Research Methods in Psycholinguistics and the Neurobiology of Language

Guides to Research Methods in Language and Linguistics

Series Editor: Li Wei, Centre for Applied Linguistics, University College London

The science of language encompasses a truly interdisciplinary field of research, with a wide range of focuses, approaches, and objectives. While linguistics has its own traditional approaches, a variety of other intellectual disciplines have contributed methodological perspectives that enrich the field as a whole. As a result, linguistics now draws on state‐of‐the‐art work from such fields as psychology, computer science, biology, neuroscience and cognitive science, sociology, music, philosophy, and anthropology.

The interdisciplinary nature of the field presents both challenges and opportu-nities to students who must understand a variety of evolving research skills and methods. The Guides to Research Methods in Language and Linguistics addresses these skills in a systematic way for advanced students and beginning researchers in language science. The books in this series focus especially on the relationships between theory, methods, and data—the understanding of which is fundamental to the successful completion of research projects and the advancement of knowledge.

1. The Blackwell Guide to Research Methods in Bilingualism and MultilingualismEdited by Li Wei and Melissa G. Moyer

2. Research Methods in Child Language: A Practical GuideEdited by Erika Hoff

3. Research Methods in Second Language Acquisition: A Practical GuideEdited by Susan M. Gass and Alison Mackey

4. Research Methods in Clinical Linguistics and Phonetics: A Practical GuideEdited by Nicole Müller and Martin J. Ball

5. Research Methods in Sociolinguistics: A Practical GuideEdited by Janet Holmes and Kirk Hazen

6. Research Methods in Sign Language Studies: A Practical GuideEdited by Eleni Orfanidou, Bencie Woll, and Gary Morgan

7. Research Methods in Language Policy and Planning: A Practical GuideEdited by Francis Hult and David Cassels Johnson

8. Research Methods in Intercultural Communication: A Practical GuideEdited by Zhu Hua

9. Research Methods in Psycholinguistics and the Neurobiology of Language: A Practical GuideEdited by Annette M. B. de Groot and Peter Hagoort

Research Methods in Psycholinguistics and the Neurobiology of Language

A Practical Guide

Edited by

Annette M. B. de Groot and Peter Hagoort

This edition first published 2018© 2018 John Wiley & Sons, Inc.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by law. Advice on how to obtain permission to reuse material from this title is available at http://www.wiley.com/go/permissions.

The right of Annette M. B. de Groot and Peter Hagoort to be identified as the authors of the editorial material in this work has been asserted in accordance with law.

Registered OfficesJohn Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA

Editorial Office9600 Garsington Road, Oxford, OX4 2DQ, UK

For details of our global editorial offices, customer services, and more information about Wiley products visit us at www.wiley.com.

Wiley also publishes its books in a variety of electronic formats and by print‐on‐demand. Some content that appears in standard print versions of this book may not be available in other formats.

Limit of Liability/Disclaimer of WarrantyWhile the publisher and authors have used their best efforts in preparing this work, they make no representations or warranties with respect to the accuracy or completeness of the contents of this work and specifically disclaim all warranties, including without limitation any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives, written sales materials or promotional statements for this work. The fact that an organization, website, or product is referred to in this work as a citation and/or potential source of further information does not mean that the publisher and authors endorse the information or services the organization, website, or product may provide or recommendations it may make. This work is sold with the understanding that the publisher is not engaged in rendering professional services. The advice and strategies contained herein may not be suitable for your situation. You should consult with a specialist where appropriate. Further, readers should be aware that websites listed in this work may have changed or disappeared between when this work was written and when it is read. Neither the publisher nor authors shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages.

Library of Congress Cataloging‐in‐Publication data applied for

Hardback: 9781119109846Paperback: 9781119109853

Cover image: (Figure) Designed by Hartebeest, Nijmegen, The NetherlandsCover design by Wiley

