ms, flip angle = 90°, band-

width 116 kHz, field of view (FOV) = 19.2 cm, in-plane resolution 3 × 3 mm3 , slice thickness 3 mm,

interslice gap 1 mm, 30 horizontal slices parallel to the intercommissural (AC-PC) line, whole-

brain coverage, 912 volumes), with a functional scan time of 30 min. Diffusion-weighted data were

acquired in a separate session with the same scanner, equipped with a 32-channel phased-array

head array coil. Images were acquired with a twice-refocused spin-echo EPI sequence (Reese et

al., 2003; TE = 100 ms, TR = 12 s, 128 × 128 image matrix, FOV = 220 × 220 mm2 , 88 axial slices

(no gap), resolution 1.72 × 1.72 × 1.7 mm3 ). Additionally, fat saturation was employed together

with 6/8 partial Fourier imaging and generalized auto-calibrating partially-parallel acquisitions

(GRAPPA; acceleration factor = 2; Griswold et al., 2002). Diffusion-weighting was isotropically

distributed along 60 diffusion-encoding gradient directions with a b-value of 1000 s/mm2 . Seven

images with no diffusion-weighting (b0) were acquired initially and interleaved after each block of

10 diffusion-weighted images as anatomical reference for offline motion correction. The dMRI sequence

lasted about 16 minutes. Anatomical T1-weighted 3D magnetization-prepared rapid gradient echo

(MP-RAGE) images (Mugler III & Brookeman, 1990, inversion time (TI) = 650 ms, TR = 1300 ms,

alpha = 10°, FOV = 256 × 240 mm2 , 2 acquisitions, 1 mm isotropic resolution) were previously

acquired with a non-slice-selective inversion pulse followed by a single excitation of each slice.

4.2.6 Data Analysis

4.2.6.1 Behavioral Data

For the behavioral data, d" -scores and reaction times were calculated. D" -scores are a more adequate

representation of participants’ performance than mean-percentage-correct scores in that they eliminate

participants’ response bias (i.e., the individual tendency to either press the yes–correct or no–incorrect

button; Macmillan & Creelman, 2005). A one-sample t-test on the difference between the d" -scores and

chance-level performance (50 % correct responses) was performed. A 2 × 2 ANOVA was run on the

response data to determine condition-specific effects. From the individual scores on the forward- and

44
 4.2. Methods

backward-digit-span subtests, a composite score—the Mahalanobis distance—was calculated, based on a

zero-centered sample with covariance corresponding to the participant sample (Tabachnick et al., 2001).

4.2.6.2 Functional-Magnetic-Resonance-Imaging Data

Functional data analysis was performed using the SPM8 software package (Wellcome Imaging Depart-

ment, University College, London, UK). Before undergoing statistical analysis, the functional data

were co-registered using the corresponding high-resolution 3-D structural images. They were resampled

to 3 × 3 × 3 mm3 voxel size. Further preprocessing was performed by realigning the functional time

series to the first image, correcting them for slice timing and field inhomogeneities. Next, normaliza-

tion to the standard MR template (gray-matter segmentation-based procedure) and smoothing using

an isotropic 8-mm kernel were applied. For statistical analysis, a participant-wise GLM was estimated

using the canonical HRF from the SPM8 software package, starting at sentence onset and spanning

the individual stimulus length (mean 4.89 s, SD 0.36 s), and treating fillers, silent trials, and comprehen-

sion questions as regressors of no interest. In addition, I included the performance measures for each

condition (see Section 4.3.1) as a regressor of no interest in order to factor out performance-related

variance from the analysis. A high pass filter of 1/100 s was used to attenuate slow global signal changes.

Contrast estimates for the four experimental conditions (compared against the global mean) were passed

into a second-level within-subjects ANOVA, in which main effects and interactions were assessed.

For thresholding of the statistical parametrical maps, an AFNI-implemented Monte-Carlo simulation

(NIMH Scientific and Statistical Computing Core, Bethesda, MD, USA) ensured that a cluster extent of

at least 51 voxels and an uncorrected voxel-wise p-value of 0.005 would protect against whole-volume

type I error at α = 0.05. For anatomical assessment of functional activations, cytoarchitectonic maps

provided with the SPM anatomy toolbox (Eickhoff et al., 2005) as well as probability maps for the

planum temporale (Westbury, Zatorre, & Evans, 1999) were used.

4.2.6.3 Correlation Analysis

Using the Marsbar toolbox (Brett, Anton, Valabregue, & Poline, 2002), the linear predictor values for

each participant and design cell for the resulting regions of interest (ROIs)—as defined by the group-peak

activation-clusters—were transformed into percentages of signal change. From these, difference values

45
4. Magnetic-Resonance-Imaging Study

for the two main manipulation effects (i.e., the ordering and storage effect) were computed and Pear-

son’s linear correlations between the individual activations and Mahalanobis-transformed composite

digit-span scores (see Section 4.2.2) were run.

4.2.6.4 Voxel-Based Morphometry

The anatomical scans of the 23 participants who remained in the statistical analysis (see Section 4.3.1)

underwent VBM analysis as implemented in SPM8. First, each participant’s gray-matter voxels were

segmented from the previously acquired T1 images using a rigid-body procedure (Ashburner & Friston,

2005). The segmented gray-matter volumes underwent DARTEL as described by Ashburner (2007): First,

a group-specific template was generated by iteratively matching the individual participants’ gray-matter

images to a group-mean template. Second, each participant’s individual gray-matter volume was warped

onto this template image. The resulting images were modulated for normalization bias (i.e., differential

non-linear warping magnitude per voxel; see Ashburner, 2009; Richardson et al., 2011), smoothed with

an isotropic 8-mm kernel, and resampled to 1.5-mm isotropic voxels.

The available cytoarchitectonically-defined brain atlases do not yet contain reliable ROI defini-

tions for the TP region; thus, our VBM analysis was confined only to the first half of our fMRI results

(see Section 4.3.2). Using the probability maps provided with the SPM anatomy toolbox (Eickhoff et

al., 2005), ROI volumes for the left and right BA 44 and 45 were generated (see Section 4.3.2). These

volumes were used to extract the gray-matter probability values for the left and right BA 44 and 45

from participants’ MNI-normalized gray-matter volumes. The same ROI volumes were used for the

extraction of the functional data using the methodology described in Section 4.2.6.3.

To check for a direct correlation between the underlying neuroanatomy of the anatomically

defined left and right IFG (i.e., BA 44 plus 45 for each hemisphere) and the functional activation during

the experiment (see Section 4.3.2), I calculated asymmetry values for the underlying gray matter as well

as the functional activations by subtracting the corresponding values for the left and right hemisphere.

From these values, the minimum for each hemisphere was subtracted, and the resulting values were

divided by the new maximum to arrive at difference values between minus one and one. Positive values

of these scores indicate a rightward asymmetry of the functional activation or gray-matter probability

values, respectively, whereas negative values indicate a leftward asymmetry.

46
 4.2. Methods

Between the structural and functional lateralization indices, a partial Pearson’s linear correlation

analysis was run, factoring out participants’ total intracranial volume (i.e., the summed volume of

each individual’s gray and white matter and cerebrospinal fluid) to avoid gross gray-matter-volumetric

differences to obscure the statistical result.

4.2.6.5 Diffusion-Tensor Imaging and Fractional Anisotropy

For 22 of the 23 participants who remained in the analysis (see Section 4.3.1), dMRI data were available

and were analyzed using LIPSIA (Max Planck Institute for Human Cognitive and Brain Sciences,

Leipzig, Germany), FSL (FMRIB, University of Oxford, United Kingdom) and SPM8 (Wellcome

Imaging Department, University College, London, UK) on a Linux workstation. T1-weighted structural

scans were skull-stripped and co-registered to Talairach space (Talairach & Tournoux, 1988). Motion

correction for the diffusion-weighted images was performed based on the 7 reference images distributed

over the entire sequence using rigid-body transformations (Jenkinson, Bannister, Brady, & Smith, 2002)

implemented in FSL. Motion correction parameters were interpolated for all 67 volumes and combined

with a global registration to the T1 anatomy. The transformations were applied for all volumes resulting

in a 1-mm isotropic voxel resolution. The gradient direction for each volume was corrected using the

rotation parameters. Finally, for each voxel, a diffusion-tensor model (Basser et al., 1994) was fitted, and

the FA was computed (Basser & Pierpaoli, 1996).

To robustly analyze an across-participants correlation between the FA and the signal change inside

the functional-activation clusters, the FA values within the skeleton of the fiber bundle connecting the

activated areas were analyzed. Following the approach of tract-based spatial statistics (TBSS; Smith

et al., 2006) implemented in FSL, all FA images were normalized into a standard brain space, and

a group-average FA image was computed. A mean FA skeleton was created from this image, which

represents the centers of all tracts common to the group. The local maxima of the individual aligned

FA data representing the individual centers of the tracts were then projected onto this skeleton.

Anatomical connectivity between the functionally activated areas was investigated by tractogra-

phy from the diffusion tensor maps to estimate the location of the corresponding fiber bundle within

the individual brain volume (Anwander, Tittgemeyer, von Cramon, Friederici, & Knösche, 2007). Here,

deterministic tractography using the entire diffusion tensor to deflect the estimated fiber trajectory was

47
4. Magnetic-Resonance-Imaging Study

used (Lazar et al., 2003) as implemented in MedINRIA (Asclepios, INRIA, Sophia Antipolis, France)

according to Fillard, Pennec, Arsigny, and Ayache (2007). The same preprocessing chain as for the

computation of the FA maps described above was used, except for the fact that the diffusion tensors

were computed with an isotropic resolution of 1.7 mm. Fiber trajectories were started in all voxels with

an FA > 0.2 resulting in a complete set of trajectories within the whole brain.

To use the functional-activation clusters (see Section 4.3.2) as symmetric seed and target regions

in deterministic fiber tracking, the group-average statistical maps were back-projected into each par-

ticipant’s native image space, and all fiber trajectories connecting the two regions in each individual

participant were selected as white-matter connections. The resulting individual tract volume was mapped

onto the voxel space of the individual anatomical scan. These tractograms were projected onto the

group-average FA data set using the transformation matrices obtained in the TBSS analysis. Voxels inside

these normalized tracts were included in an individual logical map, and these maps were summed across

participants to obtain a group-level probability map. The probability map was thresholded at p < 0.5

(i.e., at least 50 % of participants had a tract at this voxel) to leave only the core volume intact (Reich,

Ozturk, Calabresi, & Mori, 2010). This provided an across-participants search volume connecting the

functional-activation clusters and only representing white-matter voxels present across participants.

This volume was used to mask a voxel-wise multiple-regression analysis in SPM8 on the aligned

individual data as resulting from the TBSS procedure. Individual percent signal changes from the

two functional clusters (see Section 4.3.2) were used as regressors, correcting for total intracranial

volume, age, and gender. The volume for the statistical analysis encompassed only voxels inside the

final tract volume (1561 voxels), and an AFNI-implemented Monte Carlo simulation (NIMH Scientific

and Statistical Computing Core, Bethesda, MD, USA) ensured that a cluster of at least 7 voxels at a

voxel-wise p-value of 0.05 protected against type I errors at α = 0.005. All stereotaxic coordinates are

reported in MNI space.

4.3 Results

4.3.1 Behavioral Results

Mean d" -score across conditions was 0.66 (SD 0.94) with a mean response bias of c = –0.22 (SD 0.59),

and mean reaction time was 997 ms (SD 155 ms). One participant was excluded from all further analyses

48
 4.3. Results

because his average response time of 450 ms was outside of the 95-% confidence interval (CI) for the

group. Although the d" -scores indicate serious task challenges for the listeners, a one-sample t-test on

these scores showed that participants’ performance was significantly better than chance (i.e., a d" of

zero; t(22) = 3.49, p < 0.005). Furthermore, the 2 × 2 ANOVA on neither the condition-specific scores

(Figure 4.1 A) nor RTs (Figure 4.1 B) yielded any main effects or interactions (p > 0.2). This indicates

that the current design was free of processing-difficulty confounds.

A 0.8
B 1200

Reaction time (ms)

800
D−prime

0.4

400
Short
Long

Subject-first
Object-first 0 0

Figure 4.1: (A) d" -scores and (B) reaction times for all four conditions; error bars mark SEM.

4.3.2 Functional-Magnetic-Resonance-Imaging Results

For the main effects of both reordering and storage, focal and exclusive supra-threshold activations

in the left hemisphere were obtained. As shown in the panels (A) and (B) of Figure 4.2 (shown in

red), a test for a main effect of reordering elicited activation in the left pars opercularis of the IFG,

peaking at x = –54, y = 10, z = 18. According to cytoarchitectonic maps (Eickhoff et al., 2005), 64.8 %

of the obtained activation mapped onto BA 44, while 19.5 % mapped onto BA 45. Panels (B) and (C)

of Figure 4.2 (shown in blue) illustrate the activation for the main effect of storage, which peaked

at x = –42, y = –40, z = 10. This activation peak maps onto lateral superior parts of the left planum

temporale (PT) with an across-participant probability of 26–45 %, (Westbury et al., 1999). Since this

activation cluster included regions both in the temporal and inferior parietal region, it is referred to as TP

activation. The interaction between the two factors did not produce significant activations; specifically,

there was no evidence for prefrontal activation by the storage manipulation. The full set of activations

is shown in Table 4.1.

49
4. Magnetic-Resonance-Imaging Study

Table 4.1: Overview of significant clusters in the functional contrasts surviving the 51-voxel threshold
at p < 0.005 to achieve whole-volume type-I-error control at p < 0.05.

MNI coordinate
Site Cluster size (mm3 ) Z-score
X Y Z

Object-first > subject-first

Left BA 44 / pars opercularis* –54 14 13 1602 3.76

Long > short

Left planum temporale** –42 –40 10 3.66
Left supramarginal gyrus*** –45 –28 22 1485 3.27
Left superior temporal gyrus*** –57 –25 1 2.89

*According to Eickhoff et al. (2005), **according to Westbury et al. (1999), ***according to Talairach and Tournoux (1988).

A B C
0.1 0.4

% signal change
% signal change

0 0.2
Short

Short
Long

Long

−0.1 0

Subject-first Subject-first
R
Object-first Object-first

4 4

Z Z

L
1 Main effect ordering 1
Main effect storage

Figure 4.2: Brain activations and signal change (bar plots including SEM) for (A) the reordering effect
(red clusters) and (C) the storage effect (blue clusters). Activations are thresholded at p < 0.005 at a
minimum cluster size of 51 suprathreshold voxels to achieve type-I-error control at p < 0.05. For the
reordering factor, 64.8 % of the activation is in the left BA 44 (peak at x = –54, y = 14, z = 13; z = 3.76);
for the storage factor, the activation is found in the medial left TP region (peak at x = –42, y = –40,
z = 10; z = 3.66).

50
 4.3. Results

Notably, a Pearson’s linear correlation analysis found that the difference in signal change in the

left TP ROI (as defined by the functional effect in the TP region) was negatively correlated with the

combined digit-span scores (r = –0.45, p < 0.05). Participants with higher digit span show relatively

less signal change in the contrast reflecting the storage effect (Figure 4.3 A). There was no substantial

correlation between activity in the IFG in the storage contrast and the combined digit-span scores

(r = –0.22, p > 0.3; see Figure 4.3 B). In sum, individual combined digit spans were able to explain

r2 ≈ 20 % of the storage effect variance in the TP region, but only r2 ≈ 4 % of the analogue effect in

the IFG.

A B
"#2 "#2
% signal change in TP region

% signal change in IFG

$%&%'"#()
*%+%"#")
"#1 "#1
$%&%'"#,,
*%-%"#."

0 0

!"#1 !"#1
10 20 30 40 50 10 20 30 40 50
Combined digit span Combined digit span

Figure 4.3: (A) Negative correlation of storage effect in the TP region with combined digit span
(r = –0.45, p < 0.05); (B) analogue correlation between activation levels in BA 44 and combined
digit span (r = –0.22, p > 0.3).

4.3.3 Voxel-Based-Morphometry Results

While most participants in the experimental sample (n = 16) showed left-hemispheric asymmetry

(i.e., asymmetry scores > 0; see Figure 4.4) for the functional activation of the argument-reordering

effect—and the main effect of argument reordering became significant only in the left IFG on the

whole-brain level, see Section 4.3.2—, some participants (n = 7) showed a right-hemispheric asymmetry

(i.e., asymmetry scores < 0; see Figure 4.4). For the structural data, asymmetry was balanced across the

group (leftward n = 12, rightward n = 11).

51
4. Magnetic-Resonance-Imaging Study

Z = 14 Z = 18 Z = 10 Z = 20 Z = 22 Z = 26

Z = 24 Z = 14 Z = 10 Z = 24 Z = 20 Z = 24

Z = 22 Z = 26 Z = 12 Z = 18 Z = 22 Z = 20

0 2

–2 0
Z = 18 Z = 18 Z = 30 Z = 22 Z=2

Figure 4.4: Individual brain activations (slices displayed in neurological convention) during the contrast
object-first > subject-first for the 23 participants in the current analysis, masked for left and right
BA 44 and 45.

A positive relationship between functional activation and gray-matter probability holds for the

left and right hemispheric IFG ROI (r2 = 0.15 and r2 = 0.13, respectively; both p > 0.05). While the

effect sizes were too low for the individual hemispheric regressions to reach significance (Cohen’s

f2 = 0.18 and Cohen’s f2 = 0.15, respectively), the partial Pearson’s linear correlation analysis between

the lateralization values for the functional and structural data across the bilateral IFG ROI volumes

yielded a significant correlation (partial r = 0.66, p < 0.001). Figure 4.5 shows the significant correlation

between the functional and structural lateralization indices inside the left and right IFG ROI volumes

(i.e., left versus right combined BA 44 and 45).

52
 4.3. Results

Functional laterality index

"#5

"

!"#5 $%&%"#00
*%+%"#""/

!/
!/ !"#5 " "#5 /
Anatomical laterality index

Figure 4.5: Correlation between functional and structural lateralization indices; negative values indicate
rightward lateralization, positive values indicate leftward lateralization.

4.3.4 Diffusion-Tensor-Imaging and Fractional-Anisotropy Results

The correlation analysis on the FA values inside the thresholded probability map identified parts of the

arcuate fasciculus (AF)/superior longitudinal fasciculus (SLF), connecting the posterior temporal cortex

and the IFG, as differentially correlating with the experimental factors: A comparison between the

storage and reordering correlations yielded supra-threshold clusters in the middle and posterior AF/SLF.

The reverse comparison yielded a small supra-threshold cluster in the frontal AF/SLF, adjacent to the

IFG. As can be seen in Figure 4.6, the posterior part of the correlation is adjacent to the functional

cluster in the left TP region that was obtained for the main effect of storage and terminates below the

middle superior temporal gyrus (STG). Table 4.2 provides an overview of the significant clusters.

Table 4.2: Significant clusters in the reordering > storage and storage > reordering contrasts on the
FA values of the AF/SLF (thresholded at 7 voxels and p < 0.05, type-I-error-controlled at p < 0.005).

MNI coordinate
Site Cluster size (mm3 ) Z-score
X Y Z

Reordering > storage

AF/SLF –38 –2 26 9 2.13

Storage > reordering

–34 –39 21 175 3.56
AF/SLF
–38 –16 28 80 3.33

*Labels according to Mori et al. (2006).

53
4. Magnetic-Resonance-Imaging Study

Y = –2 Y = –18

Fasciculi
Reordering > storage
Storage > reordering

Figure 4.6: Correlation of FA values with signal change of the storage effect in the left TP region as
compared to the reordering effect in the IFG (blue) as well as the reordering effect as compared to the
storage effect (red). The regression analysis resulting from the averaging of the individual fiber tracts was
carried out in a volume of 1561 voxels of the AF/SLF (green). Clusters were thresholded at p < 0.05
and a minimum cluster size of 7 suprathreshold voxels to control for family-wise error at p < 0.005.

4.4 Discussion

The current MRI study set out to separate the neural correlates of reordering and storage during

sentence processing in the light of previous evidence on the neuropsychological reality of these two

concepts. I also sought to elucidate on the ongoing discussion of the role of the structural and functional

lateralization of Broca’s area concerning these processes. In the following, I will first briefly discuss the

behavioral results, before I turn in detail to the functional and structural MRI findings.

4.4.1 Behavioral Results

Although d" -scores were significantly above chance, for 7 participants, they were relatively low. I assume

that the infrequent occurrence of task trials (16.7 % of trials) and the short time window in which

participants had to respond (1500 ms) are responsible for this comparably poor performance. In addition

to the fact that the inclusion of d" -scores as a regressor of no interest in the fMRI analysis accounted

for performance-related variance in the data, it is noteworthy that an exploratory exclusion of these

participants from the fMRI analysis affected the statistical results only quantitatively, that is, the overall

pattern of activations was not changed—despite the low statistical power resulting from the small

54
 4.4. Discussion

remaining group size, both effects were robust, with the IFG effect of 179 voxels (z = 3.34, peak at

x = –57, y = 17, z = 4), and the TP-region effect encompassing multiple TP sub-peaks of 52 voxels

altogether (z = 2.71, main peaks at x = –60, y = –25, z = 46 and x = –45, y = –28, z = 22).

This is in line with results by Caplan, Chen, and Waters (2008) and Newman, Lee, and Ratliff

(2009), who observed that parts of the left IFG activate more for increased reordering demands, regardless

of the experimental task performed or task performance. Furthermore, this is in line with the finding

that none of the functional effects in the current study showed a correlation with d" -scores. Thus, I am

confident that the experimental task did keep participants’ attention directed towards the sentences,

while leaving the functional results unbiased.

4.4.2 Functional-Magnetic-Resonance-Imaging Results

The findings from the functional experiment are straightforward: I found a clear activation for

the reordering factor in the left IFG (object-first sentences leading to stronger activation than

subject-first sentences) and an as-clear but remote activation for the storage factor (long argument–verb

distances leading to stronger activation than short argument–verb distances) in the left TP region.

There was no significant interaction between the reordering and storage factor, even though visual

inspection of the signal change in the IFG and TP region may suggest differently (see Figure 4.2, A and

C). Moreover, a correlation of memory scores with the storage-related activation of the TP region, but

not the IFG, was observed (see Figure 4.3).

These findings suggest that the processing of complex sentences relies both on the storage of argu-

ments in working memory, supported by left TP regions, and the reordering of arguments, supported

by Broca’s area. The correlation between activity in left TP regions and individual memory capacity

further implies that the present “storage” manipulation did indeed tap storage. This highlights the role

of the left TP region as a neural substrate of memory storage (e.g., Buchsbaum & D’Esposito, 2008;

Jacquemot & Scott, 2006), and more importantly, it emphasizes that the TP region serves this function

also in the processing of complex sentences.

The view that Broca’s area supports sentence processing by its role in subvocal rehearsal is not

supported by the current data: In the conception of Baddeley (2012; 2009), working memory necessarily

involves both storage and rehearsal, the latter providing a constant refreshing of the content stored by the

55
4. Magnetic-Resonance-Imaging Study

former. Paulesu et al.’s (1993) and subsequent working-memory studies (Awh et al., 1996, 1995; Petrides

et al., 1993) have conceptualized the neural interplay of storage and rehearsal as a fronto-temporal

network involving the IFG (implied in rehearsal) and TP regions (implied in storage). In the current

study, the IFG did neither exhibit sensitivity to argument–verb distance (the working-memory factor

in the current paradigm), nor did its activation correlate with behavioral measures of working-memory

ability, unlike TP-region activation. To put it differently: While during verbal working memory the

IFG may underlie subvocal rehearsal in the phonological loop, it is doubtful that IFG sensitivity to an

argument-order manipulation in the present study reflects subvocal rehearsal. I will now discuss these

findings in more detail.

4.4.2.1 The Left Inferior Frontal Gyrus Activates for Reordering

The increased activity in BA 44 elicited by increased reordering demands (i.e., object-first com-

pared to subject-first sentences) in the present study is in line with previous cross-linguistic

functional-neuroimaging research from languages as diverse as German (Friederici, Fiebach, et

al., 2006; Obleser et al., 2011; Röder, Stock, Neville, Bien, & Rösler, 2002), Japanese (Kim et al., 2009;

Kinno et al., 2008), and Hebrew (Ben-Shachar et al., 2003). In all of these studies, increased activation for

object- as compared to subject-first sentences was elicited in the left IFG, but not in the left TP region.

First, in German, three studies directly contrasting object- and subject-first sentences reported the

inferior pars opercularis (BA 44) to be increasingly active as a function of increased reordering demands.

Friederici, Fiebach, et al.’s (2006) study which visually presented sentences, keeping the argument–verb

distance constant across conditions, excluded a possible explanation in terms of storage demands.

Obleser et al. (2011) used acoustic versions of the same stimuli, again finding activation in BA 44. Both

results are in line with Röder et al.’s (2002) finding that object-first argument orders in acoustically

presented sentences elicit activation in the left IFG. Second, similar evidence from Japanese strengthens

this interpretation. Both Kinno et al. (2008) and Kim et al. (2009) contrasted Japanese object- and

subject-first sentences in visual fMRI studies. The fact that Japanese allows for a constant argument–verb

distance across both object- and subject-first sentences allowed their experimental paradigms to avoid

differences in storage demands across experimental conditions. Both studies found very close areas in

the left IFG to increase in activation for object- as compared to subject-first sentences. Third, evidence

56
 4.4. Discussion

from a Hebrew study (Ben-Shachar et al., 2003) expands the cross-linguistic picture. Again, Hebrew is

a language that allows the fixing of storage while manipulating reordering demands. Ben-Shachar et

al.’s (2003) study used auditory presentation of object- or subject-first sentences, the contrast yielding

activation peaks in the left IFG in both a whole-brain and a ROI analysis.

The fact that the current results converge on this body of work strengthens the position that

BA 44 as part of Broca’s area is engaged in the reordering of arguments; it furthermore suggests that this

function is independent of a particular language and input modality, linking to English-language studies

contrasting the processing of object- and subject-first sentences across input modalities (Constable et

al., 2004) and sentences of varying processing difficulty across input modalities (Braze et al., 2011). An

overview of these converging results is given in Figure 4.7.

In spite of this convergence, English-language data have resulted in the claim that the contribution

of Broca’s area to sentence processing is that of subvocal rehearsal as part of the working-memory

network (Just, Carpenter, Keller, Eddy, & Thulborn, 1996; Rogalsky & Hickok, 2010). Unfortunately,

the data on which this view is based are not unequivocal, since the relevant studies on reordering and

storage in English contrasted subject- and object-relative clauses. While such English clauses certainly

scrutinize some aspect of working memory (by varying the number of phrases between an argument

and the verb), they also collaterally introduce a reordering manipulation: While the former (short

versus long argument–verb distance) would tax brain regions that subserve storage, the latter (object-

versus subject-first sentences) would rather tax brain regions that are concerned with the reordering of

arguments. Thus, the English results are partially ambiguous in that they might be ascribed to either

storage or reordering.

Additionally, clinical evidence demonstrates that the ability to process sentences with increased

reordering demands can be independently impaired from subvocal-rehearsal abilities (Caplan & Wa-

ters, 1999; Waters & Caplan, 1996), and imaging data suggest that subvocal rehearsal during sentence

processing does not further increase brain activation in Broca’s area (Caplan et al., 2000). Further

patient data (Martin, 1987; Martin, Blossom-Stach, Yaffee, & Wetzel, 1995; Martin & Romani, 1994)

are in line with these results in that they show that selectively impaired rehearsal abilities do not result

in sentence-comprehension deficits. The results of the current correlation analyses are in line with

these reports: If subvocal rehearsal was the driving force behind Broca’s area’s activity in the current

57
4. Magnetic-Resonance-Imaging Study

study, one would have expected activation in BA 44 to correlate with participants’ combined digit-span

scores—instead, only brain activity in the TP region was correlated with combined digit-span scores.

In contrast to the above evidence that rehearsal and reordering are conceptually and neurally

distinct, Rogalsky et al. (2008) report a decline in sentence-processing performance under conditions

of articulatory rehearsal. This result, however, was not fully conclusive, in that a control condition

(finger tapping during a sentence-processing task) also led to a selective decline in sentence-processing

performance, suggesting that the presence of a secondary task can affect sentence processing behaviorally.

Figure 4.7 puts these data into a broader perspective, listing neuroimaging findings from studies

inside and outside the sentence-processing domain. It becomes obvious that there is a relative proximity,

but no full neuroanatomical overlap between those prefrontal brain regions that support reordering

and those regions that support articulatory rehearsal. Thus, there may be a differentiation within the

prefrontal cortex with respect to reordering and the rehearsal component of working memory which,

in turn, is clearly separated from the storage component located in more posterior regions.

80

Present study
40 Storage
Ordering
MNI z−coordinate

Syntax studies
0 Storage
Ordering

Item studies
−40 Storage
Rehearsal

40 0 −40 −80
MNI y−coordinate

Figure 4.7: Overview of results from cited studies on verbal working memory in sentence-processing
and non-sentence-processing paradigms. Circles mark sentence-processing studies, diamonds mark
non-sentence-processing studies. It is visible that storage (blue) is more likely to activate TP regions,
whereas prefrontal regions dissociate between reordering (red circles) and rehearsal (green diamonds).

58
 4.4. Discussion

4.4.2.2 The Left Temporo-Parietal Region Activates for Storage

This brings us to the second main finding, a cluster in the left TP region that showed both increased

activation with increasing storage demands and a correlation with digit span. A number of studies suggest

that the left TP region (mainly the PT, but often extending into or peaking in the SMG) is critically

involved in storage processes as diverse as storing an ambiguous sentence structure, storing arguments at

increasing argument–verb distances, and storing non-sentential items in the phonological loop, making

left TP regions a strong candidate for the neural substrate of storage, both in sentence processing and

item-based tasks (see Figure 4.7 for an overview of the studies discussed here).

Amongst this work are studies that searched for the neural correlates of the phonological loop

(for review, see Baddeley, 2012), i.e., outside of sentence processing proper. Paulesu et al. (1993) suggested

the left SMG as the main region subserving storage. Other imaging studies also consider left posterior

regions to subserve storage; as mapped in Figure 4.7. However, the exact locations of the activation

peaks in these studies encompassed various regions, including left STG (Kim et al., 2002), SMG (Awh et

al., 1996; Clark et al., 2000; Paulesu et al., 1993), and left inferior and posterior parietal cortices (Awh et

al., 1995; Becker et al., 1996; Bushara et al., 1999; Clark et al., 2000; Gruber & von Cramon, 2001, 2003;

Jonides et al., 1998; Owen et al., 2005).

For sentence processing, Grossman et al. (2002) found that seniors who show difficulties in

processing sentences with increased storage demands activated left parietal cortex relatively less as

compared to younger participants. Accordingly, the authors suggest reduced storage resources in seniors

amongst the sources of their sentence-processing difficulties. The location of the reported regions (see

Figure 4.7) is in line with our finding that left TP activation correlates with digit span. The fact that

this correlation is negative fits this interpretation: In the current study, participants with relatively

better storage abilities (as tested by digit span) showed relatively less activation in the left TP region,

suggesting more efficient storage in these participants.

Additional evidence for a general storage component in left posterior cortex comes from clinical

studies on patients with damage to the left STG, SMG, or TP region. The observation that damage to

the STG causes impaired storage abilities has been made by Leff et al. (2009), who found gray-matter

integrity in the STG in stroke patients to correlate with digit span. Recently, these findings have

been augmented by studies on conduction aphasia (Buchsbaum et al., 2011; Fridriksson et al., 2010)

59
4. Magnetic-Resonance-Imaging Study

demonstrating a relation between speech-repetition problems, phonological-working-memory deficits,

and gray-matter damage to the left SMG and TP region. While such results can be ambiguous in that

conduction aphasics usually suffer from both gray- and white-matter lesions, direct causal evidence is

provided by a repetitive-transcranial-magnetic-stimulation (rTMS) study (Romero, Walsh, & Papagno,

2006) showing that rTMS applied to the SMG causes a decrease in digit span.

These somewhat heterogenous localization patterns in the posterior regions across studies may

result from distinct working-memory sub-processes tapped by the various paradigms employed, since

in addition to rehearsal and storage as discussed above, retrieval is assumed to be a separate compo-

nent of working memory. Evidence for a segregation of storage and retrieval comes from a recent

working-memory fMRI study by Ravizza et al. (2011). In a series of ROI analyses, these researchers

found storage of verbal material to correlate with brain activity in the left posterior STG, whereas

retrieval of verbal material activated the left TP junction. Functional heterogeneity in this area related

to differential working-memory processes has been previously suggested by Henson et al. (1999) and

Buchsbaum, Hickok, and Humphries (2001), who report distinct sub-peaks during a working-memory

task in the posterior STG and inferior parietal cortex. Although the current results can not decide on

the neural underpinnings of separable storage and retrieval sub-processes, the activation observed in the

left TP region involved two sub-peaks (see Table 4.1), one clearly in the PT, one in the SMG. Hence,

the current experimental manipulation of argument–verb distance may have taxed both storage and

retrieval, due to the longer retention interval and accordingly higher re-activation demands induced by

memory decay (cf. Baddeley et al., 2009).

A final and important issue that needs to be addressed is the apparent contrast between the present

data and some previous studies on working-memory demands during sentence processing. In English,

direct comparisons of either pronoun binding and argument–verb dependencies (Santi & Grodzinsky,

2007) or argument–verb dependencies and embedded sentences (Santi & Grodzinsky, 2010) yielded brain

activation in BA 45. Similarly, a German study found activity related to subject-argument–verb distance

in the inferior frontal sulcus, dorsal to and extending into BA 45 (Makuuchi et al., 2009), whereby

the argument–verb distance manipulation introduced a difference in the number of argument–verb

dependencies. Finally, a second German study by Fiebach et al. (2005) found BA-45 activation for

pronoun–verb distance, contrasting object-first sentences with an object pronoun (and a subject noun)

60
 4.4. Discussion

to subject-first sentences with a subject pronoun (and an object noun). The asymmetric comparison of

different syntactic dependencies across conditions in the above studies may have reflected the engagement

of a syntactic-working-memory system, proposed to be distinct from the working memory used in

other verbal tasks, and found to activate BA 45 (Caplan et al., 2000; Lewis et al., 2006; Van Dyke, 2007;

Van Dyke & McElree, 2006). In contrast to the above studies, the current paradigm kept the type of

syntactic dependency constant across conditions and required the storage of a given noun phrase across

a variable distance, which is corroborated by the possibility that the experimental task in the current

study was solvable using phonological strategies, which is also true for the digit-span task used in the

correlation analysis. Thus, the contrast between TP and inferior frontal brain activations may reflect

the difference between phonological and syntactic working memory, respectively. Yet, this suggestion

only counts as a future research hypothesis.

4.4.3 Individual Differences: Asymmetry of the Inferior Frontal Gyrus

The current VBM analysis found a correlation between the asymmetry of the underlying cortical

microstructure of the left and right IFG and the asymmetry of the functional activation of these

regions by a classical argument-order manipulation in a sentence-processing paradigm; in other words:

Structural asymmetry of the IFG in part predicts functional asymmetry of the IFG. The relationship

between cortical microstructure and functional activation inside the IFG complements the discussions

on the asymmetry of the IFG response during sentence processing (Bookheimer, 2002).

On the population level, a leftward asymmetry of the IFG response to increased

argument-reordering demands is strongly suggested by the cross-linguistic picture discussed above.

However, evidence for the general ability of the right IFG to support high-level language tasks comes

from studies on cortical reorganization after stroke (Cao, Vikingstad, George, Johnson, & Welch, 1999;

Rosen et al., 2000; Weiduschat et al., 2011; Winhuisen et al., 2005, 2007) as well as patients suffering

from Parkinson’s disease (Grossman et al., 2003). Cao et al. (1999), Rosen et al. (2000), Winhuisen et al.

(2005), Winhuisen et al. (2007) converge in reporting brain reorganization of patients suffering from

prefrontal brain lesions after stroke to involve increased activity in the right homologue of the left IFG

during language processing. Grossman et al.’s (2003) data support these findings in that they report

senior patients with Parkinson’s disease and accordingly-reduced sentence-processing abilities to exhibit

61
4. Magnetic-Resonance-Imaging Study

reduced brain activation in the right IFG compared to age-matched controls. A compensatory role of

the right IFG is also in line with Tyler et al.’s (2010) report of increased right IFG activity in seniors

who show relative atrophy of the left IFG. It is, however, important to point out that this functional

reorganization is not sufficient for a full recovery of sentence-processing abilities in either patients

(Rosen et al., 2000; Tyler et al., 2011; Winhuisen et al., 2005), or seniors (Tyler et al., 2010)—stressing

the leftward functional asymmetry of the IFG on the population level.

The current data leave open the possibility that such a compensatory role of the right IFG can

also account for differential results in the imaging literature on healthy participant groups—in line with

the proposal of Just et al. (1996), who assigns the right IFG a supportive role for the language-dominant

left IFG (Just et al., 1996). It is interesting to note that while most fMRI studies on argument-reordering

report left-only IFG activations on the group level (Friederici, Fiebach, et al., 2006; Kinno et al., 2008;

Obleser et al., 2011), a body of previous studies has consistently reported brain activation for increased

argument-reordering demands not only in the left IFG, but also its right-hemispheric homologue (Ben-

Shachar et al., 2003, 2004; Bornkessel-Schlesewsky, Schlesewsky, & von Cramon, 2009; Just et al., 1996;

Wilson et al., 2010).

The reason for the divergence of these results—some studies observing right IFG activity, some

not—may lie in across-studies lateralization differences between participant groups: While Obleser et al.

(2011) and the current study did match their participants for high handedness indices (i.e., laterality

quotient (LQ) > 80 in both studies; Oldfield, 1971), it is possible that Ben-Shachar et al. (2003, 2004),

Bornkessel-Schlesewsky et al. (2009), and Just et al. (1996) applied less-strict lateralization-matching

criteria to their experimental participants—also not controlling for familial handedness: Participants’

familial history of handedness may bias their reliance on syntactic processing strategies during sentence

processing, with familial right-handers relying more on syntax as compared to familial left-handers

(Townsend, Carrithers, & Bever, 2001). Handedness and familial handedness may have increased the

fraction of participants with less-strong leftward language lateralization in the respective samples (cf.

Knecht et al., 2000). While this may sound ad hoc at first glance, the motif goes well with morphometry

results by Foundas, Eure, Luevano, and Weinberger (1998), who state that structural asymmetry of both

pars opercularis and triangularis of the IFG is dependent on participants’ handedness, with left-handed

participants showing reduced left-sided lateralization when compared to right-handed participants.

62
 4.4. Discussion

In turn, the structural–functional correlation in the current results entails that the functional

involvement of the right IFG may directly depend on the asymmetry of the underlying gray matter in

the left and right IFG—which may be tilted towards the right hemisphere in left-handers. Although LQ

scores were available for 21 of our participants, these were not normally distributed due to the selection

criteria of the current study (see Section 4.2.1). Thus, we median-split our participant sample for LQ

and functional and structural lateralization indices to run Fisher’s Exact Tests on the observed cell

counts from the handedness–functional and handedness–structural pairings, which turned out highly

significant in both cases (p < 0.001). This supports our interpretation that handedness is a key predictor

of the structural lateralization of Broca’s area as well as the functional lateralization of this brain

structure during sentence processing. Figure 4.8 illustrates these relationships.

A / B /
Functional laterality index
Structural laterality index

"#5 "#5

" "

!"#5 !"#5

!/ !/
LQ < 95 LQ > 95 LQ < 95 LQ > 95
?9%&%/"@ ?9%&%/"@ ?9%&%/"@ ?9%&%/"@

Figure 4.8: Boxplots of structural (A) and functional (B) lateralization, arranged by median-split partici-
pant groups (median LQ = 95, range 80–100). Whiskers mark range; negative values indicate rightward
lateralization, whereas positive values indicate leftward lateralization. While the effect is clearly stronger
for the structural data, Fisher’s Exact Test found that significantly more strongly-right-handed partici-
pants (i.e., LQ > 95) show increased structural lateralization and left-lateralized functional engagement
of Broca’s area during sentence processing (p < 0.001).

As a cross-species aside on the relationship between handedness and leftward asymmetry of

the IFG, it is interesting to mention that a leftward asymmetry of BA 44 in great apes (bonobos,

chimpanzees, and gorillas) has been found predictive of communicative gesturing behavior using the

right hand (Cantalupo & Hopkins, 2001; Hopkins, Russell, & Cantalupo, 2007; Schenker et al., 2010;

Taglialatela, Cantalupo, & Hopkins, 2006). Combined with the fact that BA 44 and 45 exhibit cytoar-

63
4. Magnetic-Resonance-Imaging Study

chitectonic similarities between humans and other primates, such as macaques (for reviews, see Aboitiz,

García, Bosman, & Brunetti, 2006 and Petrides, Tomaiuolo, Yeterian, & Pandya, 2012), this lends some

credibility to the suggestion that human right-handedness has its evolutionary origin in the combination

of gesturing and vocalizing in pre-linguistic communication (Aboitiz & García, 2009; Corballis, 2003;

Corballis, Badzakova-Trajkov, & Häberling, 2011).

In sum, while the left IFG in humans certainly remains the dominant region for the processing of

argument-reordering tasks on the population level (cf. Amunts et al., 1999; Amunts & Zilles, 2012), a

fraction of the general population may additionally involve the right IFG in this process—or even show

right-dominant IFG activity, as is evident from the individual data shown in Figure 4.5. The current

finding brings up the future research question whether left-handedness is advantageous in recovering

from left frontal stroke lesions or language deficits observed in Parkinson’s disease.

4.4.4 Diffusion-Tensor-Imaging and Fractional-Anisotropy Results

Deterministic fiber tracking between the observed functional effects suggested a dorsal connection,

including the AF/SLF. As to the correlation analysis of the FA values in the AF/SLF with the functional

effects, a series of clusters was obtained. Clusters of FA along the superior and posterior parts of the

AF/SLF correlated with increased storage demands as compared to increased reordering demands, and a

single cluster in the more anterior part correlated with increasing reordering as compared to storage

demands. The involvement of the AF/SLF, but not the inferior longitudinal fasciculus, suggests that the

present effects are related to the dorsal pathway, linking the auditory cortex to the IFG via TP regions

(Catani, Jones, & ffytche, 2005; Parker et al., 2005; Weiller, Musso, Rijntjes, & Saur, 2009). Generally,

both the effects of reordering as compared to storage and storage as compared to reordering along the

left AF/SLF are in accordance with the view that this tract is involved in sentence processing (Friederici,

2009b). This is in line with probabilistic-fiber-tracking results, which were based on functionally defined

seeding points in the IFG (pars opercularis/BA 44; Friederici, Bahlmann, Heim, Schubotz, & Anwander,

2006) and are similar to the current data with regard to this. Specifically, it seems anatomically plausible

that the neural basis of sentence processing involves a reordering process as subserved by the IFG that

queries a storage component in TP regions through the AF/SLF.

64
 4.4. Discussion

The literature on an involvement of different fiber tracts in working memory from healthy

populations is sparse, although initial DTI work is available (Charlton, Barrick, Lawes, Markus, &

Morris, 2010). Charlton et al. (2010) used a composite working-memory score collating different

working-memory measures to explore the white-matter network underlying working memory. Their

results show a distributed array of MD and FA clusters in both hemispheres, including a tract that

connects BA 40 to inferior frontal areas. But since none of the measures employed by Charlton et al.

(2010) only tapped storage, their results do not easily map directly onto the current sentence-processing

data. Thus, to my knowledge, the current study is the first to specify the role of the AF/SLF in sentence

processing to lie in argument storage in working memory over a certain argument–verb distance and

manipulating the argument order by syntactic working memory, such that the who and the whom of the

sentence are not confused. While the present data—in combination with Charlton et al.’s (2010) results—

may suggest a common role of the AF/SLF in storage both in- and outside of the sentence-processing

domain, Section 8.4 will discuss whether such a unitary view can meet recent fiber-tracking evidence.

The common role of the AF/SLF may best be described as mediating a storage component

in TP regions to a rehearsal component located in the dorsal prefrontal region for tasks outside of

sentence processing and to a reordering system located in Broca’s area for sentence processing proper.

This interpretation fits both the finding of a correlation with storage involved in sentence processing

(as measured by brain activation during increasing argument–verb distances) and outside of sentence

processing (as measured by digit span). Also, the relatively weaker and less extensive correlations

with argument order in the anterior portion of the AF/SLF as opposed to rather extensive effects of

argument–verb distance suggest that the functional role of this tract is less specific to reordering itself

than it is general to the storage of processing-relevant information.

In addition to work on healthy participants, there is evidence on the role of the left AF/SLF in

sentence processing from patients with primary progressive aphasia (Wilson et al., 2011) and patients

with conduction aphasia, who also show reduced working-memory capacities (Buchsbaum et al., 2011;

Caramazza, Basili, Koller, & Berndt, 1981). Investigating patients with primary progressive aphasia,

Wilson et al. (2011) found that damage to the AF/SLF causes severe sentence-processing difficulties in

primary progressive aphasics, which was not the case for patients with damage to the ventrally located

extreme capsule fiber system and uncinate fasciculus. The studies on conduction aphasia are of different

65
4. Magnetic-Resonance-Imaging Study

degrees of neuroanatomical specificity. Friedmann and Gvion (2003) report that conduction aphasics

show both reduced digit span and particular problems in argument retrieval at long argument–verb

distances. Since the authors are not precise about lesion site and extent, this study does not allow

to draw strong neuroanatomical conclusions. Yamada et al. (2007), however, report a case in which

selective damage to the posterior part of the AF/SLF resulted in conduction aphasia, sparing the gray

matter in the left TP region. Finally, Fridriksson et al. (2010) found that impaired speech repetition

in conduction aphasics co-occurs with both damage to the white matter underlying the left SMG and

reduced gray-matter integrity in the left SMG and TP junction, converging on data by Quigg and

Fountain (1999) and Baldo, Klostermann, and Dronkers (2008). The current data agree with both the

gray-matter and white-matter proposal. Together, this is solid ground for the proposal that the integrity

of the left AF/SLF is crucial to the storage component of working memory and its linkage to Broca’s

area during the processing of argument–verb dependencies.

Turning to methodology, it is worth to briefly discuss the current white-matter analysis—in

particular, because the fiber-tracking results were not used as a dependent variable in the correlation

analysis, but merely as the search volume for the regression of functional data across FA maps. An

alternative approach would have been to compare the reconstructed individual tracts themselves, e.g.

the number of streamlines. However, Wakana et al. (2007) and Heiervang, Behrens, Mackay, Robson,

and Johansen-Berg (2006) have pointed out that there is no proportional relation between individual

FA values and the number of streamlines extracted by fiber tracking, and even less so at small sample

sizes. Fiber tracking is sensitive to anatomical variability along the to-be-reconstructed tract, such as

overall tract length, crossing fibers or tract diameter, inter alia. Thus, while fiber tracking is a valuable

tool for anatomical description, FA values are a functionally more relevant and statistically more reliable

measure of local microstructural properties (Reich et al., 2010).

Another issue with the applied FA analysis might be that the functional effects were used both

as seeding points in the fiber-tracking analysis and as regressors in the correlation analysis between

FA values and differences in functional activation. In principle, Kriegeskorte, Simmons, Bellgowan,

and Baker (2009) point out that such an analysis could yield invalid results due to an overlap between

dependent and independent variables. However, even though the functional-activation clusters were

used as gray-matter seeding points, the DTI procedure reconstructed only white-matter tracts. In

66
 4.5. Conclusion

addition, these tracts were masked with the TBSS skeleton, which also only contains areas of high FA

across participants, that is, white matter. Since the selection data at this point were not the functional

data themselves, but a reconstructed tract volume, I consider the dependent measure (i.e.,voxel-wise FA

of the fiber tracts connecting the IFG and TP region) to be different from the independent measure (i.e.,

the individual difference in functional-activation strength).