Set in 10/12pt Sabon by SPi Global, Pondicherry, India

10 9 8 7 6 5 4 3 2 1

Contents

List of Figures viiList of Tables ixNotes on Contributors xPreface xvi

1 Habituation Techniques 1Christopher T. Fennell

2 Visual Preference Techniques 18Roberta Michnick Golinkoff, Melanie Soderstrom, Dilara Deniz Can, and Kathy Hirsh‐Pasek

3 Assessing Receptive and Expressive Vocabulary in Child Language 40Virginia A. Marchman and Philip S. Dale

4 Eye‐Movement Tracking During Reading 68Reinhold Kliegl and Jochen Laubrock

5 The Visual World Paradigm 89Anne Pier Salverda and Michael K. Tanenhaus

6 Word Priming and Interference Paradigms 111Zeshu Shao and Antje S. Meyer

7 Structural Priming 130Holly P. Branigan and Catriona L. Gibb

8 Conversation Analysis 151Elliott M. Hoey and Kobin H. Kendrick

9 Virtual Reality 174Daniel Casasanto and Kyle M. Jasmin

10 Studying Psycholinguistics out of the Lab 190Laura J. Speed, Ewelina Wnuk, and Asifa Majid

11 Computational Modeling 208Ping Li and Xiaowei Zhao

vi Contents

12 Corpus Linguistics 230Marc Brysbaert, Paweł Mandera, and Emmanuel Keuleers

13 Electrophysiological Methods 247Joost Rommers and Kara D. Federmeier

14 Hemodynamic Methods: fMRI and fNIRS 266Roel M. Willems and Alejandrina Cristia

15 Structural Neuroimaging 288Stephanie J. Forkel and Marco Catani

16 Lesion Studies 310Juliana V. Baldo and Nina F. Dronkers

17 Molecular Genetic Methods 330Carolien G. F. de Kovel and Simon E. Fisher

Index 354

List of Figures

Figure 1.1 Examples of various infant language habituation tasks 5Figure 1.2 Mean looking times across various trial types in Fennell

and Byers‐Heinlein (2014) 12Figure 2.1 The Intermodal Preferential Looking Paradigm 22Figure 2.2 Means of single longest look in seconds to infant‐directed (IDS)

and adult‐directed (ADS) speech stimuli 25Figure 2.3 The Interactive Intermodal Preferential Looking Paradigm 26Figure 2.4 Visual fixation to original label, new label, and recovery trials

by condition 28Figure 2.5 Eye gaze shifts toward and away from target in

looking‐while‐listening task by age 30Figure 2.6 The Headturn Preference Procedure 33Figure 4.1 Typical eye tracker set up 71Figure 4.2 Illustration of the gaze‐contingent moving‐window (top) and

boundary (bottom) paradigms 73Figure 4.3 Velocity‐based saccade detection 75Figure 4.4 Determination of word boundaries with PRAAT software 80Figure 4.5 Main effect of eye‐voice span and its interaction with predictability 81Figure 5.1 Example of a screen‐based visual world paradigm experimental

set up 90Figure 5.2 Example visual display modeled after Altmann and Kamide (1999) 91Figure 5.3 Timing of target fixations for each trial, for one participant

and fixation proportions computed for same data 100Figure 5.4 Proportion of fixations over time (from target‐word onset) to

target (goat), cohort competitor (goal), and distractor in neutral and constraining verb conditions in Experiment 1 in Dahan and Tanenhaus (2004) 104

Figure 6.1 An illustration of the trial structure in Meyer and Schvaneveldt (1971) 113Figure 6.2 An illustration of the prime‐target pairs used in Glaser

and Düngelhoff (1984) 114Figure 6.3 Results obtained by Glaser and Düngelhoff (1984) 115Figure 6.4 Illustration of trial structures in the masked and unmasked

conditions in de Wit and Kinoshita (2015) 119

viii List of Figures

Figure 7.1 Example trial in a picture‐matching comprehension priming paradigm 138

Figure 7.2 Example trial in a picture‐matching and picture‐description production priming paradigm 140

Figure 7.3 Example trial in a sentence recall production priming paradigm 142Figure 10.1 Comparison of cut and break verbs in Chontal, Hindi,

and Jalonke 195Figure 11.1 The basic architecture of a Simple Recurrent Network (SRN) 213Figure 11.2 A sketch of the probabilistic model that incorporates

distributional statistics from cross‐situational observation and prosodic and attentional highlights from social gating 219

Figure 11.3 A sketch of the DevLex‐II model 221Figure 11.4 Vocabulary spurt simulated by DevLex‐II (591 target words) 223Figure 13.1 Idealized example of an event‐related potential waveform

in response to a visual stimulus, with labeled positive and negative peaks 248

Figure 13.2 Grand average ERPs from three parietal channels, elicited by the final words in the three conditions 257

Figure 13.3 Simulated EEG data illustrating the difference between ERPs and time‐frequency analyses in their sensitivity to phase‐locked (evoked) and non‐phase‐locked (induced) activity 260

Figure 14.1 An anatomical scan of the head and the brain (A), and Functional MRI images (B) 269

Figure 14.2 Example of an idealized BOLD curve, sometimes called the hemodynamic response function (HRF) 271

Figure 14.3 A statistical map overlaid on an anatomical brain scan 276Figure 14.4 Image of a 5‐month‐old infant wearing a fNIRS cap, including

a schematic illustration of the path of light between a source (star) and a detector (circle), through the scalp (dashed line) and cortical tissue (in gray) 278

Figure 14.5 Sample of signal in fNIRS studies 280Figure 15.1 Imaging of an acute patient presenting with anomia

following left inferior parietal and frontal lobe stroke 293Figure 15.2 Lesion mapping based on T1‐weighted data (A), on a diffusion

tractography atlas (B), and an example of extracting tract‐based measurements from tractography (C) 299

Figure 15.3 Anatomical variability in perisylvian white matter anatomy and its relation to post‐stroke language recovery 302

Figure 16.1 A schematic illustration showing the steps involved in a VLSM analysis 317

Figure 16.2 Overlay of patients’ lesions 320Figure 16.3 Power analysis map showing the degree of power in our sample,

given a medium effect size and alpha set at p < .05 321Figure 16.4 VLSM results showing neural correlates of auditory

word recognition with varying levels of correction 322Figure 17.1 Transmission of DNA between generations 332Figure 17.2 Visualization of Sanger sequencing results 338Figure 17.3 Next generation sequencing 339Figure 17.4 Visualization of SNP‐chip results 340

List of Tables

Table 1.1 Mock habituation data from four experiments with looking time as the dependent variable 8

Table 1.2 Steps in data collection and analyses 9Table 2.1 Visual and linguistic stimuli used to teach two novel

words in either infant‐directed or adult‐directed speech 24Table 2.2 Ten- to 12‐month‐old infants saw two types of discrimination

trials, one to test for path discrimination and one for actor discrimination 31

Table 3.1 Overview of instruments/analysis tools for studying vocabulary development in children 45

Table 3.2 Example transcript from CHILDES 48Table 4.1 Definitions of location and duration eye‐tracking measures 77Table 4.2 Practical issues related to eye‐tracking during reading 82Table 7.1 Example structural alternations studied in structural priming

experiments 134Table 7.2 Stimulus materials for a hypothetical small clause study 144Table 7.3 Hypothetical results for a small clause study 145Table 8.1 Questions and assessments from Extracts 8.1 to 8.3 161Table 12.1 Excerpt from the SUBTLEX‐US database for the word “appalled” 234Table 12.2 Stimuli used in a semantic priming experiment by de Mornay

Davies (1998) 239Table 17.1 Example of genotyping chip results for four individuals

and five polymorphisms 340

Notes on Contributors

Juliana V. Baldo is Research Scientist at Veterans Affairs Northern California Health Care System and Adjunct Professor of Psychology at California State University East Bay. She specializes in research related to language and neuropsychological disorders arising from brain injury, including both stroke and traumatic brain injury. Dr. Baldo has also published a number of articles on language impairments in aphasia and associated cognitive deficits, and has utilized various brain imaging methodologies to better understand the neural basis of these impairments.

Holly P. Branigan is Professor of Psychology of Language and Cognition at the University of Edinburgh. Her research uses a wide range of experimental psycholin-guistic methods to investigate language production in monologue and dialogue, with a particular focus on syntactic processing and representation in adults and in typically and atypically developing children.

Marc Brysbaert is Professor of Psychology at Ghent University. In recent years his word recognition research has shifted to big data, including the calculation and validation of improved word frequency measures, running megastudies to establish word processing times, collecting subjective measures of word features (concreteness, valence, arousal, age‐of‐acquisition), and investigating the use and validation of semantic vectors.

Daniel Casasanto is Associate Professor of Human Development and Psychology at Cornell University, Ithaca, NY. He studies how physical and social experiences shape our brains and minds.

Marco Catani holds a joint affiliation as clinical senior lecturer and honorary consultant psychiatrist with the Department of Forensics and Neurodevelopmental Sciences and the Department Neuroimaging at King’s College London. He studies the lateralization of human brain networks and their implications for post stroke recovery from aphasia and neglect.

Alejandrina Cristia is a researcher at the Centre National de Recherche Scientifique, affiliated with the Laboratoire de Sciences Cognitives et Psycholinguistique

Notes on Contributors xi

(ENS, EHESS, CNRS), Département d’Etudes Cognitives, Ecole Normale Supérieure, PSL Research University. She studies early language acquisition.

Philip S. Dale is Professor Emeritus of Speech & Hearing Sciences at the University of New Mexico and Visiting Professor at King’s College London. He is a co‐developer of the MacArthur‐Bates Communicative Development Inventories. He has conducted research on the assessment, genetic and environmental causes, and consequences of individual differences in early language development, with a special interest in late talkers. He also conducted research on the effectiveness of intervention programs for young children.

Annette M. B. de Groot is Professor of Psycholinguistics at the University of Amsterdam. Her early research focused on priming effects on word recognition, the structure of the mental lexicon, and the psychology of reading and spelling. Later her research shifted toward bilingualism and multilingualism, studying bilingual word recognition and word production, foreign‐language vocabulary acquisition, translation and simultaneous interpreting, and the influence of bilingualism on various aspects of verbal and non‐verbal cognition.

Carolien G. F. de Kovel is a researcher at the Max Planck Institute for Psycholinguistics in Nijmegen, the Netherlands. She studies the genetic background of lateralization in humans. Carolien received her PhD in Biology at the University of Utrecht, the Netherlands.

Dilara Deniz Can is a scientist/practitioner who has obtained her PhD and Educational Specialist degrees from the University of Delaware in School Psychology. She completed a post‐doctoral research fellowship at the University of Washington’s Institute for Learning and Brain Sciences working with vulnerable children ages 3 to 5, studying the links between brain, environment, and language development. She has worked as a school psychologist in public schools of WA State, completing psycho‐educational evaluations for young children and adolescents.