4.5 Conclusion

Our results show that, in sentence processing, the storage of arguments over increased argument–verb dis-

tance and the ordering of arguments rely on distinct neural subsystems. The direct comparison of these

factors within one stimulus set shows that Broca’s area (IFG) is mainly concerned with ordering, and less

so with storage. Storage of a single argument in working memory across a given argument–verb distance

activates the left TP region; a region classically assumed to subserve working-memory storage. The

data, moreover, provide direct evidence for the role of the left AF/SLF in interfacing working-memory

storage and ordering. The results suggest that a minimal cognitive architecture of sentence processing

can be rooted in the interplay of concrete cognitive concepts, such as reordering and storage.

67
5
PA T I E N T S T U D Y

5.1 Introduction

The initial fMRI and dMRI experiment performed in the course of this thesis (see Chapter 4) yielded

three key findings: First, the results strengthened the previous cross-linguistic evidence on an involve-

ment of the left IFG in the processing of argument-reordering tasks during sentence processing. Second,

the results showed that increasing verbal-working-memory load during sentence processing—as induced

by an increased argument–verb distance—drives brain activity in the TP region. This suggests that the

classical role of this brain region in verbal-working-memory storage carries over to sentence processing

as well. Third, the results provided direct evidence for a functional role of the AF/SLF white matter

structure in mediating working-memory storage and argument reordering.

While the fMRI and dMRI analyses so far point to a functional involvement of the AF/SLF, they

do not allow for a generalization as to whether this functional involvement is causal, that is, whether the

AF/SLF is a structure necessary for the mediation of working-memory storage and argument reordering.

While clinical work by Charlton et al. (2010) and Yamada et al. (2007) shows a general role for the

AF/SLF in verbal-working-memory storage outside the sentence-processing domain, clinical work from

the sentence-processing domain lacks the anatomical specificity to tease apart causal influences of the

TP-region gray matter and the white matter of the underlying AF/SLF: Neuroanatomically diffuse

patient groups, such as primary progressive aphasics (Wilson et al., 2011) or conduction aphasics (Baldo

et al., 2008; Buchsbaum et al., 2011; Fridriksson et al., 2010; Friedmann & Gvion, 2003), do not yield

clear conclusions in this respect.

69
5. Patient Study

While there are electro-stimulation studies targeting the dorsal language-relevant fiber tracts

(De Witt Hamer, Moritz-Gasser, Gatignol, & Duffau, 2011; Duffau, 2008; Duffau et al., 2002; Duffau,

Gatignol, Denvil, Lopes, & Capelle, 2003), their results are not straightforwardly mapped onto the

conceptual dialectics of storage and reordering: Duffau et al. (2002) and De Witt Hamer et al. (2011)

report that electrical stimulation of the posterior-most limb of the dorsally running middle longitudinal

fasciculus during surgery elicits anomia, while Duffau et al. (2003) find that a glioma involving the left

parietal operculum in the TP region causes overt speech-repetition deficits as well as reduced digit span.

Neither the overt speech task (Duffau et al., 2003), the naming tasks (De Witt Hamer et al., 2011; Duffau

et al., 2002) nor the target area of stimulation (De Witt Hamer et al., 2011) directly correspond to the

involvement of the AF/SLF in the conceptual dialectics of argument storage and reordering during

sentence comprehension.

To instantiate a direct link between the proposed role of the AF/SLF in mediating argument

storage and reordering during sentence processing, the current paradigm was used in a modified version

in a case study on a patient who suffers from a focal, rupture-like lesion in the left TP region, reaching

from the gray matter into the deep white matter underlying the region. Visual inspection of the patient’s

anatomical brain images suggested that the lesion might have caused focal damage to the AF/SLF,

providing a test case for the hypothesis that the AF/SLF is specifically involved in mediating storage

and reordering. If this were so, two hypotheses were possible with respect to our paradigm: First, it

is possible that focal damage to the AF/SLF results in a general working-memory deficit as well as

problems on sentences with long argument–verb dependencies, regardless of the relative argument order.

A second possible hypothesis is that focal damage to the AF/SLF results in more specific problems

in comprehending sentences that increase both argument-reordering and argument-storage demands

(i.e., object-first sentences with long argument–verb dependencies), because the role of the AF/SLF is

proposed to lie in the mediation of reordering and storage rather than one of the processes in particular.

These hypotheses were tested in the current patient study, using the full 2 × 2 paradigm as well as a

variety of working-memory tests.

70
 5.2. Methods

5.2 Methods

5.2.1 Participants

A single female patient was tested (age 46 years). The patient was diagnosed with a cleft-like lesion

following an intra-cerebral bleeding, spanning the gray matter of the left STG towards the SMG as

well as the underlying white matter (see anatomical data, Figure 5.1). Time post-onset was 7.5 years.

Behaviorally, the patient showed a productive deficit with phonological disruptions as well as a minor

comprehension deficit involving morphosyntactic problems (Cunitz, 2011). In addition to the patient,

an age-, gender- and education-matched control group of seven participants (mean age 47 years, SD 1.92)

took part in the study. All participants were right-handed as determined by an abridged version of the

Edinburgh inventory (Oldfield, 1971) and had German as their first language. Control participants

reported no neurological or hearing deficits. Per hour of participation, the patient received A
C8, plus a

travel reimbursement. Control participants were paid A

C7 per hour of participation.

A P L R L R

X = –34 Z = –28 Y = 18

Figure 5.1: Recent anatomical scan of patient’s brain in sagittal, axial, and coronal view; the enlarge-
ments show the lesion in the left TP region.

71
5. Patient Study

5.2.2 Working-Memory Test Battery

The patient and control participants underwent a battery of working-memory tests that focused on the

ability to store, rehearse, and reorder phonological content. The involved tests were the forward- and

backward-digit-span test (Jacobs, 1887) from the German version of the Wechsler test (Tewes, 1994), the

Vorländer syllable-span and paronomasia test (Vorländer, 1986) and the Mottier pseudo-word-repetition

test (Welte, 1981). The Vorländer syllable-span and paronomasia test (Vorländer, 1986) involves word

repetition of multi-syllable words in three stages of increasing difficulty (words with two syllables,

words with more than two syllables and one-syllable paronomasias). The three levels of the Mottier

pseudo-word test (Welte, 1981) involve the repetition of pseudo-words of increasing syllable counts.

5.2.3 Materials

To use the sentences of the current stimulus set described in Section 2.3 as stimuli in a patient study,

modifications to the stimuli were necessary to avoid overburdening the patient’s language processing

abilities. First, the conjunct clause—which had been attached to the original stimulus sentences to

control for the possible effect of a sentence-final processing slowdown (Friedman et al., 1975; for

details, see Section 2.3)—was removed from the stimuli. This also necessitated new recordings of the

stimulus set, which were now carried out at a reduced syllable frequency as compared to the original

stimuli. The same trained female German speaker as in the previous fMRI study (see Chapter 4)

recorded the stimuli, and the same audio-preprocessing pipeline described in Section 4.2.3 was used.

Again, 192 stimuli were recorded, of which a pseudo-randomized list of 96 items (24 for each of the

four conditions) was generated for each participant using MATLAB® (The MathWorks, Inc., Natick,

MA, USA) scripts. To increase the number of behavioral data points, each of these sentences introduced

a who-did-what-to-whom yes–no comprehension question to be answered within a limited time. The

proportion of yes–correct and no–incorrect questions was balanced.

5.2.4 Procedure

Stimuli were presented using Presentation® (Neurobehavioural Systems, Inc., Albany, CA, USA). Au-

ditory stimuli were presented using a pair of ELAC Cool II stereo speakers (ELAC Electroacustic

GmbH, Kiel, Germany). Visual stimuli were presented on a Sony Trinitron® Multiscan G400 CRT

72
 5.2. Methods

VGA monitor with a refresh rate of 75 Hz (Sony Corporation, Tokyo, Japan), 70 cm in front of the

participants. A sans-serif font in black letters against a gray background (size 20 px) was used.

A trial started with a fixation cross that stayed on screen for the whole trial. After a random

jitter between 500 and 1000 ms, an auditory stimulus started (mean length 4.15 s, SD 0.30 s). After

the auditory stimulus, the yes–no comprehension question would appear and stay on screen until a

response would occur. After comprehension questions, visual feedback was given for 2000 ms by a

happy green or sad red emoticon. Participants were instructed to carefully listen to the sentences and to

answer the comprehension questions via button press with either their left or right hand, with one hand

corresponding to yes and the other to no. Response button assignment was counterbalanced across

participants.

5.2.5 Magnetic-Resonance-Imaging Data Acquisition

Structural and diffusion magnetic resonance images of the patient were acquired with a 3 Tesla Siemens

TIM TRIO scanner (Siemens Healthcare, Erlangen, Germany) at the Max Planck Institute for Human

Cognitive and Brain Sciences in Leipzig, Germany.

Anatomical T1-weighted 3D MP-RAGE images (Mugler III & Brookeman, 1990, TI = 650 ms,

TR = 1300 ms, alpha = 10°, FOV = 256 × 240 mm2 , 2 acquisitions, 1 mm isotropic resolution) were

acquired with a non-slice-selective inversion pulse followed by a single excitation of each slice, and were

available for preprocessing of the dMRI data.

Diffusion-weighted data were acquired with the same scanner, equipped with a 32-channel phased-

array head array coil. Images were acquired with a twice-refocused spin-echo EPI sequence (Reese

et al., 2003; TE = 100 ms, TR = 12 s, 128 × 128 image matrix, FOV = 220 × 220 mm2 , 88 axial slices

(no gap), resolution 1.72 × 1.72 × 1.7 mm3 ). Additionally, fat saturation was employed together with

6/8 partial Fourier imaging and GRAPPA (acceleration factor = 2; Griswold et al., 2002). Diffusion-

weighting was isotropically distributed along 60 diffusion-encoding gradient directions with a b-value

of 1000 s/mm2 . Seven images with no diffusion-weighting (b0) were acquired initially and interleaved

after each block of 10 diffusion-weighted images as anatomical reference for offline motion correction.

The dMRI sequence lasted about 16 minutes.

73
5. Patient Study

5.2.6 Data Analysis

5.2.6.1 Working-Memory Test Battery

The raw working-memory measures were split for the patient and control group, and 95-% CIs based

on the control-group mean were generated using a bootstrapping approach (5000 random draws from

the control population assuming a normal distribution). Memory scores of the patient were considered

to differ from the control group when they fell outside of these CIs.

5.2.6.2 Behavioral Data

For the behavioral data, d" -scores and RTs were calculated for the patient and control participants.

The RTs were logarithmized to correct for the typically F-shaped distribution of raw RTs. From the

across-control-group d" -prime scores and logarithmized RTs, 95-% CIs were bootstrapped (based on

5000 random draws from the control population). D" -scores and RTs of the patient were considered to

differ from the control group when they fell outside of these CIs.

5.2.6.3 Diffusion-Tensor Imaging

Diffusion-MRI data were analyzed using the pipeline described in Section 4.2.6.5: T1-weighted struc-

tural scans were used for skull stripping and the brain images were co-registered into Talairach space

(Talairach & Tournoux, 1988). Motion correction was performed based on the 7 reference images

without diffusion-weighting distributed over the entire sequence using rigid-body transformations

(Jenkinson et al., 2002) implemented in FSL. The parameters were interpolated and combined with

a global registration to the T1 anatomy, computed with the same method. The application of these

transformations resulted in an isotropic voxel resolution of 1 mm. The gradient direction for each

volume was corrected using the rotation parameters. For each voxel, a diffusion tensor model (Basser et

al., 1994) was fitted and the FA was computed (Basser & Pierpaoli, 1996).

Based on prior data (Meyer, Obleser, Anwander, & Friederici, 2012; see Chapter 4), I selected

two ROIs for probabilistic fiber tracking: Since the AF/SLF runs dorsally between BA 44 and the

TP region (Catani et al., 2005; Friederici, 2009b, 2011; Weiller et al., 2009), a 5-mm sphere at the center

of the anatomically defined BA 44 as provided with the SPM anatomy toolbox (Eickhoff et al., 2005)

74
 5.3. Results

was generated as starting point for tracking using the Marsbar toolbox (Brett et al., 2002). As a control

seed region, the same procedure was used to generate a control ROI at the center of BA 45. Both ROI

volumes were back-projected into the patient’s native image space, and all fiber trajectories passing

through these respective ROI volumes were selected as white-matter projections by tractography from

the diffusion-tensor maps (Anwander et al., 2007). Here I used probabilistic tractography (Lazar et al.,

2003) as implemented in MedINRIA (Asclepios, INRIA, Sophia Antipolis, France; cf. Fillard et al.,

2007). I used the same preprocessing chain as for the computation of the FA maps mentioned above,

except for the fact that the diffusion tensors were computed with an isotropic resolution of 1.7 mm.

Fiber trajectories were started in all voxels with an FA > 0.2 resulting in a complete set of trajectories

within the whole brain.

5.3 Results

5.3.1 Working-Memory Results

On all working-memory measures, the patient was below the control-group mean, and outside of the

bootstrapped CIs for six of the eight tests (all Mottier pseudo-word sub-tests, Welte, 1981; forward digit

span, Tewes, 1994; two of three Vorländer syllable-span and paronomasia sub-tests, Vorländer, 1986).

Scores of the patient with respect to the group average and the bounds of the bootstrapped 95-% CIs are

given in Table 5.1, Figure 5.2 gives a graphical summary.

10 Control group
Patient

8
Test score

0
Mottier (A) Mottier (B) Mottier (C) Digit span (A) Digit span (B) Vorländer (A) Vorländer (B) Vorländer(C)

Figure 5.2: Working-memory-test-battery results; lines and error bars mark mean and bootstrapped
95-% confidence intervals for the control group; diamonds mark the raw scores of the patient.

75
5. Patient Study

Table 5.1: Working-memory test results; for the control group, mean and SD are given, for the patient,
individual values are provided; lower and upper 95-%-CI limits are shown in the two rightmost columns.

Control group 95-% CI

Test Patient
Mean SD Lower Upper

Mottier
Syllables 7.57 1.81 6 6.36 8.8
Pseudo-words (easy) 5.29 1.38 4 4.35 6.23
Pseudo-words (hard) 7.43 1.40 5 6.48 8.38

Digit span
Forward 6.71 1.89 4 5.43 8
Backward 5.43 1.40 5 4.46 6.39

Vorländer
Short words 6.14 1.87 4 4.86 7.41
Long words 4.57 1.62 2 3.48 5.68
Paronomasias 5.14 1.87 4 3.85 6.43

5.3.2 Sentence-Processing Results

Sentence-processing performance showed a selective picture: while the patient showed above-group RTs

for all conditions, a selectively decreased performance (i.e., below-group d" -scores) was observed only

for the long object-first conditions. For the remaining three conditions (short and long subject-first and

short object-first sentences) the patient remained inside the control-group 95-% CI.

A 4
B 4
Reaction time (log 10 ms)

2 3
D−prime

0 2

−2 1
Short
Long

Subject-first
Object-first −4 0

Figure 5.3: Sentence-processing results; (A) d" -scores, (B) RTs (log 10 ms); lines mark control-group
mean and 95-% CI, diamonds mark the patient’s values; blue color marks subject-first, red color marks
object-first argument orders, solid lines mark short, dashed lines mark long argument–verb distances.

76
 5.3. Results

Table 5.2: Sentence-processing results; for the control group, mean and SD are given, for the patient,
raw values are provided; lower and upper 95-%-CI limits are shown in the two rightmost columns.

Control group 95-% CI

Condition Dependent Patient
Mean SD Lower Upper

Subject-first
d" 1.64 1.58 1.54 0.56 2.71
Short
RT 3.30 0.12 3.43 3.22 3.39
"
d 1.85 1.89 2.64 0.56 3.13
Long
RT 3.30 0.14 3.45 3.21 3.40

Object-first
d" 0.26 1.04 0.51 -0.47 0.98
Short
RT 3.31 0.15 3.41 3.20 3.42
d" 0.89 0.99 -0.20 0.19 1.57
Long
RT 3.36 0.20 3.47 3.23 3.50

5.3.3 Diffusion-Tensor-Imaging Results

Probabilistic fiber tracking from the BA 44 and BA 45 ROIs showed a series of inter- and

intra-hemispheric streamlines for the control ROI BA 45, including a ventral connection to the

TP region. Seeding in BA 44, on the other hand, did not yield extensive streamlines—in particular, the

patient showed no dorsal fiber tract from BA 44 to the TP region (see Figure 5.4).

A B

5–mm sphere at center of BA 45 5–mm sphere at center of BA 44

Reconstructed tracts from BA 45 Reconstructed tracts from BA 44

Figure 5.4: Results of DTI procedure; (A) from a 5-mm sphere at the center of BA 45 ROI; (B) from a
5-mm sphere at the center of BA 44. While a ventral connection to the TP region from BA 45 is evident
in (A), no dorsal connection to the TP region from BA 44 is present in (B).

77
5. Patient Study

5.4 Discussion

Behaviorally, the patient showed impairments across most of the administered working-memory tests;

also, generally increased RTs in the sentence-processing task were observed, with no condition-specific

effects—pointing to a general processing slow-down in the patient. Response accuracy showed largely

spared sentence-processing performance in case of easy (that is, subject-first sentences that involve no

reordering) and working-memory non-intensive sentences (that is, short argument–verb dependen-

cies that do not involve high verbal-working-memory storage demands). The patient did, however,

show below-control group performance on long argument–verb dependencies in object-first sentences.

Anatomically, DTI found that the patient apparently does not exhibit a dorsal fronto-temporal fiber

tract from BA 44 to the TP region. The control tracking procedure from BA 45 showed a ventral

fronto-temporal fiber tract from BA 45 to the TP region, suggesting that the absence of the dorsal

BA-44–TP-region tract in the patient does not result from a global white-matter-integrity decrease in

this patient. Together, this suggests that this patient’s cleft-like lesion in the TP region (see Figure 5.1)

is the cause of an interrupted AF/SLF in the patient, resulting in a selective deficit in the processing

of sentences that tax both argument-reordering and argument-storage abilities. I will now discuss these

findings in detail.

Only indirect evidence is available for a role of the AF/SLF in mediating argument storage

and reordering, but no direct causal evidence. Duffau et al. (2002) report that electrical stimulation of

the posterior-most limb of the AF/SLF during surgery elicits anomia. Since a case study by Duffau

et al. (2003) finds that a glioma involving the left parietal operculum in the TP region causes overt

speech-repetition deficits as well as reduced digit span, it is well possible that the anomia of the patients

in Duffau et al.’s (2002) study may result from an underlying damage to the working-memory system.

This interpretation is rather speculative, because neither the overt speech task (Duffau et al., 2003) nor

the naming tasks (Duffau et al., 2002) are easily mapped onto the presumed involvement of the AF/SLF

in the conceptual dialectics of argument storage and reordering during sentence comprehension. In

addition, while the evidence delivered by electrical brain stimulation may be taken as causal, all patients

in these studies suffered from extensive gray-matter lesions, mostly encompassing the TP region, leaving

this work with the caveat of the possible attribution of the observed working-memory problems to either

gray-matter damage or white-matter stimulation (cf. Section 4.4.4). Similarly indirect and potentially

78
 5.4. Discussion

ambiguous evidence is delivered by the classical work on conduction aphasia already discussed in

Section 4.4.4 (cf. Buchsbaum et al., 2011; Caramazza et al., 1981; Friedmann & Gvion, 2003): Fridriksson

et al. (2010) and Baldo et al. (2008) converge on reporting impaired speech repetition in conduction

aphasics co-occurring with both damage to the white matter underlying the left SMG and reduced

gray-matter integrity in the TP region.

In spite of this indirect evidence, we suggest that the current clinical data constitute direct causal

evidence for a role of the AF/SLF not only in verbal working memory, but, more specifically, in

anatomically linking—and functionally mediating—a storage component in TP regions to a reordering

system located in Broca’s area during sentence processing. Yamada et al.’s (2007) case study may come

close to deliver causal evidence for a direct role of the AF/SLF in working memory—their patient

suffered from a focal lesion to the AF/SLF, presenting with a sustained language-repetition deficit.

However, this result cannot easily be mapped onto the storage–reordering dichotomy. In the light of the

above fMRI results from a healthy population (Chapter 4), it is noteworthy that the patient in the current

study did only show a sustained sentence-comprehension deficit in the long object-first sentences, with

spared comprehension of all other sentence types: If the above proposal is correct that argument

reordering during sentence processing is subserved by the IFG, while argument storage is subserved

by TP regions, a selective disconnection of the two regions should result in a selective comprehension

deficit on sentences that involve both reordering and increased storage demands—which is exactly

what the current patient data seem to suggest. It is left open for future work whether our full 2 × 2

paradigm might be able to verify the preliminary conclusion in that it could potentially show a selective

deficit in frontal-lesion patients on object-first sentences (à la Grodzinsky, 2000, 2001; Grodzinsky,

Piñango, Zurif, & Drai, 1999; Zurif & Piñango, 1999, but cf. Bastiaanse & van Zonneveld, 2006; Berndt

& Caramazza, 1999), a selective deficit in posterior-lesion patients on long argument–verb-distance

sentences (Friedmann & Gvion, 2003; Leff et al., 2009; Romero et al., 2006), and a selective deficit in

AF/SLF-patients on long-distance object-first sentences.

As a final side note, it is interesting that while a general verbal-working-memory deficit on

span tasks was observed in the patient (i.e., decreased performance on six of eight administered tests),

the sentence-processing deficit was of a more selective nature rather than general to any sentence

involving a long argument–verb dependency. I propose that this apparent conflict can be explained by

79
5. Patient Study

the dissociation of reordering and rehearsal processes in the left inferior frontal cortex: As discussed

above (Chapter 4), some authors (Just et al., 1996; Rogalsky & Hickok, 2010; Rogalsky et al., 2008) have

equated reordering processes during sentence processing and subvocal rehearsal in the phonological

loop (Baddeley, 2012; Baddeley et al., 2009). We have suggested above that this position neglects the

evidence for a functional dissociation of the left inferior frontal cortex into dorsal (rehearsal) and

more ventral (reordering) functions (cf. Figure 4.7). Given that our patient showed a general deficit

for storage–rehearsal-type tasks, but a more selective deficit for a storage–reordering-type task, the

current data point to a general involvement of the AF/SLF in storage and rehearsal outside of sentence

processing and a specific involvement in storage and reordering for sentence processing proper. To put

it differently: If reordering would boil down to rehearsal, a patient with focal damage to the AF/SLF

should exhibit a deficit with any object-first sentence—contrary to what the current data show.

5.5 Conclusion

The current patient study set out to elucidate on the functional role of the AF/SLF with respect to

the working-memory system in general as well as the storage–reordering dichotomy in sentence pro-

cessing, in the light of previous indirect evidence from healthy and patient groups as well as potentially

unequivocal electrical-stimulation studies. The current findings corroborate the conclusions from the

above fMRI study on healthy participants. The AF/SLF seems to serve differential roles with respect to

verbal-working-memory processes, depending on whether in- or outside of the sentence-processing do-

main: Inside the sentence-processing domain, a focal lesion to the AF/SLF can result in a selective deficit

in processing sentences that involve both an object-first argument order and a long argument–verb

dependency, pointing to its role in mediating reordering processes as subserved by the IFG and storage

processes as subserved by the TP region. Outside of the sentence-processing domain, damage to the

AF/SLF leads to a deficit on phonological-loop-type tasks, pointing to a functional segregation of

reordering and rehearsal in spite of their possible common link to TP regions through the AF/SLF.

80
6
EL E C T RO E N C E P H A L O G R A P H Y ST U DY

6.1 Introduction

Traditionally, it had been assumed that the human alpha rhythm (7–13 Hz; Berger, 1929) represents

an idling cortical state (Pfurtscheller, Stancák, & Neuper, 1996), based mainly on the observation that

alpha oscillations increase as a preface to sleep, during eye closure, or motor relaxation (for review,

see Klimesch, Sauseng, & Hanslmayr, 2007)15 . More recently, it has been proposed that the idea of

alpha oscillations as an idling rhythm of the cortex may not reflect the full picture, in particular in

the auditory domain (for review, see Weisz, Hartmann, Müller, Lorenz, & Obleser, 2011). Following

Lehtelä, Salmelin, and Hari’s (1997) report of a 10-Hz rhythm in primary auditory cortex which is

sensitive to changes in auditory input, a number of recent articles have pointed out the significance of

alpha oscillations for verbal working memory. For example, a magnetoencephalography (MEG) study by

Jensen, Gelfand, Kounios, and Lisman (2002) used a modified version of Sternberg’s (1966) letter-based

working-memory paradigm, finding increased alpha power over posterior electrodes with increased

verbal-working-memory load. Along these lines, Leiberg, Lutzenberger, and Kaiser (2006) reported

increased alpha activity under conditions of increased verbal-working-memory-storage demands. Finally,

Van Dijk, Nieuwenhuis, and Jensen (2010) reported increased alpha amplitude during storage of task-

relevant pitch information. In sum, there is convincing evidence that enhanced alpha oscillations are a

robust neural correlate of verbal working memory.

15
The experiment reported in this chapter has been published in a modified version (Meyer, Obleser, & Friederici, 2012).

83
6. Electroencephalography Study

Given this role of alpha oscillations in item or pitch retention, we hypothesized that also

higher-level cognitive processes in the auditory domain that exhibit increased verbal-working-memory

demands—such as sentence processing—will increase alpha activity as well. Verbal working memory is

commonly agreed upon to play an important role in sentence processing (Just & Carpenter, 1992; Ro-

galsky & Hickok, 2010; Waters & Caplan, 1996; Wingfield & Butterworth, 1984). Baddeley and Hitch’s

(1974) initial work found that concurrent memory load decreases reading-comprehension performance,

inducing that reading comprehension is subserved by a capacity-constrained verbal working memory.

More specific work showed that working-memory capacity determines the ability to store and retrieve

the arguments (both subject and object) until they can be syntactically linked to the main verb of the

sentence and the sentence meaning can be inferred (King & Just, 1991)—which is of particular relevance

in languages with sentence constructions requiring the verb to be in sentence-final position such as

German and Japanese.

While there is support for an involvement of verbal-working-memory resources during

argument–verb dependency processing (i.e., domain-specific functions), this support is difficult

to map onto the literature on alpha oscillations during verbal-working-memory storage outside

of the sentence-processing domain. Previous ERP studies on verbal working memory from the

sentence-processing domain rather focused on argument-reordering processes during sentence pro-

cessing, mostly triggered by research questions derived from theoretical linguistics. Such studies

isolated sustained negative ERP effects for object-first as compared to subject-first sentences (Felser et

al., 2003; Fiebach et al., 2001, 2002; Kluender & Kutas, 1993; Phillips et al., 2005; Ueno & Kluender,

2003). However, more general work on working memory, in particular on visual working memory,

suggests that the retention of order information may be distinct from mere (that is, order-indifferent)

storage: Hsieh, Ekstrom, and Ranganath (2011) had their participants focus on either the presence or

absence of an item or the order of items in a delayed-response paradigm, obtaining enhanced posterior

alpha for storage only, independent of the order of items. Given that behavioral work suggests that

argument retrieval in the vicinity of verbs is a mechanism common to both subject- and object-first

sentences (Nicol & Swinney, 1989), we hypothesize common oscillatory dynamics of argument storage,

independent of the relative order of arguments—as opposed to distinct sustained ERP indices which are

sensitive to the relative order of arguments.

84
 6.1. Introduction

This experiment was performed to investigate verbal-working-memory-storage processes inde-

pendent of a particular sentence structure, that is, ignoring the reordering factor from our original

paradigm: These are required for the processing of any argument–verb dependency, regardless of the

argument order. If alpha oscillations during sentence processing are independent from the processing

of order information, such a result may help in disentangling verbal-working-memory and reordering

processes during sentence processing.

While ERP findings from the sentence-processing domain are hard to link to the underlying

neuroanatomy, a number of functional-imaging studies provide information about the neural under-

pinning of verbal working memory during sentence processing. As an example, Novais-Santos et al.

(2007) reported left inferior parietal cortex to increase its activation with the retention interval for

disambiguating information in ambiguous sentences, that is, verbal-working-memory load. In addition,

Grossman et al. (2002) found an age-related decrease of brain activation related to increased argument–

verb distance in the left parietal cortex, alongside sentence-processing difficulties in seniors. The notion

of left parietal cortex as potential neural substrate of verbal working memory during sentence processing

is in line with meta-analyses, imaging studies and clinical work from outside the sentence-processing

domain (Awh et al., 1996, 1995; D’Esposito et al., 1999; Jonides et al., 1998; Leff et al., 2009; Owen et

al., 2005; Petrides et al., 1993; Smith & Jonides, 1999; Wager & Smith, 2003). However, other studies

focusing on verbal working memory during sentence processing reported BA 45 in the left prefrontal

cortex to play a role, using paradigms comparing different syntactic dependencies (Fiebach et al., 2005;

Makuuchi et al., 2009; Santi & Grodzinsky, 2007, 2010). Hence, the imaging results for verbal working

memory during sentence processing and their relation to the ERP literature are unequivocal, and a

temporally more fine-grained method may complement the discussion.

Due to the potential role of cortical alpha oscillations in higher-level cognitive tasks such as

sentence processing, we investigated cortical oscillations during the processing of sentences that involve

greater working-memory load without additionally increasing reordering demands. We hypothesized

that any argument—regardless of whether it is a subject or an object—is stored in verbal working

memory until the verb position at which retrieval of the argument becomes necessary for sentence

interpretation. Consequently, we reasoned that oscillatory activity in the alpha band should increase

with verbal-working-memory demands (Leiberg et al., 2006; Van Dijk et al., 2010) regardless of ar-

85
6. Electroencephalography Study

gument order. Testing this assumption will help to bridge the gap between the emerging literature

on alpha oscillations in verbal working memory and the supposed role of verbal working memory

during argument–verb-dependency processing. In a similar vein, we will link changes in oscillatory

power during storage-intensive sentence processing to a classical working-memory measure from the

sentence-processing domain, that is, reading span (Daneman & Carpenter, 1980). Finally, we investigate

the neural generators of the observed responses using source localization to provide a tentative link to

neuroimaging studies of sentence processing and verbal working memory.

6.2 Methods

6.2.1 Participants

Thirty-six participants took part in the study (mean age 26.4 years, SD 3 years; 18 males, all native

speakers of German). All of them were right-handed as assessed by an abridged German version of the

Edinburgh Inventory (Oldfield, 1971). They were matched for reading span (mean reading span 3.6,

SD 0.8) according to an abridged version of the reading-span test (Daneman & Carpenter, 1980). None

of the participants reported neurological or hearing deficits, and all had normal or corrected-to-normal

vision. Participants were naïve as to the purpose of the study. They were paid A
C17.50 for participating.

6.2.2 Materials

Auditory recordings of all stimuli from the stimulus set described in Section 2.3 were available for the

current EEG experiment (Section 4.2.3). For each participant, an individual pseudo-randomized list

of all 192 stimuli was generated using MATLAB® (The MathWorks, Inc., Natick, MA, USA) scripts.

As a task to maintain participants’ attention and to get a behavioral performance measure, a yes–no

comprehension question followed in 25 % of trials (e.g., Hat der Trainer den Stürmer geehrt? / Did the

coach honor the center forward?); the proportion of yes–correct and no–incorrect questions was balanced.

6.2.3 Procedure

Participants were seated in a dimly-lit, magnetically-shielded, and sound-proof room. Stimuli were

presented using the Presentation® software package (Neurobehavioral Systems, Inc., Albany, CA, USA).

Auditory stimuli were presented using a pair of Infinity® Reference I MkII stereo speakers (Harman

86
 6.2. Methods

International Industries, Inc., Stamford, CT, USA), approximately 100 cm to the left and right front

of the participants. In a quarter of trials, comprehension questions were presented visually, using a

proportional, sans-serif font (size 20 px), black characters on a light-gray background—a Sony Trinitron®

Multiscan G220 CRT VGA monitor with a refresh rate of 75 Hz (Sony Corporation, Tokyo, Japan)

was used, approximately 70 cm in front of the participants. A trial started with a green fixation cross of

a random length between 2000 and 3500 ms. After this, the fixation cross turned red, and an auditory

stimulus was presented—participants were instructed to blink only when the fixation cross showed up

green, ensuring a low amount of blink artifacts in the data.

A sequence was either followed by the next trial or—in one fourth of the stimuli—by the yes–no

comprehension question. Participants had to answer these questions by pressing one of the two buttons

of a two-button response box. Response-button assignment was counterbalanced across participants.

Prior to comprehension questions, a green fixation cross of a random length was presented to avoid

task-preparation effects during the processing of the acoustic input. Comprehension questions were

present on the screen until a button press occurred; this ensured participants were comfortable and

avoided task artifacts (Hagoort, Brown, & Groothusen, 1993). Following a comprehension question,

visual feedback was given for 800 ms in the form of a happy green or sad red emoticon. An experimental

run, consisting of 192 trials, lasted for approximately 35 minutes. Including preparation, the experi-

ment lasted approximately 1.5 hours. The EEG was recorded with a pair of Brainvision BrainAmp

direct-current (DC) amplifiers (Brain Products GmbH, Munich, Germany) from 64 tin scalp electrodes,

attached to an elastic cap (Electro-Cap International, Inc., Eaton, OH, USA). The electrodes were placed

at the standard positions based on the extended international 10–20 system. Each of the electrodes was

referenced to the left mastoid, and the setup grounded to the sternum. The vertical electrooculogram

(EOG) was recorded from electrodes located above and below the left eye. The horizontal EOG was

recorded from electrodes positioned at the outer canthus of each eye. The impedances of the electrodes

were kept below 3 kΩ. The EEG and EOG were recorded continuously with a band-pass filter from

DC to 250 Hz with a sampling rate of 500 Hz. Electrode positions were tracked using a Polhemus

FASTRAK® electromagnetic motion tracker (Polhemus, Colchester, VT, USA). In five participants,

tracking failed and mean positions of all other participants were used.

87
6. Electroencephalography Study

6.2.4 Data Analysis

All analyses were carried out using the Fieldtrip toolbox for EEG/MEG analysis (Oostenveld et al., 2011).

An epoch of 3.5 s length was defined for analysis because we were interested in sentential oscillatory

effects prior to the main verb, whose mean onset latency was 2933 ms (SD 276 ms). To resolve slow

electrode drifts, the data were high-pass filtered at 0.03 Hz with a Hamming-windowed sixth-order

two-pass finite-impulse-response (FIR) filter (Edgar et al., 2005). The experimental trials, including a

1-s pre-stimulus baseline, were then extracted from the data. For artifact rejection, EEG epochs were

off-line re-referenced to linked-mastoid electrodes, and automatic EOG- and muscle-artifact rejection

was performed on a trial-by-channel basis. Cutoffs for the EOG- and muscle-artifact rejection were set

at z = 3 and z = 7 and performed inside frequency bands of 1–14 Hz and 110–140 Hz, respectively. The

rejection procedure followed a distribution-based artifact-identification approach (as implemented in

Fieldtrip), that is, z-scores for rejection result from the amplitude distribution across trials and channels;

this resulted in a rejection rate of 34.40 % of trials, with no significant differences in rejection rates

between conditions as verified by an ANOVA. After preprocessing, TFR was carried out using Morlet

wavelets (Lachaux et al., 1999) in 50 frequency windows of 2 Hz each between 2 Hz and 100 Hz and in

adjacent time windows of 50 ms length each. A fixed time–frequency resolution m of seven cycles was

chosen. For statistical analyses, a massed cluster permutation test (Maris & Oostenveld, 2007) was carried

out on the resulting (baseline-corrected) power-change estimates inside the time–frequency subspace

from 5 Hz to 20 Hz. As outlined above, we tested our main hypothesis (higher oscillatory power during

sentences with long argument–verb distances) in this massed-permutation-test framework using a paired

t-test on data that were collapsed across the two levels of the factor deemed irrelevant to this analysis

(i.e., argument order). To ensure that collapsing across the levels of the original argument-order factor

would yield a statistically-reliable result, we also ran the identical analysis for this factor. A Monte-

Carlo simulation with 1000 repetitions was used to identify significant clusters in time–frequency space,

while controlling for false positives. We set the algorithm to first identify time–frequency bins that

showed a significant effect at p < 0.025 and then searched for time–frequency–electrode clusters that

behaved similarly, considering a minimum of three neighboring (i.e., inter-electrode distance < 6.5 cm)

electrodes as a cluster.

88
 6.2. Methods

6.2.5 Source Localization

Source localization of the significant alpha-band time–frequency cluster (see Section 6.3.2) first involved

warping participants’ individual electrode positions to the cortical mesh of a standard BEM as derived

from a standard structural-MR image using a rigid-body transform (Besl & McKay, 1992). For each

point along a 1-cm-spaced grid in this volume conductor, a forward model was estimated.

The source localization followed the workflow proposed by various previous studies using an

adaptive beamformer in the frequency domain (the “dynamic imaging of coherent sources” beamformer,

Gross et al., 2001; for applications, see e.g. Haegens, Osipova, Oostenveld, & Jensen, 2010; Jensen &

Mazaheri, 2010; Medendorp et al., 2007; Obleser & Weisz, 2011): To attain a good spatial filter for all

conditions, an additional frequency analysis on the data segments of interest (see Section 6.3.2) and

their respective baselines was carried out, centered at 10 Hz (± 2 Hz spectral smoothing) and 2500 ms

(± 250 ms relative to sentence onset, see Section 6.3.2; plus respective estimates from the –500–0 ms

baseline intervals), using a multitaper approach (Mitra & Pesaran, 1999). From this, the cross-spectral

density matrix was gathered for subsequent localization. In this way, a spatial filter for each grid point in

the volume conductor was generated. Participant- and condition-specific source-activity estimates were

derived by applying this spatial filter to the condition-specific sensor data. The resulting source-activity

volumes were corrected for activity during the baseline period, collapsed across short and long conditions

(see Section 6.1), and passed into a paired t-test. This procedure resulted in a source-level t statistic of

alpha-power change for each voxel in the volume grid, which could then be mapped back onto a

standard structural-MR image. Resulting peak coordinates in MNI space were converted into the space

of Talairach and Tournoux’ (1988) atlas using a non-linear transformation (Lacadie, Fulbright, Rajeevan,

Constable, & Papademetris, 2008) for anatomical labeling.

6.2.6 Correlation Analysis

Because the behavioral performance as determined by d" -scores was positively correlated with

reading-span-test scores (see Section 6.3.1), we also sought to further elucidate on the relation between

participants’ verbal-working-memory abilities and alpha power. To this end, we first computed

individual differences between the individual averaged source-activity volumes for the long and short

conditions (see Section 6.3.2). We masked these volumes for all voxels that had shown a significant

89
6. Electroencephalography Study

difference in source activity between the two conditions of at least t(35) > 2.5 at the group level. All

voxels inside these volumes underwent a linear regression analysis between the increase in source

activity for the long as compared to the short argument–verb conditions and individual reading-span

scores (see Section 6.3.2).

6.3 Results

6.3.1 Behavioral Results

We calculated d" -scores in order to avoiding participants’ response bias (i.e., the individual tendency to

respond either yes–correct or no–incorrect) from obscuring the behavioral result. Mean d" -scores for the

yes–no sentence-comprehension task were 3.80 (SD 0.97) for the short and 3.96 (SD 1.00) for the long

argument–verb dependencies. Mean percentage correct scores for the short and long conditions were

89.01 % (SD 7.89 %) and 92.06 % (SD 9.18 %), respectively. For comparison to d" -scores, we cleaned

these mean percentage-correct scores for the mean response bias of c = –0.24 (short condition) and

c = –0.13 (long condition) according to the procedure suggested by Macmillan and Creelman (2005).

Bias-free percentage-correct estimates of 81.82 (short condition) and 82.56 (long condition) resulted.

A dependent-samples t-test on the d" -scores found no significant difference between the two condi-

tions. Also, mean d" -scores were positively correlated with reading-span-test scores across participants

(r = 0.38, p < 0.05).

6.3.2 Time–Frequency Results

The statistical comparison between the time–frequency patterns of the short and long argument–verb

distances yielded a single, sustained difference (p < 0.025), lasting from 2.25 s to 3.20 s in the

7–13 Hz (alpha-band) range. It was most pronounced at 10 Hz, ranging from 2.25 to 2.75 s. From

2.75 s onwards (i.e., adjacent to the sentence-final verb, which began at 2.9 s on average) this cluster

broadened in frequency and ranged up to the 20-Hz limit (beta band) of our analysis window. For the

second factor in the original design (argument order), no effect was obtained. Figure 6.1 summarizes the

results: While baseline-corrected spectral power in the 4–20 Hz range is shown in the upper panel, the

experimental manipulation elicits a significant increase of alpha activity for long as compared to short

argument–verb dependencies (lower panel; shown are t-values from the statistical comparison).

90
 6.3. Results

0 1 2 3
s
20 Hz
SHORT

4 Hz

Relative change (%)

20 Hz 50

4 Hz
20 Hz
LONG

4 Hz 0
20 Hz

4 Hz

Argument, long Argument, short Verb

20 Hz

4 Hz

80 Short 3
Relative change (%)
EFFECT

Long
40
T

0
0

!("
4 8 12 16 20
2–2.25 s 2.25–2.5 s 2.5–2.75 s
Frequency (Hz)

Figure 6.1: Time–frequency results in relation to the latency of the arguments and the sentence-final
verb; upper panel shows grand-average sustained alpha activity (10 Hz) for the four conditions
subject-first short, object-first short, subject-first long, and object-first long; lower panel displays the
t-values from the paired-samples t-test on the long and short conditions, averaged across significant
sensors. The alpha cluster starts about 0.75 s prior to and ends with the main-verb onset. The bottom
left panel shows the average frequency spectrum, the bottom right panel shows the topography in the
10-Hz band in steps of 0.5 s, starting at 2 s (thresholded at p < 0.025, cluster-size corrected).

6.3.3 Source Localization of Alpha Activity: Results

Source localization on the cluster segment of interest from 2.25 to 2.75 s which showed a sustained

effect at 10 Hz yielded maxima in bilateral occipital cortices (left: x = –31, y = –88, z = 22; right: x = 7,

y = –94, z = –20; all coordinates in MNI-space) as well as in left parietal cortex (x = –60, y = –14,

z = 44). The peaks correspond to the left superior occipital gyrus, right lingual gyrus, and the transition

between the left SMG and left precentral gyrus, respectively. Statistical maps of these results are given

in Figure 6.2.

91
6. Electroencephalography Study

Y = –24
X = –58

Z = 48
T
2 4

Y = –83
X = –25

Z=0
Figure 6.2: Statistical maps of the source-space comparison between alpha activity during the short
and long conditions; upper row shows sagittal, coronal, and axial slices at local source maxima in the
left parietal cortex; lower row shows slices for the local source maxima in the bilateral occipital cortices.
(thresholded at t(35) = 2, approximating the uncorrected threshold from the statistical analysis).

Parietal (peak: x = –51, y = –16, z = 49) Occipital (peak: x = 12, y = –79, z = –4)

0.4

–0.4

80 80
Relative change (%)
Relative change (%)

40 40
r2 = 0.18
p < 0.01
r2 = 0.07
0 0 p > 0.1

–40 –40

2.5 3.5. 4.5 5.5 2.5 3.5. 4.5 5.5

Reading span Reading span

Figure 6.3: Whole-brain coefficient maps of the correlation between alpha-power change and reading
span; left panels illustrate the distribution of the correlation across the left hemisphere, the scatterplot
showing the negative correlation at the parietal peak; right panels show the correlation across bilateral
occipital cortices, the scatterplot showing the non-significant correlation at the occipital peak.

92
 6.4. Discussion

6.3.4 Results of the Correlation Analysis

As can be seen in the right panel of Figure 6.3, a voxel-wise correlation analysis of individuals’

source-level alpha-power change and reading span yielded a strong negative peak in the left parietal

cortex (peak at x = –51, y = –16, z = 49; r = –0.42, p < 0.01), while no significant correlation was

obtained in the occipital area (peak at x = 12, y = –79, z = –4; r = 0.26, p > 0.1).

6.4 Discussion

This study set out to elucidate the potential role of alpha oscillations during the processing of sentences

that place high storage demands on verbal working memory, whereby verbal-working-memory storage

was scrutinized by systematically varying the distance of an argument–verb dependency.

The results of this study show that long argument–verb distances in sentence processing elicit

stronger sustained oscillations at 10 Hz (alpha band) during the storage phase than short argument–verb

distances. This difference starts about 2 s after argument presentation with a maximum prior to memory

retrieval at the main verb in sentence-final position. This effect turns into a transient beta-band effect

(13–20 Hz) immediately at the main verb. As the time–frequency spectra in the upper panel of Figure 6.1

illustrate, the processing of all sentence types used in our paradigm elicits an alpha enhancement that

builds up throughout the sentence. The lower panel of Figure 6.1 illustrates that this increase in alpha

activity is significantly stronger for long as compared to short argument–verb dependencies.

The sources of this alpha-power increase were localized to bilateral occipital and left parietal

cortices. Only in parietal cortex did source activity correlate significantly and negatively with reading

span, a classical behavioral measure of verbal-working-memory ability from the sentence-processing

domain. Because reading span was also predictive of participants’ behavioral performance on the

experimental task, we specifically suggest that left parietal cortex may be an important neural substrate

for verbal working memory during sentence processing. We interpret these results as evidence that

the role of alpha oscillations previously shown in verbal-working-memory tasks from outside the

sentence-processing domain also applies to sentence processing. We will now discuss our time–frequency

findings from a functional, psycholinguistic, and neuroanatomical perspective.

The functional significance of alpha oscillations for verbal working memory revealed in the

present study is in line with data from various earlier working-memory studies conducted on pitch

93
6. Electroencephalography Study

(Van Dijk et al., 2010), letter (Krause, Heikki Lang, Laine, Kuusisto, & Pörn, 1996), syllable (Leiberg et al.,

2006; Luo, Husain, Horwitz, & Poeppel, 2005), and digit retention (Schack, Klimesch, & Sauseng, 2005)

in the auditory, non-sentential domain. In particular, the present result from a sentence-comprehension

task converges on a seminal study on letter retention by Jensen et al. (2002), who report alpha power

over posterior EEG sensors to increase with higher demands in the letter-retention task. Jensen et

al.’s (2002) effect was most pronounced during the late storage phase, immediately prior to memory

retrieval. In particular, this detail matches our data, since the observed alpha cluster in our data occurred

in temporal proximity to the verb, where working-memory retrieval is most likely to take place, and

was less pronounced during the early storage phase. Along these lines, Maltseva, Geissler, and Başar

(2000) observed increased alpha phase locking prior to omitted stimuli when participants were given a

series of identical auditory stimuli. Crucially, this increased alpha phase locking was observed prior

to the anticipated retrieval of a stimulus, that is, during trials where no actual retrieval took place.

Given this finding, we suggest that the increased alpha power in our results for the long argument–verb

dependency reflects increased inhibition demands for the argument immediately prior to argument

retrieval at the sentence-final verb. Since an increased likelihood for the sentence-final verb to occur

entails a potential for argument retrieval, increased alpha activity in temporal proximity to the verb

may index the inhibition of a premature argument release.

This view converges surprisingly well on a classical psycholinguistic theory of argument–verb

dependency processing, the active-filler hypothesis (Clifton & Frazier, 1988), according to which a stored

argument will be released from verbal working memory as soon as the verb which the argument will be

linked to is likely to occur. In the light of our data, this view nicely goes together with a major theoretical

approach of oscillatory brain activity that puts forward the idea of alpha oscillations as a domain-general

mechanism for the functional inhibition of neural activity (Jensen & Mazaheri, 2010; Klimesch, Sauseng,

& Hanslmayr, 2007). For the domain of working-memory processes, Klimesch, Sauseng, and Hanslmayr

(2007) propose that alpha activity increases with inhibition demands to avoid premature information

release. Under this view, our finding of a negative correlation between parietal alpha activity and reading

span might index that participants with higher reading spans need to tax their functional-inhibition

system less to achieve good sentence-comprehension performance, reflected in relatively less alpha

power in high-span as compared to low-span participants in spite of an overall group-level alpha-power

94
 6.4. Discussion

increase. This interpretation is supported by the fact that behavioral performance in our study was

positively correlated with reading-span scores, in line with Klimesch, Sauseng, and Hanslmayr (2007),

who find alpha power during task performance to be negatively correlated with cognitive abilities.