Nina F. Dronkers is a VA Research Career Scientist and Director of the Center for Aphasia and Related Disorders with the Department of Veterans Affairs Northern California Health Care System. She is also an Adjunct Professor at the University of California, Davis in the Department of Neurology. She received her interdisciplinary Ph.D. degree in Neuropsychology from the University of California, Berkeley, and has since used novel techniques to identify new brain structures that play critical roles in the processing of speech and language, and studies how these relate to other cognitive skills.

Kara D. Federmeier is a Professor in the Department of Psychology, the Program in Neuroscience, and the Beckman Institute of Advanced Science and Technology at the University of Illinois. Her research uses event‐related potentials, EEG, and eye tracking to understand the mechanisms involved in language comprehension and meaning processing, the nature of hemispheric differences in cognitive processing,

xii Notes on Contributors

the impact of age‐related changes on language and memory functioning, and the effects of literacy on cognitive processing in adulthood.

Christopher T. Fennell is an Associate Professor at the University of Ottawa and Director of the Language Development Lab. His research focuses on speech perception, phonological development, and lexical acquisition in monolingual and bilingual infants. He has published numerous articles on infant language development, many using habit-uation methods, in journals such as Child Development, Developmental Science, Infancy, and Bilingualism: Language and Cognition.

Simon E. Fisher is a Director of the Max Planck Institute for Psycholinguistics and Professor of Language and Genetics at the Donders Institute, Radboud University, in Nijmegen, the Netherlands. He obtained a Natural Sciences degree from Cambridge University, UK, followed by a DPhil in Human Genetics at Oxford University, UK. His research uses genes as molecular windows into the basis of human cognitive traits, with a particular focus on speech, language, and reading skills.

Stephanie J. Forkel is senior neuroimaging research scientist in the Department for Neuroimaging at King’s College London. She investigates the lateralization of human brain networks and their implications for post stroke recovery from aphasia and neglect.

Catriona L. Gibb was a PhD student in the Psychology Department at the University of Edinburgh. Her research used experimental methods to study the psycholinguis-tics of bilingualism, most recently focusing on the nature of syntactic processing and syntactic representation in early and late bilinguals.

Roberta Michnick Golinkoff is Unidel H. Rodney Sharp Professor at the University of Delaware. She has received numerous awards for her research and her dissemina-tion work. Funded by federal agencies, she has over 150 publications, as well as 16 books and monographs. Passionate about the dissemination of psychological science for improving our schools and families’ lives, her latest book is Becoming Brilliant: What Science Tells Us About Raising Successful Children (APA Press).

Peter Hagoort is Academy Professor of the Royal Netherlands Academy of Arts and Sciences, and Professor of Cognitive Neuroscience at Radboud University. He is a Director of the Max Planck Institute for Psycholinguistics and of the Donders Institute for Brain, Cognition, and Behaviour. His research focuses on the neurobiolog-ical infrastructure for language with the help of advanced neuroimaging methods such as fMRI, MEG, and TMS.

Kathy Hirsh‐Pasek is the Stanley and Debra Lefkowitz Faculty Fellow in the Department of Psychology at Temple University and a Senior Fellow at the Brookings Institution. Author of 14 books and hundreds of publications, she is the recipient of numerous awards, is President of the International Society for Infant Studies, and served as an Associate Editor of Child Development. An expert in early learning (language, literacy, STEM), she is dedicated to translating basic science for public consumption.

Notes on Contributors xiii

Elliott M. Hoey is a PhD student in the Language and Cognition Department at the Max Planck Institute for Psycholinguistics in Nijmegen, The Netherlands. He uses conversation analytic and interactional linguistic methods to study the multimodal constitution of mundane social settings. His recent research has addressed the inter-actional uses of sighing and drinking, and participants’ conduct during extended silences in conversation.

Kyle M. Jasmin is a postdoctoral researcher at University College London. He studies the cognitive neuroscience of language and communication, in typical and atypical populations.

Kobin H. Kendrick is a Lecturer in the Department of Language and Linguistic Science at the University of York. His research uses conversation analysis to investigate basic organizations of talk‐in‐interaction such as turn‐taking, action‐sequencing, and repair. A recent line of research has examined the multimodal practices that participants in interaction use to “recruit” others to assist them with troubles that emerge in everyday activities.

Emmanuel Keuleers is an Assistant Professor in the Department of Communication and Information Sciences at Tilburg University, the Netherlands. He has done extensive research on visual word recognition and computational modeling of morphology. In his current research he is particularly interested in effects of age and multilingualism on vocabulary growth and in the application and interpretation of crowd‐based lexical measures to language processing.

Reinhold Kliegl is Professor of Psychology, Department of Psychology, University of Potsdam, Germany. His research focuses on how the dynamics of language‐related, perceptual, and oculomotor processes subserve attentional control, using reading as one experimental venue. He also specializes in applied multivariate statistics, espe-cially linear mixed models. He is an active promoter of Open Science with the Potsdam Mind Research Repository (PMR2); (http://read.psych.uni‐potsdam.de/pmr2/).

Jochen Laubrock is Senior Research Scientist, Department of Psychology, University of Potsdam, Germany. His research focuses on how perceptual, attentional, and (oculo‐)motor processes interact in the planning of goal‐related behavior. A special interest is in the co‐operation of foveal, parafoveal, and peripheral processing for the control of saccade timing and target selection in reading, related tasks (RAN), and scene perception, and how these processes operating at quite different time‐scales cooperate when reading graphic literature.

Ping Li is Professor of Psychology, Linguistics, and Information Sciences and Technology, Associate Director of the Institute for CyberScience, and Co‐Director of the Center for Brain, Behavior, and Cognition at Pennsylvania State University. He holds a Ph.D. (1990) in psycholinguistics from the University of Leiden. His research is focused on the neurocognitive and computational mechanisms of language acquisition and bilingualism.

xiv Notes on Contributors

Asifa Majid is Professor of Language, Communication, and Cultural Cognition at the Centre for Language Studies, Radboud University Nijmegen and Affiliated Principle Investigator at the Max Planck Institute for Psycholinguistics and Donders Institute for Brain, Cognition, and Behaviour in Nijmegen, The Netherlands. She investigates concepts in language and cognition by conducting cross‐cultural and developmental studies. At the heart of her research program lie the questions: Where do our categories come from, and how widely are they shared across languages and cultures?

Paweł Mandera is a Postdoctoral Researcher at the Department of Experimental Psychology, Ghent University. In his research he brings together methods from com-puter science and psychology to study how text corpora and other sources of behavioral data can be used to advance our understanding of human language processing.

Virginia A. Marchman is a Research Associate in Psychology at Stanford University and Adjunct Associate Professor in the School of Behavioral and Brain Sciences at the University of Texas at Dallas. She is a member of the Advisory Board of the MacArthur‐Bates Communicative Development Inventories and a contributing member of Wordbank. Her research focuses on the causes and consequences of individual differences in language processing efficiency and vocabulary development in monolingual and bilingual children.

Antje S. Meyer (PhD Radboud University) is a professor at Radboud University and director at the Max Planck Institute for Psycholinguistics in Nijmegen. Before taking up her appointments in Nijmegen in 2010, she was a professor of psycholinguistics at the University of Birmingham, UK. Meyer has worked on various aspects of psycho-linguistics, in particular word and sentence production, dialogue, and the relationship between visual‐conceptual and linguistic processing.

Joost Rommers is a postdoctoral researcher in the Psychology Department and the Beckman Institute for Advanced Science and Technology, University of Illinois. His  research uses electrophysiological and eye‐tracking methods to investigate language comprehension and production. One focus concerns the mechanisms and consequences of predicting upcoming language input.

Anne Pier Salverda is a Research Associate in the Department of Brain and Cognitive Sciences at the University of Rochester. He did his graduate work at the Max‐Planck‐Institute for Psycholinguistics in Nijmegen. His research focuses on speech perception and spoken‐word recognition.

Zeshu Shao (PhD Radboud University) is a research scientist at the Max Planck Institute for Psycholinguistics in Nijmegen. She has worked on speech production, specifically on the attention control mechanism influencing the planning and pro-duction of spoken words in a wide range of populations, and the effects of social network structure on lexical choice.

Notes on Contributors xv

Melanie Soderstrom is Associate Professor and Associate Head in the Department of Psychology at the University of Manitoba. She has published a number of studies using the Headturn Preference Procedure on infants’ sensitivity to the prosodic char-acteristics of speech and their understanding of grammatical dependencies. She is currently active in initiatives to automate analysis of large scale recordings of children’s real‐world language experiences.

Laura J. Speed is Postdoctoral Researcher at the Centre for Language Studies, Radboud University Nijmegen, The Netherlands. She conducts psychological research on the interplay between language and the senses. Her PhD thesis investigated an embodied account of language comprehension: How perception and action systems contribute to the understanding of words and sentences. Her current work focuses on language and olfaction—how we talk about smell and understand language about smell, and how language and information from multiple perceptual modalities can influence odour cognition.

Michael K. Tanenhaus is the Beverly Petterson Bishop and Charles W. Bishop Professor of Brain and Cognitive Sciences at the University of Rochester and a Chair Professor of Nanjing Normal University. His research with the Visual World Paradigm has spanned topics in spoken language processing ranging from speech perception to interactive conversation.

Roel M. Willems is Associate Professor at the Centre for Language Studies and Donders Institute, Radboud University, Nijmegen. He studies the role of mental sim-ulation during narrative comprehension.

Ewelina Wnuk is Postdoctoral Researcher at the Centre for Language Studies, Radboud University Nijmegen. She conducts fieldwork‐based research among the speakers of Maniq—an Austroasiatic language spoken by a group of nomadic hunter‐gatherers in Thailand. Her research interests include semantics, grammar, the relationship between language and culture, and the language of perception. In her recent work, she has been focusing on the language of smell and its relationship to cognition.