From a psycholinguistic point of view, the characterization of alpha power as an index of func-

tional inhibition is in line with our finding that the present alpha-cluster offset closely matches the verb

onset, rising into the beta band at this point (see Figure 6.1). Evidence for the view that arguments

are retrieved in the vicinity of verbs comes from a number of behavioral cross-modal-priming studies

(McElree, 2000; McElree, Foraker, & Dyer, 2003; Nicol et al., 1994; Nicol & Swinney, 1989; Tanenhaus

et al., 1985). These studies showed that arguments which are encountered early in a sentence prime

lexically or phonologically related target words at immediately pre- or post-verbal positions, pointing

to argument reactivation in verbal working memory. Importantly, such behavioral priming effects were

present independently of whether a subject or an object was reactivated.

The psycholinguistic evidence that arguments are retrieved in the vicinity of verbs may also

explain why we also obtained more transient significant beta-band oscillations immediately at the

sentence-final main verb, in addition to the prior alpha effect. This is in line with a recent report on

the significance of beta oscillations for processes related to structural aspects of sentence processing

(Bastiaansen, Magyari, & Hagoort, 2010). Enhanced oscillatory power in the beta band was found

during the processing of syntactically-correct sentences as compared to word lists and violated sentence

structures: Beta oscillations were visible in the violation condition until the violation could be detected,

and increased during the ongoing sentence in the correct-sentence condition. The authors suggested that

this effect is linked to the integration of incoming information into the ongoing syntactic representation.

The beta effect in the current study is in line with this proposal in that it occurs at a point where a link

between the incoming verb and the stored argument is established, that is, during argument retrieval

from verbal working memory (McElree, 2000; McElree et al., 2003; Nicol et al., 1994; Nicol & Swinney,

1989; Tanenhaus et al., 1985). Evidence for this interpretation is further provided by data from Weiss et

al. (2005), who report increasing beta coherence between anterior and posterior sensors at a sentence

position where a subject argument needed to be linked to a sentence-final verb. Although not discussed

in depth by Weiss et al. (2005), their data also show sustained increased alpha coherence towards the end

of the argument-retention interval, prior to the beta-coherence increase, similar to our own effect.

95
6. Electroencephalography Study

Finally, a recent non-sentential study by Hsieh et al. (2011) gives reason to assume that the alpha

oscillations observed in the present study reflect a different underlying mechanism than that reflected in

the sustained negativity observed in previous ERP studies from the sentence-processing domain. While

Hsieh et al. (2011) show that alpha oscillations during verbal-working-memory storage are independent

of the order of memory items, the electrophysiological literature on argument–verb-dependency pro-

cessing revealed a sensitivity of the sustained negative ERP effect to the relative order of subject and

object (Clahsen & Featherston, 1999; Felser et al., 2003; Fiebach et al., 2001; Kluender & Kutas, 1993;

Nakano et al., 2002; Phillips et al., 2005; Ueno & Kluender, 2003). In contrast, and in line with Hsieh et

al.’s (2011) findings, our sentence-processing study did not show an oscillatory effect of argument order.

Such an effect could have been expected since sustained negative ERP effects are sensitive to argument

order. In sum, the evidence so far suggests that alpha-power changes can serve as an independent index

of verbal-working-memory load during sentence comprehension.

Turning now to our source-localization results, the role of occipital and left parietal alpha sources

for verbal-working-memory storage is partly corroborated by previous findings. We obtained both

bilateral occipital and left parietal sources, but only found left parietal source activity during argument

storage and prior to argument release to correlate with a behavioral measure of verbal-working-memory

performance: that is, reading span. These findings can be linked to the imaging literature from both

sentence processing and outside the sentence-processing domain, which also implies TP brain regions as

a neural substrate for verbal-working-memory storage (Awh et al., 1996, 1995; D’Esposito et al., 1999;

Grossman et al., 2002; Jonides et al., 1998; Leff et al., 2009; Meyer, Obleser, Anwander, & Friederici, 2012;

Novais-Santos et al., 2007; Owen et al., 2005; Petrides et al., 1993; Smith & Jonides, 1999; Wager & Smith,

2003). For sentence processing, Grossman et al. (2002) found that brain activity in parietal cortex, related

to increased argument–verb distance requiring increased verbal-working-memory-storage resources, is

decreased in seniors who show reduced sentence-processing performance. Additional support comes

from a study on the processing of temporarily-ambiguous sentences (Novais-Santos et al., 2007) that

reported increased left parietal activity with increased length of ambiguous sentence segments: that is,

increased verbal-working-memory-storage demands.

In contrast to these studies, other imaging work suggests that BA 45 in the inferior frontal gyrus,

sometimes extending into the inferior frontal sulcus, subserves working memory during sentence

96
 6.4. Discussion

processing (Fiebach et al., 2005; Makuuchi et al., 2009; Santi & Grodzinsky, 2007, 2010). However,

these studies compared different kinds of syntactic dependencies across conditions. Particular syntactic

dependencies in these studies may have engaged a syntactic-working-memory system, which has been

suggested to be distinct from working memory used in other verbal tasks (Lewis et al., 2006; Van Dyke,

2007) and reported to activate BA 45 (Caplan et al., 2000). In contrast to the above studies, our paradigm

kept the type of syntactic dependency constant across conditions. Furthermore, our experimental

task was solveable using phonological strategies, which is also true for the reading-span task used in

our correlation analysis. In sum, we suggest that the above contrast between TP and inferior frontal

brain activations reflects the difference between phonological and syntactic working memory. An

alternative hypothesis is that the frontal–posterior neural dichotomy rather mirrors a difference between

domain-specific syntactic and domain-general attention-driven aspects of working-memory processing:

Buchsbaum and D’Esposito (2008) point out the possibility that brain activations in the middle and

superior parietal lobe (regions above BA 40) may imply attentional mechanisms of working memory

rather than storage per se (for discussion, see Chapter 8). With respect to sentence processing, either

suggestion must count as a hypothesis for future research.

With respect to parietal involvement in working memory during sentence processing, the connec-

tion between the current results, previous EEG findings and the fMRI literature also needs discussion:

While Michels et al. (2010) reported a positive correlation between left parietal BOLD signal and left pari-

etal alpha increase during a verbal-working-memory task in a combined fMRI and EEG study, Meltzer,

Negishi, Mayes, and Constable (2007), report a negative correlation between bilateral parietal alpha and

BOLD signal during a similar task. While our data cannot settle this argument, our localization agrees

with the above studies in that it suggests a crucial role of left parietal cortex in verbal working memory.

This supports the proposal of a functional relevance of this region in verbal-working-memory-related

alpha oscillations during sentence processing.

While the data suggest a functional relevance of parietal alpha in higher-level cognitive pro-

cessing, this is less straightforward for the occipital alpha sources. As occipital activity was not

significantly correlated with reading-span-test scores, we suggest that occipital alpha activity does not

reflect similar processes as the parietal alpha. This activity may rather reflect the inhibition of sensory

bottom-up processes, potentially preventing incoming information from saturating the limited-capacity

97
6. Electroencephalography Study

verbal-working-memory system (Jensen & Mazaheri, 2010; Just & Carpenter, 1992; King & Just, 1991;

Thut, Nietzel, Brandt, & Pascual-Leone, 2006). In the context of sentence processing, such a mechanism

might serve to avoid new information from interfering with arguments already stored in verbal working

memory (Lewis, 1996; Van Dyke, 2007; Van Dyke & McElree, 2006). While such a double role for alpha

oscillations in the present study is speculative, a functional role of occipital cortices in verbal working

memory is also unlikely given the fMRI data discussed above.

6.5 Conclusion

Our results show that in sentence processing, the storage of an argument (i.e., subject or object)

over longer distances modulates and enhances alpha oscillations (10 Hz) in left parietal cortex.

These oscillations are predictive of sentence-processing performance as measured by a standardized

sentence-processing-specific verbal-working-memory test. Our results also support previous research

that links argument–verb-dependency establishment to beta oscillations. The time course and brain

topography of the alpha effect suggests that alpha oscillations play a role in functionally inhibiting the

premature release of verbal information that will subsequently be integrated. This is the first evidence

that verbal-working-memory storage during sentence processing relies on oscillatory processes similar

to those observed for domain-general working memory outside of sentence processing. Thus, oscillatory

dynamics in the alpha range are a candidate neural surrogate that links listeners’ sentence-processing

abilities to more domain-general neural and cognitive resources.

98
7
C O M B I N E D A N A LY S E S

7.1 Introduction

In psycholinguistic research, argument–verb dependencies have been one of the most fruitful fields in

studying the cognitive architecture of language processing. Following initial discoveries that concurrent

working-memory load decreases reading-comprehension performance (Baddeley & Hitch, 1974), it was

established early on that working memory plays an important role in sentence processing (Just &

Carpenter, 1992; Wingfield & Butterworth, 1984). Specifically, King and Just (1991) indicated that an

individual’s working-memory capacity in part determines the ability to store subject and object until

they can be retrieved and syntactically linked to the main verb of a sentence. This entails that the main

role of working memory during sentence processing is in argument storage and retrieval16 .

Event-related-brain-potential studies studies provided further evidence for the role of working

memory as argument storage during argument–verb-dependency processing by isolating sustained

negative ERP effects for object-first as compared to subject-first sentences (Felser et al., 2003; Fiebach

et al., 2001, 2002; Kluender & Kutas, 1993; Phillips et al., 2005; Ueno & Kluender, 2003). Most of

these studies involved a direct comparison of object-first and subject-first sentences, not considering

the influence of argument–verb distance on working-memory load independently of argument order.

In a recent study (Meyer, Obleser, & Friederici, 2012; see Chapter 4), we directly compared short and

long argument–verb dependencies irrespective of the relative order of subject and object. Oscillatory

brain activity in the alpha range increased during working-memory storage of arguments across an in-

16
The analyses of the current chapter have been submitted for publication (Meyer, Obleser, Kiebel, & Friederici, in revision).

101
7. Combined Analyses

creasing argument–verb distance; this effect was common to both subject-first and object-first sentences.

This suggests that working-memory storage is employed independently of the syntactic status of the

arguments.

In the present study, we assume that working-memory-storage demand increases for both subjects

and objects with dependency length. If so, this increase in dependency length should also increase

working-memory-retrieval demands for both subjects and objects at their subcategorizing verb (i.e.,

the verb that delivers the information about the number and kind of arguments), simply due to pro-

gressive working-memory decay (Lewis et al., 2006). Evidence that arguments occurring early in a

sentence are retrieved from working memory in the vicinity of their subcategorizing verb comes from

cross-modal-priming studies, where priming effects have been found to start at the subcategorizing verb

and continue subsequently (McElree et al., 2003; Tanenhaus et al., 1985; Van Dyke, 2007). In addition, it

was found that argument retrieval occurs for both subjects and objects during the processing of passive

sentences, irrespective of the relative argument order (Nicol, 1993; Osterhout & Swinney, 1993). Thus,

retrieval, as storage, may be indifferent to the relative order of subject and object.

The other important sentential process under consideration here, argument reordering, has

been shown to differentially increase processing load during argument–verb-dependency resolution at

subcategorizing verbs (Grodner & Gibson, 2005; Rösler, Pechmann, Streb, Röder, & Hennighausen,

1998). Both theoretical and empirical work suggests that this increased processing load may reflect

the recovery of the original subject-first argument order from the incoming argument order as stored

in working memory (Chomsky, 1981; Friederici, Fiebach, et al., 2006). From a working-memory

perspective, argument reordering is thought to be distinct from retrieval operations, and may be

conceptualized as an executive operation on the contents of working memory (Baddeley, 2012; Wingfield

& Butterworth, 1984).

Which brain areas form the neuroanatomical basis of argument retrieval and argument reordering?

In a recent fMRI study, we orthogonally manipulated argument-storage and argument-reordering de-

mands (Meyer, Obleser, Anwander, & Friederici, 2012; see Chapter 4). We found significant activity

in the left TP region for long as compared to short argument–verb distances, while the left IFG was

significantly active for object-first as compared to subject-first argument orders. These results are in line

with imaging research on the neural correlates of working memory during sentence processing (Gross-

102
 7.1. Introduction

man et al., 2002; Novais-Santos et al., 2007). Since a role of the left TP region in both working-memory

storage and retrieval has been suggested previously (Buchsbaum et al., 2011; Henson et al., 1999), these

results are also compatible with the proposal that increased argument-storage demands entail increased

argument-retrieval demands. Furthermore, the IFG responsivity to increasing argument-reordering

demands is in line with crosslinguistic data on argument-order processing (Ben-Shachar et al., 2003;

Friederici, Fiebach, et al., 2006; Kinno et al., 2008).

In sum, a possible conceptual scenario for the working-memory sub-processes of argument

retrieval and reordering is the following: Argument–verb dependency processing might involve the

initial storage of incoming arguments in working memory across the argument–verb distance; at their

subcategorizing verb, these arguments are retrieved; finally, an argument-reordering step is required to

re-establish the original, that is, intended argument order (e.g. in case of object-first sentences). A sketch

of this sequence of processes is provided in Figure 7.1.

Argument–verb
integration

Feature Argument High demands

extraction reordering for object-first

Argument Argument High demands

encoding Argument storage retrieval for long distance

Subcategorizing
Argument
verb

Time

Figure 7.1: Conceptual sketch of possible working-memory sub-components involved in

argument–verb-dependency processing: In brief, arguments are extracted from the speech stream on
encounter, encoded, and stored in working memory. At their subcategorizing verb, arguments are
retrieved and reordered into their original order based on their syntactic features; retrieval demands are
higher for long argument–verb distances due to memory decay, whereas reordering demands are higher
for object-first sentences.

In spite of the neurolinguistic evidence for an involvement of the left TP region in

working-memory storage and retrieval and the cross-linguistic evidence for a role of the left IFG in

103
7. Combined Analyses

argument reordering, it is unclear how these processes of argument retrieval and reordering map onto

the spatiotemporal neuronal dynamics of this underlying neuroanatomical network.

The current study focused on these dynamics, using a previously implemented paradigm that

orthogonally manipulated argument storage and argument reordering in German sentences. To perform

the current combined fMRI–EEG analyses, previously acquired fMRI (Meyer, Obleser, Anwander, &

Friederici, 2012) and EEG (Meyer, Obleser, & Friederici, 2012) data were re-analyzed from a sub-sample

of 14 participants who had participated in both of these studies. Our combined analysis can principally

map temporal EEG dynamics to spatially-precisely-defined regions derived from the fMRI results.

As outlined above, we expected that memory retrieval at subcategorizing verbs is harder for long

as compared to short argument–verb distances due to increased memory decay (Baddeley, 2012; Gibson,

2000). Since memory retrieval has been related to the P300 range (cf. Ergen, Yildirim, Uslu, Gurvit, &

Demiralp, 2012; see Section 7.4), we hypothesized that this difference results in early, increased brain

activity in the working-memory-retrieval-related TP region in response to the subcategorizing verb,

leading to an increased response in the TP-region EEG source time course for long as compared to

short argument–verb dependencies. At the EEG sensor level, an increased correlation of the ERP over

posterior sensors with the fMRI effect in the TP region is expected temporally proximate to the source

response. Equivalently, we hypothesized that increased argument-reordering demands (object-first as

compared to subject-first argument orders) subsequently increase brain responses in the left IFG during

the processing of subcategorizing verbs. Accordingly, this should be reflected by a response in the

IFG dipole time course (greater for object-first than for subject-first argument orders) as well as an

increased correlation of the ERP over frontal sensors with the fMRI effect in the IFG region. Previous

research has found argument-reordering demands to drive ERP effects in the P600 range (cf. Friederici,

Steinhauer, & Pfeifer, 2002; see Section 7.4); therefore, we expected both the source- and sensor-level

effects of argument reordering to occur in a similar time range. To test these hypotheses, we used two

complementary analysis techniques: Firstly, we reconstructed the time course of the TP and IFG ROIs

from ERP data to test the hypothesis about the temporal sequence of activation due to reordering and

retrieval. Secondly, we correlated, across participants, task-specific fMRI activations of these two regions

with the ERPs at the sensor level.

104
 7.2. Methods

7.2 Methods

7.2.1 Participants

Data of fourteen participants (mean age 26.7 years, SD 3.5 years, 6 females, all native speakers of Ger-

man) were analyzed. These data were taken from two larger studies of 24 participants (fMRI study) and

36 participants (EEG study), from which we chose those participants who had taken part in both exper-

iments (in counterbalanced order; minimum time between sessions was 53 days, maximum 160 days).

Participants had been matched for their reading span being in the range between 2.5 and 4.5 (sub-sample

mean 3.8, SD 0.7) according to an abridged version of the reading-span test (Daneman & Carpenter,

1980). All participants were right-handed as assessed by an abridged version of the Edinburgh Inventory

(Oldfield, 1971), reported no neurological or hearing deficits, and had normal or corrected-to-normal

vision. Participants were paid A

C14 for participation in the fMRI study and an additional A
C17.50 for

participation in the EEG study. Written informed consent was obtained from all participants. All

procedures received ethical approval by the local ethics committee (University of Leipzig).

7.2.2 Materials

As described in Section 4.2.3 and Section 6.2.2, auditory recordings of all stimuli from the stimulus

set described in Section 2.3 were used in the experiments. For the fMRI study, 192 additional filler

sentences described in Section 4.2.3 were used. For the EEG study, we had chosen to not use filler

sentences to keep the recording sessions as short as possible: Long-lasting EEG recording sessions tend

to decrease participants’ attention and increase the amount of recording artifacts towards the end of the

experimental run. As a second step to counter participants’ attention drop, we increased the percentage

of task trials to 25 % (i.e., 48 sentences per participant).

For the fMRI study, each participant received an individual list of 216 stimuli; for the EEG

study, each participant received the total stimulus set of 192 stimuli. As a task to maintain par-

ticipants’ attention in both the fMRI and the EEG study, 16.7 % (fMRI) and 25 % of trials (EEG)

introduced a who-did-what-to-whom yes–no comprehension question. The proportion of yes–correct

and no–incorrect questions was balanced. For further details on stimulus preparation, please refer to

the respective sections above.

105
7. Combined Analyses

7.2.3 Procedure

I here repeat parts of Section 4.2.4 and Section 6.2.3, so the reader can grasp how details of the acquisition

modality necessitated subtle changes to the respective procedures. The order of participation in the

fMRI and EEG was counterbalanced across participants. For the fMRI study, the auditory stimuli

were presented using air-conduction headphones (Resonance Technology, Inc., Northridge, CA, USA).

Visual stimuli were presented on a Sanyo PLC-XP50L LCD XGA mirror-projection system with a

refresh rate of 100 Hz (Sanyo Electric Co., Ltd., Moriguchi, Japan), mounted onto the headcoil. In the

EEG study, auditory stimuli were presented using a pair of Infinity® Reference I MkII stereo speakers

(Harman International Industries, Inc., Stamford, CT, USA). Visual stimuli were presented using a

Sony Trinitron® Multiscan G220 CRT VGA monitor with a refresh rate of 75 Hz (Sony Corporation,

Tokyo, Japan). Across recording modalities, visual stimuli appeared in a sans-serif font in black letters

against a gray background (font size 20 px).

In the fMRI study, a trial started with a fixation cross that stayed on screen for the whole

trial. After a random jitter of 0, 500, 1000 or 1500 ms, an auditory stimulus started (mean length

4.9 s, SD 0.36 s). To keep the number of acquired volumes (and thus the signal-to-noise ratio) constant

across conditions, we interpolated a silent period and an on-screen fixation cross between stimulus and

trial ending to arrive at a constant trial duration of 8 s. Such a sequence was either followed by the

next trial or by a visual comprehension question. The question remained on screen for 1500 ms, and

participants were instructed to answer the question as quickly as possible during this time period. Visual

feedback was given for 1000 ms by a green happy or red sad emoticon. To also keep the duration of the

comprehension probes constant, silence and an on-screen fixation cross were interpolated, such that

each comprehension probe would last 4 s. Participants were instructed to answer the comprehension

questions via button press with either their left or right hand, with one hand corresponding to yes

and the other to no. Response button assignment was counterbalanced across participants. In the EEG

study, a trial started with a green fixation cross of a random length between 2000 and 3500 ms. After

this, the fixation cross turned red, and an auditory stimulus was presented. This extended prologue

was used to avoid oculomotor artifacts, which else would threaten to decrease the signal-to-noise ratio.

Additionally, participants were instructed to blink only when the fixation cross showed up green. Either

the next trial followed, or a yes–no comprehension question. Participants had to answer these questions

106
 7.2. Methods

by using either their left or right hand to press one of the two buttons of a two-button response box,

whereby response-button assignment was counterbalanced across participants. Prior to comprehension

questions, a green fixation cross of a random length was presented to avoid task-preparation effects. To

ensure participants’ comfort and avoid task artifacts (Hagoort et al., 1993), comprehension questions

stayed on screen until a button press occurred. Like in the fMRI procedure, visual feedback was given

after comprehension questions for 800 ms in the form of a happy green or sad red emoticon. An

experimental run, consisting of 192 trials, lasted for approximately 35 minutes. Including preparation

and electromagnetic position tracking (see below), the experiment lasted approximately 1.5 hours.

7.2.4 Data Acquisition

Details of structural- and functional-MRI acquisition protocols can be found in Section 4.2.5; the details

for the EEG data acquisition can be found in Section 6.2.3. The participant sub-sample of the current

analyses did not contain individuals for whom electromagnetic electrode-position determination failed.

7.2.5 Data Analysis

For assessment of behavioral performance in both the fMRI and EEG study, d" -scores were available

from the original experiments (see Section 4.2.6.1 and Section 6.3.1). For both the fMRI and EEG

study, a one-sample t-test on the difference between the mean d" -scores and chance-level performance

(d" = 0, i.e., 50 % correct responses) was performed. A 2 (fMRI versus EEG) × 2 (subject-first versus

object-first) × 2 (short versus long) ANOVA was run on the response data.

Preprocessing and analysis of the fMRI data is described in Section 4.2.6.2. The resulting statistical

maps were thresholded at p < 0.005. Details of the EEG data analysis are described in Section 6.2.4. For

the participant sub-sample used in the present analyses, the overall trial-rejection rate was 16.3 %.

7.2.6 Combined Analyses in Sensor and Source Space

For the combined fMRI–EEG analysis in sensor space, fMRI-informed topographical correlation anal-

yses were performed. To use the fMRI results as a regressor for the EEG data, we first calculated the

individual signal change inside the functionally defined fMRI ROIs (see Section 7.3.2) for each of the four

experimental levels (subject-first and object-first as well as short and long). To reduce the amount of data,

107
7. Combined Analyses

we down-sampled the ERPs and their respective baseline periods to 100 Hz. We then calculated four

mean ERPs for each participant, corresponding to the four design levels extracted from the fMRI data,

resulting in an individual average ERP for the subject-first (short and long), object-first (short and long),

short (subject-first and object-first) and long (subject-first and object-first) conditions. The fourteen

individual ERPs for each of these four levels underwent across-participants electrode- and sample-wise

Pearson’s linear regression analyses with the fMRI-based regressors (see above), resulting in a time series

of 120 (200 ms baseline period plus 1 s ERP) topographical coefficient maps for the correlation between

the fMRI signal change and the ERPs for the four respective levels. To retrieve the final statistical maps

for the two main effects of our design, we z-transformed these maps for the four levels and computed

a difference map for the main effect of reordering (object-first minus subject-first) and retrieval (long

minus short). The resulting difference values were divided by the difference between the standard errors,

and converted to p-values, which underwent false-discovery-rate (FDR) correction across samples and

electrodes (Benjamini & Hochberg, 1995) to control for the inflated type-I-error risk. For the combined

fMRI–EEG analysis in source space, ERP-informed dipole-time-course analyses were performed in

SPM8, using spatially precise fMRI priors (Daunizeau, Laufs, & Friston, 2010). First, the individual

high-resolution anatomical scans were normalized to MNI space using both Fieldtrip and SPM8. After

unified segmentation, individual BEMs (Besl & McKay, 1992) were generated, to which the individually

determined electrode positions were co-registered. A lead-field matrix for each point in this volume

conductor was generated using the Fieldtrip toolbox for EEG/MEG analysis (Oostenveld et al., 2011).

For the source-time-course analysis, a variational-Bayesian equivalent-current dipole (VB-ECD) proce-

dure (Kiebel, Daunizeau, Phillips, & Friston, 2008) was applied. We used subject-specific fMRI-based

location priors for the IFG and TP region (see Section 7.3.2) to derive source time courses. Location

priors were derived by determining the MNI coordinate of the individual statistical peak voxel inside

the respective group-level ROI (see Section 7.3.2) where we used a first-level t-contrast for the respective

main effect, masked for the respective ROI (i.e., object-first greater subject-first for the IFG ROI, long

greater short for the TP-region ROI, see Section 7.3.2). The dipoles at these prior positions were allowed

to relocate freely inside a radius of 0.5 cm and to change their orientation and moment, following the

analysis strategy used by Friederici et al. (2000). The VB-ECD algorithm was set to perform iterations

of minimizing the negative free energy F to fit the dipole locations and orientations to the actual sensor

108
 7.3. Results

data across trials and conditions. The dipole locations and orientations were determined at convergence,

and the individual EEG sensor data were projected onto these dipoles through the individual lead-field

matrix—resulting in final time courses of dipole moments Qx , Qy , and Qz in all three spatial directions.

For each ROI, the first eigenvariate from a principal component analysis on these three dipole moments

was used to arrive at a single time course for each trial. Finally, time courses for each ROI and participant

were averaged and t-tests were performed on specific time windows (see Section 7.3.2).

7.3 Results

7.3.1 Behavioral Results

For the fMRI data, mean d" -score was 0.6 (SD 1.0), which was significantly different from chance

(t(13) = 2.17, p < 0.05). For the EEG data, mean d" -score was 2.8 (SD 0.5), which was significantly dif-

ferent from chance as well (t(13) = 21.05, p < 0.001). The ANOVA on the condition-specific scores for

the fMRI and EEG experiment showed a main effect of experiment (fMRI versus EEG; F(2,13) = 116.60,

p < 0.001), with no other main effects or interactions present.

7.3.2 Combined Analyses

As functionally defined spatial priors for the combined fMRI–EEG analyses, two main activation foci

were determined from the fMRI data, applying stimulus functions using the duration of the whole

sentences (see above), shown in Figure 7.2.

A
1
Z
4
B 1
Z
4
123%4!566$789:;<
123%4!566$789:;<

." !,"

,) !,)

," !."

/) !.)

!0" !)) !)" !() !(" !)" !() !(" !.) !."
123%G!566$789:;< 123%G!566$789:;<
=%&%/0 =%&%,,

Figure 7.2: Brain activations for the main effects of argument order (magenta cluster in panel A)
and argument–verb distance (green cluster in panel B). Activations are thresholded at p < 0.005 at a
minimum cluster size of 20 supra-threshold voxels. The respective coordinate systems (left) illustrate
the distribution of the individual dipole locations and orientations inside the IFG (A) and TP region (B)
group-activation clusters in the axial plane after relocation.

109
7. Combined Analyses

A test for the main effect of reordering (object-first sentences leading to more activation than

subject-first sentences) yielded a peak in the left IFG (group-level peak at x = –45, y = 14, z = 16;

peak z-score 3.21; cluster size 127 voxels), shown in panel A of Figure 7.2. The activation focus for

the main effect of retrieval (long argument–verb distances leading to stronger activation than short

argument–verb distances due to increasing memory decay and accordingly increasing retrieval demands)

was obtained in the left TP region, more specifically the Rolandic operculum (group-level peak at

x = –42, y = –25, z = 22; peak z-score 3.38; cluster size 23 voxels), shown in panel B of Figure 7.2.

Because no significant interactions were found on the whole-brain level, we restricted our combined

analyses to the main effects of reordering (object-first versus subject-first) and retrieval (long versus

short). Table 7.1 summarizes the activations (minimal cluster size 20 voxels).

Table 7.1: List of significant clusters in the fMRI contrasts, thresholded at p < 0.005 and a minimum
cluster extent of 20 voxels*.

MNI coordinate
Site Cluster size (mm3 ) Z-score
X Y Z

Object-first > subject-first

–45 14 16 3.21
IFG / BA 44* –60 20 13 1143 3.14
–51 20 22 3.13
IFG / BA 45* –42 5 52 243 3.01

Long > short

Rolandic operculum* –42 –25 22 207 3.38
–18 –34 49 2.92
Postcentral gyrus / area 3a* –30 –31 –52 198 2.61
–27 –40 46 2.60

*According to Eickhoff et al. (2005).

The fMRI-informed topographical correlation analysis in EEG sensor space at the subcategorizing

verb yielded a significant late left frontal correlation for the reordering factor (i.e., a significant difference

in correlations between the object-first and subject-first sentences with the signal change in the IFG

ROI). This correlation difference was present at electrode AF7 (0–250 ms), electrodes FC5, F7, AF7, FP1,

FP2 and F8 (251–500 ms), electrodes FC5, F5, F7, C3, FC3, F3, AF3, AF7 and CP3 (501–750 ms), and

110
 7.3. Results

electrode F7 (751–1000 ms). In the average time course over these sensors the significant difference lasts

from 300 ms to 740 ms (Figure 7.3 A; p < 0.05, FDR-corrected). The analog analysis for the retrieval

factor (i.e., the difference in correlations between the long and short argument–verb dependencies with

the signal change in the TP region ROI) yielded a significant early left posterior correlation at electrodes

CP5, TP7, TP9, PO7 and O1 (0– 250 ms), and electrodes PO7 and O1 (251–500 ms), lasting from 10 ms

to 470 ms in the average over these sensors (Figure 7.3 B; p < 0.05; FDR-corrected).

A –2 μV Significant difference
B –2 μV Significant difference

1s 1s

Subject-first Short
2 μV Object-first 2 μV Long
0.5–0.75 s 0–0.25 s
Z Z
1 3 IFG correlations 1 3 TP correlations

Figure 7.3: Results of the fMRI-informed topographical correlation analysis for the IFG (A) and
TP region (B), respectively. Circles mark EEG sensors at which significant fMRI–EEG correlation
differences were obtained, whereby the topography represents the average correlation across the named
time window. Waveforms show the average ERPs across these EEG sensors. Gray lines mark significant
sample-wise correlation differences (p < 0.05, FDR-corrected).

The dipole-time-course analysis in MRI source space for the main-verb epoch involved minor

fMRI-prior relocations by the VB-ECD algorithm (< 1 mm), resulting in an average relocated IFG

dipole position of x = –53, y = 21, z = 20 and an average relocated TP-region dipole position of x = –42,

y = –26, z = 20 (individual dipole locations after relocation are shown in Figure 7.2). Statistical analyses

of the dipole time courses (object-first and subject-first across the individual IFG dipole positions,

long and short across the individual TP-region dipole positions) in two time windows defined by

our prior hypotheses (300–500 ms for the IFG dipole time course, 200–300 ms for the TP-region time

course) showed significantly increased dipole activity in the IFG for the object-first as compared to

the subject-first argument orders from 300 to 500 ms. While our hypothesis-based time window for the

TP-region time course did not yield a significant difference between the short and long conditions, visual

inspection (see Figure 7.4 B) informed us about a second, earlier time window showing a difference

(75 samples from 35 to 180 ms, see discussion). Although selected post-hoc and as such not a statistically

significant result, for this time window we found t(13) = 2.57, p < 0.05 for the long as compared to the

short argument–verb distances (Figure 7.4 B).

111
7. Combined Analyses

A 2 nAm Subject-first
Object-first B 2 nAm Short
Long

1s 1s

–2 nAm Significant difference –2 nAm Significant difference

IFG dipoles TP region dipoles

Figure 7.4: Results of the ERP-informed dipole-time-course analysis in MRI source space for the
IFG (A) and TP-region (B) dipoles. Dots in the brain renderings mark the final dipole positions; the
respective waveforms illustrate the reconstructed grand-average dipole time courses across IFG (A) and
TP-region (B) dipoles; gray areas mark significant source-activation differences (p < 0.05, uncorrected).
The TP region shows an early difference between the short and long argument–verb distances, surfacing
as an early left posterior positivity. The IFG shows a late difference between the subject-first and
object-first argument orders, surfacing as a late left frontal positivity.

7.4 Discussion

7.4.1 Behavioral Data

The statistical analysis on the behavioral data from our fourteen participants suggests that while behav-

ioral performance in both the fMRI and the EEG study was significantly above chance, performance

in the fMRI study was significantly worse than it was in the EEG study. As a second finding, there

were no condition-specific effects in either the fMRI or the EEG study. While the second speaks in

favor of a balanced design and no processing-difficulty confounds, the first means that the participants

were able to process and comprehend all four conditions of our experimental paradigm. Given the

overall better performance in the EEG study, the difference in performance between the fMRI and

the EEG study may be explained by the different task procedures: The response window in the fMRI

study was strongly time-constrained, whereas in the EEG study it was not. Given participants’ ability

to process our stimuli in the EEG study, we are confident that the task in the fMRI study served to keep

participants’ attention directed towards the sentences, rendering the results valid.

7.4.2 Combined Analyses

The current fMRI analyses are in line with the results from the full 24-participant sample (Meyer,

Obleser, Anwander, & Friederici, 2012; see Chapter 4): The left IFG is sensitive to the rela-

tive order of arguments in transitive German sentences, while the left TP region is sensitive to

112
 7.4. Discussion

argument–verb-dependency length. More importantly, the combined fMRI–EEG analyses at the

subcategorizing verb—the main focus of the current study—suggest that IFG source activity occurs

relatively late, whereas TP-region activity was only post-hoc found to occur earlier than predicted by

our prior hypotheses (Figure 7.3). Topographical correlation analyses showed that IFG activity surfaces

in a left frontal late positivity, while TP-region activity surfaces in an early left posterior positivity

(Figure 7.4). Both correlation and source-space results speak in favor of a plausible role of the IFG

and TP-region dipoles in the generation of the observed ERP response: Based on previous reports

(Friederici, Fiebach, et al., 2006; Novais-Santos et al., 2007), we suggest that the reconstructed source

time courses of the IFG and TP region, in response to the subcategorizing verb, reflect the activation

time course of the neuroanatomical substrates of early argument retrieval (TP region) and late argument

reordering (IFG) at subcategorizing verbs. Furthermore, we suggest that the sensor-level fMRI–EEG

correlations mirror this functional course of retrieval and reordering. Before we turn to the combined

fMRI–EEG findings, we will first briefly discuss the fMRI results.

Increased activation in the left IFG for object-first as compared to subject-first argument orders

has been reported in previous imaging work from Hebrew, German, and Japanese (Ben-Shachar et

al., 2003; Friederici, Fiebach, et al., 2006; Kinno et al., 2008; Obleser & Weisz, 2011), independent

of working-memory demands (Caplan & Waters, 1999; Waters & Caplan, 1996). Increased activation

in the TP region, on the other hand, suggests a role in retrieval of the argument at the main verb,

which is plausible given previous reports of a role of the left SMG and inferior parietal regions during

verbal-working-memory retrieval outside of sentence processing (Buchsbaum et al., 2001; Henson et al.,

1999; Ravizza, Delgado, Chein, Becker, & Fiez, 2004; Ravizza et al., 2011). Furthermore, the posterior

STG has been found active for verbal-working-memory storage during the processing of ambiguous

sentences with long retention intervals for the disambiguating information (Novais-Santos et al., 2007).

In addition, reduced verbal-working-memory-related parietal brain activity has been suggested as a

source of sentence-processing difficulties in seniors (Grossman et al., 2002). For a more comprehensive

discussion, we refer the reader to our previous study (Meyer, Obleser, Anwander, & Friederici, 2012).

In the following, we will focus on the combined fMRI–EEG findings.

The increased TP-region activity for increasing argument–verb distances, surfacing as an early left

temporo-parietal positivity in the current topographical correlation analysis, matches our hypothesis

113
7. Combined Analyses

that long argument–verb distances result in increased argument-retrieval demands at the subcategorizing

verb. Furthermore, we suggest that the combined fMRI–EEG result provides a possible link between

the evidence on TP-region involvement in verbal working memory discussed above and previous

reports of early positive ERP responses during verbal-working-memory retrieval (Donchin & Coles,

1988; Grossberg, 1984): The present results show at the sensor-level an early positive correlation with

the fMRI TP-region activity. While there has been a previous proposal along these lines (Birbaumer,

Elbert, Canavan, & Rockstroh, 1990), there remained a lack of fMRI–EEG studies that merit a direct

connection.

In line with this proposal, a previous study reports a P300 during tone retrieval from work-

ing memory, whereby the latency of this effect was positively correlated with digit-span-test scores

(Polich, Howard, & Starr, 1983). Ergen et al. (2012) found matching results for letter retrieval. Re-

view studies stress these convergences (Friedman & Johnson, 2000; Polich, 2007; Rugg & Curran,

2007). Studies from the sentence-processing domain also observed positivities in the P300 range

during verbal-working-memory retrieval: Friedman et al. (1975) performed a study in which the

meaning of early words in a sentence could only be resolved at later words—requiring sentence-final

working-memory retrieval, giving rise to a bilateral P300 response. Later work observed a parietally

distributed positive response around 345 ms (P345) during ambiguity resolution at subcategorizing verbs

(Friederici, Steinhauer, Mecklinger, & Meyer, 1998; Mecklinger, Schriefers, Steinhauer, & Friederici,

1995). Crucially, this response was larger in amplitude when the status of an ambiguous argument had to

be revised from subject to object (Mecklinger et al., 1995). Also, its occurrence was tied to participants’

reading span (Friederici et al., 1998). Hence, it is possible that the P345 indexes argument retrieval

from verbal working memory for later modification, which is in line with Phillips et al.’s (2005) report

of a positivity occurring as early as 300 ms during the resolution of an argument–verb dependency

at a main verb. The positivity in the present findings has an even shorter latency. Congruent with

this, previous results suggest that argument–verb-dependency resolution in verb-final languages may

be initiated already prior to the main verb and continue throughout the verb (Aoshima et al., 2004;

Friederici & Mecklinger, 1996; Phillips et al., 2005). This is also compatible with Fiebach et al.’s (2001)

findings of a pre-verbal positivity during the resolution of argument–verb dependencies in German

sentences, and may explain why the TP-region dipole time course effect occurred earlier than predicted.

114
 7.4. Discussion

An alternative explanation of the early topographical correlation effect is that it does rather

not reflect a P300, but a negative ERP component peaking around 400 ms (N400), classically obtained

for increased semantic-integration demands during sentence processing (Van Petten, 1993). A reduced

N400 amplitude for the long as compared to the short argument–verb dependencies (Van Petten &

Luka, 2012) is predicted under an anticipation-based parsing account (Konieczny & Döring, 2003; Levy,

2008): Here, an increased argument–verb distance should decrease processing load at the subcategorizing

verb due to cumulative lexical pre-activation and subsequently-facilitated lexical access (Hagoort, Hald,

Bastiaansen, & Petersson, 2004; Van Petten, 1993). This is compatible with both the ERP and the source

time course in the TP region. For three reasons, however, we consider this interpretation less likely:

First, the cortical generators of the N400 are rarely found in the left TP region, but rather involve

middle temporal cortices (Johnson & Hamm, 2000; Maess, Herrmann, Hahne, Nakamura, & Friederici,

2006; Silva-Pereyra et al., 2003; Simos, Basile, & Papanicolaou, 1997) or a left-lateralized network of

middle and inferior temporal cortices (Halgren et al., 2002; for review, see Lau, Phillips, & Poeppel,

2008). Second, the classical scalp distribution of the N400 during sentence processing involves bilateral

parieto-occipital sensors, with a slight tilt towards the right hemisphere (Kutas & Van Petten, 1994;

Lau et al., 2008). The ERP component at which we observed a stronger fMRI–EEG correlation for

the long as compared to the short argument–verb distances had, however, a clearly left-lateralized

posterior distribution. Third, the differential fMRI–EEG correlation for long as compared to short

argument–verb dependencies occurred too early to index a lexical-semantic response.

In sum, it is most plausible that the short latency of the observed response is related to the

pre-verbal initiation of argument retrieval (Fiebach et al., 2001; Phillips et al., 2005), triggered by a

pre-head-attachment mechanism as described in the active-filler hypothesis (Clifton & Frazier, 1988).

It is, however, impossible that a semantic N400 response to the subcategorizing verb occurs prior

to the verb itself. Nevertheless, a partial overlap between a working-memory-retrieval-related P300

and an increased lexical-activation-related N400 for the short as compared to the long argument–verb

dependencies is possible for a later time window—in line with an antagonism of lexically induced

processing facilitation and working-memory-induced processing difficulty.

This leads us to the later brain activity in the left IFG that was sensitive to reordering demands.

Considering that the topographical correlations showed a late-left-frontal-positivity correlate of IFG

115
7. Combined Analyses

activity, we suggest that our combined fMRI–EEG analysis provides evidence for a close relationship

between reordering-related IFG activity and reordering-related late positive ERP components occurring

around 600 ms (P600). While the late effect in the dipole time course of the IFG is in line with this

proposal, it is very unlikely that the IFG is the single generator of the P600: Service, Helenius, Maury,

and Salmelin (2007) have shown that P600 generators may span from bilateral middle temporal to left

inferior frontal generators across participants.

Inferior-frontal-gyrus activity can nevertheless account at least partly for the sensor-level time

courses observed in the present data, and its involvement is directly implied by the source-level time

courses. While responses in the P600 range were first observed in response to syntactic violations

(Friederici, Pfeifer, & Hahne, 1993; Hagoort et al., 1993; Osterhout & Holcomb, 1992; Osterhout &

Swinney, 1993), later work found that a P600 response at subcategorizing verbs does also occur in

ambiguous sentences. These do not contain violations, but instead require the revision of an initial

interpretation (Friederici et al., 1998). More recently, Kaan, Harris, Gibson, and Holcomb (2000)

proposed to interpret the P600 as a general index of syntactic-processing difficulty, with sentential

complexity giving rise to an anterior scalp distribution and revision processes giving rise to a posterior

scalp distribution (Kaan & Swaab, 2003).

While we suggested above that early posterior positivities rather reflect retrieval from working

memory, our proposal that late anterior P600 effects may more specifically index argument reordering

is compatible with Kaan and Swaab’s (2003) suggestion. Furthermore, the reordering proposal is in

line with previous German data by Rösler et al. (1998), who report a P600 response for object-first as

compared to subject-first argument orders, starting before and continuing during the occurrence of the

verb. Converging on these data, Friederici et al. (2002) report a fronto-centrally distributed P600 for

object-first as compared to subject-first sentences at subcategorizing verbs.

The interpretation of the current results in terms of a dissociation of late reordering-related

fronto-central P600 components and the early retrieval-related posterior positive ERP responses is

in line with the results of Vos, Gunther, Schriefers, and Friederici (2001). In their ERP study, they

compared object-first and subject-first argument orders at the subcategorizing verb. The researchers

find an increased late frontal positivity for object-first as compared to subject-first argument orders

only for low-span participants; for high-span participants, the authors find an early posterior positivity

116
 7.5. Conclusion

instead. This dissociation fits our results in that it may reflect low-span participants’ relative reliance

on reordering processes (as indexed by the P600), whereas high-span participants may rely relatively

stronger on their working-memory capacity (as indexed by the early positivity).

7.5 Conclusion

Based on combined fMRI–EEG results as well as evidence from the fMRI and EEG literature, we

propose that working-memory retrieval of arguments and argument reordering are core neurocog-

nitive functions of argument–verb-dependency resolution in sentence processing. During argument

retrieval at the subcategorizing verb, the left TP region supports initial argument retrieval, while later

argument reordering is subserved by left IFG. The data and preliminary framework presented here

generate testable hypotheses for both behavioral and neuroimaging studies; they demonstrate that joint

fMRI–EEG analyses provide the explanatory power to reconcile models from cognitive neuropsychol-

ogy and psycholinguistics with our increasing knowledge on human functional neuroanatomy.

117
8
GENERAL DISCUSSION

This thesis was concerned with the spatiotemporal brain dynamics of the following conceptualization:

The working memory of argument–verb-dependency processing involves argument storage across

the argument–verb distance; at their subcategorizing verb, arguments are retrieved, and in case of a

relative argument order that deviates from the idiosyncratic order, argument reordering takes place.

To address the according research questions, the 2 × 2 factorial design crossed the factors argument

reordering (subject-first versus object-first argument order) and argument storage and retrieval (short

versus long argument–verb distance). Across results, three leitmotifs were identified, which will be

discussed after summarizing the experimental findings in Section 8.1. Section 8.2 will discuss whether

the IFG is involved in reordering, rehearsal, or both; Section 8.3 will investigate the question whether

the TP region is involved in storage, retrieval, or both. Section 8.4 will focus on the role of the AF/SLF

across healthy and clinical populations.

8.1 Summary of Experimental Findings

Chapter 3 asked about the behavioral relationship between argument storage and reordering. Contrary

to our hypotheses, the results show an interaction of the two factors, both using an on-line SPR and

an off-line rating methodology: Increasing argument–verb distance exacerbates processing only for

subject-first argument orders; for object-first argument orders, long argument–verb distance facilitates

processing. Two possible interpretations were discussed. First, argument reordering and storage may

interact behaviorally, but operate independently on a neural level: If distinct brain regions were respon-

sible for each argument reordering and argument storage, a link between these regions could selectively

119
8. General Discussion

facilitate the processing of sentences that necessitate both argument reordering and argument storage.

Second, the two processes may also interact on the neural level, driven by a common underlying mecha-

nism: Anticipation of a verb based on argument information might be easier in object-first sentences,

since virtually any verb subcategorizes a subject, but not necessarily an object. The two competing

interpretations necessitated a combined fMRI–dMRI study.

Chapter 4 asked about the functional neuroanatomy of argument storage and reordering, testing

the current paradigm in a combined fMRI–dMRI study. The fMRI results show an activation for

argument reordering in the left IFG, an activation for argument storage in the left TP region, and

no interaction. Activation of the TP region, but not of the IFG, correlated with working-memory

span measures. The dMRI results found the AF/SLF to connect the IFG and TP region, and correla-

tion analyses stressed the functional significance of the AF/SLF. The additional VBM analysis found

that the functional asymmetry of the reordering-related IFG activation across the left and right hemi-

spheres is correlated with the underlying structural asymmetry, which is predicted by participants’

handedness. In sum, these findings suggest that argument storage and reordering are independent on

the neural level, but that the involved regions are linked by long-range white-matter bundles—giving

rise to the behavioral pattern observed in Chapter 3. The absence of a functional interaction stresses

that storage occurs independently of argument order, and that reordering occurs independently of

argument–verb distance. The correlation between span measures and TP-region activation suggests that

an isomorphism between the reordering–storage and rehearsal–storage dichotomies is restricted to stor-

age, entailing a dissociation between reordering and rehearsal in the left inferior prefrontal cortex. The

functional–structural–behavioral connection in the VBM results points to a general leftward structural

asymmetry of the IFG and its functional leftward asymmetry during sentence processing; it also implies

that left-handers may involve the right IFG in this process.