Xiaowei Zhao is Associate Professor of Psychology at Emmanuel College, Boston. He holds a B.S. (1998) as well as a Ph.D. (2003) in physics from Nankai University in China. Dr. Zhao works in the field of computational modeling of language development, knowledge representation, and bilingualism. He is the President‐elect of the Society for Computers in Psychology (term 2016‐2017).

Preface

In many aspects the human language system is a unique support system for communication and thinking. Ways to investigate this complex cognitive capacity were traditionally restricted to observational and behavioral methods in healthy people and neuropsychological patients with a language disorder. In recent decades this picture has changed dramatically. Partly due to technological developments and partly as a result of developments in other fields of research, methods to study language and communication have seen a vast increase in number and level of sophistication. Due to the technological progress in computing power, we are now able to build way more advanced computational models of language processing than ever before. Thanks to developments in neuroimaging and genetic sequencing, we are able to study the neural basis and the genetic underpinnings of the language‐ready brain in an unprecedented manner. These developments, however, come at a price. To be able to appreciate research findings or actively participate in this research field, one has to be acutely aware of the ins and outs of the research methods that are currently available. Until now a volume that summarizes and discusses all available methods in this field of research was missing. Research Methods in Psycholinguistics and the Neurobiology of Language intends to fill this gap. It provides a comprehensive overview of all relevant methods currently used in research on human language and communication. Some of them have their roots in psycholinguistics, others were introduced from other fields of science such as the biological sciences. Some require highly specialized technical knowledge and skills, whereas others take little time and effort to learn. For some methods a modest, inexpensive laboratory infrastructure suffices, whereas others depend on equipment that takes millions to acquire and interdisciplinary groups of specialists to operate. Some are offline methods that only measure the outcome of mental processing, whereas others continuously monitor mental processes as they unfold in real time, producing information‐rich and dense datasets. Presenting this diverse collection of methods, we anticipate that this book will be a useful guide for doctoral students, postdocs, and active researchers in our field who would want to inform themselves about the basics, the advantages and disadvantages of available research methods, and to get for each one of them pointers to additional method‐related information and best practice examples.

While conceiving this book we wondered how the great diversity of methods used in the study of language—its acquisition, use, neural and genetic basis, and disor-ders—could be covered within the limited space available. The solution was to not

Preface xvii

focus on the specific type of research methods called tasks, of which an innumerable variety exists, but on a broader notion of what research methods are. A task is what participants in an experiment are asked to do, for instance, to name the objects on a set of pictures shown to them. The participants’ behavioral and/or brain responses are registered and constitute the database from which the researcher subsequently extracts information. A research method in the more general sense that we had in mind for this volume is a much broader construct, one that covers a complex of procedures to study the question of interest (e.g., designing a study, constructing stimulus materials, and collecting and analysing the data), and that also includes the technical apparatus, tools, and instruments that support these procedures. Although many methods in this broad sense include data gathering by having participants per-form some task, other methods do without this altogether because the data already exist (corpus linguistics; Chapter 12) or because the method produces artificially generated data (computational modeling; Chapter 11). There are also methods that elicit data from participants without the latter explicitly being asked to perform some task (e.g., the habituation techniques and visual preference techniques presented in Chapters 1 and 2, respectively). Other methods can be combined with a multitude of different tasks (e.g., word priming and interference paradigms, Chapter 6; struc-tural priming, Chapter 7; the electrophysiological and hemodynamic neuroimaging methods presented in Chapters 13 and 14, respectively). All this shows that tasks and methods are not the same things.

A feature that characterizes many methods in the broad sense of the word is that they are domain‐nonspecific. Those developed within psycholinguistics can typically be used in various of its sub‐fields: They are suitable to address questions concerning more than one, or all three, of psycholinguistics’ main areas of study (language acqui-sition, comprehension, and production) and/or to answer questions about multiple linguistic domains (e.g., phonology, morphology, syntax, and semantics). The neuro-biological methods included in this volume are even more multipurpose, not being restricted to studying language but domain‐general pur sang, also applicable in studying other areas of cognition and other aspects of human (and animal) behavior.