Chapter 5 asked whether selective damage to the AF/SLF results in a selective processing deficit

in sentences that demand both argument reordering and argument storage. A patient for whom DTI

revealed a disconnected AF/SLF was tested on our paradigm and a working-memory test battery. The

patient showed reduced performance across working-memory tests and a selective deficit on sentences

that jointly tax argument reordering and storage. The results reconcile the behavioral interaction in

Chapter 3 with the functional main effects and their structural connection in Chapter 4: Argument

120
 8.1. Summary of Experimental Findings

reordering (IFG) and storage (TP region) operate independently on the neural level, but their anatomical

connection facilitates the processing of object-first argument orders at long argument–verb distances

only. The combination of a general working-memory deficit and a selective sentence-processing deficit

implies the AF/SLF in a dual reordering–rehearsal role, entailing the functional segregation of the two.

Chapter 6 investigated the neuroelectric substrate of argument storage, regardless of argument

order, since previous EEG research on working memory during argument–verb-dependency processing

does not speak to the reordering–storage dichotomy. Short and long argument–verb conditions were

contrasted in a TFR analysis. Long distances elicited stronger sustained alpha oscillations during the

storage phase as well as a beta-band effect immediately at the main verb. The alpha-power increase was

localized to bilateral occipital and left parietal cortices, whereby the latter source activity correlated

with reading span. Reading span was also predictive of participants’ behavioral performance. The results

suggest that left parietal alpha oscillations be an important neural substrate of argument storage in

working memory during argument–verb-dependency processing, potentially relating to the TP-region

BOLD effect from Chapter 4. The alpha oscillations may serve a functional-inhibition mechanism of a

premature filler release prior to an appropriate gap position, while the beta effect at the main verb may

index argument retrieval for argument–verb linking.

Chapter 7 asked about the spatiotemporal dynamics of argument retrieval and argument

reordering at subcategorizing verbs inside the functional-neuroanatomical brain network established

in Chapter 4 and Chapter 5—as opposed to argument storage as investigated in Chapter 6. These

fMRI–EEG analyses combined a sub-sample of data from Chapter 4 and Chapter 6 into two methods

of joint analysis, focusing on the sentence-final main verb. Dipole time courses for the IFG and the

TP region projected from sensor-level ERP data show increased early TP-region activity for increased

argument-retrieval demands, followed by later IFG activity for increased argument-reordering demands.

Topographical correlation analyses substantiate these findings, showing an early left posterior positiv-

ity for increased argument-retrieval demands, followed by a late left frontal positivity for increased

argument-reordering demands. The results suggest that the reconstructed source time courses and

correlated ERP effects in response to the subcategorizing verb reflect the spatiotemporal dynamics of

early argument retrieval—involving the TP region—and late argument reordering—involving the IFG.

121
8. General Discussion

8.2 The Inferior Frontal Gyrus: Reordering, Rehearsal, or Both?

The first leitmotif of this thesis’ experimental results is the distinction between reordering and rehearsal:

Reordering is an executive operation on working-memory content, whereas rehearsal denotes the

refreshing of working-memory content to counteract degenerative factors such as decay and interference.

While there is evidence for a role of the IFG in both reordering and rehearsal, a confrontation remains

between authors who maintain the conceptual and neuroanatomical distinction between reordering

and rehearsal and those who propose their identity: On the one side, cross-linguistic fMRI research

has found brain activity in BA 44 to increase with argument-reordering demands (Ben-Shachar et al.,

2003; Friederici, Fiebach, et al., 2006; Kim et al., 2009). On the other side, rehearsal of working-memory

content was found to increase brain activity in BA 44 and supplementary motor regions superior and

posterior to BA 44, such as BA 6 (Awh et al., 1996; Paulesu et al., 1993; Petrides et al., 1993).

These similarities led some authors to propose the identity of reordering and rehearsal. First, Just

et al.’s (1996) contrasting of English subject- and object-relative clauses yielded BA-44 activity, which

Rogalsky and Hickok (2010) reviewed as reflecting rehearsal, due to the fact that English object-first

sentences increase the argument–verb distance over their subject-first counterparts. Second, Rogalsky

et al. (2008) report a performance decline in argument-reordering-intensive sentence processing during

concurrent rehearsal, with a partial activation overlap of reordering and rehearsal in BA 44. However,

both lines of argument are ambiguous: First, Just et al.’s (1996) English object-first sentences confounded

argument–verb distance with argument order. Second, Rogalsky et al. (2008) report interference between

a finger-tapping control task and sentence-processing performance in addition to reordering–rehearsal

interference; this suggests that a secondary task per se decreases performance. In addition, other studies

have demonstrated that reordering can be impaired independently of rehearsal (Caplan & Waters,

1999), and that concurrent rehearsal does not further increase reordering-related BA-44 activation

(Caplan et al., 2000). Finally, there is clear evidence for a role of BA 44 in reordering from outside the

sentence-processing domain: Gerton et al.’s (2004) PET study investigated the functional neuroanatomy

of the forward- and backward-digit-span sub-tests—both of which need rehearsal, but only backward

digit span needs reordering. While brain activity in the inferior parietal lobule and the superior medial

vicinity of Broca’s area was common to forward and backward digit span and reminiscent of the

storage–rehearsal network, reordering-intensive backward digit span additionally activated BA 44.

122
 8.2. The Inferior Frontal Gyrus: Reordering, Rehearsal, or Both?

This thesis adds three arguments against the identity of reordering and rehearsal. First, the

correlation analysis on the fMRI data (see Chapter 4) yielded a correlation between digit-span scores

and argument–verb-distance-related TP-region activation, but no such correlation was found for the

argument-reordering-related BA-44 activation. The latter would have been expected if reordering

and rehearsal were identical. As a second argument, the patient with an AF/SLF disconnection

(see Chapter 5) showed a general working-memory deficit, but a specific processing deficit only

on sentences that jointly demand reordering and storage. These results suggest that both the

reordering–storage and the rehearsal–storage dichotomy recur to the AF/SLF to enable connective

processing by left inferior frontal and TP regions. Still, a general working-memory deficit in com-

bination with a selective sentence-processing deficit is only predicted by a conceptual distinction

between reordering and rehearsal, as well as their neuroanatomical separation in the IFG; on the

contrary, Rogalsky and Hickok’s (2010) proposal would predict decreased performance on any

long-argument–verb-distance sentence, regardless of argument order. As a third piece of evidence,

consider the EEG-source-localization and -source-space-correlation data (see Chapter 6): First, the

sources of the argument–verb-distance-related alpha-power increase were localized to bilateral occipital

and left parietal cortices, with no significant BA-44 involvement. Second, only left parietal sources were

correlated with reading-span scores. Rogalsky and Hickok (2010) predict the opposite of both findings.

The results of the current thesis join the cross-linguistic sentence-processing studies, the

rehearsal–reordering non-interference data as well as the digit-span imaging data in underlining the

conceptual and neuroanatomical distinction between reordering and rehearsal. The scope of the

reordering concept with respect to BA 44 is, however, unclear—but the available data allow for

speculation: Since the reordering of arguments (Meyer, Obleser, Anwander, & Friederici, 2012) and the

reordering of a series of arbitrary digits (Gerton et al., 2004) both activate the dorsal BA 44, this region

may play a general role in the reordering of verbal-working-memory content. Behavioral pilot data in

favor of this proposal are available. In an experiment crossing the two long conditions (subject-first and

object-first) from the current paradigm with the forward- and backward-digit-span task, participants

heard a series of digits that was either to be rehearsed or—additionally—reordered during the subsequent

auditory presentation of a sentence. The concurrent processing of object-first sentences was expected to

selectively influence backward digit span. Preliminary results are shown in Figure 8.1.

123
8. General Discussion

Levenshtein distance
0.2
Backward
Forward

0.4

Subject-first
Object-first

Figure 8.1: Results of the pilot study (n = 30); bars mark Levenshtein distance (smaller values in-
dicate better performance, i.e., small difference between original and repeated digit series); error
bars represent SEM; blue bars mark subject-first sentences, red bars mark object-first sentences;
solid lines mark forward-digit-span (rehearsal-only) conditions, dashed lines mark backward-digit-span
(rehearsal-plus-reordering) conditions; a significant interaction was obtained (p < 0.05).

As expected, performance was worse for backward digit span. Paradoxically, while rehearsal

during forward digit span left the classical processing exacerbation of object-first argument orders

untouched, digit reordering during backward digit span facilitated concurrent argument reordering.

The facilitation may be captured by the general notions of task or structural priming (Scheepers et al.,

2011; Waszak & Hommel, 2007) or the specific notion of syntactic priming17 (for review, see Branigan,

2007; Pickering & Ferreira, 2008): In the sentence-processing domain, Bock (1986) reports that sentence

structures from a prior comprehension task constrain subsequently-produced sentence structures. More

recently, Scheepers et al. (2011) found across-domains structural priming, showing that mathematical

formulae of adequate gestalt constrain participants’ sentence continuations. Priming may have reduced

BA-44 activity, causing the concurrent backward-digit-span task to facilitate the processing of object-first

sentences: First, Makuuchi, Bahlmann, and Friederici (2012) show that structural processing in BA 44

may be common to language, working-memory, and mathematical processes; second, Wagner, Koutstaal,

Maril, Schacter, and Buckner (2000) find across-tasks repetition priming to reduce BA-44 activity.

To finalize these speculations, the current VBM data (see Chapter 4) hint at an overarching role of

Broca’s area in reordering: The structural asymmetry of Broca’s area predicts its functional asymmetry

during argument reordering, with both functional and structural asymmetry predicted by handedness.

17
Kaan, Wijnen, and Swaab (2004) use the related notion of syntactic recycling to denote the partial re-use of
previously-processed phrases in ellipsis resolution.

124
 8.3. The Temporo-Parietal Region: Storage, Retrieval, or Both?

This is suggestive considering reports in the great-apes literature that tie Broca’s area’s structural and

functional asymmetry to handedness during communicative gesturing (Cantalupo & Hopkins, 2001;

Hopkins et al., 2007; Schenker et al., 2010; Taglialatela et al., 2006). Imaging data in humans find Broca’s

area involved in deaf signers’ gesture comprehension (MacSweeney, Brammer, Waters, & Goswami,

2009; Petitto et al., 2000), potentially reflecting internal reordering of gestural actions: In a recent study,

Clerget, Badets, Duqué, and Olivier (2011) found that theta-burst transcranial magnetic stimulation

(TMS) over BA 44 decreased performance during action sequencing. A conclusion would be premature,

but it is possible that the dorsal BA 44 plays an overarching role in reordering during communication

(cf. Kemmerer, 2012; Nishitani, Schurmann, Amunts, & Hari, 2005; Pastra & Aloimonos, 2012).

8.3 The Temporo-Parietal Region: Storage, Retrieval, or Both?

The TP region is the second leitmotif across the current experimental results. The fMRI (Chap-

ter 4), EEG (Chapter 6), and combined analyses (Chapter 7) strongly suggest that the TP region is

involved in working-memory storage and retrieval, both during sentence processing and outside the

sentence-processing domain. While the fMRI and EEG alpha effects may reflect argument storage,

the EEG beta effect, the early TP-region dipole activity as well as the related P300 in the combined

fMRI–EEG analyses may index argument retrieval. These findings need clarification: First, the spatial

commonalities between the BOLD and alpha effect might entail a positive BOLD–alpha relationship.

Second, the BOLD and alpha effect suggest the TP region as a storage substrate, whereas the beta

effect and the combined analyses suggest that TP-region activity reflects retrieval. Third, some previous

findings suggest BA 45 rather than the TP region to subserve argument storage, which the current data

argue against. Fourth, the imaging literature has implied the TP region in various cognitive functions

other than working-memory storage and retrieval. These points will be addressed next.

One possibility for a joint interpretation of the BOLD and alpha effect is that both reflect

TP-region activity in inhibiting a premature argument release. This relies on two assumptions: The

BOLD and alpha sources must be at the same location, and their activity must be positively coupled.

The first is possible given the method: Source localization is spatially less precise than fMRI, and reduced

correspondence between the individual electrode positions and the canonical BEM may have further

decreased its precision in our analyses. The second assumption of a positive BOLD–alpha coupling

125
8. General Discussion

is less clear: During rest, Laufs et al. (2006) report negative coupling in both a parietal–occipital and

parietal–frontal network, while Gonçalves et al. (2006) report that some participants show negative

coupling across the brain, but some show positive coupling in frontal, left parietal and occipital regions.

During verbal-working-memory tasks, the picture is unequivocal as well: Michels et al. (2010) report

positive parietal coupling, Scheeringa et al. (2009) find no coupling in the working-memory network,

and Meltzer et al. (2007) report negative bilateral parietal coupling.

According to a second interpretation, BOLD and alpha may reflect two distinct left parietal

processes during working-memory storage. Such a dualism appeared in the respective discussions

of the behavioral (see Chapter 3) and the combined-fMRI–EEG results (see Chapter 7): While long

argument–verb dependencies expose the argument to increased storage and retrieval demands (cf. Gib-

son, 2000), they have also been argued to facilitate verb processing due to the cumulatively reduced

number of possible sentence continuations (cf. Levy, 2008). For the storage phase, these approaches

cannot be disentangled, since BOLD and alpha both occurred during this interval. But for retrieval, the

two approaches may be captured by the combined analyses: While the sensor-space counterpart of the

TP-region dipole effect was interpreted as an argument-retrieval-related P300, a partial overlap with a

lexical-activation-related N400 was considered. Argument retrieval at the sentence-final main verb may

trigger a P300 (Ergen et al., 2012; Friedman et al., 1975), but stronger expectations for the main verb

towards sentence ending may trigger an N400 reduction (Van Petten, 1993; Van Petten & Luka, 2012);

the two may be closely related, yet neuroanatomically different.

Of these two interpretations, it is more plausible that BOLD and alpha reflect the same mecha-

nism: First, both BOLD and alpha were negatively correlated with working-memory span, stressing

that both are related to working memory proper rather than lexical facilitation. Second, both the BOLD

and alpha effect were sustained, making it illogical to link either to the transient effects in the combined

analyses. Lexical facilitation can be further ruled out by the design of the experimental stimuli used

across the current thesis: All four-sentence sets were balanced for aspects of lexical facilitation by keeping

the main-verb position constant across conditions, precisely to achieve identical lexical facilitation and

identical N400 reduction across conditions; for the same reason, the argument–verb sets of all stimulus

items were controlled for argument–verb co-occurrence (see Chapter 2). In sum, BOLD and alpha may

126
 8.3. The Temporo-Parietal Region: Storage, Retrieval, or Both?

both reflect verbal-working-memory storage through a cortical-inhibition mechanism that enables a

pinpoint argument release, the spatial differences being due to methodological factors.

Does the TP region, in addition to argument storage, also serve argument retrieval? While both

BOLD (Chapter 4) and alpha (Chapter 6) may index storage, both beta (Chapter 6) and P300 (Chapter 7)

may index retrieval—in spite of all relying on the TP region. Four arguments support this. First, the

ambiguity of storage and retrieval is built into the current paradigm: If decay is the underlying reason

for the processing exacerbation induced by long as compared to short argument–verb distances (cf.

Gibson, 2000; Yngwe, 1960), distance increases both storage and retrieval demands. Second, the sluggish

BOLD response and the temporal precision of the TFR analyses (see Chapter 2) imply that neither

BOLD nor sustained alpha can reflect transient retrieval effects; hence, distinct neural substrates must

be assumed. Third, the link between P300, TP region and verbal-working-memory retrieval is in line

with previous findings (Ergen et al., 2012; Phillips et al., 2005), and may carry over to the beta burst:

Given their synchronicity at the main verb, the fifty-milliseconds beta burst can simply be the P300’s

mirror image in the TFR domain. While the beta burst was not source-localized, the beta band has been

implied in argument–verb linking, which presupposes retrieval (Bastiaansen et al., 2010; Wang et al.,

2012). Fourth and last, reconsider the fMRI result: TP-region BOLD exhibited both peak-location and

magnitude variance. Since the fMRI data were averaged across whole sentences, the differential SNR

may have resulted in a strong—sustained—storage effect and a weak—transient—retrieval effect, each

at distinct coordinates. In line with this, within-experiment distinct storage and rehearsal peaks in the

TP region have been reported previously (Henson et al., 1999; Ravizza et al., 2011).

Next, I need to clarify the contrast between the current findings and previous work that

suggests BA 45 in the IFG, sometimes extending into the inferior frontal sulcus, to subserve working

memory during sentence processing, rather than the TP region. These reports must be taken with

caution, since they compared different syntactic dependencies across conditions: Fiebach et al. (2005)

contrasted pronoun–verb and noun–verb dependencies, while Santi and Grodzinsky (2007) contrasted

pronoun–noun and noun–verb dependencies. Makuuchi et al. (2009) confounded argument–verb

distance with the number of argument–verb dependencies, and did not control for additional

pronoun–noun dependencies across conditions. Santi and Grodzinsky (2010) contrasted embedded

sentences to sentences containing a single argument–verb dependency. Potential reconciliation is offered

127
8. General Discussion

by Makuuchi, Grodzinsky, Amunts, Santi, and Friederici’s (2012) recent study: Across dependencies,

increased argument–verb distance activated the inferior frontal sulcus, the pars opercularis of BA 44, and

the intraparietal sulcus. While the paradigm was well-designed in using noun–verb dependencies only,

it did confound argument–verb distance with argument order. Hence, it is possible that the BA-44 and

the inferior-frontal-sulcus activations in Makuuchi, Grodzinsky, et al.’s (2012) study do not stem from

argument–verb distance per se. This may instead be true of their intraparietal-sulcus activation, which

is close to the current TP-region peak. Given this interpretation as well as the current results, it seems

plausible that argument storage relies more strongly on TP rather than inferior-frontal regions—in line

with a vast body of research from outside the sentence-processing domain (see Chapter 4; for review,

see Buchsbaum & D’Esposito, 2008; Jacquemot & Scott, 2006; Müller & Knight, 2006; Owen et al.,

2005; Smith & Jonides, 1998; Wager et al., 2005).

The final issue to speculate on is how the working-memory role of the TP region relates to other

cognitive processes. Based, for instance, on findings that the processing of items that are similar to

items already stored in working memory is facilitated (for review, see Cabeza et al., 2012a; Gazzaley &

Nobre, 2011), the diverse cognitive functions of the ventral parietal cortex—such as memory retrieval,

personal-memory, and language processing—have been proposed to share a process of selective attention

(for discussion, see Cabeza, Ciaramelli, & Moscovitch, 2012b; Nelson, McDermott, & Petersen, 2012).

How would such a general process relate to the specific notion of verbal-working-memory storage and

retrieval of arguments during sentence processing? A conceivable relationship recurs to Anders Ericsson

and Kintsch’s (1995) conceptualization of long-term working memory, according to which working

memory is an attentional mechanism that selectively activates long-term memory content (see Chapter 1;

for discussion, see Cowan, 2008). Since nouns—such as the arguments in the current experiments—are

stored in the mental lexicon, they are represented in long-term memory. If working memory involves

the selective-attentional activation of long-term-memory content, and the ventral parietal cortex is

involved in selective attention across cognitive domains, the TP-region activation in our study may

ultimately result from selective-attentional activation of arguments’ long-term-memory representations.

Imaging work indeed implies the SMG in the information-structural attentional focusing of arguments

during sentence processing (Lœvenbruck, Baciu, Segebarth, & Abry, 2005) as well as the TP-region in

the mental lexicon (for review, see Binder, Desai, Graves, & Conant, 2009).

128
 8.4. The Dorsal Tract: Syntax, Working Memory, or Both?

8.4 The Dorsal Tract: Syntax, Working Memory, or Both?

As its third leitmotif , this thesis shelled out the role of the AF/SLF in working memory in- and outside

of sentence processing. While the neural independence of argument reordering and storage (Chapter 4)

contradicted their behavioral interaction (Chapter 3), the DTI analyses on a healthy sample (Chapter 4)

and a patient (Chapter 5) reconciled the results: As structural link between BA 44 and the TP region,

the AF/SLF is the functional mediator of reordering and storage, driving their behavioral interaction in

spite of functional-neuroanatomical independence. Furthermore, the DTI (Chapter 4) and patient data

(Chapter 5) suggest that the AF/SLF plays a dual role of linking storage in the TP region to reordering in

BA 44 for sentence processing and to rehearsal in the dorsal IFG for tasks outside of sentence processing.

This section will discuss three related questions: First, how can damage to the AF/SLF cause a selective

sentence-processing deficit? Second, how can the AF/SLF selectively subserve sentence processing, but

generally subserve working-memory? Third, is there an alternative to the dual-role hypothesis on the

role of the AF/SLF in the brain circuitry of both sentence processing and working memory?

First, disruption of the AF/SLF caused the patient’s selective sentence-processing deficit, which

may result from the AF/SLF being the functional mediator between reordering and storage by linking

BA 44 and the TP region (Chapter 5). If the tract played a reordering-only or storage-only role, the

patient would be impaired on either all object-first or all long-distance sentences, respectively. This

is corroborated by a comparison of the results of the patient study and the behavioral study (see

Figure 8.2): Contrary to the patient, long-distance object-first sentences benefit over their short-distance

counterparts in participants with an intact AF/SLF18 . While the control-group pattern on object-first

sentences in the patient study resembled the pattern in the behavioral study, the patient’s performance

on the long object-first sentences dropped below her own and the control group’s performance on

the short object-first sentences. In the patient study, the reduced difference between the short- and

long-distance subject-first sentences both in the control group and the patient can be explained by

the overall reduced working-memory load: The conjunct clause used in all current studies had been

removed from the sentences to decrease processing difficulty for the elderly participants.

18
Albeit not significant and constrained by the limited time window, the same pattern of results was obtained for the behavioral
results in the fMRI study (see Chapter 4).

129
8. General Discussion

A 1
B 4

2
2

D−prime
Rating

3
−2
Short
Long

Subject-first
Object-first 6 −4

Figure 8.2: Contrast between behavioral results; (A) results of the behavioral rating study; an in-
teraction between reordering and storage is present: Long-distance subject-first sentences exacerbate
processing over their short-distance counterparts, object-first sentences generally exacerbate processing
over subject-first sentences, and long-distance object-first sentences facilitate processing over their
short-distance counterparts; (B) results from the patient yes–no comprehension study; the patient
was selectively impaired in processing long-distance object-first sentences.

This leads to the second variation on the AF/SLF theme, the proposal that the tract subserves

both the storage–reordering and the storage–rehearsal circuitry. While some authors stress the for-

mer (Caramazza et al., 1981; Friederici, 2011) and others the latter (Baldo et al., 2008; Yamada et al.,

2007), the current patient data (Chapter 5) have reconciling implications. The patient was impaired

across working-memory-span measures, but not across working-memory-intensive long argument–verb

dependencies. Friedmann and Gvion’s (2003) conduction-aphasia study and Wilson et al.’s (2011)

primary-progressive-aphasia study find damage to the AF/SLF to impair sentence processing. Both

studies used English subject- and object-relative sentences—which jointly tax reordering and storage,

resembling the current findings. Friedmann and Gvion (2003) also report their subjects to exhibit general

working-memory deficits—in line with classical findings of an impaired storage–rehearsal circuitry (for

review, see Bernal & Ardila, 2009; Buchsbaum et al., 2011), which can result from AF/SLF-only damage

(Yamada et al., 2007; but see Geldmacher, Quigg, & Elias, 2007). While Wilson et al. (2011) do not report

working-memory deficits in primary progressive aphasics with AF/SLF damage, their sample overrep-

resented non-fluent- and underrepresented logopenic-variant patients, exaggerating sentence-processing

deficits and understating working-memory deficits: As opposed to non-fluent patients, logopenic pa-

tients suffer from TP-regional degeneration and according working-memory deficits (Galantucci et al.,

2011; Leyton et al., 2011; but see Rogalski et al., 2011).

130
 8.4. The Dorsal Tract: Syntax, Working Memory, or Both?

In addition to the clinical data, the FA correlation analyses on the AF/SLF in healthy participants

(Chapter 4) lend plausibility to the dual-role hypothesis: While both reordering and storage were found

to correlate with the FA at differential sites along the tract, storage correlations were stronger and

more extensive. Since the BOLD effect of storage in the TP region was found to correlate with digit

span, the BOLD–FA correlations may reflect storage common to working memory in- and outside of

the sentence-processing domain. Critically, the current findings from a healthy sample transcend the

structure-to-behavior mapping of previous studies and suggest a direct structure-to-function-to-behavior

mapping in healthy participants: While there is a body of fiber-tracking evidence on the presence of

the AF/SLF in humans, these studies infer the tract’s sentence-processing role from the functional

involvement of its termination points in BA 44 and/or the TP region (Catani et al., 2005; Friederici,

Bahlmann, et al., 2006; Glasser & Rilling, 2008; Parker et al., 2005; Saur et al., 2010; Weiller et al., 2009).

Without assessment of the underlying diffusion characteristics, this does only allow for an indirect

conclusion (cf. Friederici, 2009a; Heiervang et al., 2006; Wakana et al., 2007). To my knowledge, there are

few exceptions that used an FA regression method to assign functional roles in sentence processing and

working memory to the AF/SLF in healthy adults: While not easily mapped onto the reordering–storage

dichotomy, the relevance of the AF/SLF in sentence processing is in line with Flöel, De Vries, Scholz,

Breitenstein, and Johansen-Berg (2009) and Antonenko, Meinzer, Lindenberg, Witte, and Flöel (2012),

who find the FA of dorsal fiber tracts starting in the IFG to predict artificial-grammar-learning success.

A second role of the AF/SLF in mediating rehearsal and storage is suggested by Charlton et al. (2010),

who find mean diffusivity and FA of a tract that connects BA 40 to inferior frontal areas to correlate

with working-memory-span measures.

Still, this section’s final paragraph provides an alternative to the proposed dual-role hypothesis

of the AF/SLF: Sub-divisions may distinctly subserve sentence processing and working memory (for

review, see Gierhan, in revision). In the current DTI study (Chapter 4), fiber tracking from the extensive

IFG and TP-region seeding points may have erroneously lumped together AF and SLF, and the patient’s

extensive lesion (Chapter 5) may have affected both AF and SLF. Dorsal and ventral sub-divisions of

the dorsal tract were first described in the rhesus monkey, where projections from superior and medial

parietal cortex to the dorsal area 6 were separated from projections from inferior parietal cortex to

131
8. General Discussion

the ventral areas 6 and 4619 (Petrides & Pandya, 1984). More recent work separated AF and SLF in

humans too, but their respective frontal and posterior termination points are still unclear: On the one

hand, Frey et al. (2008) interpret their human DTI results from the monkey perspective, proposing

that BA 6 and BA 44’s pars opercularis are the AF’s and SLF’s respective frontal termination points,

connecting to the inferior parietal lobe and the posterior STG, respectively. While less precise on the

frontal termination points, Catani et al. (2005) suggest that the AF connects inferior frontal regions

directly to the posterior STG and SMG, while the SLF runs from the inferior frontal lobe indirectly

through the inferior frontal lobe into posterior STG and SMG.

As discussed by Catani et al. (2005), the distinction of AF and SLF may explain the distinction of

conduction aphasia and transcortical motor aphasia: Deep TP-region lesions may result in AF damage,

conduction aphasia and predominant working-memory deficits; but superficial TP-region lesions may

instead result in SLF damage, transcortical motor aphasia and predominant higher-level language deficits

(McCarthy & Warrington, 1984). While the former is in line with the classical view of conduction

aphasia (Wernicke, 1874; for discussion, see Buchsbaum et al., 2011), the latter is in line with Dogil,

Haider, Schaner-Wolles, and Husmann’s (1995) report on a case of transcortical sensory aphasia who

showed impaired repetition, but spared syntactic comprehension. Furthermore, Galantucci et al.’s

(2011) study found non-fluent (i.e., sentence-processing impaired) primary progressive aphasics to suffer

from SLF degeneration, but logopenic (i.e., working-memory impaired) primary progressive aphasics

to show degeneration of the posterior segment of the indirect AF.

Given the proposal that rehearsal activates the dorsal IFG (Awh et al., 1996; Paulesu et al., 1993;

Petrides et al., 1993) and reordering activates the ventral IFG (Ben-Shachar et al., 2003; Friederici,

Fiebach, et al., 2006; Kim et al., 2009; Meyer, Obleser, Anwander, & Friederici, 2012), the data allow for

the following speculation: The AF underlies verbal working memory, and the SLF underlies sentence

processing proper. Due to methodological limitations, the current data cannot definitely decide in favor

of this proposal20 .

19
Rhesus-monkey area 46 lies anterior to and below area 6; recent cytoarchitectonic analyses suggest area 46 as a potential
homologue of the human BA 44 (cf. Frey, Campbell, Pike, & Petrides, 2008; for review, see Petrides et al., 2012; Thiebaut de
Schotten, Dell’Acqua, Valabregue, & Catani, 2012).
20
The interested reader is referred to the related proposal that the white matter of the language network emerged from a
memory connectome during evolution (Aboitiz, Aboitiz, & García, 2010; Aboitiz & García, 2009; Aboitiz et al., 2006).

132
9
T O WA R D S A T E S T A B L E F R A M E W O R K

Before proposing a testable neurocognitive framework of argument–verb-dependency processing, I will

recapitulate its underlying psycholinguistic and cognitive conceptualizations. In a sentence, arguments

symbolize the who and whom, while a verb symbolizes an action jointly involving the two. To infer

sentence meaning, arguments and their verb must be linked. Argument–verb linking takes place in the

vicinity of subcategorizing verbs, in an order idiosyncratic to any particular language. The number and

type of required arguments become known with the verb, and speaker’s linguistic knowledge determines

their desired idiosyncratic order. Thus, argument-order deviance and argument–verb non-adjacency

each increase cognitive efforts: Due to increasing working-memory decay, non-adjacency of arguments

and their subcategorizer increases argument-storage demands across the argument–verb distance; like-

wise, retrieval demands increase. Second, the deviance between the encountered argument order and the

language-specific idiosyncratic order requires an executive mechanism of argument reordering to match

the stored and desired argument orders.

To inform the framework, the following conclusions were drawn from this thesis’ three leitmotifs:

The TP region is a core functional-neuroanatomical substrate of working-memory storage of arguments:

Increased BOLD responses to sentences that tax storage established the TP region’s role, and EEG source

localization showed this region to increase alpha oscillations during storage in response to the same

sentences, clarifying its spatiotemporal dynamics. Since both BOLD and alpha effect were correlated

with working-memory span measures, both appear to be genuine working-memory effects, potentially

reflecting storage common to sentence processing and the verbal domain outside of sentence processing.

On encounter of the stored argument’s subcategorizing verb, combined fMRI–EEG analyses suggest

135
9. Towards a Testable Framework

that the TP region underlies argument retrieval, visible both in a scalp-level P300 and early positive

TP-region dipole activity, potentially echoed in a beta burst in the EEG-TFR analyses. Following

retrieval, IFG-dipole activity and a scalp-level P600 were linked to argument reordering. Because an IFG

BOLD response had been observed previously, this clarified the spatiotemporal dynamics of the IFG’s

crucial role in argument reordering. The AF/SLF mediates the functional-neuroanatomical task-share

between the independently-operating IFG and the TP region. More generally speaking, the IFG’s

reordering process exhibits top-down executive control, dynamically querying argument representations

stored and retrieved by the TP region. Figure 9.1 links the linguistic description, cognitive processes,

and functional neuroanatomy of argument–verb dependencies.

A = Speech sound

A B = Linguistic description
C = Cognitive processes
D = Functional neuroanatomy

B
Argument: Verb:
syntactic subcategorization
features information

C Encoding Storage
Argument–verb
matching
Retrieval
Order
matching
Reordering
Argument–verb
linking

D
TP-region BOLD; beta burst; IFG BOLD;
parietal alpha TP-region dipole activity; P300 IFG dipole activity; P600

AF/SLF
TP
IFG

IFG = Inferior frontal gyrus

TP = Temporo-parietal region
AF/SLF = Arcuate-/superior-longitudinal fasciculus

Figure 9.1: Functional neuroanatomy of storage, retrieval, and reordering; (A) spectrogram of an
example long-distance object-first sentence; (B) linguistic description of the relevant information; (C)
serial representation of cognitive processes involved in storage, retrieval, and reordering; (D) relation of
these cognitive processes to the subserving spatiotemporal brain dynamics.

136
 9.1. Reordering: Beyond Sentence Processing

From a cognitive-neuropsychologist point of view, this framework has four critical advantages:

First, the framework closely relates the linguistic, cognitive, and functional-neuroanatomical levels,

relating concrete aspects of the linguistic description of speech to concrete aspects of brain structure.

Second, the framework matches conceptual and empirical granularity: While state-of-the-art neurosci-

entific analysis methods exhibit increasing temporal and spatial resolution, the granularity of most

psycholinguistic sentence-processing theories is beyond this resolution. Third, the framework is ter-

minologically open: The current results suggest that some of the atomic cognitive processes involved

in sentence processing share characteristics of working-memory processes; thus, neuroscientists from

outside the sentence-processing domain can directly relate to the framework. Fourth, the fact that some

of the brain phenomena measured in the current thesis appear across a multitude of cognitive processes

suggests that it is possible to re-inform the psycholinguist’s conceptual inventory: If a brain region

or function is involved in a variety of tasks from various domains, these tasks may share an atomic

cognitive process. Eventually, this approach may help in replacing abstract, descriptive linguistic notions

with concrete, empirically sound cognitive concepts. The proposed framework generates hypotheses

for future research, which I will address next.

9.1 Reordering: Beyond Sentence Processing

While it has independently been suggested that BA 44 be involved in argument reordering, action se-

quencing, and gesture comprehension in the deaf, none of the possible pairings have been systematically

compared within participants. Hence, there is no positive evidence that BA 44 plays an overarching

reordering role. Inspired by Gerton et al.’s (2004) and Makuuchi, Bahlmann, and Friederici’s (2012) re-

sults, initial behavioral evidence was presented for structural priming from the processing of object-first

argument orders to the processing of the backward-digit-span task (see Section 8.2). The experimental

paradigm used may be suitable for an fMRI study: Judging from the behavioral results as well as the

imaging literature that found across-tasks repetition priming to reduce BA-44 activity (Wagner et al.,

2000), structural priming should yield decreasing BA-44 activation for the interaction contrast between

argument and digit reordering. Such an effect would speak in favor of a general role in reordering across

verbal-working-memory tasks and would form the basis for within-participants studies crossing gesture

and action reordering with argument reordering.

137
9. Towards a Testable Framework

9.2 The Inferior Parietal Cortex: Beyond Working Memory

Second, it was speculated that the working-memory role of the TP region may mirror an

attention-related mechanism (Section 8.3): Argument storage, information-structural argument

focusing, and the attentional activation of arguments’ long-term-memory representations may open a

window into how general attentional mechanisms underlie specific sentence-processing mechanisms.

From a linguist’s viewpoint, this falls into the realm of information structure (for an overview, see

Krifka, 2008). From a cognitive neuropsychologist’s viewpoint, it could link the sentence-processing,

working-memory, and attention domains. A suitable fMRI paradigm would manipulate argument–verb

distance as working-memory factor, while an attention factor would aim at directing attention towards

a stored argument. To this end, previous studies have selectively cued stored items for later reuse in a

delayed-response task (e.g., Lepsien, Thornton, & Nobre, 2011). For the spatial and object domains,

fMRI studies have found shifts of selective attention amongst stored items to modulate parietal brain

activation during working-memory storage (for review, see Gazzaley & Nobre, 2011). A possibly

abridging to the sentence-processing domain is sketched in Figure 9.2.

Stimulus
In der Kontroverse hat der Demokrat den Minister wegen des Parteitags ausgelacht.

Subject cue Object cue

SUBJECT or OBJECT

Subject probe Object probe

Er hatte das Versagen geahnt. or Ihn hatte der Kanzler entlassen.

Figure 9.2: Paradigm proposal for modulating working-memory-related TP-region activation by atten-
tion; upper panel shows a stimulus sentence with the subject in blue and the object in red; middle panel
shows subject and object cues to selectively direct attention amongst the two stored arguments; lower
panel shows probe sentence, on which a delayed-response task is to be carried out.

In the proposed paradigm, both subject and object of a transitive verb must be stored in working

memory, but only either subject or object is cued for the processing of the delayed-response task.

The hypothesis for such an experiment would thus be that participants respond faster when the cue

and the sentence-initial pronoun in the probe sentence match. The match is expected to modulate

argument-storage-related TP-region BOLD during the delay interval.

138
 9.3. Disentangling Reordering and Rehearsal

9.3 Disentangling Reordering and Rehearsal

The dissociation of the reordering–storage and rehearsal–storage dichotomies (see Section 8.2) footed

on the previous imaging literature, as well as negative experimental evidence: reordering-related BA-44

activation was not correlated with working-memory span, working-memory-related alpha generators

did not include prefrontal sources, and the patient did not show a generalized deficit across long

argument–verb dependencies. To provide positive evidence on a left-lateral-prefrontal dissociation into

BA 44 subserving reordering and BA 6 subserving rehearsal, an experimental paradigm would have

to fully cross the two. Storage-only, rehearsal-only, and rehearsal-plus-reordering conditions would be

included, adapting a classical paradigm for the sentence-processing domain: Paulesu et al. (1993) disentan-

gled storage and rehearsal by having on-screen letters being followed by a probe letter that either occurred

amongst (storage condition) or rhymed with one of (rehearsal condition) the previously-presented dig-

its. An orthogonal reordering dimension would be introduced by using short-distance subject- and

object-first sentences (see Figure 9.3).

Stimulus
In der Kontroverse wegen des Parteitags hat der Demokrat den Minister ausgelacht.
In der Kontroverse wegen des Parteitags hat den Minister der Demokrat ausgelacht.
Subject-first Object-first
… der Demokrat den Minister … … der Kandidat den Magister …
”Yes“
… den Minister der Demokrat … … den Magister der Kandidat …
Rhyme
Match

… der Sozialist den Sekretär … … der Sozialist den Sekretär …

”No“ … den Sekretär der Sozialist …
… den Sekretär der Sozialist …

Figure 9.3: Paradigm proposal for dissociating reordering and rehearsal; upper panel shows subject-first
(in blue) and object-first (in red) stimulus sentences; bottom left panel shows probe items for the match
condition, bottom right panel shows probe items for the rhyme condition.

During the task interval, object-first sentences would conjointly tax reordering and activate

BA 44, but rhyming tasks would conjointly tax rehearsal and activate BA 6; the matching task serves

as a high-level baseline. The hypothesized prefrontal dissociation would be an excellent basis for

disentangling the functional roles of the proposed sub-components of the dorsal fiber tract, with

reordering relying on the SLF, but rehearsal relying on the AF.

139
9. Towards a Testable Framework

9.4 Brain Oscillations and Fronto-Temporal Connectivity

One result of the current thesis is that the working memory of argument–verb dependencies involves

gray-matter structures in the IFG and TP region, their white-matter connection through the AF/SLF,

as well as alpha oscillations. Sarnthein et al. (1998; cf. Buzsáki, 2006) have underlined that remote

brain regions’ joint processing necessitates the oscillatory coupling of either their phases or amplitudes,

either within or across frequency bands (cf. de Lange, Jensen, Bauer, & Toni, 2008). Thus, one must

hypothesize that argument–verb-dependency processing involves oscillatory coupling of the IFG and

TP region, and that the AF/SLF white matter is functionally determined to mediate coupling.

The underlying rationale of the current paradigm shows that argument reordering and storage

are an excellent test for such hypotheses; a similar setup may suffice to show that executive processes

(reordering) and representational processes (storage) of sentence processing interact by coupling their

fronto-parietal network. Source-space analyses of MEG data from the IFG and TP region, combined

with high-resolution fiber tracking, may be suited to tackle such an experiment: The accurate estimation

of oscillatory coupling in source space (for recent demonstrations, see de Lange et al., 2008; Keil, Müller,

Ihssen, & Weisz, 2012) and the advanced fiber-tracking methods now becoming available (Descoteaux et

al., 2009) may be able to link the IFG–TP-region-coupling parameters to white-matter microstructural

parameters to close the gap between indirect structural (i.e., DTI) and functional (i.e., BOLD) measures.

140
REFERENCES

Aboitiz, F., Aboitiz, S., & García, R. (2010). The Phonological Loop. A Key Innovation in Human

Evolution. Current Anthropology, 51(Supplement 1), 55–65.

Aboitiz, F., & García, R. (2009). Merging of phonological and gestural circuits in early language

evolution. Reviews in the Neurosciences, 20(1), 71–84.

Aboitiz, F., García, R., Bosman, C., & Brunetti, E. (2006). Cortical memory mechanisms and language

origins. Brain and Language, 98(1), 40–56.

Aguirre, G. K., Zarahn, E., & D’Esposito, M. (1998). The variability of human, BOLD hemodynamic

responses. NeuroImage, 8(4), 360–369.

Allen, J. (1977). Short term spectral analysis, synthesis, and modification by discrete Fourier transform.

IEEE Transactions on Acoustics, Speech and Signal Processing, 25(3), 235–238.

Allen, M., Badecker, W., & Osterhout, L. (2003). Morphological analysis in sentence processing: An

ERP study. Language and Cognitive Processes, 18(4), 405–430.

Allison, T., Wood, C. C., & McCarthy, G. (1986). The central nervous system. In M. G. H. Coles,

E. Donchin, & S. W. Porges (Eds.), Psychophysiology: systems, processes, and applications (pp. 5–25).

New York: Guilford.

Altmann, E., & Schunn, C. (2002). Integrating decay and interference: a new look at an old interaction.

In Proceedings of the 24th annual conference of the Cognitive Science Society (pp. 65–70).

Ambarzumian, V. (1929). Über eine Frage der Eigenwerttheorie. Zeitschrift für Physik, 53(9), 690–695.

Amunts, K., Schleicher, A., Bürgel, U., Mohlberg, H., Uylings, H., & Zilles, K. (1999). Broca’s region

revisited: cytoarchitecture and intersubject variability. The Journal of Comparative Neurology,

412(2), 319–341.

143
References

Amunts, K., & Zilles, K. (2012). Architecture and organizational principles of Broca’s region. Trends in

Cognitive Sciences, 16(8), 418–426.

Anders Ericsson, K., & Kintsch, W. (1995). Long-term working memory. Psychological Review, 102(2),

211–245.

Anderson, J., Reder, L., & Lebiere, C. (1996). Working Memory: Activation Limitations on Retrieval.

Cognitive Psychology, 30(3), 221–256.

Antonenko, D., Meinzer, M., Lindenberg, R., Witte, A., & Flöel, A. (2012). Grammar learning in older

adults is linked to white matter microstructure and functional connectivity. NeuroImage, 62(3),

1667–1674.

Anwander, A., Tittgemeyer, M., von Cramon, D., Friederici, A. D., & Knösche, T. (2007). Connectivity-

based parcellation of Broca’s area. Cerebral Cortex, 17(4), 816–825.

Aoshima, S., Phillips, C., & Weinberg, A. (2004). Processing filler-gap dependencies in a head-final

language. Journal of Memory and Language, 51(1), 23–54.

Ashburner, J. (2007). A fast diffeomorphic image registration algorithm. NeuroImage, 38(1), 95–113.

Ashburner, J. (2009). Computational anatomy with the SPM software. Magnetic Resonance Imaging,

27(8), 1163–1174.

Ashburner, J., & Friston, K. (2001). Why voxel-based morphometry should be used. NeuroImage,

14(6), 1238–1243.

Ashburner, J., & Friston, K. J. (2005). Unified segmentation. NeuroImage, 26(3), 839–851.

Awh, E., Jonides, J., Smith, E., Schumacher, E., Koeppe, R., & Katz, S. (1996). Dissociation of storage

and rehearsal in working memory: PET evidence. Psychological Science, 7(1), 25–31.

Awh, E., Smith, E., & Jonides, J. (1995). Human rehearsal processes and the frontal lobes: PET evidence.

Annals of the New York Academy of Sciences, 769(1), 97–118.

Baayen, H. (2008). Analyzing linguistic data: a practical introduction to statistics using R. Cambridge:

Cambridge University Press.

Baayen, H., Gulikers, L., & Piepenbrock, R. (1995). The CELEX lexical database. Philadelphia:

Linguistic Data Consortium.

Babyonyshev, M., & Gibson, E. (1999). The complexity of nested structures in Japanese. Language,

75(3), 423–450.

144
 References

Baddeley, A. (1996). Exploring the central executive. Quarterly Journal of Experimental Psychology,

49(1), 5–28.

Baddeley, A. (2012). Working memory: theories, models, and controversies. Annual Review of

Psychology, 63, 1–29.

Baddeley, A., Eysenck, M., & Anderson, M. (2009). Memory. Hove: Psychology Press.

Baddeley, A., & Hitch, G. (1974). Working memory. In G. Bower (Ed.), The psychology of learning and

motivation (pp. 47–89). New York: Academic Press.

Baddeley, A., & Larsen, J. (2007). The phonological loop unmasked? A comment on the evidence for a

perceptual-gestural alternative. Quarterly Journal of Experimental Psychology, 60(4), 497–504.

Bader, M., & Bayer, J. (2006). Case and Linking in Language Comprehension: Evidence from German.

Dordrecht: Kluwer.

Bader, M., & Lasser, I. (1994). German verb-final clauses and sentence processing: Evidence for

immediate attachment. In C. Clifton, L. Frazier, & K. Rayner (Eds.), Perspectives on Sentence

Processing (pp. 225–242). Hillsdale: Lawrence Erlbaum Associates.

Baldo, J., Klostermann, E., & Dronkers, N. (2008). It’s either a cook or a baker: Patients with conduction

aphasia get the gist but lose the trace. Brain and Language, 105(2), 134–140.

Başar, E. (1998). Brain function and oscillations. Brain oscillations. Principles and approaches. Berlin:

Springer.

Başar, E., Demiralp, T., Schürmann, M., Başar-Eroglu, C., & Ademoglu, A. (1999). Oscillatory brain

dynamics, wavelet analysis, and cognition. Brain and Language, 66(1), 146–183.

Basser, P., Mattiello, J., & Le Bihan, D. (1994). Estimation of the effective self-diffusion tensor from the

NMR spin echo. Journal of Magnetic Resonance Series B, 103(3), 247–247.

Basser, P., & Pierpaoli, C. (1996). Microstructural and physiological features of tissues elucidated by

quantitative-diffusion-tensor MRI. Journal of Magnetic Resonance Series B, 111(3), 209–219.

Bastiaanse, R., Edwards, S., Mass, E., & Rispens, J. (2003). Assessing comprehension and production of

verbs and sentences: The Verb and Sentence Test (VAST). Aphasiology, 17(1), 49–73.

Bastiaanse, R., & van Zonneveld, R. (2006). Comprehension of passives in Broca’s aphasia. Brain and

Language, 96(2), 135–142.

145
References

Bastiaansen, M., & Hagoort, P. (2006). Oscillatory neuronal dynamics during language comprehension.

In C. Neuper & W. Klimesch (Eds.), Progress in Brain Research. Event-Related Dynamics of Brain

Oscillations (pp. 179–196). Amsterdam: Elsevier.

Bastiaansen, M., Magyari, L., & Hagoort, P. (2010). Syntactic unification operations are reflected in

oscillatory dynamics during on-line sentence comprehension. Journal of Cognitive Neuroscience,

22(7), 1333–1347.

Becker, J. T., Mintun, M. A., Aleva, K., Wiseman, M. B., Nichols, T., & DeKosky, S. T. (1996).

Compensatory reallocation of brain resources supporting verbal episodic memory in Alzheimer’s

disease. Neurology, 46(3), 692–700.

Behaghel, O. (1923). Deutsche Syntax: Eine geschichtliche Darstellung. Heidelberg: Winter.

Bell, A. J., & Sejnowski, T. J. (1995). An information-maximization approach to blind separation and

blind deconvolution. Neural Computation, 7(6), 1129–1159.

Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: a practical and powerful

approach to multiple testing. Journal of the Royal Statistical Society, Series B, 57(1), 289–300.

Ben-Shachar, M., Hendler, T., Kahn, I., Ben-Bashat, D., & Grodzinsky, Y. (2003). The neural reality of

syntactic transformations. Psychological Science, 14(5), 433–440.