While the majority of the 17 contributions to this book present domain‐nonspecific methods, a couple of them deal with domain‐specific methods: Chapter 3 presents three approved methods for assessing vocabulary in children (language sampling, parent report, and direct assessment); Chapter 4 discusses the ins and outs of the presumably most ecologically valid behavioral research method for examining the reading process: the tracking of eye‐movements; Chapter 8 exclusively deals with conversation analysis. But even these domain‐specific methods allow variability in how they are used and are thus able to inform multiple aspects of language processing. For instance, having the participants read complete paragraphs is what qualifies eye‐movement tracking as an ecologically valid method to study reading, but the stimulus does not need to be a whole paragraph. Sentences, even single words, may also serve as stimuli and, when they do, inform accounts of syntactic parsing, semantic analysis, and word recognition. Similarly, though the primary goal of conversation analysis is to study human social interactions and how people perform actions through talking, the database on which the analyses are done, often a corpus of naturally occurring conversations, contains information on all aspects of the conversational partners’ language use and, thus, on phonology, vocabulary, and more.

xviii Preface

In addition to guaranteeing a broad coverage of relevant research methods by predominantly selecting domain‐general methods, the volume’s coverage was increased yet further by inviting authors to present several related methods within a single chapter, directing the readers to these methods’ similarities and differences. For instance, the authors of Chapter 2 contrast multiple conceptually related variants of the visual‐preference technique to study language development in very young children, at an age at which they do not yet produce language or their verbal produc-tions are still incomprehensible. The differences between the various implementa-tions of the general method are often subtle and could easily escape readers if not presented in opposition. Similarly, the authors of Chapter 14 discuss two non‐invasive functional neuroimaging methods, fMRI and fNIRS, that both make use of the fact that neural activity leads to changes in the local cerebral blood flow in the brain and that can both reveal which parts of the brain are activated while participants per-form a particular task. Contrasting the pros and cons of these two related techniques within a single chapter will help readers to make a well‐informed choice between the two during the planning of their own research project. Likewise, after detailing the specifics of the EEG/ERP methodology, in which electrical brain activity can be measured with a temporal resolution in the order of milliseconds, the authors of Chapter 13 contrast it with MEG, which provides a record of the magnetic activity of the brain. Chapter 15 differentiates multiple non‐invasive techniques for structural neuroimaging based on MRI, which reveals the neuroanatomy of language with good spatial resolution. Among the presented methods is tractography, a novel tech-nique for visualizing white matter pathways in the living human brain. Chapter 16 also presents various structural neuroimaging methods, but whereas in Chapter 15 the major focus is on the healthy brain, in this contribution the emphasis is on the lesioned brain. Yet another example of a chapter that presents several related methods is Chapter 17, where inter‐individual variability in language skills is linked to genetic variation. The specific method used depends on whether the studied trait is suspected to be monogenetic (due to a single genetic variant) or multifactorial (resulting from the combined effects of multiple genes). Finally, the chapters dealing with the Visual World Paradigm (Chapter 5) and priming (Chapters 6 and 7) actually concern families of related methods (e.g., masked priming and cross‐modal priming).

The inevitable consequence of choosing domain‐nonspecific methods as themes for the separate chapters was that ways of organizing them that appeared obvious at first sight turned out to be neither feasible nor appropriate on second thoughts: The chapters could not be organized according to the main areas of language study, input and output modalities, or the various structural subsystems that languages consist of. After all, most of the presented methods are not specifically tied to any such subdi-vision of study. A presentation according to the type of measures used, behavioral or neurobiological, would be more appropriate and feasible but is complicated by the fact that studies using neurobiological methods generally encompass behavioral measures as well, and the opposite also occurs. This is shown in many of the chapters, for instance in Chapter 16, where the authors illustrate the “two‐pronged” nature of most lesion studies, which combine structural neuroimaging data and a diversity of behavioral data that index patients’ linguistic performance. Another example con-cerns Chapter 7 on structural priming. Though in its early days this method only involved behavioral measures, it increasingly uses brain measures such as ERPs and the BOLD response that indexes brain activation in fMRI. Still, for most language

Preface xix

researchers it makes sense to qualify methods as behavioral or neurobiological (and computational as a third category), so this is how we ordered the chapters, from primarily behavioral (Chapters 1‐10) and computational methods (Chapters 11 and 12) to neurobiological methods (Chapters 13‐17). But because the partitions between these classes of methods are not clear‐cut and a continued growth in inter-disciplinary research will likely result in their further integration, we have decided against explicitly labeling these three subsections in the table of contents.

In the preceding paragraphs almost all chapters have been introduced, however briefly. The exceptions were Chapter 9, on virtual reality, and Chapter 10, which presents ways for studying language outside the laboratory. These chapters were saved for now, where we mention two limitations of many traditional methods for studying language processing: Their ecological validity and external validity are often low; that is, their findings cannot easily be generalized to real‐world settings and to other populations and situations. The main reason why much traditional research lacks ecological validity is that in order to obtain reliable data and make sense of them, strict control over the experimental variables is required. Such con-trol can generally only be secured by using laboratory tasks that are impoverished substitutes of the real phenomena under study, the latter being stripped of many of their essentials, including the context in which they take place. The authors of Chapter 9 show how with virtual‐reality techniques it is possible to realize ecolo gical validity in the laboratory while at the same time controlling numerous experimental variables. The authors of Chapter 10 describe ways to enhance ecological validity and external validity by, for example, taking the experiment out of the laboratory into institutionalized public spaces such as museums, by crowdsourcing data on the internet, or by conducting cross‐cultural fieldwork. But unlike in research that makes use of virtual reality, in such studies maintaining experimental control is a real challenge.