Ben-Shachar, M., Palti, D., & Grodzinsky, Y. (2004). Neural correlates of syntactic movement: converg-

ing evidence from two fMRI experiments. NeuroImage, 21(4), 1320–1336.

Berg, P., & Scherg, M. (1994). A fast method for forward computation of multiple-shell spherical head

models. Electroencephalography and Clinical Neurophysiology, 90(1), 58–64.

Berger, H. (1929). Über das Elektrenkephalogramm des Menschen. Archiv für Psychiatrie und

Nervenkrankheiten, 87, 527–570.

Bernal, B., & Ardila, A. (2009). The role of the arcuate fasciculus in conduction aphasia. Brain, 132(9),

2309–2316.

Berndt, R. S., & Caramazza, A. (1999). How "regular" is sentence comprehension in Broca’s aphasia? It

depends on how you select the patients. Brain and Language, 67(3), 242–247.

Besl, P., & McKay, N. (1992). A method for registration of 3-D shapes. IEEE Transactions on Pattern

Analysis and Machine Intelligence, 14(2), 239–256.

146
 References

Bever, T., & McElree, B. (1988). Empty categories access their antecedents during comprehension.

Linguistic Inquiry, 19(1), 35–43.

Biemann, C., Bordag, S., Quasthoff, U., & Wolff, C. (2004). Web services for language resources

and language technology applications. In Proceedings of the Fourth International Conference on

Language Resources and Evaluation.

Binder, J., Desai, R., Graves, W., & Conant, L. (2009). Where is the semantic system? A critical review

and meta-analysis of 120 functional neuroimaging studies. Cerebral Cortex, 19(12), 2767–2796.

Binder, K., Duffy, S., & Rayner, K. (2001). The effects of thematic fit and discourse context on syntactic

ambiguity resolution. Journal of Memory and Language, 44(2), 297–324.

Birbaumer, N., Elbert, T., Canavan, A., & Rockstroh, B. (1990). Slow potentials of the cerebral cortex

and behavior. Physiological Reviews, 70(1), 1–41.

Blume, W., Kaibara, M., & Young, G. (1995). Atlas of adult electroencephalography. New York: Raven

Press.

Bock, J. (1986). Syntactic persistence in language production. Cognitive psychology, 18(3), 355–387.

Boersma, P., & Weenink, D. (2001). Praat, a system for doing phonetics by computer. Glot International,

5(9/10), 341–345.

Boland, J. (1993). The role of verb argument structure in sentence processing: Distinguishing between

syntactic and semantic effects. Journal of Psycholinguistic Research, 22(2), 133–152.

Boland, J., Tanenhaus, M., Garnsey, S., & Carlson, G. (1995). Verb argument structure in parsing and

interpretation: evidence from wh-questions. Journal of Memory and Language, 34(6), 774–806.

Bookheimer, S. (2002). Functional MRI of language: new approaches to understanding the cortical

organization of semantic processing. Annual Review of Neuroscience, 25, 151–188.

Bookstein, F. (2001). Voxel-based morphometry should not be used with imperfectly registered images.

NeuroImage, 14(6), 1454–1462.

Bornkessel, I., Zysset, S., Friederici, A. D., von Cramon, D., & Schlesewsky, M. (2005). Who did what

to whom? The neural basis of argument hierarchies during language comprehension. NeuroImage,

26(1), 221–233.

Bornkessel-Schlesewsky, I., Schlesewsky, M., & von Cramon, D. (2009). Word order and Broca’s region:

Evidence for a supra-syntactic perspective. Brain and Language, 111(3), 125–139.

147
References

Branigan, H. (2007). Syntactic priming. Language and Linguistics Compass, 1(1–2), 1–16.

Braze, D., Mencl, W. E., Tabor, W., Pugh, K. R., Todd Constable, R., Fulbright, R. K., et al. (2011).

Unification of sentence processing via ear and eye: An fMRI study. Cortex, 47(4), 416–431.

Bremer, F. (1958). Cerebral and cerebellar potentials. Physiological Reviews, 38(3), 357–388.

Brett, M., Anton, J., Valabregue, R., & Poline, J. (2002). Region of interest analysis using an SPM

toolbox. NeuroImage, 16(2), 1140–1141.

Brown, J. (1958). Some tests of the decay theory of immediate memory. Quarterly Journal of

Experimental Psychology, 10(1), 12–21.

Brown, R. (1827). A brief account of microscopical observations and on the general existence of active

molecules in organic and inorganic bodies. Philosophical Magazine, 4, 161–173.

Buchsbaum, B. R., Baldo, J., Okada, K., Berman, K. F., Dronkers, N., D’Esposito, M., et al. (2011).

Conduction aphasia, sensory-motor integration, and phonological short-term memory – an

aggregate analysis of lesion and fMRI data. Brain and Language, 119(3), 119–128.

Buchsbaum, B. R., & D’Esposito, M. (2008). The search for the phonological store: from loop to

convolution. Journal of Cognitive Neuroscience, 20(5), 762–778.

Buchsbaum, B. R., Hickok, G., & Humphries, C. (2001). Role of left posterior superior temporal

gyrus in phonological processing for speech perception and production. Cognitive Science, 25(5),

663–678.

Bühler, K. (1934). Sprachtheorie. Die Darstellungsfunktion der Sprache. Jena: Gustav Fischer.

Bushara, K., Weeks, R., Ishii, K., Catalan, M., Tian, B., Rauschecker, J., et al. (1999). Modality-

specific frontal and parietal areas for auditory and visual spatial localization in humans. Nature

Neuroscience, 2(8), 759–766.

Buxton, R., Wong, E., & Frank, L. (1998). Dynamics of blood flow and oxygenation changes during

brain activation: the balloon model. Magnetic Resonance in Medicine, 39(6), 855–864.

Buzsáki, G. (2006). Rhythms of the Brain. New York: Oxford University Press.

Cabeza, R., Ciaramelli, E., & Moscovitch, M. (2012a). Cognitive contributions of the ventral parietal

cortex: an integrative theoretical account. Trends in Cognitive Sciences, 16(6), 338–352.

148
 References

Cabeza, R., Ciaramelli, E., & Moscovitch, M. (2012b). Response to Nelson et al.: ventral parietal

subdivisions are not incompatible with an overarching function. Trends in Cognitive Sciences,

16(8), 400–401.

Campoy, G. (2012). Evidence for decay in verbal short-term memory: a commentary on Berman,

Jonides, and Lewis (2009). Journal of Experimental Psychology: Learning, Memory, and Cognition,

38(4), 1129–1136.

Cantalupo, C., & Hopkins, W. (2001). Asymmetric Broca’s area in great apes: a region of the ape brain

is uncannily similar to one linked with speech in humans. Nature, 414(6863), 505.

Cao, Y., Vikingstad, E., George, K., Johnson, A., & Welch, K. (1999). Cortical language activation in

stroke patients recovering from aphasia with functional MRI. Stroke, 30(11), 2331–2340.

Caplan, D. (2009). Experimental design and interpretation of functional neuroimaging studies of

cognitive processes. Human Brain Mapping, 30(1), 59–77.

Caplan, D., Alpert, N., Waters, G., & Olivieri, A. (2000). Activation of Broca’s area by syntactic

processing under conditions of concurrent articulation. Human Brain Mapping, 9(2), 65–71.

Caplan, D., Chen, E., & Waters, G. (2008). Task-dependent and task-independent neurovascular

responses to syntactic processing. Cortex, 44(3), 257–275.

Caplan, D., & Waters, G. (1999). Verbal working memory and sentence comprehension. Behavioral

and Brain Sciences, 22(1), 77–94.

Caramazza, A., Basili, A. G., Koller, J. J., & Berndt, R. S. (1981). An investigation of repetition and

language processing in a case of conduction aphasia. Brain and Language, 14(2), 235–271.

Carpenter, P., & Just, M. (1989). The role of working memory in language comprehension. In D. Klahr

& K. Kotovsky (Eds.), Complex information processing: The impact of Herbert A. Simon (pp.

31–68). Hillsdale: Lawrence Erlbaum Associates.

Caspari, I., Parkinson, S., LaPointe, L., & Katz, R. (1998). Working Memory and Aphasia. Brain and

Cognition, 37(2), 205–223.

Catani, M., Jones, D., & ffytche, D. (2005). Perisylvian language networks of the human brain. Annals

of Neurology, 57(1), 8–16.

Charlton, R., Barrick, T., Lawes, I., Markus, H., & Morris, R. (2010). White matter pathways associated

with working memory in normal aging. Cortex, 46(4), 474–489.

149
References

Chomsky, N. (1955). The logical structure of linguistic theory. New York: Plenum press.

Chomsky, N. (1965). Aspects of the theory of syntax. Cambridge: MIT Press.

Chomsky, N. (1981). Lectures on government and binding. Dordrecht: Foris.

Clahsen, H., & Featherston, S. (1999). Antecedent priming at trace positions: evidence from german

scrambling. Journal of Psycholinguistic Research, 28(4), 415–437.

Clark, C., Egan, G., McFarlane, A., Morris, P., Weber, D., Sonkkilla, C., et al. (2000). Updating working

memory for words: a PET activation study. Human Brain Mapping, 9(1), 42–54.

Clerget, E., Badets, A., Duqué, J., & Olivier, E. (2011). Role of Broca’s area in motor sequence

programming: a cTBS study. NeuroReport, 22(18), 965–969.

Clifton, C., & Frazier, L. (1988). Comprehending sentences with long-distance dependencies. In

G. Carlson & M. K. Tanenhaus (Eds.), Linguistic structure in language processing (pp. 273–317).

Dordrecht: Reidel Press.

Clifton, C., et al. (1984). Lexical expectations in sentence comprehension. Journal of Verbal Learning

and Verbal Behavior, 23(6), 696–708.

Comrie, B. (1993). Argument structure. In J. Jacobs, A. von Stechow, W. Sternefeld, & T. Vennemann

(Eds.), Syntax. Ein internationales Handbuch zeitgenössischer Forschung (pp. 905–914). Berlin: de

Gruyter.

Conrad, R. (1967). Interference or decay over short retention intervals? Journal of Verbal Learning and

Verbal Behavior, 6(1), 49–54.

Conrad, R., & Hull, A. (1964). Information, acoustic confusion and memory span. British Journal of

Psychology, 55(4), 429–432.

Constable, R., Pugh, K., Berroya, E., Mencl, W., Westerveld, M., Ni, W., et al. (2004). Sentence

complexity and input modality effects in sentence comprehension: an fMRI study. NeuroImage,

22(1), 11–21.

Corballis, M. C. (2003). From mouth to hand: gesture, speech, and the evolution of right-handedness.

Behavioral and Brain Sciences, 26(2), 199–208.

Corballis, M. C., Badzakova-Trajkov, G., & Häberling, I. S. (2011). Right hand, left brain: genetic and

evolutionary bases of cerebral asymmetries for language and manual action. Cognitive Science,

3(1), 1–17.

150
 References

Corbett, A., & Chang, F. (1983). Pronoun disambiguation: Accessing potential antecedents. Memory &

Cognition, 11(3), 283–294.

Cowan, N. (2008). What are the differences between long-term, short-term, and working memory? In

W. Sossin, J.-C. Lacaille, V. Castellucci, & S. Belleville (Eds.), Progress in Brain Research. Essence of

Memory (pp. 323–338). Amsterdam: Elsevier.

Cowper, E. (1976). Constraints on sentence complexity: a model for syntactic processing. Unpublished

doctoral dissertation, Brown University.

Cunitz, K. (2011). Die funktionelle Rolle des Fasciculus arcuatus. Zwei Einzelfallstudien mit Patienten.

Unpublished diploma thesis, Universität Bremen.

Daneman, M., & Carpenter, P. (1980). Individual differences in working memory and reading. Journal

of Verbal Learning and Verbal Behavior, 19(4), 450–466.

Daunizeau, J., Laufs, H., & Friston, K. J. (2010). EEG–fMRI. In C. Mulert & L. Lemieux (Eds.),

EEG–fMRI Information Fusion: Biophysics and Data Analysis (pp. 511–526). Heidelberg: Springer.

de Groot, A. (1978). Thought and choice in chess. Den Haag: de Gruyter.

de Lange, F., Jensen, O., Bauer, M., & Toni, I. (2008). Interactions between posterior gamma and frontal

alpha/beta oscillations during imagined actions. Frontiers in Human Neuroscience, 2, 7.

de Munck, J., Van Dijk, B., & Spekreijse, H. (1988). Mathematical dipoles are adequate to describe

realistic generators of human brain activity. IEEE Transactions on Biomedical Engineering, 35(11),

960–966.

Della-Maggiore, V., Chau, W., Peres-Neto, P., & McIntosh, A. (2002). An empirical comparison of SPM

preprocessing parameters to the analysis of fMRI data. NeuroImage, 17(1), 19–28.

Delorme, A., Sejnowski, T., & Makeig, S. (2007). Enhanced detection of artifacts in EEG data using

higher-order statistics and independent component analysis. NeuroImage, 34(4), 1443–1449.

Demberg, V., & Keller, F. (2008). Data from eye-tracking corpora as evidence for theories of syntactic

processing complexity. Cognition, 109(2), 193–210.

Descoteaux, M., Deriche, R., Knösche, T., & Anwander, A. (2009). Deterministic and probabilistic

tractography based on complex fibre orientation distributions. IEEE Transactions on Medical

Imaging, 28(2), 269–286.

151
References

D’Esposito, M., Postle, B. R., Ballard, D., & Lease, J. (1999). Maintenance versus manipulation of

information held in working memory: an event-related fMRI study. Brain and Cognition, 41(1),

66–86.

De Witt Hamer, P. C., Moritz-Gasser, S., Gatignol, P., & Duffau, H. (2011). Is the human left middle

longitudinal fascicle essential for language? A brain electrostimulation study. Human Brain

Mapping, 32(6), 962–973.

Dogil, G., Haider, H., Schaner-Wolles, C., & Husmann, R. (1995). Radical autonomy of syntax:

Evidence from transcortical sensory aphasia. Aphasiology, 9(6), 577–602.

Donchin, E., & Coles, M. (1988). Is the P300 component a manifestation of context updating? Behavioral

and Brain Sciences, 11(3), 357–427.

Donchin, E., & Heffley. (1978). Multivariate analysis of event-related potential data: a tutorial review. In

D. Otto (Ed.), Multidisciplinary perspectives in event-related brain potential research (pp. 555–572).

Washington: US Government Printing Office.

Du Bois, J. (1987). The discourse basis of ergativity. Language, 63(4), 805–855.

Duffau, H. (2008). The anatomo-functional connectivity of language revisited:: New insights provided

by electrostimulation and tractography. Neuropsychologia, 46(4), 927–934.

Duffau, H., Capelle, L., Sichez, N., Denvil, D., Lopes, M., Sichez, J.-P., et al. (2002). Intraoperative

mapping of the subcortical language pathways using direct stimulations. An anatomo-functional

study. Brain, 125(1), 199–214.

Duffau, H., Gatignol, P., Denvil, D., Lopes, M., & Capelle, L. (2003). The articulatory loop: study of

the subcortical connectivity by electrostimulation. NeuroReport, 14(15), 2005–2008.

Edgar, J., Stewart, J., & Miller, G. (2005). Digital filters in ERP research. In T. C. Handy (Ed.),

Event-related potentials: A handbook (pp. 85–113). Cambridge: MIT Press.

Eickhoff, S., Stephan, K., Mohlberg, H., Grefkes, C., Fink, G., Amunts, K., et al. (2005). A new

SPM toolbox for combining probabilistic cytoarchitectonic maps and functional imaging data.

NeuroImage, 25(4), 1325–1335.

Einstein, A. (1905). Über die von der molekularkinetischen Theorie der Wärme geforderte Bewegung

von in ruhenden Flüssigkeiten suspendierten Teilchen. Annalen der Physik, 322(8), 549–560.

152
 References

Ergen, M., Yildirim, E., Uslu, A., Gurvit, H., & Demiralp, T. (2012). P3 response during short-term

memory retrieval revisited by a spatio-temporal analysis. International Journal of Psychophysiology,

84(2), 205–210.

Ericsson, K., & Chase, W. (1982). Exceptional memory: Extraordinary feats of memory can be matched

or surpassed by people with average memories that have been improved by training. American

Scientist, 70(6), 607–615.

Featherston, S. (2001). Empty categories in sentence processing. Amsterdam: John Benjamins.

Fedorenko, E., Gibson, E., & Rohde, D. (2006). The nature of working memory capacity in sentence

comprehension: Evidence against domain-specific working memory resources. Journal of Memory

and Language, 54(4), 541–553.

Fedorenko, E., Gibson, E., & Rohde, D. (2007). The nature of working memory in linguistic, arithmetic

and spatial integration processes. Journal of Memory and Language, 56(2), 246–269.

Fell, J., Dietl, T., Grunwald, T., Kurthen, M., Klaver, P., Trautner, P., et al. (2004). Neural bases of

cognitive ERPs: more than phase reset. Journal of Cognitive Neuroscience, 16(9), 1595–1604.

Felser, C., Clahsen, H., & Münte, T. (2003). Storage and integration in the processing of filler-gap

dependencies: An ERP study of topicalization and wh-movement in German. Brain and Language,

87(3), 345–354.

Ferreira, F., & Henderson, J. (1993). Reading processes during syntactic analysis and reanalysis.

Canadian Journal of Experimental Psychology, 47(2), 247–247.

Fiebach, C., Friederici, A., Smith, E., & Swinney, D. (2007). Lateral inferotemporal cortex main-

tains conceptual-semantic representations in verbal working memory. Journal of Cognitive

Neuroscience, 19(12), 2035–2049.

Fiebach, C., Schlesewsky, M., & Friederici, A. D. (2001). Syntactic working memory and the estab-

lishment of filler-gap dependencies: Insights from ERPs and fMRI. Journal of Psycholinguistic

Research, 30(3), 321–338.

Fiebach, C., Schlesewsky, M., & Friederici, A. D. (2002). Separating syntactic memory costs and

syntactic integration costs during parsing: the processing of German WH-questions. Journal of

Memory and Language, 47(2), 250–272.

153
References

Fiebach, C., Schlesewsky, M., Lohmann, G., von Cramon, D., & Friederici, A. D. (2005). Revisiting

the role of Broca’s area in sentence processing: Syntactic integration versus syntactic working

memory. Human Brain Mapping, 24(2), 79–91.

Fillard, P., Pennec, X., Arsigny, V., & Ayache, N. (2007). Clinical DT-MRI estimation, smoothing,

and fiber tracking with log-Euclidean metrics. IEEE Transactions on Medical Imaging, 26(11),

1472–1482.

Flöel, A., De Vries, M., Scholz, J., Breitenstein, C., & Johansen-Berg, H. (2009). White matter integrity

in the vicinity of Broca’s area predicts grammar learning success. NeuroImage, 47(4), 1974–1981.

Fodor, J. (1978). Parsing strategies and constraints on transformations. Linguistic Inquiry, 9(3), 427–473.

Fodor, J., Garrett, M., & Bever, T. (1968). Some syntactic determinants of sentential complexity, II:

Verb structure. Perception & Psychophysics, 3(6), 453–461.

Foundas, A., Eure, K., Luevano, L., & Weinberger, D. (1998). MRI asymmetries of Broca’s area: the

pars triangularis and pars opercularis. Brain and Language, 64(3), 282–296.

Fourier, J. (1822). Théorie analytique de la Chaleur. Paris: Didot.

Frank, E. (1952). Electric potential produced by two point current sources in a homogeneous conducting

sphere. Journal of Applied Physics, 23(11), 1225–1228.

Frauenfelder, U., Segui, J., & Mehler, J. (1980). Monitoring around the relative clause. Journal of Verbal

Learning and Verbal Behavior, 19(3), 328–337.

Frazier, L. (1987). Sentence processing: a tutorial review. In M. Coltheart (Ed.), Attention and

Performance XII: The Psychology of Reading (pp. 559–586). London: Lawrence Erlbaum Associates.

Frazier, L., Clifton, C., & Randall, J. (1983). Filling gaps: Decision principles and structure in sentence

comprehension. Cognition, 13(2), 187–222.

Frege, G. (1879). Begriffsschrift: eine der arithmetischen nachgebildete Formelsprache des reinen Denkens.

Halle: Nebert.

Frey, S., Campbell, J. S. W., Pike, G. B., & Petrides, M. (2008). Dissociating the human language

pathways with high angular resolution diffusion fiber tractography. Journal of Neuroscience,

28(45), 11435–11444.

154
 References

Fridriksson, J., Kjartansson, O., Morgan, P. S., Hjaltason, H., Magnusdottir, S., Bonilha, L., et al.

(2010). Impaired speech repetition and left parietal lobe damage. Journal of Neuroscience, 30(33),

11057–11061.

Friederici, A. D. (2009a). Allocating functions to fiber tracts: facing its indirectness. Trends in Cognitive

Sciences, 13(9), 370–371.

Friederici, A. D. (2009b). Pathways to language: fiber tracts in the human brain. Trends in Cognitive

Sciences, 13(4), 175–181.

Friederici, A. D. (2011). The brain basis of language processing: from structure to function. Physiological

Reviews, 91(4), 1357–1392.

Friederici, A. D., Bahlmann, J., Heim, S., Schubotz, R., & Anwander, A. (2006). The brain differentiates

human and non-human grammars: functional localization and structural connectivity. Proceedings

of the National Academy of Sciences of the USA, 103(7), 2458–2463.

Friederici, A. D., Fiebach, C., Schlesewsky, M., Bornkessel, I., & von Cramon, D. (2006). Processing

linguistic complexity and grammaticality in the left frontal cortex. Cerebral Cortex, 16(12),

1709–1717.

Friederici, A. D., & Frisch, S. (2000). Verb argument structure processing: The role of verb-specific and

argument-specific information. Journal of Memory and Language, 43(3), 476–507.

Friederici, A. D., & Mecklinger, A. (1996). Syntactic parsing as revealed by brain responses: first-pass

and second-pass parsing processes. Journal of Psycholinguistic Research, 25(1), 157–176.

Friederici, A. D., Pfeifer, E., & Hahne, A. (1993). Event-related brain potentials during natural speech

processing: Effects of semantic, morphological and syntactic violations. Cognitive Brain Research,

1(3), 183–192.

Friederici, A. D., Steinhauer, K., Mecklinger, A., & Meyer, M. (1998). Working memory constraints

on syntactic ambiguity resolution as revealed by electrical brain responses. Biological Psychology,

47(3), 193–221.

Friederici, A. D., Steinhauer, K., & Pfeifer, E. (2002). Brain signatures of artificial language processing:

Evidence challenging the critical period hypothesis. Proceedings of the National Academy of

Sciences of the USA, 99(1), 529–534.

155
References

Friederici, A. D., Wang, Y., Herrmann, C., Maess, B., & Oertel, U. (2000). Localization of early

syntactic processes in frontal and temporal cortical areas: a magnetoencephalographic study.

Human Brain Mapping, 11(1), 1–11.

Friedman, D., & Johnson, R. (2000). Event-related potential (ERP) studies of memory encoding and

retrieval: a selective review. Microscopy Research and Technique, 51(1), 6–28.

Friedman, D., Simson, R., Ritter, W., & Rapin, I. (1975). The late positive component (P300) and

information processing in sentences. Glot International, 38(3), 255–262.

Friedmann, N., & Gvion, A. (2003). Sentence comprehension and working memory limitation in

aphasia: adissociation between semantic-syntactic and phonological reactivation. Brain and

Language, 86(1), 23–39.

Friston, K., Frith, C., Frackowiak, R., & Turner, R. (1995). Characterizing dynamic brain responses

with fMRI: a multivariate approach. NeuroImage, 2(2), 166–172.

Fuentemilla, L., Marco-Pallarés, J., & Grau, C. (2006). Modulation of spectral power and of phase

resetting of EEG contributes differentially to the generation of auditory event-related potentials.

NeuroImage, 30(3), 909–916.

Fuster, J. (1999). Memory in the cerebral cortex: An empirical approach to neural networks in the human

and nonhuman primate. Cambridge: MIT Press.

Galantucci, S., Tartaglia, M. C., Wilson, S. M., Henry, M. L., Filippi, M., Agosta, F., et al. (2011). White

matter damage in primary progressive aphasias: a diffusion tensor tractography study. Brain,

31(10), 3011–3029.

Garnsey, S., Pearlmutter, N., Myers, E., & Lotocky, M. (1997). The contributions of verb bias and

plausibility to the comprehension of temporarily ambiguous sentences. Journal of Memory and

Language, 37(1), 58–93.

Gazzaley, A., & Nobre, A. (2011). Top-down modulation: bridging selective attention and working

memory. Trends in Cognitive Sciences.

Geldmacher, D., Quigg, M., & Elias, W. (2007). MR tractography depicting damage to the arcuate

fasciculus in a patient with conduction aphasia. Neurology, 69(3), 321–322.

156
 References

Gerton, B., Brown, T., Meyer-Lindenberg, A., Kohn, P., Holt, J., Olsen, R., et al. (2004). Shared and

distinct neurophysiological components of the digits forward and backward tasks as revealed by

functional neuroimaging. Neuropsychologia, 42(13), 1781–1787.

Gibson, E. (2000). The dependency locality theory: a distance-based theory of linguistic complexity. In

Y. Miyashita, A. Marantz, & W. O’Neil (Eds.), Image, Language, Brain (pp. 95–126). Cambridge:

MIT Press.

Gibson, E., & Thomas, J. (1999). Memory limitations and structural forgetting: the perception of

complex ungrammatical sentences as grammatical. Language and Cognitive Processes, 14(3),

225–248.

Gierhan, S. (in revision). Connections for language in the human brain. Brain and Language.

Glasser, M., & Rilling, J. (2008). DTI tractography of the human brain’s language pathways. Cerebral

Cortex, 18(11), 2471–2482.

Glover, G. H. (1999). Deconvolution of impulse response in event-related BOLD fMRI. NeuroImage,

9(4), 416–429.

Goldberg, A. (1992). The inherent semantics of argument structure: the case of the English ditransitive

construction. Cognitive Linguistics, 3(1), 37–74.

Gonçalves, S., De Munck, J., Pouwels, P., Schoonhoven, R., Kuijer, J., Maurits, N., et al. (2006). Cor-

relating the alpha rhythm to BOLD using simultaneous EEG/fMRI: inter-subject variability.

NeuroImage, 30(1), 203–213.

Good, C. D., Johnsrude, I. S., Ashburner, J., Henson, R. N., Friston, K. J., & Frackowiak, R. S. (2001).

A voxel-based morphometric study of ageing in 465 normal adult human brains. NeuroImage,

14(1), 21–36.

Gordon, P., Hendrick, R., & Johnson, M. (2001). Memory interference during language processing.

Journal of Experimental Psychology: Learning, Memory, and Cognition, 27(6), 1411–1423.

Gordon, P., Hendrick, R., Johnson, M., & Lee, Y. (2006). Similarity-based interference during language

comprehension: evidence from eye tracking during reading. Journal of Experimental Psychology:

Learning, Memory, and Cognition, 32(6), 1304–1321.

Gordon, P., Hendrick, R., & Levine, W. (2002). Memory-load interference in syntactic processing.

Psychological Science, 13(5), 425–430.

157
References

Grech, R., Cassar, T., Muscat, J., Camilleri, K., Fabri, S., Zervakis, M., et al. (2008). Review on solving

the inverse problem in EEG source analysis. Journal of NeuroEngineering and Rehabilitation, 5,

25.

Grewe, T., Bornkessel, I., Zysset, S., Wiese, R., von Cramon, D., & Schlesewsky, M. (2005). The

emergence of the unmarked: a new perspective on the language-specific function of Broca’s area.

Human Brain Mapping, 26(3), 178–190.

Griswold, M., Jakob, P., Heidemann, R., Nittka, M., Jellus, V., Wang, J., et al. (2002). Generalized

autocalibrating partially parallel acquisitions (GRAPPA). Magnetic Resonance in Medicine, 47(6),

1202–1210.

Grodner, D., & Gibson, E. (2005). Consequences of the serial nature of linguistic input for sentenial

complexity. Cognitive Science, 29(2), 261–290.

Grodzinsky, Y. (2000). Overarching agrammatism. In Y. Grodzinsky, L. Shapiro, & D. Swinney (Eds.),

Language and the brain: representation and processing (pp. 73–86). San Diego: Academic Press.

Grodzinsky, Y. (2001). The neurology of syntax: Language use without Broca’s area. Behavioral and

Brain Sciences, 23(1), 1–71.

Grodzinsky, Y., Piñango, M., Zurif, E., & Drai, D. (1999). The critical role of group studies in

neuropsychology: comprehension regularities in Broca’s aphasia. Brain and Language, 67(2),

134–147.

Gross, J., Kujala, J., Hämäläinen, M., Timmermann, L., Schnitzler, A., & Salmelin, R. (2001). Dynamic

imaging of coherent sources: studying neural interactions in the human brain. Proceedings of the

National Academy of Sciences of the USA, 98(2), 694–699.

Grossberg, S. (1984). Some psychophysiological and pharmacological correlates of a developmental,

cognitive and motivational theory. Annals of the New York Academy of Sciences, 425(1), 58–151.

Grossman, M., Cooke, A., DeVita, C., Alsop, D., Detre, J., Chen, W., et al. (2002). Age-related changes in

working memory during sentence comprehension: an fMRI study. NeuroImage, 15(2), 302–317.

Grossman, M., Cooke, A., DeVita, C., Lee, C., Alsop, D., Detre, J., et al. (2003). Grammatical and

resource components of sentence processing in Parkinson’s disease. Neurology, 60(5), 775–781.

Grossmann, A., & Morlet, J. (1984). Decomposition of Hardy functions into square integrable wavelets

of constant shape. SIAM Journal on Mathematical Analysis, 15(4), 723–736.

158
 References

Gruber, O., & von Cramon, D. (2001). Domain-specific distribution of working memory processes

along human prefrontal and parietal cortices: a functional magnetic resonance imaging study.

Neuroscience Letters, 297(1), 29–32.

Gruber, O., & von Cramon, D. (2003). The functional neuroanatomy of human working memory

revisited: Evidence from 3-T fMRI studies using classical domain-specific interference tasks.

NeuroImage, 19(3), 797–809.

Haegens, S., Osipova, D., Oostenveld, R., & Jensen, O. (2010). Somatosensory working memory per-

formance in humans depends on both engagement and disengagement of regions in a distributed

network. Human Brain Mapping, 31(1), 26–35.

Hagoort, P., Brown, C., & Groothusen, J. (1993). The syntactic positive shift (SPS) as an ERP measure

of syntactic processing. Language and Cognitive Processes, 8(4), 439–483.

Hagoort, P., Hald, L., Bastiaansen, M., & Petersson, K. (2004). Integration of word meaning and world

knowledge in language comprehension. Science, 304(5669), 438–441.

Haider, H. (1993). Deutsche Syntax, generativ: Vorstudien zur Theorie einer projektiven Gram-

matikalisierung. Tübingen: Narr.

Halgren, E., Dhond, R. P., Christensen, N., Van Petten, C., Marinkovic, K., Lewine, J. D., et al. (2002).

N400-like magnetoencephalography responses modulated by semantic context, word frequency,

and lexical class in sentences. NeuroImage, 17(3), 1101–1116.

Hayasaka, S., Phan, K. L., Liberzon, I., Worsley, K. J., & Nichols, T. E. (2004). Nonstationary

cluster-size inference with random field and permutation methods. NeuroImage, 22(2), 676–687.

He, B., Musha, T., Okamoto, Y., Homma, S., Nakajima, Y., & Sato, T. (1987). Electric dipole tracing

in the brain by means of the boundary element method and its accuracy. IEEE Transactions on

Biomedical Engineering, 34(6), 406–414.

Heiervang, E., Behrens, T., Mackay, C., Robson, M., & Johansen-Berg, H. (2006). Between session

reproducibility and between subject variability of diffusion MR and tractography measures.

NeuroImage, 33(3), 867–877.

Heisenberg, W. (1927). Über den anschaulichen Inhalt der quantentheoretischen Kinematik und

Mechanik. Zeitschrift für Physik, 43(3), 172–198.

159
References

Helmholtz, H. (1881). Ueber die auf das Innere magnetisch oder diëlektrisch polarisirter Körper

wirkenden Kräfte. Annalen der Physik, 249(7), 385–406.

Henson, R. N., Burgess, N., & Frith, C. D. (1999). Recoding, storage, rehearsal and grouping in verbal

short-term memory: an fMRI study. Neuropsychologia, 38(4), 426–440.

Holmes, V. (1987). Syntactic parsing: In search of the garden path. In M. Coltheart (Ed.), Attention and

performance XII: the psychology of reading (pp. 587–599). London: Lawrence Erlbaum Associates.

Holmes, V., Stowe, L., & Cupples, L. (1989). Lexical expectations in parsing complement-verb sentences.

Journal of Memory and Language, 28(6), 668–689.

Hopkins, W. D., Russell, J. L., & Cantalupo, C. (2007). Neuroanatomical correlates of handedness for

tool use in chimpanzees (Pan troglodytes): implication for theories on the evolution of language.

Psychological Science, 18(11), 971–977.

Hsiao, F., & Gibson, E. (2003). Processing relative clauses in Chinese. Cognition, 90(1), 3–27.

Hsieh, L. T., Ekstrom, A. D., & Ranganath, C. (2011). Neural oscillations associated with item and

temporal order maintenance in working memory. Journal of Neuroscience, 31(30), 10803–10810.

Huettel, S., Song, A., & McCarthy, G. (2004). Functional magnetic resonance imaging. Sunderland:

Sinauer Associates.

Hyönä, J., & Hujanen, H. (1997). Effects of case marking and word order on sentence parsing in

Finnish: An eye fixation analysis. Quarterly Journal of Experimental Psychology, 50(4), 841–858.

Jacobs, J. (1887). Experiments in prehension. Mind, 12(45), 75–79.

Jacquemot, C., & Scott, S. (2006). What is the relationship between phonological short-term memory

and speech processing? Trends in Cognitive Sciences, 10(11), 480–486.

Jaeger, F., Fedorenko, E., & Gibson, E. (2005). Dissociation between production and comprehension

complexity. Poster Presentation at the 18th CUNY Sentence Processing Conference, Tucson.

Jenkinson, M., Bannister, P., Brady, M., & Smith, S. (2002). Improved optimization for the robust and

accurate linear registration and motion correction of brain images. NeuroImage, 17(2), 825–841.

Jensen, O., Gelfand, J., Kounios, J., & Lisman, J. (2002). Oscillations in the alpha band (9–12 Hz)

increase with memory load during retention in a short-term memory task. Cerebral Cortex,

12(8), 877–882.

160
 References

Jensen, O., & Mazaheri, A. (2010). Shaping functional architecture by oscillatory alpha activity: gating

by inhibition. Frontiers in Human Neuroscience, 4, 186.

Jezzard, P., & Balaban, R. (1995). Correction for geometric distortion in echo planar images from B0

field variations. Magnetic Resonance in Medicine, 34(1), 65–73.

Jian, B., & Vemuri, B. (2007). A unified computational framework for deconvolution to reconstruct

multiple fibers from diffusion weighted MRI. IEEE Transactions on Medical Imaging, 26(11),

1464–1471.

Johnson, B., & Hamm, J. (2000). High-density mapping in an N400 paradigm: evidence for bilateral

temporal lobe generators. Clinical Neurophysiology, 111(3), 532–545.

Johnson, M., Mitchell, K., Raye, C., & Greene, E. (2004). An age-related deficit in prefrontal cortical

function associated with refreshing information. Psychological Science, 15(2), 127–132.

Johnson, M., Reeder, J., Raye, C., & Mitchell, K. (2002). Second thoughts versus second looks: An

age-related deficit in reflectively refreshing just-activated information. Psychological Science, 13(1),

64–67.

Johnson, R. (1995). Event-related potential insights into the neurobiology of memory systems. In

F. Bollen & J. Grafman (Eds.), Handbook of Neuropsychology (pp. 135–135). Amsterdam: Elsevier.

Jones, D., Hughes, R., & Macken, W. (2007). The phonological store abandoned. Quarterly Journal of

Experimental Psychology, 60(4), 505–511.

Jonides, J., Schumacher, E., Smith, E., Koeppe, R., Awh, E., Reuter-Lorenz, P., et al. (1998). The role of

parietal cortex in verbal working memory. Journal of Neuroscience, 18(13), 5026–5034.

Jung, T., Makeig, S., Westerfield, M., Townsend, J., Courchesne, E., & Sejnowski, T. (2001). Analysis

and visualization of single-trial event-related potentials. Human Brain Mapping, 14(3), 166–185.

Just, M., & Carpenter, P. (1992). A capacity theory of comprehension: Individual differences in working

memory. Psychological Review, 99(1), 122–149.

Just, M., Carpenter, P., Keller, T., Eddy, W., & Thulborn, K. (1996). Brain activation modulated by

sentence comprehension. Science, 274(5284), 114–116.

Kaan, E., Harris, A., Gibson, E., & Holcomb, P. (2000). The P600 as an index of syntactic integration

difficulty. Language and Cognitive Processes, 15(2), 159–201.

161
References

Kaan, E., & Swaab, T. (2003). Repair, revision, and complexity in syntactic analysis: An electrophysio-

logical differentiation. Journal of Cognitive Neuroscience, 15(1), 98–110.

Kaan, E., Wijnen, F., & Swaab, T. Y. (2004). Gapping: Electrophysiological evidence for immediate

processing of “missing” verbs in sentence comprehension. Brain and Language, 89(3), 584–592.

Kamide, Y., Altmann, G., & Haywood, S. (2000). Predictive eye-movements in incremental processing

of head-final structures. Poster Presentation at the 20th CUNY Sentence Processing Conference, La

Jolla.

Kamide, Y., & Mitchell, D. (1999). Incremental pre-head attachment in Japanese parsing. Language and

Cognitive Processes, 14(5/6), 631–662.

Kamide, Y., Scheepers, C., & Altmann, G. (2003). Integration of syntactic and semantic informa-

tion in predictive processing: cross-linguistic evidence from German and English. Journal of

Psycholinguistic Research, 32(1), 37–55.

Kamp, A., Pfurtscheller, G., Edlinger, G., & Silva, F. Lopes da. (1993). Technological basis of EEG

recording. In E. Niedermeyer & F. Lopes da Silva (Eds.), Electroencephalography. Basic principles,

clinical application and related fields (pp. 127–138). München: Urban & Schwarzenberg.

Keil, J., Müller, N., Ihssen, N., & Weisz, N. (2012). On the variability of the McGurk effect: Audiovisual

integration depends on prestimulus brain states. Cerebral Cortex, 22(1), 221–231.

Kemmerer, D. (2012). The cross-linguistic prevalence of SOV and SVO word orders reflects the

sequential and hierarchical representation of action in Broca’s area. Language and Linguistics

Compass, 6(1), 50–66.

Keppel, G., & Underwood, B. (1962). Proactive inhibition in short-term retention of single items.

Journal of Verbal Learning and Verbal Behavior, 1(3), 153–161.

Kiebel, S., Daunizeau, J., Phillips, C., & Friston, K. (2008). Variational Bayesian inversion of the

equivalent current dipole model in EEG/MEG. NeuroImage, 39(2), 728–741.

Kim, J., Kim, M. S., Lee, J. S., Lee, D. S., Lee, M. C., & Kwon, J. S. (2002). Dissociation of working

memory processing associated with native and second languages: PET investigation. NeuroImage,

15(4), 879–891.

162
 References

Kim, J., Koizumi, M., Ikuta, N., Fukumitsu, Y., Kimura, N., Iwata, K., et al. (2009). Scrambling

effects on the processing of Japanese sentences: An fMRI study. Journal of Neurolinguistics, 22(2),

151–166.

King, J., & Just, M. (1991). Individual differences in syntactic processing: The role of working memory.

Journal of Memory and Language, 30(5), 580–602.

Kinno, R., Kawamura, M., Shioda, S., & Sakai, K. (2008). Neural correlates of noncanonical syntactic

processing revealed by a picture-sentence matching task. Human Brain Mapping, 29(9), 1015–1027.

Kintsch, W., & Van Dijk, T. (1978). Toward a model of text comprehension and production. Psychological

Review, 85(5), 363–394.

Klein, A., Andersson, J., Ardekani, B., Ashburner, J., Avants, B., Chiang, M., et al. (2009). Evaluation

of 14 nonlinear deformation algorithms applied to human brain MRI registration. NeuroImage,

46(3), 786–802.

Klein, W. (2000). An analysis of the German Perfekt. Language, 76(2), 358–382.

Klimesch, W., Sauseng, P., & Hanslmayr, S. (2007). EEG alpha oscillations: the inhibition-timing

hypothesis. Brain Research Reviews, 53(1), 63–88.

Klimesch, W., Sauseng, P., Hanslmayr, S., Gruber, W., & Freunberger, R. (2007). Event-related phase

reorganization may explain evoked neural dynamics. Neuroscience and Biobehavioral Reviews,

31(7), 1003–1016.

Kluender, R., & Kutas, M. (1993). Bridging the gap: Evidence from ERPs on the processing of

unbounded dependencies. Journal of Cognitive Neuroscience, 5(2), 196–214.

Knecht, S., Dräger, B., Deppe, M., Bobe, L., Lohmann, H., Flöel, A., et al. (2000). Handedness and

hemispheric language dominance in healthy humans. Brain, 123(12), 2512–2518.

Konieczny, L. (2000). Locality and parsing complexity. Journal of Psycholinguistic Research, 29(6),

627–645.

Konieczny, L., & Döring, P. (2003). Anticipation of clause-final heads: Evidence from eye-tracking and

SRNs. In Proceedings of ICCS/ASCS.

Krause, C., Heikki Lang, A., Laine, M., Kuusisto, M., & Pörn, B. (1996). Event-related. EEG desyn-

chronization and synchronization during an auditory memory task. Electroencephalography and

Clinical Neurophysiology, 98(4), 319–326.

163
References

Kriegeskorte, N., Simmons, W., Bellgowan, P., & Baker, C. (2009). Circular analysis in systems

neuroscience – the dangers of double dipping. Nature Neuroscience, 12(5), 535–540.

Krifka, M. (2008). Basic notions of information structure. Acta Linguistica Hungarica, 55(3), 243–276.

Kutas, M., Lindamood, T., & Hillyard, S. (1984). Word expectancy and event-related brain potentials

during sentence processing. In M. Kutas, T. E. Lindamood, S. A. Hillyard, S. Kornblum, &

J. Requin (Eds.), Preparatory states and processes (pp. 217–237). Hillsdale: Erlbaum.

Kutas, M., & Van Petten, C. (1994). Psycholinguistics electrified. In M. A. Gernsbacher (Ed.), Handbook

of psycholinguistics (pp. 83–143). San Diego: Academic Press.

Kutas, M., Van Petten, C., & Kluender, R. (2005). Psycholinguistics electrified II: 1994–2005. In

M. J. Traxler & M. A. Gernsbacher (Eds.), Handbook of psycholinguistics (pp. 659–724). San

Diego: Academic Press.

Lacadie, C., Fulbright, R., Rajeevan, N., Constable, R., & Papademetris, X. (2008). More accurate

Talairach coordinates for neuroimaging using non-linear registration. NeuroImage, 42(2), 717–

725.

Lachaux, J., Rodriguez, E., Martinerie, J., & Varela, F. (1999). Measuring phase synchrony in brain

signals. Human Brain Mapping, 8(4), 194–208.

Lapata, M., Keller, F., & Walde, S. Schulte im. (2001). Verb frame frequency as a predictor of verb bias.

Journal of Psycholinguistic Research, 30(4), 419–435.

Lau, E. F., Phillips, C., & Poeppel, D. (2008). A cortical network for semantics:(de) constructing the

N400. Nature Reviews Neuroscience, 9(12), 920–933.

Laufs, H., Holt, J., Elfont, R., Krams, M., Paul, J., Krakow, K., et al. (2006). Where the BOLD signal

goes when alpha EEG leaves. NeuroImage, 31(4), 1408–1418.

Lazar, M., Weinstein, D., Tsuruda, J., Hasan, K., Arfanakis, K., Meyerand, M., et al. (2003). White

matter tractography using diffusion tensor deflection. Human Brain Mapping, 18(4), 306–321.

Le Bihan, D., Breton, E., Lallemand, D., Grenier, P., Cabanis, E., Laval-Jeantet, M., et al. (1986). MR

imaging of intravoxel incoherent motions: application to diffusion and perfusion in neurologic

disorders. Radiology, 161(2), 401–407.

164
 References

Leff, A. P., Schofield, T. M., Crinion, J. T., Seghier, M. L., Grogan, A., Green, D. W., et al. (2009). The

left superior temporal gyrus is a shared substrate for auditory short-term memory and speech

comprehension: evidence from 210 patients with stroke. Brain, 132(12), 3401–3410.

Lehtelä, L., Salmelin, R., & Hari, R. (1997). Evidence for reactive magnetic 10-Hz rhythm in the human

auditory cortex. Neuroscience Letters, 222(2), 111–114.

Leiberg, S., Lutzenberger, W., & Kaiser, J. (2006). Effects of memory load on cortical oscillatory activity

during auditory pattern working memory. Brain Research, 1120(1), 131–140.

Lepsien, J., Thornton, I., & Nobre, A. (2011). Modulation of working-memory maintenance by directed

attention. Neuropsychologia, 49(6), 1569–1577.

Levy, R. (2008). Expectation-based syntactic comprehension. Cognition, 106(3), 1126–1177.

Lewandowsky, S., Geiger, S. M., & Oberauer, K. (2008). Interference-based forgetting in verbal

short-term memory. Journal of Memory and Language, 59(2), 200–222.

Lewandowsky, S., Oberauer, K., & Brown, G. (2009). No temporal decay in verbal short-term memory.

Trends in Cognitive Sciences, 13(3), 120–126.

Lewis, R. (1996). Interference in short-term memory: The magical number two (or three) in sentence

processing. Journal of Psycholinguistic Research, 25(1), 93–115.

Lewis, R., & Vasishth, S. (2005). An activation-based model of sentence processing as skilled memory

retrieval. Cognitive Science, 29(3), 375–419.

Lewis, R., Vasishth, S., & Van Dyke, J. (2006). Computational principles of working memory in

sentence comprehension. Trends in Cognitive Sciences, 10(10), 447–454.

Leyton, C. E., Villemagne, V. L., Savage, S., Pike, K. E., Ballard, K. J., Piguet, O., et al. (2011). Subtypes

of progressive aphasia: application of the International Consensus Criteria and validation using

beta-amyloid imaging. Brain, 134(10), 3030–3043.

Lisman, J., & Idiart, M. (1995). Storage of 7+/-2 short-term memories in oscillatory subcycles. Science,

267(5203), 1512–1515.

Lœvenbruck, H., Baciu, M., Segebarth, C., & Abry, C. (2005). The left inferior frontal gyrus under

focus: an fMRI study of the production of deixis via syntactic extraction and prosodic focus.

Journal of Neurolinguistics, 18(3), 237–258.

165
References

Logothetis, N., Pauls, J., Augath, M., Trinath, T., Oeltermann, A., et al. (2001). Neurophysiological

investigation of the basis of the fMRI signal. Nature, 412(6843), 150–157.

Love, T., & Swinney, D. (1996). Coreference processing and levels of analysis in object-relative

constructions; demonstration of antecedent reactivation with the cross-modal priming paradigm.

Journal of Psycholinguistic Research, 25(1), 5–24.

Luo, H., Husain, F. T., Horwitz, B., & Poeppel, D. (2005). Discrimination and categorization of speech

and non-speech sounds in an MEG delayed-match-to-sample study. NeuroImage, 28(1), 59–71.

Macmillan, N., & Creelman, C. (2005). Detection theory: a user’s guide (2nd ed.). Mahwan: Lawrence

Erlbaum Associates.

MacSweeney, M., Brammer, M., Waters, D., & Goswami, U. (2009). Enhanced activation of the left

inferior frontal gyrus in deaf and dyslexic adults during rhyming. Brain, 132(7), 1928–1940.

MacWhinney, B., & Pléh, C. (1988). The processing of restrictive relative clauses in Hungarian.

Cognition, 29(2), 95–141.

Maess, B., Herrmann, C. S., Hahne, A., Nakamura, A., & Friederici, A. D. (2006). Localizing the

distributed language network responsible for the N400 measured by MEG during auditory

sentence processing. Brain Research, 1096(1), 163–172.

Makeig, S., Debener, S., Onton, J., & Delorme, A. (2004). Mining event-related brain dynamics. Trends

in Cognitive Sciences, 8(5), 204–210.

Mäkinen, V., Tiitinen, H., & May, P. (2005). Auditory event-related responses are generated indepen-

dently of ongoing brain activity. NeuroImage, 24(4), 961–968.

Makuuchi, M., Bahlmann, J., Anwander, A., & Friederici, A. D. (2009). Segregating the core computa-

tional faculty of human language from working memory. Proceedings of the National Academy of

Sciences of the USA, 106(20), 8362–8367.

Makuuchi, M., Bahlmann, J., & Friederici, A. (2012). An approach to separating the levels of hierarchical

structure building in language and mathematics. Philosophical Transactions of the Royal Society B:

Biological Sciences, 367(1598), 2033–2045.

Makuuchi, M., Grodzinsky, Y., Amunts, K., Santi, A., & Friederici, A. (2012). Processing noncanonical

sentences in Broca’s region: reflections of movement distance and type. Cerebral Cortex.

166
 References

Maltseva, I., Geissler, H., & Başar, E. (2000). Alpha oscillations as an indicator of dynamic memory

operations–anticipation of omitted stimuli. International Journal of Psychophysiology, 36(3),

185–197.

Maris, E., & Oostenveld, R. (2007). Nonparametric statistical testing of EEG- and MEG-data. Journal

of Neuroscience Methods, 164(1), 177–190.

Marslen-Wilson, W. (1973). Linguistic structure and speech shadowing at very short latencies. Nature,

244(5417), 522–523.

Martin, R. (1987). Articulatory and phonological deficits in short-term memory and their relation to

syntactic processing. Brain and Language, 32(1), 159–192.

Martin, R., Blossom-Stach, C., Yaffee, L., & Wetzel, W. (1995). Consequences of a motor programming

deficit for rehearsal and written sentence comprehension. Quarterly Journal of Experimental

Psychology, 48(3), 536–572.

Martin, R., & He, T. (2004). Semantic short-term memory and its role in sentence processing:

areplication. Brain and Language, 89(1), 76–82.

Martin, R., & Romani, C. (1994). Verbal working memory and sentence comprehension: a multiple-

components view. Neuropsychology, 8(4), 506–506.

Martin, R., Shelton, J., & Yaffee, L. (1994). Language processing and working memory: Neuropsy-

chological evidence for separate phonological and semantic capacities. Journal of Memory and

Language, 33(1), 83–111.

Matzke, M., Mai, H., Nager, W., Rüsseler, J., & Münte, T. (2002). The costs of freedom: An ERP –

study of non-canonical sentences. Clinical Neurophysiology, 113(6), 844–852.

Mazuka, R., Itoh, K., & Kondo, T. (2002). Costs of scrambling in Japanese sentence processing. In

M. Nakayama (Ed.), Sentence processing in east-Asian languages (pp. 131–166). Stanford: CSLI

Publications.

McCarthy, R., & Warrington, E. (1984). A two-route model of speech production. Brain, 107(2),

463–485.

McElree, B. (2000). Sentence comprehension is mediated by content-addressable memory structures.

Journal of Psycholinguistic Research, 29(2), 111–123.

167
References

McElree, B., & Dosher, B. (1993). Serial retrieval processes in the recovery of order information. Journal

of Experimental Psychology: General, 122(3), 291–291.

McElree, B., Foraker, S., & Dyer, L. (2003). Memory structures that subserve sentence comprehension.

Journal of Memory and Language, 48(1), 67–91.

McElree, B., & Griffith, T. (1995). Syntactic and thematic processing in sentence comprehension:

evidence for a temporal dissociation. Journal of Experimental Psychology: Learning, Memory, and

Cognition, 21(1), 134–134.

McGeoch, J. (1932). Forgetting and the law of disuse. Psychological Review, 39(4), 352–370.

Mechelli, A., Price, C., Friston, K., & Ashburner, J. (2005). Voxel-based morphometry of the human

brain: methods and applications. Current Medical Imaging Reviews, 1(2), 105–113.

Mecklinger, A., Schriefers, H., Steinhauer, K., & Friederici, A. (1995). Processing relative clauses

varying on syntactic and semantic dimensions: an analysis with event-related potentials. Memory

& Cognition, 23(4), 477–494.

Medendorp, W., Kramer, G., Jensen, O., Oostenveld, R., Schoffelen, J., & Fries, P. (2007). Oscillatory

activity in human parietal and occipital cortex shows hemispheric lateralization and memory

effects in a delayed double-step saccade task. Cerebral Cortex, 17(10), 2364–2374.

Meltzer, J., Negishi, M., Mayes, L., & Constable, R. (2007). Individual differences in EEG theta and

alpha dynamics during working memory correlate with fMRI responses across subjects. Clinical

Neurophysiology, 118(11), 2419–2436.

Meyer, L. (2009). Distance phenomena in german sentence processing. Unpublished master’s thesis,

Universität Potsdam.

Meyer, L., Obleser, J., Anwander, A., & Friederici, A. D. (2012). Linking ordering in Broca’s area

to storage in left temporo-parietal regions: The case of sentence processing. NeuroImage, 62(3),

1987–1998.

Meyer, L., Obleser, J., & Friederici, A. D. (2012). Left parietal alpha enhancement during working

memory-intensive sentence processing. Cortex.

Meyer, L., Obleser, J., Kiebel, S., & Friederici, A. (in revision). Spatiotemporal dynamics of argument

retrieval and reordering: an fMRI and EEG study on sentence processing. Frontiers in Language

Sciences.

168
 References

Michels, L., Bucher, K., Lüchinger, R., Klaver, P., Martin, E., Jeanmonod, D., et al. (2010). Simultaneous

EEG-fMRI during a working memory task: modulations in low and high frequency bands. PloS

ONE, 5(4), e10298–e10298.

Miller, G. (1956). The magical number seven, plus or minus two: some limits on our capacity for

processing information. Psychological Review, 63(2), 81–97.

Mitra, P., & Pesaran, B. (1999). Analysis of dynamic brain imaging data. Biophysical Journal, 76(2),

691–708.

Miyake, A., & Shah, P. (1999). Toward unified theories of working memory: Emerging general

consensus, unresolved theoretical issues, and future research directions. In A. Miyake & P. Shah

(Eds.), Models of working memory: Mechanisms of active maintenance and executive control (pp.

442–481). Cambridge: Cambridge University Press.

Miyamoto, E., & Takahashi, S. (2002). Sources of difficulty in processing scrambling in Japanese. In

M. Nakayama (Ed.), Sentence processing in East-Asian languages (pp. 167–188). Stanford: CSLI

Publications.

Mori, S., Wakana, S., Nagae-Poetscher, L., & Van Zijl, P. (2006). MRI atlas of human white matter.

Amsterdam: Elsevier.

Mugler III, J., & Brookeman, J. (1990). Three dimensional magnetization prepared rapid gradient echo

imaging (3D MP RAGE). Magnetic Resonance in Medicine, 15(1), 152–157.

Müller, N., & Knight, R. (2006). The functional neuroanatomy of working memory: contributions of

human brain lesion studies. Neuroscience, 139(1), 51–58.

Nakano, Y., Felser, C., & Clahsen, H. (2002). Antecedent priming at trace positions in Japanese

long-distance scrambling. Journal of Psycholinguistic Research, 31(5), 531–571.

Nakatani, K., & Gibson, E. (2008). Distinguishing theories of syntactic storage costs: Evidence from

Japanese. Linguistics, 46(1), 63–87.

Nakatani, K., & Gibson, E. (2010). An on-line study of Japanese nesting complexity. Cognitive Science,

34(1), 94–112.

Neisser, U. (1967). Cognitive psychology. New York: Appleton-Century Crofts.

Nelson, S., McDermott, K., & Petersen, S. (2012). In favor of a ’fractionation’ view of ventral parietal

cortex: comment on Cabeza et al. Trends in Cognitive Sciences, 16(8), 399–400.

169
References

Newman, S., Lee, D., & Ratliff, K. (2009). Off-line sentence processing: What is involved in answering

a comprehension probe? Human Brain Mapping, 30(8), 2499–2511.

Nicholas Nagel, H., Shapiro, L., & Nawy, R. (1994). Prosody and the processing of filler-gap sentences.

Journal of Psycholinguistic Research, 23(6), 473–485.

Nicol, J. (1993). Reconsidering reactivation. In A. Altmann & R. Shillcock (Eds.), Cognitive models of

speech processing (pp. 321–347). Hove: Lawrence Erlbaum Associates.

Nicol, J., Fodor, J., & Swinney, D. (1994). Using cross-modal lexical decision tasks to investigate

sentence processing. Journal of Experimental Psychology: Learning, Memory, and Cognition, 20(5),

1229–1238.

Nicol, J., & Swinney, D. (1989). The role of structure in coreference assignment during sentence

comprehension. Journal of Psycholinguistic Research, 18(1), 5–19.

Niedermeyer, E., & Lopes da Silva, F. (1993). Electroencephalography. Basic principles, clinical application

and related fields. München: Urban & Schwarzenberg.

Nishitani, N., Schurmann, M., Amunts, K., & Hari, R. (2005). Broca’s region: from action to language.

Physiology, 20(1), 60–69.

Novais-Santos, S., Gee, J., Shah, M., Troiani, V., Work, M., & Grossman, M. (2007). Resolving sentence

ambiguity with planning and working memory resources: Evidence from fMRI. NeuroImage,

37(1), 361–378.

Nunez, P., & Srinivasan, R. (1981). Electric fields of the brain: the neurophysics of EEG. New York:

Oxford University Press.

Oberauer, K., & Lange, E. (2008). Interference in verbal working memory: Distinguishing similarity-

based confusion, feature overwriting, and feature migration. Journal of Memory and Language,

58(3), 730–745.

Oberauer, K., Süß, H. M., Wilhelm, O., & Wittman, W. W. (2003). The multiple faces of working

memory: storage, processing, supervision, and coordination. Intelligence, 31(2), 167–193.

Obleser, J., Meyer, L., & Friederici, A. D. (2011). Dynamic assignment of neural resources in auditory

comprehension of complex sentences. NeuroImage, 56(4), 2310–2320.

Obleser, J., & Weisz, N. (2011). Suppressed alpha oscillations predict intelligibility of speech and its

acoustic details. Cerebral Cortex.

170
 References

Oldfield, R. (1971). The assessment and analysis of handedness: the Edinburgh inventory. Neuropsy-

chologia, 9(1), 97–113.

Oostenveld, R., Fries, P., Maris, E., & Schoffelen, J. (2011). FieldTrip: open source software for advanced

analysis of MEG, EEG, and invasive electrophysiological data. Computational Intelligence and

Neuroscience, 2011, 1.

Osterhout, L., & Holcomb, P. (1992). Event-related brain potentials elicited by syntactic anomaly.

Journal of Memory and Language, 31(6), 785–785.

Osterhout, L., & Holcomb, P. (1993). Event-related potentials and syntactic anomaly: Evidence of

anomaly detection during the perception of continuous speech. Language and Cognitive Processes,

8(4), 413–437.

Osterhout, L., & Swinney, D. (1993). On the temporal course of gap-filling during comprehension of

verbal passives. Journal of Psycholinguistic Research, 22(2), 273–286.

Owen, A. M., McMillan, K. M., Laird, A. R., & Bullmore, E. (2005). N-back working memory

paradigm: a meta-analysis of normative functional neuroimaging studies. Human Brain Mapping,

25(1), 46–59.

Papoutsi, M., Stamatakis, E. A., Griffiths, J., Marslen-Wilson, W. D., & Tyler, L. K. (2011). Is left fronto-

temporal connectivity essential for syntax? Effective connectivity, tractography and performance

in left-hemisphere damaged patients. NeuroImage, 58(2), 656–664.

Pappert, S., Schließer, J., & Pechmann, T. (2008). Effects of local context on argument number and

verb type expectations. In A. Steube (Ed.), The discourse potential of underspecified structures (pp.

193–205). Berlin: de Gruyter.

Parker, G. J. M., Luzzi, S., Alexander, D. C., Wheeler-Kingshott, C. A. M., Ciccarelli, O., & Lam-

bon Ralph, M. A. (2005). Lateralization of ventral and dorsal auditory-language pathways in the

human brain. NeuroImage, 24(3), 656–666.

Parkin, A. (2001). The structure and mechanisms of memory. In B. Rapp (Ed.), The handbook of cognitive

neuropsychology. What deficits reveal about the human mind (pp. 399–422). Hove: Psychology

Press.

Pastra, K., & Aloimonos, Y. (2012). The minimalist grammar of action. Philosophical Transactions of

the Royal Society B: Biological Sciences, 367(1585), 103–117.

171
References

Paulesu, E., Frith, C., & Frackowiak, R. (1993). The neural correlates of the verbal component of

working memory. Nature, 362(6418), 342–345.

Pauling, L. (1935). The oxygen equilibrium of hemoglobin and its structural interpretation. Proceedings

of the National Academy of Sciences of the USA, 21(4), 186–191.

Peterson, L., & Peterson, M. (1959). Short-term retention of individual verbal items. Journal of

Experimental Psychology, 58(3), 193–198.

Petitto, L., Zatorre, R., Gauna, K., Nikelski, E., Dostie, D., & Evans, A. (2000). Speech-like cerebral

activity in profoundly deaf people processing signed languages: implications for the neural basis of

human language. Proceedings of the National Academy of Sciences of the USA, 97(25), 13961–13966.

Petrides, M., Alivisatos, B., Meyer, E., & Evans, A. (1993). Functional activation of the human frontal

cortex during the performance of verbal working memory tasks. Proceedings of the National

Academy of Sciences of the USA, 90(3), 878–882.

Petrides, M., & Pandya, D. (1984). Projections to the frontal cortex from the posterior parietal region

in the rhesus monkey. The Journal of Comparative Neurology, 228(1), 105–116.

Petrides, M., Tomaiuolo, F., Yeterian, E. H., & Pandya, D. N. (2012). The prefrontal cortex: comparative

architectonic organization in the human and the macaque monkey brains. Cortex, 48(1), 46–57.

Pfurtscheller, G., Stancák, A., & Neuper, C. (1996). Event-related synchronization (ERS) in the alpha

band—an electrophysiological correlate of cortical idling: a review. International Journal of

Psychophysiology, 24(1), 39–46.

Phillips, C., Kazanina, N., & Abada, S. (2005). ERP effects of the processing of syntactic long-distance

dependencies. Cognitive Brain Research, 22(3), 407–428.

Phillips, O. R., Clark, K. A., Woods, R. P., Subotnik, K. L., Asarnow, R. F., Nuechterlein, K. H., et

al. (2010). Topographical relationships between arcuate fasciculus connectivity and cortical

thickness. Human Brain Mapping, 32(11), 1799–1801.

Pickering, M., & Ferreira, V. (2008). Structural priming: a critical review. Psychological Bulletin, 134(3),

427.

Polich, J. (2007). Updating P300: an integrative theory of P3a and P3b. Clinical Neurophysiology,

118(10), 2128–2148.

172
 References

Polich, J., Howard, L., & Starr, A. (1983). P300 latency correlates with digit span. Psychophysiology,

20(6), 665–669.

Potter, M. (1976). Short-term conceptual memory for pictures. Journal of Experimental Psychology:

Human Learning and Memory, 2(5), 509–522.

Quigg, M., & Fountain, N. (1999). Conduction aphasia elicited by stimulation of the left posterior

superior temporal gyrus. Journal of Neurology, Neurosurgery, and Psychiatry, 66(3), 393–356.

Quiroga, R. (1998). Quantitative analysis of EEG signals: Time-frequency methods and Chaos theory.

Unpublished doctoral dissertation, Medical University Lübeck.

Radford, A. (1997). Syntax: a minimalist introduction. Cambridge: Cambridge University Press.

Ramírez, R., Wipf, D., & Baillet, S. (2010). Neuroelectromagnetic source imaging of brain dynamics.

Computational Neuroscience, 38(2), 127–155.

Ravizza, S. M., Delgado, M. R., Chein, J. M., Becker, J. T., & Fiez, J. A. (2004). Functional dissociations

within the inferior parietal cortex in verbal working memory. NeuroImage, 22(2), 562–573.

Ravizza, S. M., Hazeltine, E., Ruiz, S., & Zhu, D. C. (2011). Left TPJ activity in verbal working memory:

Implications for storage- and sensory-specific models of short term memory. NeuroImage, 55(4),

1836–1846.

Reese, T., Heid, O., Weisskoff, R., & Wedeen, V. (2003). Reduction of eddy current induced distortion

in diffusion MRI using a twice refocused spin echo. Magnetic Resonance in Medicine, 49(1),

177–182.

Regan, D. (1989). Human brain electrophysiology: evoked potentials and evoked magnetic fields in science

and medicine. New York: Elsevier.

Reich, D., Ozturk, A., Calabresi, P., & Mori, S. (2010). Automated vs. conventional tractography in

multiple sclerosis: Variability and correlation with disability. NeuroImage, 49(4), 3047–3056.

Richardson, F. M., Ramsden, S., Ellis, C., Burnett, S., Megnin, O., Catmur, C., et al. (2011). Auditory

short-term memory capacity correlates with gray matter density in the left posterior STS in

cognitively normal and dyslexic adults. Journal of Cognitive Neuroscience, 23(12), 3746–3756.

Ridgway, G. R., Henley, S. M. D., Rohrer, J. D., Scahill, R. I., Warren, J. D., & Fox, N. C. (2008). Ten

simple rules for reporting voxel-based morphometry studies. NeuroImage, 40(4), 1429–1435.

173
References

Roach, B. J., & Mathalon, D. H. (2008). Event-related EEG time-frequency analysis: an overview of

measures and an analysis of early gamma band phase locking in schizophrenia. Schizophrenia

Bulletin, 34(5), 907–926.

Röder, B., Stock, O., Neville, H., Bien, S., & Rösler, F. (2002). Brain activation modulated by the com-

prehension of normal and pseudo-word sentences of different processing demands: a functional

magnetic resonance imaging study. NeuroImage, 15(4), 1003–1014.

Rogalski, E., Cobia, D., Harrison, T., Wieneke, C., Thompson, C., Weintraub, S., et al. (2011). Anatomy

of language impairments in primary progressive aphasia. The Journal of Neuroscience, 31(9), 3344–

3350.

Rogalsky, C., & Hickok, G. (2010). The role of Broca’s area in sentence comprehension. Journal of

Cognitive Neuroscience, 23(7), 1664–1680.

Rogalsky, C., Matchin, W., & Hickok, G. (2008). Broca’s area, sentence comprehension, and working

memory: an fMRI study. Frontiers in Human Neuroscience, 2, 14.

Romero, L., Walsh, V., & Papagno, C. (2006). The neural correlates of phonological short-term

memory: a repetitive transcranial magnetic stimulation study. Journal of Cognitive Neuroscience,

18(7), 1147–1155.

Rosen, H., Petersen, S., Linenweber, M., Snyder, A., White, D., Chapman, L., et al. (2000). Neural

correlates of recovery from aphasia after damage to left inferior frontal cortex. Neurology, 55(12),

1883–1894.

Rösler, F., Pechmann, T., Streb, J., Röder, B., & Hennighausen, E. (1998). Parsing of sentences in a

language with varying word order: word-by-word variations of processing demands are revealed

by event-related brain potentials. Journal of Memory and Language, 38(2), 150–176.

Ruchkin, D., Berndt, R., Johnson, R., Grafman, J., Ritter, W., Canoune, H., et al. (1999). Lexical con-

tributions to retention of verbal information in working memory: Event-related brain potential

evidence. Journal of Memory and Language, 41(3), 345–364.

Ruchkin, D., Johnson, R., Grafman, J., Canoune, H., Ritter, W., et al. (1992). Distinctions and

similarities among working memory processes: An event-related potential study. Cognitive Brain

Research, 1(1), 53–66.

174
 References

Rugg, M. D., & Coles, M. G. H. (1995). Electrophysiology of mind: Event-related brain potentials and

cognition. New York: Oxford University Press.

Rugg, M. D., & Curran, T. (2007). Event-related potentials and recognition memory. Trends in

Cognitive Sciences, 11(6), 251–257.

Rumelhart, D. (1980). Schemata: The building blocks of cognition. In R. J. Spiro, B. C. Bruce, &

W. F. Brewer (Eds.), Theoretical issues in reading comprehension (pp. 161–188). Hillsdale: Erlbaum.

Rumelhart, D., Lindsay, P., & Norman, D. (1972). A process model for long-term memory. In

E. Tulving, W. Donaldson, & G. Bower (Eds.), Organization of memory. New York: Academic

Press.

Samar, V., Bopardikar, A., Rao, R., & Swartz, K. (1999). Wavelet analysis of neuroelectric waveforms: a

conceptual tutorial. Brain and Language, 66(1), 7–60.

Santi, A., & Grodzinsky, Y. (2007). Working memory and syntax interact in Broca’s area. NeuroImage,

37(1), 8–17.

Santi, A., & Grodzinsky, Y. (2010). fMRI adaptation dissociates syntactic complexity dimensions.

NeuroImage, 51(4), 1285–1293.

Sarnthein, J., Petsche, H., Rappelsberger, P., Shaw, G. L., & von Stein, A. (1998). Synchronization

between prefrontal and posterior association cortex during human working memory. Proceedings

of the National Academy of Sciences of the USA, 95(12), 7092–7096.

Saur, D., Schelter, B., Schnell, S., Kratochvil, D., Küpper, H., Kellmeyer, P., et al. (2010). Combining

functional and anatomical connectivity reveals brain networks for auditory language comprehen-

sion. NeuroImage, 49(4), 3187–3197.

Schack, B., Klimesch, W., & Sauseng, P. (2005). Phase synchronization between theta and upper alpha

oscillations in a working memory task. International Journal of Psychophysiology, 57(2), 105–114.

Schacter, D. L., Buckner, R. L., Koutstaal, W., Dale, A. M., & Rosen, B. R. (1997). Late onset of anterior

prefrontal activity during true and false recognition: an event-related fMRI study. NeuroImage,

6(4), 259–269.

Scheepers, C., Sturt, P., Martin, C. J., Myachykov, A., Teevan, K., & Viskupova, I. (2011). Struc-

tural priming across cognitive domains: from simple arithmetic to relative-clause attachment.

Psychological Science, 22(10), 1319–1326.

175
References

Scheeringa, R., Petersson, K., Oostenveld, R., Norris, D., Hagoort, P., & Bastiaansen, M. (2009). Trial-

by-trial coupling between EEG and BOLD identifies networks related to alpha and theta EEG

power increases during working memory maintenance. NeuroImage, 44(3), 1224–1238.

Schenker, N., Hopkins, W., Spocter, M., Garrison, A., Stimpson, C., Erwin, J., et al. (2010). Broca’s area

homologue in chimpanzees (pan troglodytes): probabilistic mapping, asymmetry, and comparison

to humans. Cerebral Cortex, 20(3), 730–742.

Schlesewsky, M., & Bornkessel, I. (2004). On incremental interpretation: degrees of meaning accessed

during sentence comprehension. Lingua, 114(9–10), 1213–1234.

Seghier, M. L., Ramlackhansingh, A., Crinion, J., Leff, A. P., & Price, C. J. (2008). Lesion identification

using unified segmentation-normalisation models and fuzzy clustering. NeuroImage, 41(4),

1253–1266.

Service, E., Helenius, P., Maury, S., & Salmelin, R. (2007). Localization of syntactic and semantic brain

responses using magnetoencephalography. Journal of Cognitive Neuroscience, 19(7), 1193–1205.

Shah, P., & Miyake, A. (1999). Models of working memory: Mechanisms of active maintenance and

executive control. Cambridge: Cambridge University Press.

Shapiro, L., Brookins, B., Gordon, B., & Nagel, N. (1991). Verb effects during sentence processing.

Journal of Experimental Psychology: Learning, Memory, and Cognition, 17(5), 983–996.

Shapiro, L., Gordon, B., Hack, N., & Killackey, J. (1993). Verb-argument structure processing in

complex sentences in Broca’s and Wernicke’s aphasia. Brain and Language, 45(3), 423–447.

Shapiro, L., Zurif, E., & Grimshaw, J. (1989). Verb processing during sentence comprehension:

Contextual impenetrability. Journal of Psycholinguistic Research, 18(2), 223–243.

Shetreet, E., Palti, D., Friedmann, N., & Hadar, U. (2007). Cortical representation of verb processing

in sentence comprehension: number of complements, subcategorization, and thematic frames.

Cereb Cortex, 17(8), 1958–1969.

Shivde, G., & Anderson, M. (2011). On the existence of semantic working memory: Evidence for direct

semantic maintenance. Journal of Experimental Psychology: Learning, Memory, and Cognition,

37(6), 1342–1370.

Shivde, G., & Thompson-Schill, S. (2004). Dissociating semantic and phonological maintenance using

fMRI. Cognitive, Affective, & Behavioral Neuroscience, 4(1), 10–19.

176
 References

Shulman, H. (1970). Similarity effects in short-term memory. Psychological Bulletin, 75, 399–414.

Silva-Pereyra, J., Rivera-Gaxiola, M., Aubert, E., Bosch, J., Galán, L., & Salazar, A. (2003). N400 during

lexical decision tasks: a current source localization study. Clinical Neurophysiology, 114(12),

2469–2486.

Simos, P., Basile, L., & Papanicolaou, A. (1997). Source localization of the N400 response in a sentence-

reading paradigm using evoked magnetic fields and magnetic resonance imaging. Brain Research,

762(1–2), 29–39.

Sladky, R., Friston, K., Tröstl, J., Cunnington, R., Moser, E., & Windischberger, C. (2011). Slice-timing

effects and their correction in functional MRI. NeuroImage, 58(2), 588–594.

Smith, E. E., & Jonides, J. (1998). Neuroimaging analyses of human working memory. Proceedings of

the National Academy of Sciences of the USA, 95(20), 12061–12068.

Smith, E. E., & Jonides, J. (1999). Storage and executive processes in the frontal lobes. Science, 283(5408),

1657–1661.

Smith, S., Jenkinson, M., Johansen-Berg, H., Rueckert, D., Nichols, T., Mackay, C., et al. (2006). Tract-

based spatial statistics: voxelwise analysis of multi-subject diffusion data. NeuroImage, 31(4),

1487–1505.

Speckmann, E.-J., & Elger, C. (1993). Neurophysiological basis of the EEG and of DC potentials.

In E. Niedermeyer & F. Lopes da Silva (Eds.), Electroencephalography. Basic principles, clinical

application and related fields (pp. 15–26). München: Urban & Schwarzenberg.

Sternberg, S. (1966). High-speed scanning in human memory. Science, 153(3736), 652–654.

Swinney, D., Ford, M., Frauenfelder, U., & Bresnan, J. (1988). On the temporal course of gap-filling

and antecedent assignment during sentence comprehension. Unpublished manuscript.

Tabachnick, B., Fidell, L., & Osterlind, S. (2001). Using multivariate statistics. New York: Harper-

Collins.

Taglialatela, J. P., Cantalupo, C., & Hopkins, W. D. (2006). Gesture handedness predicts asymmetry in

the chimpanzee inferior frontal gyrus. NeuroReport, 17(9), 923–927.

Takao, H., Abe, O., & Ohtomo, K. (2010). Computational analysis of cerebral cortex. Neuroradiology,

52(8), 691–698.

Talairach, J., & Tournoux, P. (1988). Co-planar stereotaxic atlas of the human brain. Stuttgart: Thieme.

177
References

Tanenhaus, M., Carlson, G., & Seidenberg, M. (1985). Do listeners compute linguistic representa-

tions. In D. Dowty, L. Karttunen, & A. Zwicky (Eds.), Natural language parsing: Psychological,

computational, and theoretical perspectives (pp. 359–408). New York: Cambridge University Press.

Temperley, D. (2007). Minimization of dependency length in written English. Cognition, 105(2),

300–333.

Tewes, U. (1994). Hamburg-Wechsler-Intelligenztest für Erwachsene. Bern: Huber.

Thiebaut de Schotten, M., Dell’Acqua, F., Valabregue, R., & Catani, M. (2012). Monkey to human

comparative anatomy of the frontal lobe association tracts. Cortex, 48(1), 82–96.

Thulborn, K., Waterton, J., Matthews, P., & Radda, G. (1982). Oxygenation dependence of the

transverse relaxation time of water protons in whole blood at high field. Biochimica et Biophysica

Acta (BBA) - General Subjects, 714(2), 265–270.

Thut, G., Nietzel, A., Brandt, S., & Pascual-Leone, A. (2006). α-Band electroencephalographic activity

over occipital cortex indexes visuospatial attention bias and predicts visual target detection.

Journal of Neuroscience, 26(37), 9494–9502.

Tomasello, M., & Merriman, W. (1995). Beyond names for things: Young children’s acquisition of verbs.

Hillsdale: Lawrence Erlbaum Associates.

Townsend, D., Carrithers, C., & Bever, T. (2001). Familial handedness and access to words, meaning,

and syntax during sentence comprehension. Brain and Language, 78(3), 308–331.

Trueswell, J., & Kim, A. (1998). How to prune a garden path by nipping it in the bud: fast priming of

verb argument structure. Journal of Memory and Language, 39(1), 102–123.

Trueswell, J., Tanenhaus, M., & Kello, C. (1993). Verb-specific constraints in sentence processing:

Separating effects of lexical preference from garden-paths. Journal of Experimental Psychology:

Learning, Memory, and Cognition, 19(3), 528.

Tsuzuki, T., Uchida, T., Yukihiro, R., Hisano, M., & Tsuzuki, K. (2004). Effects of syntactic information

on semantic access of ambiguous verbs in spoken language comprehension: evidence from a

cross-modal priming experiment. Japanese Psychological Research, 46(1), 31–43.

Turi, G., Gotthardt, S., Singer, W., Anh Vuong, T., Munk, M., & Wibral, M. (2012). Quantifying

additive evoked contributions to the event-related potential. NeuroImage, 59(3), 2607–2624.

178
 References

Tyler, L., Marslen-Wilson, W., Randall, B., Wright, P., Devereux, B., Zhuang, J., et al. (2011). Left inferior

frontal cortex and syntax: function, structure and behaviour in patients with left hemisphere

damage. Brain, 134(2), 415–431.

Tyler, L., Shafto, M., Randall, B., Wright, P., Marslen-Wilson, W., & Stamatakis, E. (2010). Preserving

syntactic processing across the adult life span: the modulation of the frontotemporal language

system in the context of age-related atrophy. Cerebral Cortex, 20(2), 352–364.

Ueno, M., & Kluender, R. (2003). Event-related brain indices of Japanese scrambling. Brain and

Language, 86(2), 243–271.

Unsworth, N., & Engle, R. (2007). The nature of individual differences in working memory capac-

ity: active maintenance in primary memory and controlled search from secondary memory.

Psychological Review, 114(1), 104–132.

Van Dijk, H., Nieuwenhuis, I., & Jensen, O. (2010). Left temporal alpha band activity increases during

working memory retention of pitches. European Journal of Neuroscience, 31(9), 1701–1707.

Van Dyke, J. (2007). Interference effects from grammatically unavailable constituents during sentence

processing. Journal of Experimental Psychology: Learning, Memory, and Cognition, 33(2), 407–430.

Van Dyke, J., & McElree, B. (2006). Retrieval interference in sentence comprehension. Journal of

Memory and Language, 55(2), 157–166.

Van Petten, C. (1993). A comparison of lexical and sentence-level context effects in event-related

potentials. Language and Cognitive Processes, 8(4), 485–531.

Van Petten, C., Kutas, M., Kluender, R., Mitchiner, M., & McIsaac, H. (1991). Fractionating the word

repetition effect with event-related potentials. Journal of Cognitive Neuroscience, 3(2), 131–150.

Van Petten, C., & Luka, B. J. (2012). Prediction during language comprehension: Benefits, costs, and

ERP components. International Journal of Psychophysiology, 83(2), 176–190.

Vasishth, S. (2003). Working memory in sentence comprehension: processing Hindi center embeddings.

New York: Routledge.

Vasishth, S., & Lewis, R. (2006). Argument-head distance and processing complexity: Explaining both

locality and antilocality effects. Language, 82(4), 767–794.

Vorländer, T. (1986). Konzeption eines klinischen Kurzzeitgedächtnistests für hirngeschädigte Patienten.

Unpublished diploma thesis, Universität Trier.

179
References

Vos, S., Gunther, T., Schriefers, H., & Friederici, A. D. (2001). Syntactic parsing and working memory:

The effects of syntactic complexity, reading span, and concurrent load. Language and Cognitive

Processes, 16(1), 65–103.

Wager, T., Keller, M., Lacey, S., & Jonides, J. (2005). Increased sensitivity in neuroimaging analyses

using robust regression. NeuroImage, 26(1), 99–113.

Wager, T., & Smith, E. (2003). Neuroimaging studies of working memory: a meta-analysis. Cognitive,

Affective, & Behavioral Neuroscience, 3(4), 255–274.

Wagner, A., Koutstaal, W., Maril, A., Schacter, D., & Buckner, R. (2000). Task-specific repetition

priming in left inferior prefrontal cortex. Cerebral Cortex, 10(12), 1176–1184.

Wakana, S., Caprihan, A., Panzenboeck, M., Fallon, J., Perry, M., Gollub, R., et al. (2007). Repro-

ducibility of quantitative tractography methods applied to cerebral white matter. NeuroImage,

36(3), 630–644.

Wang, L., Jensen, O., Brink, D. van den, Weder, N., Schoffelen, J., Magyari, L., et al. (2012). Beta

oscillations relate to the N400m during language comprehension. Human Brain Mapping.

Wang, X. (2010). Neurophysiological and computational principles of cortical rhythms in cognition.

Physiological Reviews, 90(3), 1195–1268.

Wanner, E., & Maratsos, M. (1979). An ATN approach to comprehension. In M. Halle, J. Bresnan, &

G. Miller (Eds.), Linguistic theory and psychological reality (pp. 119–161). Cambridge: MIT Press.

Waszak, F., & Hommel, B. (2007). The costs and benefits of cross-task priming. Memory & Cognition,

35(5), 1175–1186.

Waters, G., & Caplan, D. (1996). The measurement of verbal working memory capacity and its relation

to reading comprehension. Quarterly Journal of Experimental Psychology, 49(1), 51–79.

Weiduschat, N., Thiel, A., Rubi-Fessen, I., Hartmann, A., Kessler, J., Merl, P., et al. (2011). Effects of

repetitive transcranial magnetic stimulation in aphasic stroke. Stroke, 42(2), 409–415.

Weiller, C., Musso, M., Rijntjes, M., & Saur, D. (2009). Please don’t underestimate the ventral pathway

in language. Trends in Cognitive Sciences, 13(9), 369–370.

Weiss, S., Mueller, H. M., Schack, B., King, J. W., Kutas, M., & Rappelsberger, P. (2005). Increased

neuronal communication accompanying sentence comprehension. International Journal of

Psychophysiology, 57(2), 129–141.

180
 References

Weisz, N., Hartmann, T., Müller, N., Lorenz, I., & Obleser, J. (2011). Alpha rhythms in audition:

cognitive and clinical perspectives. Frontiers in Psychology, 2, 73.

Welte, V. (1981). Der Mottier-Test, ein Prüfmittel für die Lautdifferenzierungsfähigkeit und die auditive

Merkfähigkeit. Sprache-Stimme-Gehör, 5(3), 121–125.

Wernicke, C. (1874). Der aphasische Symptomencomplex: eine psychologische Studie auf anatomischer Basis.

Breslau: Cohn & Weigert.

Westbury, C., Zatorre, R., & Evans, A. (1999). Quantifying variability in the planum temporale: a

probability map. Cerebral Cortex, 9(4), 392–405.

Wilson, S. M., Dronkers, N. F., Ogar, J. M., Jang, J., Growdon, M. E., Agosta, F., et al. (2010). Neural

correlates of syntactic processing in the nonfluent variant of primary progressive aphasia. Journal

of Neuroscience, 30(50), 16845–16854.

Wilson, S. M., Galantucci, S., Tartaglia, M. C., Rising, K., Patterson, D. K., Henry, M. L., et al. (2011).

Syntactic processing depends on dorsal language tracts. Neuron, 72(2), 397–403.

Wingfield, A., & Butterworth, B. (1984). Running memory for sentences and parts of sentences:

Syntactic parsing as a control function in working memory. In H. Bouma & D. Bouwhuis (Eds.),

Attention and performance X: Control of language processes (pp. 351–364). London: Lawrence

Erlbaum Associates.

Winhuisen, L., Thiel, A., Schumacher, B., Kessler, J., Rudolf, J., Haupt, W., et al. (2005). Role of the

contralateral inferior frontal gyrus in recovery of language function in poststroke aphasia. Stroke,

36(8), 1759–1763.

Winhuisen, L., Thiel, A., Schumacher, B., Kessler, J., Rudolf, J., Haupt, W., et al. (2007). The right

inferior frontal gyrus and poststroke aphasia. Stroke, 38(4), 1286–1292.

Woody, C. (1967). Characterization of an adaptive filter for the analysis of variable latency neuroelectric

signals. Medical and Biological Engineering and Computing, 5(6), 539–554.

Worsley, K. J., Andermann, M., Koulis, T., MacDonald, D., & Evans, A. C. (1999). Detecting changes

in nonisotropic images. Human Brain Mapping, 8(2–3), 98–101.

Wright, I. C., McGuire, P. K., Poline, J. B., Travere, J. M., Murray, R. M., Frith, C. D., et al. (1995).

A voxel-based method for the statistical analysis of gray and white matter density applied to

schizophrenia. NeuroImage, 2(4), 244–252.

181
References

Yamada, K., Nagakane, Y., Mizuno, T., Hosomi, A., Nakagawa, M., & Nishimura, T. (2007). MR

tractography depicting damage to the arcuate fasciculus in a patient with conduction aphasia.

Neurology, 69(3), 321.

Yamashita, H. (1997). The effects of word-order and case marking information on the processing of

Japanese. Journal of Psycholinguistic Research, 26(2), 163–188.

Yassa, M., Stark, S., Bakker, A., Albert, M., Gallagher, M., & Stark, C. (2010). High-resolution

structural and functional MRI of hippocampal CA3 and dentate gyrus in patients with amnestic

Mild Cognitive Impairment. NeuroImage, 51(3), 1242–1252.

Yngwe, V. (1960). A model and a hypothesis for language structure. Proceedings of the American

Philosophical Society, 104(5), 444–466.

Zurif, E., & Piñango, M. (1999). The existence of comprehension patterns in Broca’s aphasia. Brain

and Language, 70(1), 133–138.

182
LIST OF FIGURES

2.1 Segmentation, template-generation, and co-registration steps in the VBM pipeline . . . . 16

2.2 Overview of DWI analysis steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

2.3 Overview of time–frequency analysis steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

2.4 Overview of experimental materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

3.1 Average ratings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

3.2 Position-wise grand-average reading times . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

4.1 D" -scores and reaction times for all four conditions . . . . . . . . . . . . . . . . . . . . . . . . 49

4.2 Brain activations and signal change for the reordering effect and the storage effect . . . . 50

4.3 Negative correlation of storage effect in the TP region with combined digit span . . . . 51

4.4 Individual brain activations during the contrast object-first > subject-first . . . . . . . . . 52

4.5 Correlation between functional and structural lateralization indices . . . . . . . . . . . . . 53

4.6 Correlation of FA values with signal change . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

4.7 Overview of critical group results from cited studies on verbal working memory . . . . 58

4.8 Boxplots of functional and structural asymmetry, median-split by LQ scores . . . . . . . 63

5.1 Anatomical scan of patient’s brain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

5.2 Working-memory-test-battery results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

5.3 Sentence-processing results from the patient study. . . . . . . . . . . . . . . . . . . . . . . . . 76

5.4 Results of DTI procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

6.1 Time–frequency results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91

185
List of Figures

6.2 Comparison between the short and long conditions in source space . . . . . . . . . . . . . 92

6.3 Statistical maps of the correlation analysis between alpha power and reading span . . . . 92

7.1 Working-memory sub-components of argument–verb-dependency processing . . . . . . 103

7.2 Brain activations for the sub-sample fMRI main effects . . . . . . . . . . . . . . . . . . . . . 109

7.3 Results of the combined fMRI–EEG analyses: sensor space . . . . . . . . . . . . . . . . . . . 111

7.4 Results of the combined fMRI–EEG analyses: source space . . . . . . . . . . . . . . . . . . . 112

8.1 Results of the interference pilot study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124

8.2 Contrast between behavioral results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130

9.1 Functional neuroanatomy of storage, retrieval, and reordering . . . . . . . . . . . . . . . . 136

9.2 Paradigm proposal for modulating TP-region activation by attention . . . . . . . . . . . . 138

9.3 Paradigm proposal for dissociating reordering and rehearsal . . . . . . . . . . . . . . . . . . 139

186
L I S T O F TA B L E S

3.1 Significant position-wise RT effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

4.1 Overview of significant clusters in the functional contrasts . . . . . . . . . . . . . . . . . . . 50

4.2 Overview of significant clusters in the contrasts on the FA values . . . . . . . . . . . . . . 53

5.1 Working-memory test results from the patient study . . . . . . . . . . . . . . . . . . . . . . . 76

5.2 Sentence-processing results from patient study . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

7.1 List of significant clusters in the fMRI contrasts . . . . . . . . . . . . . . . . . . . . . . . . . . 110

189
AB B R E V I AT I O N S

AC-PC intercommissural line

AF arcuate fasciculus

ANOVA analysis of variance

BA Brodmann area

BEM boundary-element model

BOLD blood-oxygen-level-dependent

CI confidence interval

CRT cathode-ray tube

DARTEL diffeomorphic anatomical registration using exponentiated lie algebra

DC direct current

dMRI diffusion magnetic resonance imaging

DTI diffusion-tensor imaging

EEG electroencephalography

EOG electrooculogram

EPI echo planar imaging

ERP event-related brain potential

FA fractional anisotropy

FDR false-discovery rate

FIR finite impulse response

fMRI functional magnetic resonance imaging

FOV field of view

191
Abbreviations

GLM general linear model

GRAPPA generalized auto-calibrating partially-parallel acquisition

HRF hemodynamic response function

ICA independent-component analysis

IFG inferior frontal gyrus

LCD liquid crystal display

LQ laterality quotient

MEG magnetoencephalography

MNI Montreal Neurological Institute

MP-RAGE magnetization-prepared rapid gradient echo

MR magnetic resonance

MRI magnetic resonance imaging

PT planum temporale

ROI region of interest

RT reaction time

rTMS repetitive transcranial magnetic stimulation

SD standard deviation

SEM standard error of the mean

SLF superior longitudinal fasciculus

SMG supramarginal gyrus

SNR signal-to-noise ratio

SPR self-paced reading

STG superior temporal gyrus

TBSS tract-based spatial statistics

TE echo time

TFR time–frequency analysis

TI inversion time

TMS transcranial magnetic stimulation

TP temporo-parietal

192
 Abbreviations

TPM tissue probability map

TR repetition time

VB-ECD variational-Bayesian equivalent-current dipole

VBM voxel-based morphometry

VGA video graphics array

XGA extended graphics array

193
A
S T I M U L I F R O M B E H AV I O R A L , F M R I A N D E EG S T U D I E S *

1. SF–SD Während der Sitzung in der Hauptstadt hat der Minister den Sprecher getroffen und das Ergebnis berichtet.
SF–LD Der Minister hat während der Sitzung in der Hauptstadt den Sprecher getroffen und das Ergebnis berichtet.
OF–SD Während der Sitzung in der Hauptstadt hat den Sprecher der Minister getroffen und das Ergebnis berichtet.
OF–LD Den Sprecher hat während der Sitzung in der Hauptstadt der Minister getroffen und das Ergebnis berichtet.

2. SF–SD Nach einer Saison in der Bundesliga hat der Trainer den Stürmer gewürdigt und die Entwicklung bestätigt.
SF–LD Der Trainer hat nach einer Saison in der Bundesliga den Stürmer gewürdigt und die Entwicklung bestätigt.
OF–SD Nach einer Saison in der Bundesliga hat den Stürmer der Trainer gewürdigt und die Entwicklung bestätigt.
OF–LD Den Stürmer hat nach einer Saison in der Bundesliga der Trainer gewürdigt und die Entwicklung bestätigt.

3. SF–SD In der Verhandlung mit der Gewerkschaft hat der Arbeitgeber den Mitarbeiter angehört und den Arbeitskampf vermieden.
SF–LD Der Arbeitgeber hat in der Verhandlung mit der Gewerkschaft den Mitarbeiter angehört und den Arbeitskampf vermieden.
OF–SD In der Verhandlung mit der Gewerkschaft hat den Mitarbeiter der Arbeitgeber angehört und den Arbeitskampf vermieden.
OF–LD Den Mitarbeiter hat in der Verhandlung mit der Gewerkschaft der Arbeitgeber angehört und den Arbeitskampf vermieden.

4. SF–SD Vor der Veranstaltung für die Elternschaft hat der Lehrer den Hausmeister überzeugt und das Klassenzimmer vorbereitet.
SF–LD Der Lehrer hat vor der Veranstaltung für die Elternschaft den Hausmeister überzeugt und das Klassenzimmer vorbereitet.
OF–SD Vor der Veranstaltung für die Elternschaft hat den Hausmeister der Lehrer überzeugt und das Klassenzimmer vorbereitet.
OF–LD Den Hausmeister hat vor der Veranstaltung für die Elternschaft der Lehrer überzeugt und das Klassenzimmer vorbereitet.

5. SF–SD Wegen der Leistung in der Berufsschule hat der Meister den Lehrling entlassen und die Entscheidung bedauert.
SF–LD Der Meister hat wegen der Leistung in der Berufsschule den Lehrling entlassen und die Entscheidung bedauert.
OF–SD Wegen der Leistung in der Berufsschule hat den Lehrling der Meister entlassen und die Entscheidung bedauert.
OF–LD Den Lehrling hat wegen der Leistung in der Berufsschule der Meister entlassen und die Entscheidung bedauert.

6. SF–SD Vor der Lage an dem Fachbereich hat der Professor den Doktor unterstützt und das Projekt fortgeführt.
SF–LD Der Professor hat vor der Lage an dem Fachbereich den Doktor unterstützt und das Projekt fortgeführt.
OF–SD Vor der Lage an dem Fachbereich hat den Doktor der Professor unterstützt und das Projekt fortgeführt.
OF–LD Den Doktor hat vor der Lage an dem Fachbereich der Professor unterstützt und das Projekt fortgeführt.

7. SF–SD Nach dem Trainingslager mit dem Fußballverein hat der Spieler den Torwart verabschiedet und die Ehefrau begrüßt.
SF–LD Der Spieler hat nach dem Trainingslager mit dem Fußballverein den Torwart verabschiedet und die Ehefrau begrüßt.
OF–SD Nach dem Trainingslager mit dem Fußballverein hat den Torwart der Spieler verabschiedet und die Ehefrau begrüßt.
OF–LD Den Torwart hat nach dem Trainingslager mit dem Fußballverein der Spieler verabschiedet und die Ehefrau begrüßt.

*
Subject-first (SF), object-first (OF), short distance (SD), long distance (LD).

A-1
Appendix A. Stimuli from Behavioral, fMRI and EEG Studies

8. SF–SD Während des Prozesses an dem Gericht hat der Richter den Zeugen vernommen und den Zusammenhang aufgeklärt.
SF–LD Der Richter hat während des Prozesses an dem Gericht den Zeugen vernommen und den Zusammenhang aufgeklärt.
OF–SD Während des Prozesses an dem Gericht hat den Zeugen der Richter vernommen und den Zusammenhang aufgeklärt.
OF–LD Den Zeugen hat während des Prozesses an dem Gericht der Richter vernommen und den Zusammenhang aufgeklärt.

9. SF–SD Nach dem Unfall auf der Landstraße hat der Notarzt den Verletzten behandelt und die Diagnose gestellt.
SF–LD Der Notarzt hat nach dem Unfall auf der Landstraße den Verletzten behandelt und die Diagnose gestellt.
OF–SD Nach dem Unfall auf der Landstraße hat den Verletzten der Notarzt behandelt und die Diagnose gestellt.
OF–LD Den Verletzten hat nach dem Unfall auf der Landstraße der Notarzt behandelt und die Diagnose gestellt.

10. SF–SD An dem Jahrestag für die Belegschaft hat der Firmenchef den Arbeiter befördert und die Anstrengung gelobt.
SF–LD Der Firmenchef hat an dem Jahrestag für die Belegschaft den Arbeiter befördert und die Anstrengung gelobt.
OF–SD An dem Jahrestag für die Belegschaft hat den Arbeiter der Firmenchef befördert und die Anstrengung gelobt.
OF–LD Den Arbeiter hat an dem Jahrestag für die Belegschaft der Firmenchef befördert und die Anstrengung gelobt.

11. SF–SD Während des Wettbewerbs in der Sporthalle hat der Athlet den Funktionär attackiert und den Leistungssport erschüttert.
SF–LD Der Athlet hat während des Wettbewerbs in der Sporthalle den Funktionär attackiert und den Leistungssport erschüttert.
OF–SD Während des Wettbewerbs in der Sporthalle hat den Funktionär der Athlet attackiert und den Leistungssport erschüttert.
OF–LD Den Funktionär hat während des Wettbewerbs in der Sporthalle der Athlet attackiert und den Leistungssport erschüttert.

12. SF–SD In der Kontroverse wegen des Parteitags hat der Demokrat den Delegierten ausgelacht und die Parteiführung gestärkt.
SF–LD Der Demokrat hat in der Kontroverse wegen des Parteitags den Delegierten ausgelacht und die Parteiführung gestärkt.
OF–SD In der Kontroverse wegen des Parteitags hat den Delegierten der Demokrat ausgelacht und die Parteiführung gestärkt.
OF–LD Den Delegierten hat in der Kontroverse wegen des Parteitags der Demokrat ausgelacht und die Parteiführung gestärkt.

13. SF–SD Vor der Kundgebung gegen den Straßenbau hat der Investor den Bauleiter informiert und die Befürchtungen geschildert.
SF–LD Der Investor hat vor der Kundgebung gegen den Straßenbau den Bauleiter informiert und die Befürchtungen geschildert.
OF–SD Vor der Kundgebung gegen den Straßenbau hat den Bauleiter der Investor informiert und die Befürchtungen geschildert.
OF–LD Den Bauleiter hat vor der Kundgebung gegen den Straßenbau der Investor informiert und die Befürchtungen geschildert.

14. SF–SD Nach dem Essen in der Kantine hat den Chefkoch der Gourmet kritisiert und die Speisekarte bemängelt.
SF–LD Der Gourmet hat nach dem Essen in der Kantine den Chefkoch kritisiert und die Speisekarte bemängelt.
OF–SD Nach dem Essen in der Kantine hat der Gourmet den Chefkoch kritisiert und die Speisekarte bemängelt.
OF–LD Den Chefkoch hat nach dem Essen in der Kantine der Gourmet kritisiert und die Speisekarte bemängelt.

15. SF–SD Bei dem Kampf an der Grenze hat der General den Offizier geopfert und die Armeeführung empört.
SF–LD Der General hat bei dem Kampf an der Grenze den Offizier geopfert und die Armeeführung empört.
OF–SD Bei dem Kampf an der Grenze hat den Offizier der General geopfert und die Armeeführung empört.
OF–LD Den Offizier hat bei dem Kampf an der Grenze der General geopfert und die Armeeführung empört.

16. SF–SD Während der Diskussion zwischen den Künstlern hat der Bildhauer den Maler geachtet und die Auffassung geteilt.
SF–LD Der Bildhauer hat während der Diskussion zwischen den Künstlern den Maler geachtet und die Auffassung geteilt.
OF–SD Während der Diskussion zwischen den Künstlern hat den Maler der Bildhauer geachtet und die Auffassung geteilt.
OF–LD Den Maler hat während der Diskussion zwischen den Künstlern der Bildhauer geachtet und die Auffassung geteilt.

17. SF–SD Vor dem Interview mit der Zeitung hat der Präsident den Journalist empfangen und die Reform dargestellt.
SF–LD Der Präsident hat vor dem Interview mit der Zeitung den Journalist empfangen und die Reform dargestellt.
OF–SD Vor dem Interview mit der Zeitung hat den Journalist der Präsident empfangen und die Reform dargestellt.
OF–LD Den Journalist hat vor dem Interview mit der Zeitung der Präsident empfangen und die Reform dargestellt.

18. SF–SD Vor dem Vortrag an der Universität hat der Wissenschaftler den Experten aufgesucht und die Methode verteidigt.
SF–LD Der Wissenschaftler hat vor dem Vortrag an der Universität den Experten aufgesucht und die Methode verteidigt.
OF–SD Vor dem Vortrag an der Universität hat den Experten der Wissenschaftler aufgesucht und die Methode verteidigt.
OF–LD Den Experten hat vor dem Vortrag an der Universität der Wissenschaftler aufgesucht und die Methode verteidigt.

A-2
 Appendix A. Stimuli from Behavioral, fMRI and EEG Studies

19. SF–SD Nach dem Skandal auf dem Bauernhof hat der Naturschützer den Landwirt beleidigt und die Schließung gefordert.
SF–LD Der Naturschützer hat nach dem Skandal auf dem Bauernhof den Landwirt beleidigt und die Schließung gefordert.
OF–SD Nach dem Skandal auf dem Bauernhof hat den Landwirt der Naturschützer beleidigt und die Schließung gefordert.
OF–LD Den Landwirt hat nach dem Skandal auf dem Bauernhof der Naturschützer beleidigt und die Schließung gefordert.

20. SF–SD Wegen des Zustands in der Abteilung hat der Kollege den Personalrat benachrichtigt und die Wahrheit eröffnet.
SF–LD Der Kollege hat wegen des Zustands in der Abteilung den Personalrat benachrichtigt und die Wahrheit eröffnet.
OF–SD Wegen des Zustands in der Abteilung hat den Personalrat der Kollege benachrichtigt und die Wahrheit eröffnet.
OF–LD Den Personalrat hat wegen des Zustands in der Abteilung der Kollege benachrichtigt und die Wahrheit eröffnet.

21. SF–SD Nach der Halbzeit in dem Länderspiel hat der Schiedsrichter den Reporter abgewiesen und die Stellungnahme verweigert.
SF–LD Der Schiedsrichter hat nach der Halbzeit in dem Länderspiel den Reporter abgewiesen und die Stellungnahme verweigert.
OF–SD Nach der Halbzeit in dem Länderspiel hat den Reporter der Schiedsrichter abgewiesen und die Stellungnahme verweigert.
OF–LD Den Reporter hat nach der Halbzeit in dem Länderspiel der Schiedsrichter abgewiesen und die Stellungnahme verweigert.

22. SF–SD Bei dem Anschlag auf die Botschaft hat der Terrorist den Diplomat ermordet und das Gebäude gesprengt.
SF–LD Der Terrorist hat bei dem Anschlag auf die Botschaft den Diplomat ermordet und das Gebäude gesprengt.
OF–SD Bei dem Anschlag auf die Botschaft hat den Diplomat der Terrorist ermordet und das Gebäude gesprengt.
OF–LD Den Diplomat hat bei dem Anschlag auf die Botschaft der Terrorist ermordet und das Gebäude gesprengt.

23. SF–SD Vor dem Ausflug an die Küste hat der Reiseleiter den Busfahrer begrüßt und das Fahrzeug überprüft.
SF–LD Der Reiseleiter hat vor dem Ausflug an die Küste den Busfahrer begrüßt und das Fahrzeug überprüft.
OF–SD Vor dem Ausflug an die Küste hat den Busfahrer der Reiseleiter begrüßt und das Fahrzeug überprüft.
OF–LD Den Busfahrer hat vor dem Ausflug an die Küste der Reiseleiter begrüßt und das Fahrzeug überprüft.

24. SF–SD In der Vorstellung mit dem Orchester hat der Dirigent den Pianist irritiert und das Konzert unterbrochen.
SF–LD Der Dirigent hat in der Vorstellung mit dem Orchester den Pianist irritiert und das Konzert unterbrochen.
OF–SD In der Vorstellung mit dem Orchester hat den Pianist der Dirigent irritiert und das Konzert unterbrochen.
OF–LD Den Pianist hat in der Vorstellung mit dem Orchester der Dirigent irritiert und das Konzert unterbrochen.

25. SF–SD Während der Besprechung in der Firma hat der Geschäftsführer den Leiter berufen und den Erfolg gesichert.
SF–LD Der Geschäftsführer hat während der Besprechung in der Firma den Leiter berufen und den Erfolg gesichert.
OF–SD Während der Besprechung in der Firma hat den Leiter der Geschäftsführer berufen und den Erfolg gesichert.
OF–LD Den Leiter hat während der Besprechung in der Firma der Geschäftsführer berufen und den Erfolg gesichert.

26. SF–SD Nach der Krise wegen des Wahlversprechens hat der Redakteur den Kanzler verurteilt und den Neuanfang verlangt.
SF–LD Der Redakteur hat nach der Krise wegen des Wahlversprechens den Kanzler verurteilt und den Neuanfang verlangt.
OF–SD Nach der Krise wegen des Wahlversprechens hat den Kanzler der Redakteur verurteilt und den Neuanfang verlangt.
OF–LD Den Kanzler hat nach der Krise wegen des Wahlversprechens der Redakteur verurteilt und den Neuanfang verlangt.

27. SF–SD In der Debatte um die Finanzierung hat der Befürworter den Gegner überzeugt und die Niederlage vermieden.
SF–LD Der Befürworter hat in der Debatte um die Finanzierung den Gegner überzeugt und die Niederlage vermieden.
OF–SD In der Debatte um die Finanzierung hat den Gegner der Befürworter überzeugt und die Niederlage vermieden.
OF–LD Den Gegner hat in der Debatte um die Finanzierung der Befürworter überzeugt und die Niederlage vermieden.

28. SF–SD Vor der Ausstellung in der Galerie hat der Kunde den Maler angerufen und die Preise erfragt.
SF–LD Der Kunde hat vor der Ausstellung in der Galerie den Maler angerufen und die Preise erfragt.
OF–SD Vor der Ausstellung in der Galerie hat den Maler der Kunde angerufen und die Preise erfragt.
OF–LD Den Maler hat vor der Ausstellung in der Galerie der Kunde angerufen und die Preise erfragt.

29. SF–SD Bei dem Treffen über die Ausbildung hat der Bewerber den Direktor beeindruckt und die Stelle bekommen.
SF–LD Der Bewerber hat bei dem Treffen über die Ausbildung den Direktor beeindruckt und die Stelle bekommen.
OF–SD Bei dem Treffen über die Ausbildung hat den Direktor der Bewerber beeindruckt und die Stelle bekommen.
OF–LD Den Direktor hat bei dem Treffen über die Ausbildung der Bewerber beeindruckt und die Stelle bekommen.

A-3
Appendix A. Stimuli from Behavioral, fMRI and EEG Studies

30. SF–SD Vor der Lesung in der Buchhandlung hat der Verleger den Schriftsteller vorgestellt und den Roman präsentiert.
SF–LD Der Verleger hat vor der Lesung in der Buchhandlung den Schriftsteller vorgestellt und den Roman präsentiert.
OF–SD Vor der Lesung in der Buchhandlung hat den Schriftsteller der Verleger vorgestellt und den Roman präsentiert.
OF–LD Den Schriftsteller hat vor der Lesung in der Buchhandlung der Verleger vorgestellt und den Roman präsentiert.

31. SF–SD Wegen der Erfahrung in der Klinik hat der Pfleger den Patienten gemieden und die Station verlassen.
SF–LD Der Pfleger hat wegen der Erfahrung in der Klinik den Patienten gemieden und die Station verlassen.
OF–SD Wegen der Erfahrung in der Klinik hat den Patienten der Pfleger gemieden und die Station verlassen.
OF–LD Den Patienten hat wegen der Erfahrung in der Klinik der Pfleger gemieden und die Station verlassen.

32. SF–SD Während der Tagung an der Hochschule hat der Forscher den Besucher begeistert und das Fachpublikum enttäuscht.
SF–LD Der Forscher hat während der Tagung an der Hochschule den Besucher begeistert und das Fachpublikum enttäuscht.
OF–SD Während der Tagung an der Hochschule hat den Besucher der Forscher begeistert und das Fachpublikum enttäuscht.
OF–LD Den Besucher hat während der Tagung an der Hochschule der Forscher begeistert und das Fachpublikum enttäuscht.

33. SF–SD Nach der Begegnung in dem Studio hat der Regisseur den Schauspieler engagiert und das Drehbuch geändert.
SF–LD Der Regisseur hat nach der Begegnung in dem Studio den Schauspieler engagiert und das Drehbuch geändert.
OF–SD Nach der Begegnung in dem Studio hat den Schauspieler der Regisseur engagiert und das Drehbuch geändert.
OF–LD Den Schauspieler hat nach der Begegnung in dem Studio der Regisseur engagiert und das Drehbuch geändert.

34. SF–SD Bei dem Jubiläum in dem Finanzamt hat der Vorgänger den Nachfolger eingeführt und die Aufgaben erklärt.
SF–LD Der Vorgänger hat bei dem Jubiläum in dem Finanzamt den Nachfolger eingeführt und die Aufgaben erklärt.
OF–SD Bei dem Jubiläum in dem Finanzamt hat den Nachfolger der Vorgänger eingeführt und die Aufgaben erklärt.
OF–LD Den Nachfolger hat bei dem Jubiläum in dem Finanzamt der Vorgänger eingeführt und die Aufgaben erklärt.

35. SF–SD Bei dem Sommerfest in der Siedlung hat der Nachbar den Bürgermeister erwartet und den Empfang vorbereitet.
SF–LD Der Nachbar hat bei dem Sommerfest in der Siedlung den Bürgermeister erwartet und den Empfang vorbereitet.
OF–SD Bei dem Sommerfest in der Siedlung hat den Bürgermeister der Nachbar erwartet und den Empfang vorbereitet.
OF–LD Den Bürgermeister hat bei dem Sommerfest in der Siedlung der Nachbar erwartet und den Empfang vorbereitet.

36. SF–SD Nach der Ansprache an das Publikum hat der Leser den Autor bewundert und die Geschichte verschlungen.
SF–LD Der Leser hat nach der Ansprache an das Publikum den Autor bewundert und die Geschichte verschlungen.
OF–SD Nach der Ansprache an das Publikum hat den Autor der Leser bewundert und die Geschichte verschlungen.
OF–LD Den Autor hat nach der Ansprache an das Publikum der Leser bewundert und die Geschichte verschlungen.

37. SF–SD Vor dem Antrag gegen das Parlament hat der Politiker den Konkurrenten gefürchtet und die Wiederwahl bezweifelt.
SF–LD Der Politiker hat vor dem Antrag gegen das Parlament den Konkurrenten gefürchtet und die Wiederwahl bezweifelt.
OF–SD Vor dem Antrag gegen das Parlament hat den Konkurrenten der Politiker gefürchtet und die Wiederwahl bezweifelt.
OF–LD Den Konkurrenten hat vor dem Antrag gegen das Parlament der Politiker gefürchtet und die Wiederwahl bezweifelt.

38. SF–SD Vor der Übereinkunft mit den Eltern hat der Jugendliche den Erzieher ignoriert und das Jugendamt beunruhigt.
SF–LD Der Jugendliche hat vor der Übereinkunft mit den Eltern den Erzieher ignoriert und das Jugendamt beunruhigt.
OF–SD Vor der Übereinkunft mit den Eltern hat den Erzieher der Jugendliche ignoriert und das Jugendamt beunruhigt.
OF–LD Den Erzieher hat vor der Übereinkunft mit den Eltern der Jugendliche ignoriert und das Jugendamt beunruhigt.

39. SF–SD Während der Planung für die Initiative hat der Veranstalter den Teilnehmer beruhigt und die Aufregung verringert.
SF–LD Der Veranstalter hat während der Planung für die Initiative den Teilnehmer beruhigt und die Aufregung verringert.
OF–SD Während der Planung für die Initiative hat den Teilnehmer der Veranstalter beruhigt und die Aufregung verringert.
OF–LD Den Teilnehmer hat während der Planung für die Initiative der Veranstalter beruhigt und die Aufregung verringert.

40. SF–SD Wegen der Probleme bei der Produktion hat der Unternehmer den Lieferanten gewechselt und die Qualität verbessert.
SF–LD Der Unternehmer hat wegen der Probleme bei der Produktion den Lieferanten gewechselt und die Qualität verbessert.
OF–SD Wegen der Probleme bei der Produktion hat den Lieferanten der Unternehmer gewechselt und die Qualität verbessert.
OF–LD Den Lieferanten hat wegen der Probleme bei der Produktion der Unternehmer gewechselt und die Qualität verbessert.

A-4
 Appendix A. Stimuli from Behavioral, fMRI and EEG Studies

41. SF–SD Nach der Ermittlung in dem Milieu hat der Polizist den Verbrecher beschuldigt und den Verdacht erhärtet.
SF–LD Der Polizist hat nach der Ermittlung in dem Milieu den Verbrecher beschuldigt und den Verdacht erhärtet.
OF–SD Nach der Ermittlung in dem Milieu hat den Verbrecher der Polizist beschuldigt und den Verdacht erhärtet.
OF–LD Den Verbrecher hat nach der Ermittlung in dem Milieu der Polizist beschuldigt und den Verdacht erhärtet.

42. SF–SD Mit der Mitteilung wegen des Zeugnisses hat der Schulleiter den Abiturient erschrocken und die Einschätzung unterstrichen.
SF–LD Der Schulleiter hat mit der Mitteilung wegen des Zeugnisses den Abiturient erschrocken und die Einschätzung unterstrichen.
OF–SD Mit der Mitteilung wegen des Zeugnisses hat den Abiturient der Schulleiter erschrocken und die Einschätzung unterstrichen.
OF–LD Den Abiturient hat mit der Mitteilung wegen des Zeugnisses der Schulleiter erschrocken und die Einschätzung unterstrichen.

43. SF–SD Mit der Premiere in dem Theater hat der Dichter den Kritiker überrascht und die Ansprüche erfüllt.
SF–LD Der Dichter hat mit der Premiere in dem Theater den Kritiker überrascht und die Ansprüche erfüllt.
OF–SD Mit der Premiere in dem Theater hat den Kritiker der Dichter überrascht und die Ansprüche erfüllt.
OF–LD Den Kritiker hat mit der Premiere in dem Theater der Dichter überrascht und die Ansprüche erfüllt.

44. SF–SD Wegen der Bilanz an der Börse hat der Beamte den Sekretär verhaftet und die Festnahme verteidigt.
SF–LD Der Beamte hat wegen der Bilanz an der Börse den Sekretär verhaftet und die Festnahme verteidigt.
OF–SD Wegen der Bilanz an der Börse hat den Sekretär der Beamte verhaftet und die Festnahme verteidigt.
OF–LD Den Sekretär hat wegen der Bilanz an der Börse der Beamte verhaftet und die Festnahme verteidigt.

45. SF–SD Nach dem Urteil in dem Verfahren hat der Anwalt den Angeklagten aufgegeben und den Widerstand beendet.
SF–LD Der Anwalt hat nach dem Urteil in dem Verfahren den Angeklagten aufgegeben und den Widerstand beendet.
OF–SD Nach dem Urteil in dem Verfahren hat den Angeklagten der Anwalt aufgegeben und den Widerstand beendet.
OF–LD Den Angeklagten hat nach dem Urteil in dem Verfahren der Anwalt aufgegeben und den Widerstand beendet.

46. SF–SD Vor der Entlassung aus dem Gefängnis hat der Psychologe den Häftling untersucht und die Einstellung überprüft.
SF–LD Der Psychologe hat vor der Entlassung aus dem Gefängnis den Häftling untersucht und die Einstellung überprüft.
OF–SD Vor der Entlassung aus dem Gefängnis hat den Häftling der Psychologe untersucht und die Einstellung überprüft.
OF–LD Den Häftling hat vor der Entlassung aus dem Gefängnis der Psychologe untersucht und die Einstellung überprüft.

47. SF–SD Wegen des Verlusts bei den Geschäften hat der Kaufmann den Bankier kontaktiert und den Kredit aufgenommen.
SF–LD Der Kaufmann hat wegen des Verlusts bei den Geschäften den Bankier kontaktiert und den Kredit aufgenommen.
OF–SD Wegen des Verlusts bei den Geschäften hat den Bankier der Kaufmann kontaktiert und den Kredit aufgenommen.
OF–LD Den Bankier hat wegen des Verlusts bei den Geschäften der Kaufmann kontaktiert und den Kredit aufgenommen.

48. SF–SD Nach der Aussprache mit der Familie hat der Bruder den Vater angelächelt und den Kompromiss akzeptiert.
SF–LD Der Bruder hat nach der Aussprache mit der Familie den Vater angelächelt und den Kompromiss akzeptiert.
OF–SD Nach der Aussprache mit der Familie hat den Vater der Bruder angelächelt und den Kompromiss akzeptiert.
OF–LD Den Vater hat nach der Aussprache mit der Familie der Bruder angelächelt und den Kompromiss akzeptiert.

A-5
B
S T I M U L I F R O M PA T I E N T S T U D Y †

1. SF–SD Während der Sitzung in der Hauptstadt hat der Minister den Sprecher getroffen.
SF–LD Der Minister hat während der Sitzung in der Hauptstadt den Sprecher getroffen.
OF–SD Während der Sitzung in der Hauptstadt hat den Sprecher der Minister getroffen.
OF–LD Den Sprecher hat während der Sitzung in der Hauptstadt der Minister getroffen.

2. SF–SD Nach einer Saison in der Bundesliga hat der Trainer den Stürmer gewürdigt.
SF–LD Der Trainer hat nach einer Saison in der Bundesliga den Stürmer gewürdigt.
OF–SD Nach einer Saison in der Bundesliga hat den Stürmer der Trainer gewürdigt.
OF–LD Den Stürmer hat nach einer Saison in der Bundesliga der Trainer gewürdigt.

3. SF–SD In der Verhandlung mit der Gewerkschaft hat der Arbeitgeber den Mitarbeiter angehört.
SF–LD Der Arbeitgeber hat in der Verhandlung mit der Gewerkschaft den Mitarbeiter angehört.
OF–SD In der Verhandlung mit der Gewerkschaft hat den Mitarbeiter der Arbeitgeber angehört.
OF–LD Den Mitarbeiter hat in der Verhandlung mit der Gewerkschaft der Arbeitgeber angehört.

4. SF–SD Vor der Veranstaltung für die Elternschaft hat der Lehrer den Hausmeister überzeugt.
SF–LD Der Lehrer hat vor der Veranstaltung für die Elternschaft den Hausmeister überzeugt.
OF–SD Vor der Veranstaltung für die Elternschaft hat den Hausmeister der Lehrer überzeugt.
OF–LD Den Hausmeister hat vor der Veranstaltung für die Elternschaft der Lehrer überzeugt.

5. SF–SD Wegen der Leistung in der Berufsschule hat der Meister den Lehrling entlassen.
SF–LD Der Meister hat wegen der Leistung in der Berufsschule den Lehrling entlassen.
OF–SD Wegen der Leistung in der Berufsschule hat den Lehrling der Meister entlassen.
OF–LD Den Lehrling hat wegen der Leistung in der Berufsschule der Meister entlassen.

6. SF–SD Vor der Lage an dem Fachbereich hat der Professor den Doktor unterstützt.
SF–LD Der Professor hat vor der Lage an dem Fachbereich den Doktor unterstützt.
OF–SD Vor der Lage an dem Fachbereich hat den Doktor der Professor unterstützt.
OF–LD Den Doktor hat vor der Lage an dem Fachbereich der Professor unterstützt.

7. SF–SD Nach dem Trainingslager mit dem Fußballverein hat der Spieler den Torwart verabschiedet.
SF–LD Der Spieler hat nach dem Trainingslager mit dem Fußballverein den Torwart verabschiedet.
OF–SD Nach dem Trainingslager mit dem Fußballverein hat den Torwart der Spieler verabschiedet.
OF–LD Den Torwart hat nach dem Trainingslager mit dem Fußballverein der Spieler verabschiedet.

†
Subject-first (SF), object-first (OF), short distance (SD), long distance (LD).

B-7
Appendix B. Stimuli from Patient Study

8. SF–SD Während des Prozesses an dem Gericht hat der Richter den Zeugen vernommen.
SF–LD Der Richter hat während des Prozesses an dem Gericht den Zeugen vernommen.
OF–SD Während des Prozesses an dem Gericht hat den Zeugen der Richter vernommen.
OF–LD Den Zeugen hat während des Prozesses an dem Gericht der Richter vernommen.

9. SF–SD Nach dem Unfall auf der Landstraße hat der Notarzt den Verletzten behandelt.
SF–LD Der Notarzt hat nach dem Unfall auf der Landstraße den Verletzten behandelt.
OF–SD Nach dem Unfall auf der Landstraße hat den Verletzten der Notarzt behandelt.
OF–LD Den Verletzten hat nach dem Unfall auf der Landstraße der Notarzt behandelt.

10. SF–SD An dem Jahrestag für die Belegschaft hat der Firmenchef den Arbeiter befördert.
SF–LD Der Firmenchef hat an dem Jahrestag für die Belegschaft den Arbeiter befördert.
OF–SD An dem Jahrestag für die Belegschaft hat den Arbeiter der Firmenchef befördert.
OF–LD Den Arbeiter hat an dem Jahrestag für die Belegschaft der Firmenchef befördert.

11. SF–SD Während des Wettbewerbs in der Sporthalle hat der Athlet den Funktionär attackiert.
SF–LD Der Athlet hat während des Wettbewerbs in der Sporthalle den Funktionär attackiert.
OF–SD Während des Wettbewerbs in der Sporthalle hat den Funktionär der Athlet attackiert.
OF–LD Den Funktionär hat während des Wettbewerbs in der Sporthalle der Athlet attackiert.

12. SF–SD In der Kontroverse wegen des Parteitags hat der Demokrat den Delegierten ausgelacht.
SF–LD Der Demokrat hat in der Kontroverse wegen des Parteitags den Delegierten ausgelacht.
OF–SD In der Kontroverse wegen des Parteitags hat den Delegierten der Demokrat ausgelacht.
OF–LD Den Delegierten hat in der Kontroverse wegen des Parteitags der Demokrat ausgelacht.

13. SF–SD Vor der Kundgebung gegen den Straßenbau hat der Investor den Bauleiter informiert.
SF–LD Der Investor hat vor der Kundgebung gegen den Straßenbau den Bauleiter informiert.
OF–SD Vor der Kundgebung gegen den Straßenbau hat den Bauleiter der Investor informiert.
OF–LD Den Bauleiter hat vor der Kundgebung gegen den Straßenbau der Investor informiert.

14. SF–SD Nach dem Essen in der Kantine hat den Chefkoch der Gourmet kritisiert.
SF–LD Der Gourmet hat nach dem Essen in der Kantine den Chefkoch kritisiert.
OF–SD Nach dem Essen in der Kantine hat der Gourmet den Chefkoch kritisiert.
OF–LD Den Chefkoch hat nach dem Essen in der Kantine der Gourmet kritisiert.

15. SF–SD Bei dem Kampf an der Grenze hat der General den Offizier geopfert.
SF–LD Der General hat bei dem Kampf an der Grenze den Offizier geopfert.
OF–SD Bei dem Kampf an der Grenze hat den Offizier der General geopfert.
OF–LD Den Offizier hat bei dem Kampf an der Grenze der General geopfert.

16. SF–SD Während der Diskussion zwischen den Künstlern hat der Bildhauer den Maler geachtet.
SF–LD Der Bildhauer hat während der Diskussion zwischen den Künstlern den Maler geachtet.
OF–SD Während der Diskussion zwischen den Künstlern hat den Maler der Bildhauer geachtet.
OF–LD Den Maler hat während der Diskussion zwischen den Künstlern der Bildhauer geachtet.

17. SF–SD Vor dem Interview mit der Zeitung hat der Präsident den Journalist empfangen.
SF–LD Der Präsident hat vor dem Interview mit der Zeitung den Journalist empfangen.
OF–SD Vor dem Interview mit der Zeitung hat den Journalist der Präsident empfangen.
OF–LD Den Journalist hat vor dem Interview mit der Zeitung der Präsident empfangen.

18. SF–SD Vor dem Vortrag an der Universität hat der Wissenschaftler den Experten aufgesucht.
SF–LD Der Wissenschaftler hat vor dem Vortrag an der Universität den Experten aufgesucht.
OF–SD Vor dem Vortrag an der Universität hat den Experten der Wissenschaftler aufgesucht.
OF–LD Den Experten hat vor dem Vortrag an der Universität der Wissenschaftler aufgesucht.

B-8
 Appendix B. Stimuli from Patient Study

19. SF–SD Nach dem Skandal auf dem Bauernhof hat der Naturschützer den Landwirt beleidigt.
SF–LD Der Naturschützer hat nach dem Skandal auf dem Bauernhof den Landwirt beleidigt.
OF–SD Nach dem Skandal auf dem Bauernhof hat den Landwirt der Naturschützer beleidigt.
OF–LD Den Landwirt hat nach dem Skandal auf dem Bauernhof der Naturschützer beleidigt.

20. SF–SD Wegen des Zustands in der Abteilung hat der Kollege den Personalrat benachrichtigt.
SF–LD Der Kollege hat wegen des Zustands in der Abteilung den Personalrat benachrichtigt.
OF–SD Wegen des Zustands in der Abteilung hat den Personalrat der Kollege benachrichtigt.
OF–LD Den Personalrat hat wegen des Zustands in der Abteilung der Kollege benachrichtigt.

21. SF–SD Nach der Halbzeit in dem Länderspiel hat der Schiedsrichter den Reporter abgewiesen.
SF–LD Der Schiedsrichter hat nach der Halbzeit in dem Länderspiel den Reporter abgewiesen.
OF–SD Nach der Halbzeit in dem Länderspiel hat den Reporter der Schiedsrichter abgewiesen.
OF–LD Den Reporter hat nach der Halbzeit in dem Länderspiel der Schiedsrichter abgewiesen.

22. SF–SD Bei dem Anschlag auf die Botschaft hat der Terrorist den Diplomat ermordet.
SF–LD Der Terrorist hat bei dem Anschlag auf die Botschaft den Diplomat ermordet.
OF–SD Bei dem Anschlag auf die Botschaft hat den Diplomat der Terrorist ermordet.
OF–LD Den Diplomat hat bei dem Anschlag auf die Botschaft der Terrorist ermordet.

23. SF–SD Vor dem Ausflug an die Küste hat der Reiseleiter den Busfahrer begrüßt.
SF–LD Der Reiseleiter hat vor dem Ausflug an die Küste den Busfahrer begrüßt.
OF–SD Vor dem Ausflug an die Küste hat den Busfahrer der Reiseleiter begrüßt.
OF–LD Den Busfahrer hat vor dem Ausflug an die Küste der Reiseleiter begrüßt.

24. SF–SD In der Vorstellung mit dem Orchester hat der Dirigent den Pianist irritiert.
SF–LD Der Dirigent hat in der Vorstellung mit dem Orchester den Pianist irritiert.
OF–SD In der Vorstellung mit dem Orchester hat den Pianist der Dirigent irritiert.
OF–LD Den Pianist hat in der Vorstellung mit dem Orchester der Dirigent irritiert.

25. SF–SD Während der Besprechung in der Firma hat der Geschäftsführer den Leiter berufen.
SF–LD Der Geschäftsführer hat während der Besprechung in der Firma den Leiter berufen.
OF–SD Während der Besprechung in der Firma hat den Leiter der Geschäftsführer berufen.
OF–LD Den Leiter hat während der Besprechung in der Firma der Geschäftsführer berufen.

26. SF–SD Nach der Krise wegen des Wahlversprechens hat der Redakteur den Kanzler verurteilt.
SF–LD Der Redakteur hat nach der Krise wegen des Wahlversprechens den Kanzler verurteilt.
OF–SD Nach der Krise wegen des Wahlversprechens hat den Kanzler der Redakteur verurteilt.
OF–LD Den Kanzler hat nach der Krise wegen des Wahlversprechens der Redakteur verurteilt.

27. SF–SD In der Debatte um die Finanzierung hat der Befürworter den Gegner überzeugt.
SF–LD Der Befürworter hat in der Debatte um die Finanzierung den Gegner überzeugt.
OF–SD In der Debatte um die Finanzierung hat den Gegner der Befürworter überzeugt.
OF–LD Den Gegner hat in der Debatte um die Finanzierung der Befürworter überzeugt.

28. SF–SD Vor der Ausstellung in der Galerie hat der Kunde den Maler angerufen.
SF–LD Der Kunde hat vor der Ausstellung in der Galerie den Maler angerufen.
OF–SD Vor der Ausstellung in der Galerie hat den Maler der Kunde angerufen.
OF–LD Den Maler hat vor der Ausstellung in der Galerie der Kunde angerufen.

29. SF–SD Bei dem Treffen über die Ausbildung hat der Bewerber den Direktor beeindruckt.
SF–LD Der Bewerber hat bei dem Treffen über die Ausbildung den Direktor beeindruckt.
OF–SD Bei dem Treffen über die Ausbildung hat den Direktor der Bewerber beeindruckt.
OF–LD Den Direktor hat bei dem Treffen über die Ausbildung der Bewerber beeindruckt.

B-9
Appendix B. Stimuli from Patient Study

30. SF–SD Vor der Lesung in der Buchhandlung hat der Verleger den Schriftsteller vorgestellt.
SF–LD Der Verleger hat vor der Lesung in der Buchhandlung den Schriftsteller vorgestellt.
OF–SD Vor der Lesung in der Buchhandlung hat den Schriftsteller der Verleger vorgestellt.
OF–LD Den Schriftsteller hat vor der Lesung in der Buchhandlung der Verleger vorgestellt.

31. SF–SD Wegen der Erfahrung in der Klinik hat der Pfleger den Patienten gemieden.
SF–LD Der Pfleger hat wegen der Erfahrung in der Klinik den Patienten gemieden.
OF–SD Wegen der Erfahrung in der Klinik hat den Patienten der Pfleger gemieden.
OF–LD Den Patienten hat wegen der Erfahrung in der Klinik der Pfleger gemieden.

32. SF–SD Während der Tagung an der Hochschule hat der Forscher den Besucher begeistert.
SF–LD Der Forscher hat während der Tagung an der Hochschule den Besucher begeistert.
OF–SD Während der Tagung an der Hochschule hat den Besucher der Forscher begeistert.
OF–LD Den Besucher hat während der Tagung an der Hochschule der Forscher begeistert.

33. SF–SD Nach der Begegnung in dem Studio hat der Regisseur den Schauspieler engagiert.
SF–LD Der Regisseur hat nach der Begegnung in dem Studio den Schauspieler engagiert.
OF–SD Nach der Begegnung in dem Studio hat den Schauspieler der Regisseur engagiert.
OF–LD Den Schauspieler hat nach der Begegnung in dem Studio der Regisseur engagiert.

34. SF–SD Bei dem Jubiläum in dem Finanzamt hat der Vorgänger den Nachfolger eingeführt.
SF–LD Der Vorgänger hat bei dem Jubiläum in dem Finanzamt den Nachfolger eingeführt.
OF–SD Bei dem Jubiläum in dem Finanzamt hat den Nachfolger der Vorgänger eingeführt.
OF–LD Den Nachfolger hat bei dem Jubiläum in dem Finanzamt der Vorgänger eingeführt.

35. SF–SD Bei dem Sommerfest in der Siedlung hat der Nachbar den Bürgermeister erwartet.
SF–LD Der Nachbar hat bei dem Sommerfest in der Siedlung den Bürgermeister erwartet.
OF–SD Bei dem Sommerfest in der Siedlung hat den Bürgermeister der Nachbar erwartet.
OF–LD Den Bürgermeister hat bei dem Sommerfest in der Siedlung der Nachbar erwartet.

36. SF–SD Nach der Ansprache an das Publikum hat der Leser den Autor bewundert.
SF–LD Der Leser hat nach der Ansprache an das Publikum den Autor bewundert.
OF–SD Nach der Ansprache an das Publikum hat den Autor der Leser bewundert.
OF–LD Den Autor hat nach der Ansprache an das Publikum der Leser bewundert.

37. SF–SD Vor dem Antrag gegen das Parlament hat der Politiker den Konkurrenten gefürchtet.
SF–LD Der Politiker hat vor dem Antrag gegen das Parlament den Konkurrenten gefürchtet.
OF–SD Vor dem Antrag gegen das Parlament hat den Konkurrenten der Politiker gefürchtet.
OF–LD Den Konkurrenten hat vor dem Antrag gegen das Parlament der Politiker gefürchtet.

38. SF–SD Vor der Übereinkunft mit den Eltern hat der Jugendliche den Erzieher ignoriert.
SF–LD Der Jugendliche hat vor der Übereinkunft mit den Eltern den Erzieher ignoriert.
OF–SD Vor der Übereinkunft mit den Eltern hat den Erzieher der Jugendliche ignoriert.
OF–LD Den Erzieher hat vor der Übereinkunft mit den Eltern der Jugendliche ignoriert.

39. SF–SD Während der Planung für die Initiative hat der Veranstalter den Teilnehmer beruhigt.
SF–LD Der Veranstalter hat während der Planung für die Initiative den Teilnehmer beruhigt.
OF–SD Während der Planung für die Initiative hat den Teilnehmer der Veranstalter beruhigt.
OF–LD Den Teilnehmer hat während der Planung für die Initiative der Veranstalter beruhigt.

40. SF–SD Wegen der Probleme bei der Produktion hat der Unternehmer den Lieferanten gewechselt.
SF–LD Der Unternehmer hat wegen der Probleme bei der Produktion den Lieferanten gewechselt.
OF–SD Wegen der Probleme bei der Produktion hat den Lieferanten der Unternehmer gewechselt.
OF–LD Den Lieferanten hat wegen der Probleme bei der Produktion der Unternehmer gewechselt.

B-10
 Appendix B. Stimuli from Patient Study

41. SF–SD Nach der Ermittlung in dem Milieu hat der Polizist den Verbrecher beschuldigt.
SF–LD Der Polizist hat nach der Ermittlung in dem Milieu den Verbrecher beschuldigt.
OF–SD Nach der Ermittlung in dem Milieu hat den Verbrecher der Polizist beschuldigt.
OF–LD Den Verbrecher hat nach der Ermittlung in dem Milieu der Polizist beschuldigt.

42. SF–SD Mit der Mitteilung wegen des Zeugnisses hat der Schulleiter den Abiturient erschrocken.
SF–LD Der Schulleiter hat mit der Mitteilung wegen des Zeugnisses den Abiturient erschrocken.
OF–SD Mit der Mitteilung wegen des Zeugnisses hat den Abiturient der Schulleiter erschrocken.
OF–LD Den Abiturient hat mit der Mitteilung wegen des Zeugnisses der Schulleiter erschrocken.

43. SF–SD Mit der Premiere in dem Theater hat der Dichter den Kritiker überrascht.
SF–LD Der Dichter hat mit der Premiere in dem Theater den Kritiker überrascht.
OF–SD Mit der Premiere in dem Theater hat den Kritiker der Dichter überrascht.
OF–LD Den Kritiker hat mit der Premiere in dem Theater der Dichter überrascht.

44. SF–SD Wegen der Bilanz an der Börse hat der Beamte den Sekretär verhaftet.
SF–LD Der Beamte hat wegen der Bilanz an der Börse den Sekretär verhaftet.
OF–SD Wegen der Bilanz an der Börse hat den Sekretär der Beamte verhaftet.
OF–LD Den Sekretär hat wegen der Bilanz an der Börse der Beamte verhaftet.

45. SF–SD Nach dem Urteil in dem Verfahren hat der Anwalt den Angeklagten aufgegeben.
SF–LD Der Anwalt hat nach dem Urteil in dem Verfahren den Angeklagten aufgegeben.
OF–SD Nach dem Urteil in dem Verfahren hat den Angeklagten der Anwalt aufgegeben.
OF–LD Den Angeklagten hat nach dem Urteil in dem Verfahren der Anwalt aufgegeben.

46. SF–SD Vor der Entlassung aus dem Gefängnis hat der Psychologe den Häftling untersucht.
SF–LD Der Psychologe hat vor der Entlassung aus dem Gefängnis den Häftling untersucht.
OF–SD Vor der Entlassung aus dem Gefängnis hat den Häftling der Psychologe untersucht.
OF–LD Den Häftling hat vor der Entlassung aus dem Gefängnis der Psychologe untersucht.

47. SF–SD Wegen des Verlusts bei den Geschäften hat der Kaufmann den Bankier kontaktiert.
SF–LD Der Kaufmann hat wegen des Verlusts bei den Geschäften den Bankier kontaktiert.
OF–SD Wegen des Verlusts bei den Geschäften hat den Bankier der Kaufmann kontaktiert.
OF–LD Den Bankier hat wegen des Verlusts bei den Geschäften der Kaufmann kontaktiert.

48. SF–SD Nach der Aussprache mit der Familie hat der Bruder den Vater angelächelt.
SF–LD Der Bruder hat nach der Aussprache mit der Familie den Vater angelächelt.
OF–SD Nach der Aussprache mit der Familie hat den Vater der Bruder angelächelt.
OF–LD Den Vater hat nach der Aussprache mit der Familie der Bruder angelächelt.

B-11
C
F I L L E R S F R O M B E H AV I O R A L S T U D Y

1. A Der kranke Vorsitzende hat nach der Konferenz endlich den Job gekündigt und in ein Restaurant investiert.
B Nach der Konferenz hat der kranke Vorsitzende endlich den Job gekündigt und in ein Restaurant investiert.
C Der kranke Vorsitzende der seine Arbeit hasste hat nach der Konferenz mit den gierigen Aktionären endlich den Job gekündigt und
in ein Restaurant investiert.
D Nach der Konferenz mit den gierigen Aktionären hat der kranke Vorsitzende der seine Arbeit hasste endlich den Job gekündigt und
in ein Restaurant investiert.

2. A Der fröhliche Schaffner hat an dem Bahnhof täglich die Regionalbahn bestiegen und auf seine Uhr geschaut.
B An dem Bahnhof hat der fröhliche Schaffner täglich die Regionalbahn bestiegen und auf seine Uhr geschaut.
C Der fröhliche Schaffner der seinen Beruf liebte hat auf dem Bahnhof in der kleinen Stadt täglich die Regionalbahn bestiegen und
auf seine Uhr geschaut.
D Auf dem Bahnhof in der kleinen Stadt hat der fröhliche Schaffner der seinen Beruf liebte täglich die Regionalbahn bestiegen und
auf seine Uhr geschaut.

3. A Die fleißige Maus hat in einer Scheune heimlich am Saatgut geknabbert und von den Würsten genascht.
B In einer Scheune hat die fleißige Maus heimlich am Saatgut geknabbert und von den Würsten genascht.
C Die fleißige Maus die gerne Abenteuer erlebte hat in einer Scheune hinter der verlassenen Mühle heimlich am Saatgut geknabbert
und von den Würsten genascht.
D In einer Scheune hinter der verlassenen Mühle hat die fleißige Maus die gerne Abenteuer erlebte heimlich am Saatgut geknabbert
und von den Würsten genascht.

4. A Die genervten Schüler sind nach dem Unterricht sofort zum Strandbad gefahren und in den See gesprungen.
B Nach dem Unterricht sind die genervten Schüler sofort zum Strandbad gefahren und in den See gesprungen.
C Die genervten Schüler die große Langeweile hatten sind nach dem Unterricht bei dem strengen Lehrer sofort zum Strandbad
gefahren und in den See gesprungen.
D Nach dem Unterricht bei dem strengen Lehrer sind die genervten Schüler die große Langeweile hatten sofort zum Strandbad
gefahren und in den See gesprungen.

5. A Der kluge Bauer hat am letzten Freitag bereits die Kartoffeln geerntet und auf dem Wochenmarkt verkauft.
B Am letzten Freitag hat der kluge Bauer bereits die Kartoffeln geerntet und auf dem Wochenmarkt verkauft.
C Der kluge Bauer der aus Sachsen stammte hat am letzten Freitag nach dem großen Sturm bereits die Kartoffeln geerntet und auf
dem Wochenmarkt verkauft.
D Am letzten Freitag nach dem großen Sturm hat der kluge Bauer der aus Sachsen stammte bereits die Kartoffeln geerntet und auf
dem Wochenmarkt verkauft.

6. A Das junge Ehepaar hat in der Nacht erneut über Umzugspläne gestritten und um einen Kompromiss gekämpft.
B In der Nacht hat das junge Ehepaar erneut über Umzugspläne gestritten und um einen Kompromiss gekämpft.
C Das junge Ehepaar das immer Streit hatte hat in der Nacht vor der ersehnten Reise erneut über Umzugspläne gestritten und um
einen Kompromiss gekämpft.
D In der Nacht vor der ersehnten Reise hat das junge Ehepaar das immer Streit hatte erneut über Umzugspläne gestritten und um
einen Kompromiss gekämpft.

C-13
Appendix C. Fillers from Behavioral Study

7. A Ein riesiges Schiff ist an der Ostküste plötzlich auf Grund gelaufen und in einem Unwetter gekentert.
B An der Ostküste ist ein riesiges Schiff plötzlich auf Grund gelaufen und in einem Unwetter gekentert.
C Ein riesiges Schiff das aus Russland kam ist an der Ostküste in einer nebligen Nacht plötzlich auf Grund gelaufen und in einem
Unwetter gekentert.
D An der Ostküste in einer nebligen Nacht ist ein riesiges Schiff das aus Russland kam plötzlich auf Grund gelaufen und in einem
Unwetter gekentert.

8. A Der faule Student hat in einem Seminar wieder ein Referat gehalten und gegen die Formalien verstoßen.
B In einem Seminar hat der faule Student wieder ein Referat gehalten und gegen die Formalien verstoßen.
C Der faule Student der die Uni vernachlässigte hat in einem Seminar vor den ersehnten Ferien wieder ein Referat gehalten und gegen
die Formalien verstoßen.
D In einem Seminar vor den ersehnten Ferien hat der faule Student der die Uni vernachlässigte wieder ein Referat gehalten und gegen
die Formalien verstoßen.

9. A Der beliebte Kanzler hat vor dem Parlament immer die Abgeordneten begrüßt und auf den Tagesplan hingewiesen.
B Vor dem Parlament hat der beliebte Kanzler immer die Abgeordneten begrüßt und auf den Tagesplan hingewiesen.
C Der beliebte Kanzler der sein Amt schätzte hat vor dem Parlament in der deutschen Hauptstadt immer die Abgeordneten begrüßt
und auf den Tagesplan hingewiesen.
D Vor dem Parlament in der deutschen Hauptstadt hat der beliebte Kanzler der sein Amt schätzte immer die Abgeordneten begrüßt
und auf den Tagesplan hingewiesen.

10. A Ein wilder Braunbär hat in den Bergen gestern ein Schaf angegriffen und in den Abgrund getrieben.
B In den Bergen hat ein wilder Braunbär gestern ein Schaf angegriffen und in den Abgrund getrieben.
C Ein wilder Braunbär der aus Österreich kam hat in den Bergen nahe einer abgelegenen Weide gestern ein Schaf angegriffen und in
den Abgrund getrieben.
D In den Bergen nahe einer abgelegenen Weide hat ein wilder Braunbär der aus Österreich kam gestern ein Schaf angegriffen und in
den Abgrund getrieben.

11. A Der geniale Professor ist nach der Arbeit unerwartet vom Balkon gestürzt und in der Arztpraxis aufgewacht.
B Nach der Arbeit ist der geniale Professor unerwartet vom Balkon gestürzt und in der Arztpraxis aufgewacht.
C Der geniale Professor der die Forschung mochte ist nach der Arbeit in seinem neuen Labor unerwartet vom Balkon gestürzt und in
der Arztpraxis aufgewacht.
D Nach der Arbeit in seinem neuen Labor ist der geniale Professor der die Forschung mochte unerwartet vom Balkon gestürzt und in
der Arztpraxis aufgewacht.

12. A Der mürrische Winzer hat nach der Weinlese stets den Gottesdienst besucht und für die Kirche gespendet.
B Nach der Weinlese hat der mürrische Winzer stets den Gottesdienst besucht und für die Kirche gespendet.
C Der mürrische Winzer der lange Witwer war hat nach der Weinlese in den herbstlichen Hügeln stets den Gottesdienst besucht und
für die Kirche gespendet.
D Nach der Weinlese in den herbstlichen Hügeln hat der mürrische Winzer der lange Witwer war stets den Gottesdienst besucht und
für die Kirche gespendet.

13. A Der übermüdete Flugkapitän ist nach seiner Landung gleich ins Hotel gefahren und in sein Bett gestiegen.
B Nach seiner Landung ist der übermüdete Flugkapitän gleich ins Hotel gefahren und in sein Bett gestiegen.
C Der übermüdete Flugkapitän dem die Augen zufielen ist nach seiner Landung auf dem fernen Flughafen gleich ins Hotel gefahren
und in sein Bett gestiegen.
D Nach seiner Landung auf dem fernen Flughafen ist der übermüdete Flugkapitän dem die Augen zufielen gleich ins Hotel gefahren
und in sein Bett gestiegen.

14. A Der erfolgreiche Stürmer hat in jedem Spiel mindestens ein Tor geschossen und für den Sieg gesorgt.
B In jedem Spiel hat der erfolgreiche Stürmer mindestens ein Tor geschossen und für den Sieg gesorgt.
C Der erfolgreiche Stürmer der ein Volksheld war hat in jedem Spiel in seinem heimatlichen Stadion mindestens ein Tor geschossen
und für den Sieg gesorgt.
D In jedem Spiel in seinem heimatlichen Stadion hat der erfolgreiche Stürmer der ein Volksheld war mindestens ein Tor geschossen
und für den Sieg gesorgt.

C-14
 Appendix C. Fillers from Behavioral Study

15. A Die besten Gärtner haben trotz aller Vorsorge schon mit Maulwürfen gerungen und mit ihren Tricks gehadert.
B Trotz aller Vorsorge haben die besten Gärtner schon mit Maulwürfen gerungen und mit ihren Tricks gehadert.
C Die besten Gärtner die jedem Schädling trotzen haben trotz aller Vorsorge gegen die dreisten Nager schon mit Maulwürfen gerungen
und mit ihren Tricks gehadert.
D Trotz aller Vorsorge gegen die dreisten Nager haben die besten Gärtner die jedem Schädling trotzen schon mit Maulwürfen gerungen
und mit ihren Tricks gehadert.

16. A Die meisten Tänzer haben nach ihrem Ausscheiden weiter am Theater gearbeitet und auf der Bühne geholfen.
B Nach ihrem Ausscheiden haben die meisten Tänzer weiter am Theater gearbeitet und auf der Bühne geholfen.
C Die meisten Tänzer die ihre Karriere beendeten haben nach ihrem Ausscheiden aus der aktiven Laufbahn weiter am Theater gear-
beitet und auf der Bühne geholfen.
D Nach ihrem Ausscheiden aus der aktiven Laufbahn haben die meisten Tänzer die ihre Karriere beendeten weiter am Theater gear-
beitet und auf der Bühne geholfen.

17. A Ein guter Polizist hat bei der Streife besonders kleine Details beachtet und in seinen Bericht geschrieben.
B Bei der Streife hat ein guter Polizist besondern kleine Details beachtet und in seinen Bericht geschrieben.
C Ein guter Polizist der die Verbrecher kennt hat bei der Streife dank seiner langen Erfahrung besonders kleine Details beachtet und
in seinen Bericht geschrieben.
D Bei der Streife dank seiner langen Erfahrung hat ein guter Polizist der die Verbrecher kennt besonders kleine Details beachtet und
in seinen Bericht geschrieben.

18. A Ein starkes Gewitter hat bei einer Wanderung vorgestern fünf Bergsteiger überrumpelt und zu einem Umweg gezwungen.
B Bei einer Wanderung hat ein starkes Gewitter vorgestern fünf Bergsteiger überrumpelt und zu einem Umweg gezwungen.
C Ein starkes Gewitter das am Nachmittag aufzog hat bei einer Wanderung auf einer neuen Route vorgestern fünf Bergsteiger über-
rascht und zu einem Umweg gezwungen.
D Bei einer Wanderung auf einer neuen Route hat ein starkes Gewitter das am Nachmittag aufzog vorgestern fünf Bergsteiger über-
rascht und zu einem Umweg gezwungen.

19. A Ein gefährliches Virus hat in dem Internat zuletzt zehn Schüler infiziert und vom normalen Betrieb ausgeschlossen.
B In dem Internat hat ein gefährliches Virus zuletzt zehn Schüler infiziert und vom normalen Betrieb ausgeschlossen.
C Ein gefährliches Virus das kein Arzt kannte hat in dem Internat in der entlegenen Gegend zuletzt zehn Schüler infiziert und vom
normalen Betrieb ausgeschlossen.
D An dem Internat in der entlegenen Gegend hat ein gefährliches Virus das kein Arzt kannte zuletzt zehn Schüler infiziert und vom
normalen Betrieb ausgeschlossen.

20. A Eine düstere Wolke hat während des Festes drohend den Himmel verdunkelt und über dem Rathaus angehalten.
B Während des Festes hat eine düstere Wolke drohend den Himmel verdunkelt und über dem Rathaus angehalten.
C Eine düstere Wolke die die Stimmung drückte hat während des Festes für die siegreichen Fußballer drohend den Himmel verdunkelt
und über dem Rathaus angehalten.
D Während des Festes für die siegreichen Fußballer hat eine düstere Wolke die die Stimmung drückte drohend den Himmel verdunkelt
und über dem Rathaus angehalten.

21. A Eine dreiste Dränglerin hat auf der Autobahn abends einen Audi abgedrängt und gegen die Leitplanke gedrückt.
B Auf der Autobahn hat eine dreiste Dränglerin abends einen Audi abgedrängt und gegen die Leitplanke gedrückt.
C Eine dreiste Dränglerin die einen Porsche fuhr hat auf der Autobahn an einer engen Stelle abends einen Audi abgedrängt und gegen
die Leitplanke gedrückt.
D Auf der Autobahn an einer engen Stelle hat eine dreiste Dränglerin die einen Porsche fuhr abends einen Audi abgedrängt und gegen
die Leitplanke gedrückt.

22. A Ein gefährlicher Hai hat auf dem Meer offenbar einen Fischer überrascht und von seinem Boot gerissen.
B Auf dem Meer hat ein gefährlicher Hai offenbar einen Fischer überrascht und von seinem Boot gerissen.
C Ein gefährlicher Hai der im Atlantik lebt hat auf dem Meer in den frühen Morgenstunden offenbar einen Fischer überrascht und
von seinem Boot gerissen.
D Auf dem Meer in den frühen Morgenstunden hat ein gefährlicher Hai der im Atlantik lebt offenbar einen Fischer überrascht und
von seinem Boot gerissen.

C-15
Appendix C. Fillers from Behavioral Study

23. A Eine glückliche Mutter hat in der Klinik erstmals einen Jungen geboren und unter vielen Tränen umarmt.
B In der Klinik hat eine glückliche Mutter erstmals einen Jungen geboren und unter vielen Tränen umarmt.
C Eine glückliche Mutter die aus Hamburg kam hat in der Klinik auf der neuen Säuglingsstation erstmals einen Jungen geboren und
unter vielen Tränen umarmt.
D In der Klinik auf der neuen Säuglingsstation hat eine glückliche Mutter die aus Hamburg kam erstmals einen Jungen geboren und
unter vielen Tränen umarmt.

24. A Ein arbeitsloser Schlosser hat in der Tombola zufällig eine Million gewonnen und in der Spielhalle verzockt.
B In der Tombola hat ein arbeitsloser Schlosser zufällig eine Million gewonnen und in der Spielhalle verzockt.
C Ein arbeitsloser Schlosser der immer Pech hatte hat in der Tombola auf der jährlichen Kirmes zufällig eine Million gewonnen und
in der Spielhalle verzockt.
D In der Tombola auf der jährlichen Kirmes hat ein arbeitsloser Schlosser der immer Pech hatte zufällig eine Million gewonnen und
in der Spielhalle verzockt.

C-16
D
F I L L E R S F R O M B E H AV I O R A L A N D M R I S T U D I E S

1. A Gestern hat der Vater dem Sohn den Schnuller gegeben und den Kopf gestreichelt.
B Gestern hat dem Sohn der Vater den Schnuller gegeben und den Kopf gestreichelt.
C Gestern hat dem Sohn den Schnuller der Vater gegeben und den Kopf gestreichelt.
D Gestern hat der Vater gegeben dem Sohn den Schnuller und den Kopf gestreichelt.

2. A Heute hat der Opa dem Enkel den Lolli gekauft und das Eis geschenkt.
B Heute hat dem Enkel der Opa den Lolli gekauft und das Eis geschenkt.
C Heute hat dem Enkel den Lolli der Opa gekauft und das Eis geschenkt.
D Heute hat der Opa gekauft dem Enkel den Lolli und das Eis geschenkt.

3. A Dort hat der Dieb dem Anwalt den Wagen zerkratzt und die Rache genossen.
B Dort hat dem Anwalt der Dieb den Wagen zerkratzt und die Rache genossen.
C Dort hat dem Anwalt den Wagen der Dieb zerkratzt und die Rache genossen.
D Dort hat der Dieb zerkratzt dem Anwalt den Wagen und die Rache genossen.

4. A Gestern hat der Leser dem Bibliothekar den Artikel zurückgegeben und die Gebühren bezahlt.
B Gestern hat dem Bibliothekar der Leser den Artikel zurückgegeben und die Gebühren bezahlt.
C Gestern hat dem Bibliothekar den Artikel der Leser zurückgegeben und die Gebühren bezahlt.
D Gestern hat der Leser zurückgegeben dem Bibliothekar den Artikel und die Gebühren bezahlt.

5. A Dann hat der Aufseher dem Besuch den Ausgang gezeigt und die Tür geöffnet.
B Dann hat dem Besuch der Aufseher den Ausgang gezeigt und die Tür geöffnet.
C Dann hat dem Besuch den Ausgang der Aufseher gezeigt und die Tür geöffnet.
D Dann hat der Aufseher gezeigt dem Besuch den Ausgang und die Tür geöffnet.

6. A Heute hat der Mechaniker dem Rennfahrer den Motor repariert und die Reifen gewechselt.
B Heute hat dem Rennfahrer der Mechaniker den Motor repariert und die Reifen gewechselt.
C Heute hat dem Rennfahrer den Motor der Mechaniker repariert und die Reifen gewechselt.
D Heute hat der Mechaniker repariert dem Rennfahrer den Motor und die Reifen gewechselt.

7. A Vielleicht hat der Chef dem Klempner den Vertrag verlängert und den Lohn erhöht.
B Vielleicht hat dem Klempner der Chef den Vertrag verlängert und den Lohn erhöht.
C Vielleicht hat dem Klempner den Vertrag der Chef verlängert und den Lohn erhöht.
D Vielleicht hat der Chef verlängert dem Klempner den Vertrag und den Lohn erhöht.

D-19
Appendix D. Fillers from Behavioral and MRI Studies

8. A Offensichtlich hat der Arzt dem Bergsteiger den Arm verbunden und die Beine geschient.
B Offensichtlich hat dem Bergsteiger der Arzt den Arm verbunden und die Beine geschient.
C Offensichtlich hat dem Bergsteiger den Arm der Arzt verbunden und die Beine geschient.
D Offensichtlich hat der Arzt verbunden dem Bergsteiger den Arm und die Beine geschient.

9. A Schliesslich hat der Bruder dem Bräutigam den Ring überreicht und die Hand geschüttelt.
B Schliesslich hat dem Bräutigam der Bruder den Ring überreicht und die Hand geschüttelt.
C Schliesslich hat dem Bräutigam den Ring der Bruder überreicht und die Hand geschüttelt.
D Schliesslich hat der Bruder überreicht dem Bräutigam den Ring und die Hand geschüttelt.

10. A Überraschenderweise hat der Sekretär dem Fahrer den Umschlag anvertraut und das Beste gehofft.
B Überraschenderweise hat dem Fahrer der Sekretär den Umschlag anvertraut und das Beste gehofft.
C Überraschenderweise hat dem Fahrer den Umschlag der Sekretär anvertraut und das Beste gehofft.
D Überraschenderweise hat der Sekretär anvertraut dem Fahrer den Umschlag und das Beste gehofft.

11. A Gestern hat der Portier dem Manager den Koffer getragen und das Trinkgeld angenommen.
B Gestern hat dem Manager der Portier den Koffer getragen und das Trinkgeld angenommen.
C Gestern hat dem Manager den Koffer der Portier getragen und das Trinkgeld angenommen.
D Gestern hat der Portier getragen dem Manager den Koffer und das Trinkgeld angenommen.

12. A Dann hat der Friseur dem Metzger den Bart rasiert und die Haare geschnitten.
B Dann hat dem Metzger der Friseur den Bart rasiert und die Haare geschnitten.
C Dann hat dem Metzger den Bart der Friseur rasiert und die Haare geschnitten.
D Dann hat der Friseur rasiert dem Metzger den Bart und die Haare geschnitten.

13. A Schliesslich hat der Designer dem Teddy den Hut aufgesetzt und die Schaufenster dekoriert.
B Schliesslich hat dem Teddy der Designer den Hut aufgesetzt und die Schaufenster dekoriert.
C Schliesslich hat dem Teddy den Hut der Designer aufgesetzt und die Schaufenster dekoriert.
D Schliesslich hat der Designer aufgesetzt dem Teddy den Hut und die Schaufenster dekoriert.

14. A Dort hat der Hausmeister dem Mieter den Schlüssel überreicht und die Verträge erklärt.
B Dort hat dem Mieter der Hausmeister den Schlüssel überreicht und die Verträge erklärt.
C Dort hat dem Mieter den Schlüssel der Hausmeister überreicht und die Verträge erklärt.
D Dort hat der Hausmeister überreicht dem Mieter den Schlüssel und die Verträge erklärt.

15. A Überraschenderweise hat der Zauberer dem Zuschauer den Trick verraten und den Vorgang wiederholt.
B Überraschenderweise hat dem Zuschauer der Zauberer den Trick verraten und den Vorgang wiederholt.
C Überraschenderweise hat dem Zuschauer den Trick der Zauberer verraten und den Vorgang wiederholt.
D Überraschenderweise hat der Zauberer verraten dem Zuschauer den Trick und den Vorgang wiederholt.

16. A Offensichtlich hat der Kellner dem Gast den Braten serviert und die Getränke geholt.
B Offensichtlich hat dem Gast der Kellner den Braten serviert und die Getränke geholt.
C Offensichtlich hat dem Gast den Braten der Kellner serviert und die Getränke geholt.
D Offensichtlich hat der Kellner serviert dem Gast den Braten und die Getränke geholt.

17. A Schliesslich hat der Ober dem Urlauber den Saft eingeschenkt und die Flaschen hingestellt.
B Schliesslich hat dem Urlauber der Ober den Saft eingeschenkt und die Flaschen hingestellt.
C Schliesslich hat dem Urlauber den Saft der Ober eingeschenkt und die Flaschen hingestellt.
D Schliesslich hat der Ober eingeschenkt dem Urlauber den Saft und die Flaschen hingestellt.

18. A Dann hat der Chirurg dem Chefarzt den Befund vorgelesen und die Operationen vorgeschlagen.
B Dann hat dem Chefarzt der Chirurg den Befund vorgelesen und die Operationen vorgeschlagen.
C Dann hat dem Chefarzt den Befund der Chirurg vorgelesen und die Operationen vorgeschlagen.
D Dann hat der Chirurg vorgelesen dem Chefarzt den Befund und die Operationen vorgeschlagen.

D-20
 Appendix D. Fillers from Behavioral and MRI Studies

19. A Gestern hat der Sportler dem Richter den Unfall geschildert und die Angeklagten erkannt.
B Gestern hat dem Richter der Sportler den Unfall geschildert und die Angeklagten erkannt.
C Gestern hat dem Richter den Unfall der Sportler geschildert und die Angeklagten erkannt.
D Gestern hat der Sportler geschildert dem Richter den Unfall und die Angeklagten erkannt.

20. A Vielleicht hat der Gärtner dem Freund den Rasenmäher ausgeliehen und die Blumen gebracht.
B Vielleicht hat dem Freund der Gärtner den Rasenmäher ausgeliehen und die Blumen gebracht.
C Vielleicht hat dem Freund den Rasenmäher der Gärtner ausgeliehen und die Blumen gebracht.
D Vielleicht hat der Gärtner ausgeliehen dem Freund den Rasenmäher und die Blumen gebracht.

21. A Morgen wird der Bauleiter dem Maurer den Plan vorlegen und das Werk beginnen.
B Morgen wird dem Maurer der Bauleiter den Plan vorlegen und das Werk beginnen.
C Morgen wird dem Maurer den Plan der Bauleiter vorlegen und das Werk beginnen.
D Morgen wird der Bauleiter vorlegen dem Maurer den Plan und das Werk beginnen.

22. A Heute wird der Detektiv dem Komissar den Fundort beschreiben und den Fall abschliessen.
B Heute wird dem Komissar der Detektiv den Fundort beschreiben und den Fall abschliessen.
C Heute wird dem Komissar den Fundort der Detektiv beschreiben und den Fall abschliessen.
D Heute wird der Detektiv beschreiben dem Komissar den Fundort und den Fall abschliessen.

23. A Schliesslich wird der Schüler dem Lehrer den Aufsatz vorlesen und die Note erfahren.
B Schliesslich wird dem Lehrer der Schüler den Aufsatz vorlesen und die Note erfahren.
C Schliesslich wird dem Lehrer den Aufsatz der Schüler vorlesen und die Note erfahren.
D Schliesslich wird der Schüler vorlesen dem Lehrer den Aufsatz und die Note erfahren.

24. A Morgen wird der Vertreter dem Rentner den Staubsauger verkaufen und die Raten festlegen.
B Morgen wird dem Rentner der Vertreter den Staubsauger verkaufen und die Raten festlegen.
C Morgen wird dem Rentner den Staubsauger der Vertreter verkaufen und die Raten festlegen.
D Morgen wird der Vertreter verkaufen dem Rentner den Staubsauger und die Raten festlegen.

25. A Vielleicht wird der Einbrecher dem Wirt den Fernseher stehlen und die Möbel zerstören.
B Vielleicht wird dem Wirt der Einbrecher den Fernseher stehlen und die Möbel zerstören.
C Vielleicht wird dem Wirt den Fernseher der Einbrecher stehlen und die Möbel zerstören.
D Vielleicht wird der Einbrecher stehlen dem Wirt den Fernseher und die Möbel zerstören.

26. A Dann wird der Lehrling dem Schreiner den Hammer reichen und die Nägel nehmen.
B Dann wird dem Schreiner der Lehrling den Hammer reichen und die Nägel nehmen.
C Dann wird dem Schreiner den Hammer der Lehrling reichen und die Nägel nehmen.
D Dann wird der Lehrling reichen dem Schreiner den Hammer und die Nägel nehmen.

27. A Überraschenderweise wird der Millionär dem Diener den Besitz vererben und die Kinder vergessen.
B Überraschenderweise wird dem Diener der Millionär den Besitz vererben und die Kinder vergessen.
C Überraschenderweise wird dem Diener den Besitz der Millionär vererben und die Kinder vergessen.
D Überraschenderweise wird der Millionär vererben dem Diener der Besitz und die Kinder vergessen.

28. A Morgen wird der Künstler dem Bürgermeister den Entwurf präsentieren und die Kritik anhören.
B Morgen wird dem Bürgermeister der Künstler den Entwurf präsentieren und die Kritik anhören.
C Morgen wird dem Bürgermeister den Entwurf der Künstler präsentieren und die Kritik anhören.
D Morgen wird der Künstler präsentieren dem Bürgermeister den Entwurf und die Kritik anhören.

29. A Vielleicht wird der Kanzler dem Minister den Fehltritt verzeihen und die Probleme lösen.
B Vielleicht wird dem Minister der Kanzler den Fehltritt verzeihen und die Probleme lösen.
C Vielleicht wird dem Minister den Fehltritt der Kanzler verzeihen und die Probleme lösen.
D Vielleicht wird der Kanzler verzeihen dem Minister den Fehltritt und die Probleme lösen.

D-21
Appendix D. Fillers from Behavioral and MRI Studies

30. A Dort wird der Leibwächter dem Erpresser den Revolver wegnehmen und die Polizei rufen.
B Dort wird dem Erpresser der Leibwächter den Revolver wegnehmen und die Polizei rufen.
C Dort wird dem Erpresser den Revolver der Leibwächter wegnehmen und die Polizei rufen.
D Dort wird der Leibwächter wegnehmen dem Erpresser den Revolver und die Polizei rufen.

31. A Heute wird der Schauspieler dem Chauffeur den Vertrag kündigen und die Köche entlassen.
B Heute wird dem Chauffeur der Schauspieler den Vertrag kündigen und die Köche entlassen.
C Heute wird dem Chauffeur den Vertrag der Schauspieler kündigen und die Köche entlassen.
D Heute wird der Schauspieler kündigen dem Chauffeur den Vertrag und die Köche entlassen.

32. A Morgen wird der Tenor dem Zuhörer den Platz reservieren und die Lieder singen.
B Morgen wird dem Zuhörer der Tenor den Platz reservieren und die Lieder singen.
C Morgen wird demZ uhörer den Platz der Tenor reservieren und die Lieder singen.
D Morgen wird der Tenor reservieren dem Zuhörer den Platz und die Lieder singen.

33. A Überraschenderweise wird der Regisseur dem Darsteller den Film widmen und die Auszeichnungen überlassen.
B Überraschenderweise wird dem Darsteller der Regisseur den Film widmen und die Auszeichnungen überlassen.
C Überraschenderweise wird dem Darsteller den Film der Regisseur widmen und die Auszeichnungen überlassen.
D Überraschenderweise wird der Regisseur widmen dem Darsteller den Film und die Auszeichnungen überlassen.

34. A Jetzt wird der Mörder dem Wanderer den Fuss abhacken und die Beweise vernichten.
B Jetzt wird dem Wanderer der Mörder den Fuss abhacken und die Beweise vernichten.
C Jetzt wird dem Wanderer den Fuss der Mörder abhacken und die Beweise vernichten.
D Jetzt wird der Mörder abhacken dem Wanderer den Fuss und die Beweise vernichten.

35. A Jetzt wird der Gastgeber dem Pfarrer den Mantel umlegen und den Hut reichen.
B Jetzt wird dem Pfarrer der Gastgeber den Mantel umlegen und den Hut reichen.
C Jetzt wird dem Pfarrer den Mantel der Gastgeber umlegen und den Hut reichen.
D Jetzt wird der Gastgeber umlegen dem Pfarrer den Mantel und den Hut reichen.

36. A Offensichtlich wird der Senator dem Richter den Betrug gestehen und die Ämter aufgeben.
B Offensichtlich wird dem Richter der Senator den Betrug gestehen und die Ämter aufgeben.
C Offensichtlich wird dem Richter den Betrug der Senator gestehen und die Ämter aufgeben.
D Offensichtlich wird der Senator gestehen dem Richter den Betrug und die Ämter aufgeben.

37. A Offensichtlich wird der Schriftsteller dem Kritiker den Inhalt schildern und die Verbesserungen nennen.
B Offensichtlich wird dem Kritiker der Schriftsteller den Inhalt schildern und die Verbesserungen nennen.
C Offensichtlich wird dem Kritiker den Inhalt der Schriftsteller schildern und die Verbesserungen nennen.
D Offensichtlich wird der Schriftsteller schildern dem Kritiker den Inhalt und die Verbesserungen nennen.

38. A Dort wird der Torwart dem Stürmer den Ball zuspielen und das Lob erhalten.
B Dort wird dem Stürmer der Torwart den Ball zuspielen und das Lob erhalten.
C Dort wird dem Stürmer den Ball der Torwart zuspielen und das Lob erhalten.
D Dort wird der Torwart zuspielen dem Stürmer den Ball und das Lob erhalten.

39. A Jetzt wird der Steuerberater dem Gutachter den Fall vorlegen und die Entscheidungen abwarten.
B Jetzt wird dem Gutachter der Steuerberater den Fall vorlegen und die Entscheidungen abwarten.
C Jetzt wird dem Gutachter den Fall der Steuerberater vorlegen und die Entscheidungen abwarten.
D Jetzt wird der Steuerberater vorlegen dem Gutachter den Fall und die Entscheidungen abwarten.

40. A Jetzt wird der Ehemann dem Schwager den Wein einschenken und das Gespräch anfangen.
B Jetzt wird dem Schwager der Ehemann den Wein einschenken und das Gespräch anfangen.
C Jetzt wird dem Schwager den Wein der Ehemann einschenken und das Gespräch anfangen.
D Jetzt wird der Ehemann einschenken dem Schwager den Wein und das Gespräch anfangen.

D-22
 Appendix D. Fillers from Behavioral and MRI Studies

41. A Dennoch wird der Offizier dem Admiral den Kapitän opfern und das Schiff retten.
B Dennoch wird dem Admiral der Offizier den Kapitän opfern und das Schiff retten.
C Dennoch wird dem Admiral den Kapitän der Offizier opfern und das Schiff retten.
D Dennoch wird der Offizier opfern dem Admiral den Kapitän und das Schiff retten.

42. A Vielleicht wird der Arbeiter dem Bankier den Lehrling ausreden und den Lohn sparen.
B Vielleicht wird dem Bankier der Arbeiter den Lehrling ausreden und den Lohn sparen.
C Vielleicht wird dem Bankier den Lehrling der Arbeiter ausreden und den Lohn sparen.
D Vielleicht wird der Arbeiter ausreden dem Bankier den Lehrling und den Lohn sparen.

43. A Anscheinend wird der Kommissar dem Priester den Flüchtling anvertrauen und die Familien schützen.
B Anscheinend wird dem Priester der Kommissar den Flüchtling anvertrauen und die Familien schützen.
C Anscheinend wird dem Priester den Flüchtling der Kommissar anvertrauen und die Familien schützen.
D Anscheinend wird der Kommissar anvertrauen dem Priester den Flüchtling und die Familien schützen.

44. A Danach wird der Fluggast dem Stadtrat den Begleiter vorstellen und die Maschine verlassen.
B Danach wird dem Stadtrat der Fluggast den Begleiter vorstellen und die Maschine verlassen.
C Danach wird dem Stadtrat den Begleiter der Fluggast vorstellen und die Maschine verlassen.
D Danach wird der Fluggast vorstellen dem Stadtrat den Begleiter und die Maschine verlassen.

45. A Offensichtlich wird dem Redakteur der Berater den Schreiber unterschieben und den Betrug begehen.
B Offensichtlich wird dem Redakteur den Schreiber der Berater unterschieben und den Betrug begehen.
C Offensichtlich wird der Berater dem Redakteur den Schreiber unterschieben und den Betrug begehen.
D Offensichtlich wird der Berater unterschieben dem Redakteur den Schreiber und den Betrug begehen.

46. A Gestern hat der Zahnarzt dem Onkel den Zahn gezogen und die Schmerzen beendet.
B Gestern hat dem Onkel der Zahnarzt den Zahn gezogen und die Schmerzen beendet.
C Gestern hat dem Onkel den Zahn der Zahnarzt gezogen und die Schmerzen beendet.
D Gestern hat der Zahnarzt gezogen dem Onkel den Zahn und die Schmerzen beendet.

47. A Ausserdem hat der Witwer dem Wahrsager den Schmuck geschenkt und das Geld bezahlt.
B Ausserdem hat dem Wahrsager der Witwer den Schmuck geschenkt und das Geld bezahlt.
C Ausserdem hat dem Wahrsager den Schmuck der Witwer geschenkt und das Geld bezahlt.
D Ausserdem hat der Witwer geschenkt dem Wahrsager den Schmuck und das Geld bezahlt.

48. A Danach hat der Betreuer dem Rollstuhlfahrer den Schuh angezogen und den Weg erklärt.
B Danach hat dem Rollstuhlfahrer der Betreuer den Schuh angezogen und den Weg erklärt.
C Danach hat dem Rollstuhlfahrer den Schuh der Betreuer angezogen und den Weg erklärt.
D Danach hat der Betreuer angezogen dem Rollstuhlfahrer den Schuh und den Weg erklärt.

D-23
MPI Series in Human Cognitive and Brain Sciences:
1 Anja Hahne 19 Silke Urban
Charakteristika syntaktischer und semantischer Prozesse bei der auditi- Verbinformationen im Satzverstehen
ven Sprachverarbeitung: Evidenz aus ereigniskorrelierten Potentialstudien
20 Katja Werheid
2 Ricarda Schubotz Implizites Sequenzlernen bei Morbus Parkinson
Erinnern kurzer Zeitdauern: Behaviorale und neurophysiologische
Korrelate einer Arbeitsgedächtnisfunktion 21 Doreen Nessler
Is it Memory or Illusion? Electrophysiological Characteristics of True and
3 Volker Bosch False Recognition
Das Halten von Information im Arbeitsgedächtnis: Dissoziationen
langsamer corticaler Potentiale 22 Christoph Herrmann
Die Bedeutung von 40-Hz-Oszillationen für kognitive Prozesse
4 Jorge Jovicich
An investigation of the use of Gradient- and Spin-Echo (GRASE) imaging 23 Christian Fiebach
for functional MRI of the human brain Working Memory and Syntax during Sentence Processing.
A neurocognitive investigation with event-related brain potentials and
5 Rosemary C. Dymond functional magnetic resonance imaging
Spatial Specificity and Temporal Accuracy in Functional Magnetic
Resonance Investigations 24 Grit Hein
Lokalisation von Doppelaufgabendefiziten bei gesunden älteren
6 Stefan Zysset Personen und neurologischen Patienten
Eine experimentalpsychologische Studie zu Gedächtnisabrufprozessen
unter Verwendung der funktionellen Magnetresonanztomographie 25 Monica de Filippis
Die visuelle Verarbeitung unbeachteter Wörter. Ein elektrophysiologischer
7 Ulrich Hartmann Ansatz
Ein mechanisches Finite-Elemente-Modell des menschlichen Kopfes
26 Ulrich Müller
8 Bertram Opitz Die katecholaminerge Modulation präfrontaler kognitiver Funktionen
Funktionelle Neuroanatomie der Verarbeitung einfacher und komplexer beim Menschen
akustischer Reize: Integration haemodynamischer und elektrophysiolo-
gischer Maße 27 Kristina Uhl
Kontrollfunktion des Arbeitsgedächtnisses über interferierende Information
9 Gisela Müller-Plath
Formale Modellierung visueller Suchstrategien mit Anwendungen bei der 28 Ina Bornkessel
Lokalisation von Hirnfunktionen und in der Diagnostik von Aufmerksam- The Argument Dependency Model: A Neurocognitive Approach to
keitsstörungen Incremental Interpretation
10 Thomas Jacobsen 29 Sonja Lattner
Characteristics of processing morphological structural and inherent case Neurophysiologische Untersuchungen zur auditorischen Verarbeitung
in language comprehension von Stimminformationen
11 Stefan Kölsch 30 Christin Grünewald
Brain and Music Die Rolle motorischer Schemata bei der Objektrepräsentation: Untersu-
A contribution to the investigation of central auditory processing with a chungen mit funktioneller Magnetresonanztomographie
new electrophysiological approach 31 Annett Schirmer
12 Stefan Frisch Emotional Speech Perception: Electrophysiological Insights into the
Verb-Argument-Struktur, Kasus und thematische Interpretation beim Processing of Emotional Prosody and Word Valence in Men and Women
Sprachverstehen 32 André J. Szameitat
13 Markus Ullsperger Die Funktionalität des lateral-präfrontalen Cortex für die Verarbeitung
The role of retrieval inhibition in directed forgetting – an event-related von Doppelaufgaben
brain potential analysis 33 Susanne Wagner
14 Martin Koch Verbales Arbeitsgedächtnis und die Verarbeitung ambiger Wörter in
Measurement of the Self-Diffusion Tensor of Water in the Human Brain Wort- und Satzkontexten
15 Axel Hutt 34 Sophie Manthey
Methoden zur Untersuchung der Dynamik raumzeitlicher Signale Hirn und Handlung: Untersuchung der Handlungsrepräsentation im
ventralen prämotorischen Cortex mit Hilfe der funktionellen Magnet-
16 Frithjof Kruggel Resonanz-Tomographie
Detektion und Quantifizierung von Hirnaktivität mit der funktionellen
Magnetresonanztomographie 35 Stefan Heim
Towards a Common Neural Network Model of Language Production and
17 Anja Dove Comprehension: fMRI Evidence for the Processing of Phonological and
Lokalisierung an internen Kontrollprozessen beteiligter Hirngebiete Syntactic Information in Single Words
mithilfe des Aufgabenwechselparadigmas und der ereigniskorrelierten
funktionellen Magnetresonanztomographie 36 Claudia Friedrich
Prosody and spoken word recognition: Behavioral and ERP correlates
18 Karsten Steinhauer
Hirnphysiologische Korrelate prosodischer Satzverarbeitung bei gespro- 37 Ulrike Lex
chener und geschriebener Sprache Sprachlateralisierung bei Rechts- und Linkshändern mit funktioneller
Magnetresonanztomographie
38 Thomas Arnold neurologischen Erkrankungen gemessen mit funktioneller Magnetreso-
Computergestützte Befundung klinischer Elektroenzephalogramme nanztomographie – Einflüsse von Händigkeit, Läsion, Performanz und
Perfusion
39 Carsten H. Wolters
Influence of Tissue Conductivity Inhomogeneity and Anisotropy on EEG/ 58 Jutta L. Mueller
MEG based Source Localization in the Human Brain Mechanisms of auditory sentence comprehension in first and second
language: An electrophysiological miniature grammar study
40 Ansgar Hantsch
Fisch oder Karpfen? Lexikale Aktivierung von Benennungsalternative bei 59 Franziska Biedermann
der Objektbenennung Auditorische Diskriminationsleistungen nach unilateralen Läsionen im
Di- und Telenzephalon
41 Peggy Bungert
Zentralnervöse Verarbeitung akustischer Informationen 60 Shirley-Ann Rüschemeyer
Signalidentifikation, Signallateralisation und zeitgebundene Informati- The Processing of Lexical Semantic and Syntactic Information in Spoken
onsverarbeitung bei Patienten mit erworbenen Hirnschädigungen Sentences: Neuroimaging and Behavioral Studies of Native and Non-
Native Speakers
42 Daniel Senkowski
Neuronal correlates of selective attention: An investigation of electro- 61 Kerstin Leuckefeld
physiological brain responses in the EEG and MEG The Development of Argument Processing Mechanisms in German.
An Electrophysiological Investigation with School-Aged Children and
43 Gert Wollny Adults
Analysis of Changes in Temporal Series of Medical Images
62 Axel Christian Kühn
44 Angelika Wolf Bestimmung der Lateralisierung von Sprachprozessen unter besondere
Sprachverstehen mit Cochlea-Implantat: EKP-Studien mit postlingual Berücksichtigung des temporalen Cortex, gemessen mit fMRT
ertaubten erwachsenen CI-Trägern
63 Ann Pannekamp
45 Kirsten G. Volz Prosodische Informationsverarbeitung bei normalsprachlichem und
Brain correlates of uncertain decisions: Types and degrees of uncertainty deviantem Satzmaterial: Untersuchungen mit ereigniskorrelierten
46 Hagen Huttner Hirnpotentialen
Magnetresonanztomographische Untersuchungen über die anatomische 64 Jan Derrfuß
Variabilität des Frontallappens des menschlichen Großhirns Functional specialization in the lateral frontal cortex: The role of the
47 Dirk Köster inferior frontal junction in cognitive control
Morphology and Spoken Word Comprehension: Electrophysiological 65 Andrea Mona Philipp
Investigations of Internal Compound Structure The cognitive representation of tasks – Exploring the role of response
48 Claudia A. Hruska modalities using the task-switching paradigm
Einflüsse kontextueller und prosodischer Informationen in der audito- 66 Ulrike Toepel
rischen Satzverarbeitung: Untersuchungen mit ereigniskorrelierten Contrastive Topic and Focus Information in Discourse – Prosodic
Hirnpotentialen Realisation and Electrophysiological Brain Correlates
49 Hannes Ruge 67 Karsten Müller
Eine Analyse des raum-zeitlichen Musters neuronaler Aktivierung im Die Anwendung von Spektral- und Waveletanalyse zur Untersuchung
Aufgabenwechselparadigma zur Untersuchung handlungssteuernder der Dynamik von BOLD-Zeitreihen verschiedener Hirnareale
Prozesse
68 Sonja A.Kotz
50 Ricarda I. Schubotz The role of the basal ganglia in auditory language processing: Evidence
Human premotor cortex: Beyond motor performance from ERP lesion studies and functional neuroimaging
51 Clemens von Zerssen 69 Sonja Rossi
Bewusstes Erinnern und falsches Wiedererkennen: Eine funktionelle MRT The role of proficiency in syntactic second language processing: Evidence
Studie neuroanatomischer Gedächtniskorrelate from event-related brain potentials in German and Italian
52 Christiane Weber 70 Birte U. Forstmann
Rhythm is gonna get you. Behavioral and neural correlates of endogenous control processes in task
Electrophysiological markers of rhythmic processing in infants with and switching
without risk for Specific Language Impairment (SLI)
71 Silke Paulmann
53 Marc Schönwiesner Electrophysiological Evidence on the Processing of Emotional Prosody:
Functional Mapping of Basic Acoustic Parameters in the Human Central Insights from Healthy and Patient Populations
Auditory System
72 Matthias L. Schroeter
54 Katja Fiehler Enlightening the Brain – Optical Imaging in Cognitive Neuroscience
Temporospatial characteristics of error correction
73 Julia Reinholz
55 Britta Stolterfoht Interhemispheric interaction in object- and word-related visual areas
Processing Word Order Variations and Ellipses: The Interplay of Syntax
and Information Structure during Sentence Comprehension 74 Evelyn C. Ferstl
The Functional Neuroanatomy of Text Comprehension
56 Claudia Danielmeier
Neuronale Grundlagen der Interferenz zwischen Handlung und visueller 75 Miriam Gade
Wahrnehmung Aufgabeninhibition als Mechanismus der Konfliktreduktion zwischen
Aufgabenrepräsentationen
57 Margret Hund-Georgiadis
Die Organisation von Sprache und ihre Reorganisation bei ausgewählten,
76 Juliane Hofmann 95 Henning Holle
Phonological, Morphological, and Semantic Aspects of Grammatical The Comprehension of Co-Speech Iconic Gestures: Behavioral, Electrophy-
Gender Processing in German siological and Neuroimaging Studies
77 Petra Augurzky 96 Marcel Braß
Attaching Relative Clauses in German – The Role of Implicit and Explicit Das inferior frontale Kreuzungsareal und seine Rolle bei der kognitiven
Prosody in Sentence Processing Kontrolle unseres Verhaltens
78 Uta Wolfensteller 97 Anna S. Hasting
Habituelle und arbiträre sensomotorische Verknüpfungen im lateralen Syntax in a blink: Early and automatic processing of syntactic rules as
prämotorischen Kortex des Menschen revealed by event-related brain potentials
79 Päivi Sivonen 98 Sebastian Jentschke
Event-related brain activation in speech perception: From sensory to Neural Correlates of Processing Syntax in Music and Language – Influ-
cognitive processes ences of Development, Musical Training and Language Impairment
80 Yun Nan 99 Amelie Mahlstedt
Music phrase structure perception: the neural basis, the effects of The Acquisition of Case marking Information as a Cue to Argument
acculturation and musical training Interpretation in German
An Electrophysiological Investigation with Pre-school Children
81 Katrin Schulze
Neural Correlates of Working Memory for Verbal and Tonal Stimuli in 100 Nikolaus Steinbeis
Nonmusicians and Musicians With and Without Absolute Pitch Investigating the meaning of music using EEG and fMRI
82 Korinna Eckstein 101 Tilmann A. Klein
Interaktion von Syntax und Prosodie beim Sprachverstehen: Untersu- Learning from errors: Genetic evidence for a central role of dopamine in
chungen anhand ereigniskorrelierter Hirnpotentiale human performance monitoring
83 Florian Th. Siebörger 102 Franziska Maria Korb
Funktionelle Neuroanatomie des Textverstehens: Kohärenzbildung bei Die funktionelle Spezialisierung des lateralen präfrontalen Cortex:
Witzen und anderen ungewöhnlichen Texten Untersuchungen mittels funktioneller Magnetresonanztomographie
84 Diana Böttger 103 Sonja Fleischhauer
Aktivität im Gamma-Frequenzbereich des EEG: Einfluss demographischer Neuronale Verarbeitung emotionaler Prosodie und Syntax: die Rolle des
Faktoren und kognitiver Korrelate verbalen Arbeitsgedächtnisses
85 Jörg Bahlmann 104 Friederike Sophie Haupt
Neural correlates of the processing of linear and hierarchical artificial The component mapping problem: An investigation of grammatical
grammar rules: Electrophysiological and neuroimaging studies function reanalysis in differing experimental contexts using eventrelated
brain potentials
86 Jan Zwickel
Specific Interference Effects Between Temporally Overlapping Action and 105 Jens Brauer
Perception Functional development and structural maturation in the brain‘s neural
network underlying language comprehension
87 Markus Ullsperger
Functional Neuroanatomy of Performance Monitoring: fMRI, ERP, and 106 Philipp Kanske
Patient Studies Exploring executive attention in emotion: ERP and fMRI evidence
88 Susanne Dietrich 107 Julia Grieser Painter
Vom Brüllen zum Wort – MRT-Studien zur kognitiven Verarbeitung Music, meaning, and a semantic space for musical sounds
emotionaler Vokalisationen
108 Daniela Sammler
89 Maren Schmidt-Kassow The Neuroanatomical Overlap of Syntax Processing in Music and
What‘s Beat got to do with ist? The Influence of Meter on Syntactic Language - Evidence from Lesion and Intracranial ERP Studies
Processing: ERP Evidence from Healthy and Patient populations
109 Norbert Zmyj
90 Monika Lück Selective Imitation in One-Year-Olds: How a Model‘s Characteristics
Die Verarbeitung morphologisch komplexer Wörter bei Kindern im Influence Imitation
Schulalter: Neurophysiologische Korrelate der Entwicklung
110 Thomas Fritz
91 Diana P. Szameitat Emotion investigated with music of variable valence – neurophysiology
Perzeption und akustische Eigenschaften von Emotionen in mensch- and cultural influence
lichem Lachen
111 Stefanie Regel
92 Beate Sabisch The comprehension of figurative language: Electrophysiological evidence
Mechanisms of auditory sentence comprehension in children with on the processing of irony
specific language impairment and children with developmental dyslexia:
A neurophysiological investigation 112 Miriam Beisert
Transformation Rules in Tool Use
93 Regine Oberecker
Grammatikverarbeitung im Kindesalter: EKP-Studien zum auditorischen 113 Veronika Krieghoff
Satzverstehen Neural correlates of Intentional Actions

94 Şükrü Barış Demiral 114 Andreja Bubić

Incremental Argument Interpretation in Turkish Sentence Comprehension Violation of expectations in sequence processing
115 Claudia Männel 135 Eugenia Solano-Castiella
Prosodic processing during language acquisition: Electrophysiological In vivo anatomical segmentation of the human amygdala and parcellati-
studies on intonational phrase processing on of emotional processing
116 Konstanze Albrecht 136 Marco Taubert
Brain correlates of cognitive processes underlying intertemporal choice for Plastizität im sensomotorischen System – Lerninduzierte Veränderungen
self and other in der Struktur und Funktion des menschlichen Gehirns
117 Katrin Sakreida 137 Patricia Garrido Vásquez
Nicht-motorische Funktionen des prämotorischen Kortex: Emotion Processing in Parkinson’s Disease:
Patientenstudien und funktionelle Bildgebung The Role of Motor Symptom Asymmetry
118 Susann Wolff 138 Michael Schwartze
The interplay of free word order and pro-drop in incremental sentence Adaptation to temporal structure
processing: Neurophysiological evidence from Japanese
139 Christine S. Schipke
119 Tim Raettig Processing Mechanisms of Argument Structure and Case-marking in
The Cortical Infrastructure of Language Processing: Evidence from Child Development: Neural Correlates and Behavioral Evidence
Functional and Anatomical Neuroimaging
140 Sarah Jessen
120 Maria Golde Emotion Perception in the Multisensory Brain
Premotor cortex contributions to abstract and action-related relational
processing 141 Jane Neumann
Beyond activation detection: Advancing computational techniques for
121 Daniel S. Margulies the analysis of functional MRI data
Resting-State Functional Connectivity fMRI: A new approach for asses-
sing functional neuroanatomy in humans with applications to neuroa- 142 Franziska Knolle
natomical, developmental and clinical questions Knowing what’s next: The role of the cerebellum in generating
predictions
122 Franziska Süß
The interplay between attention and syntactic processes in the adult and 143 Michael Skeide
developing brain: ERP evidences Syntax and semantics networks in the developing brain

123 Stefan Bode 144 Sarah M. E. Gierhan

From stimuli to motor responses: Decoding rules and decision mecha- Brain networks for language
nisms in the human brain Anatomy and functional roles of neural pathways supporting language
comprehension and repetition
124 Christiane Diefenbach
Interactions between sentence comprehension and concurrent action: 145 Lars Meyer
The role of movement effects and timing The Working Memory of Argument-Verb Dependencies
Spatiotemporal Brain Dynamics during Sentence Processing
125 Moritz M. Daum
Mechanismen der frühkindlichen Entwicklung des Handlungsverständ-
nisses
126 Jürgen Dukart
Contribution of FDG-PET and MRI to improve Understanding, Detection
and Differentiation of Dementia
127 Kamal Kumar Choudhary
Incremental Argument Interpretation in a Split Ergative Language:
Neurophysiological Evidence from Hindi
128 Peggy Sparenberg
Filling the Gap: Temporal and Motor Aspects of the Mental Simulation of
Occluded Actions
129 Luming Wang
The Influence of Animacy and Context on Word Order Processing: Neuro-
physiological Evidence from Mandarin Chinese
130 Barbara Ettrich
Beeinträchtigung frontomedianer Funktionen bei Schädel-Hirn-Trauma
131 Sandra Dietrich
Coordination of Unimanual Continuous Movements with External Events
132 R. Muralikrishnan
An Electrophysiological Investigation Of Tamil Dative-Subject Construc-
tions
133 Christian Obermeier
Exploring the significance of task, timing and background noise on
gesture-speech integration
134 Björn Herrmann
Grammar and perception: Dissociation of early auditory processes in the
brain