A final feature that characterizes this volume is that many of its chapters contain the same or very similar sections, this resulting from our instructions to the authors. They were asked to explain the underlying assumptions and rationale of “their” method, to describe the required apparatus, the nature of the stimuli and data, the way the data are collected and analysed, and what the method’s strengths and weaknesses are in comparison to related methods. We also asked them to illustrate the method with an exemplary study so that the actual research practices and tools could be more vividly pictured, and to provide a glossary for easy accessibility of the method’s central concepts and features.

We are confident that the broad collection of research methods presented in this volume is varied enough for all beginning researchers interested in human language processing to find a topic to their liking and get going, and for researchers already active in language studies to become familiar with techniques they have not yet prac-ticed themselves.

Annette M. B. de Groot and Peter Hagoort

Research Methods in Psycholinguistics and the Neurobiology of Language: A Practical Guide, First Edition. Edited by Annette M. B. de Groot and Peter Hagoort. © 2018 John Wiley & Sons, Inc. Published 2018 by John Wiley & Sons, Inc.

Assumptions and Rationale

One of the biggest challenges of determining what an infant knows about language is actually tied to language itself. Unlike Piaget (1926), who famously asked older children to reflect on and discuss their understanding of the meaning of words, we have no such luxury of interviewing a 12‐month‐old regarding their word‐referent links. Even for developmentally simpler skills, we cannot get a 6‐month‐old to give a simple yes or no answer to the question of whether they discriminate two language sounds. It is somewhat paradoxical that language itself is a barrier to understanding

1 Habituation Techniques

Christopher T. Fennell

Abstract

This chapter presents the general aspects of the habituation technique. This technique has helped to address various language acquisition questions over the past half-century. While discussing implementations using different behavioural responses, the chapter focuses on the most common measure of habituation in language acquisition research: looking time (LT). Issues in implementing the method and potential prob-lems are discussed. The simplicity of the habituation procedure in both its design and implementation, along with its long history in the field, makes this method one of the fundamental tools that psycholinguists can use to uncover nascent, emerging, and maturing language skills during infancy and early childhood.

2 Research Methods in Psycholinguistics and the Neurobiology of Language

language development in infants. The fact that infants have little or no lexical pro-duction requires researchers to often turn to tasks that require no language output from the child. Further, infants’ limited motor skills restrict the measures that can reveal underlying linguistic abilities. Tasks must take advantage of gross motor abilities, such as full head turns (fine pointing or manual selection are difficult); con-genitally organized behaviors that infants have strong control over since birth, such as looking or sucking; or, basic psychophysiological responses, such as heart rate.

One of the most valid and reliable tools we have to examine the perceptual skills related to infant language is the habituation task. Habituation is a decrease in a response to a stimulus after repeated presentations. This produces what is termed the habituation curve, a monotonically decreasing behavior in response to a repeated target stimulus. It is a task with a very long history in our field, stretching back to the nineteenth century (for a review, see Thompson, 2009). Indeed, Thompson highlights that the concept is reflected in antiquity: in Aesop’s fables, a fox is quite frightened of a lion upon first meeting him, but becomes less alarmed upon each subsequent view-ing. Perhaps Aesop’s example was prescient. Habituation tasks were primarily used for decades with animals (and continue to be used with these populations), with everything from amoebas to dogs showing habituation responses (Harris, 1943).

Considering the long history of the task and ubiquitous nature of the habituation response across other non‐verbal beings (i.e., animals), it is unsurprising that the method was extended to infants in the early twentieth century (see Humphrey, 1933). However, simply habituating an infant to a stimulus is necessarily a bit limiting with respect to what one can say about learning. If, for example, an 8‐month‐old had a reduced behavioral response to the repeated presentation of a phoneme, one could argue that they have formed a memory of that particular sound. But it could also be that the infant is simply tiring. The key to demonstrating that the infant has formed a representation of or learned something about the presented stimulus is dishabituation—an increase in behavioral response to a novel stimulus.

Sokolov’s (1963) comparator model is the classic formulation of this approach. The infant (or adult) has an orienting response to a novel or unexpected non‐threat-ening stimulus (e.g., becoming still, looking at the stimulus, reduced heart rate). As it repeats, the infant builds an internal representation of the stimulus. The increasing strength of the representation leads to a greater match between the internal percept and the repeating external stimulus. The initially large orienting response correspond-ingly reduces as the internal/external match increases. But, if the external stimulus does not match the established internal representation (i.e., a novel stimulus), the infant’s orienting response should reoccur.

Thus, habituation is one of the optimal tasks for testing pre‐verbal infants as it does not rely on overt productions, but rather on implicit cognitive measures such as those mentioned earlier (e.g., looking time, sucking, heart rate, among others). Further, based on the comparator model, it allows researchers to determine the nature of infants’ per-cepts and concepts by testing differing levels of novelty from the habituated stimulus (e.g., changing a habituated word form by one phoneme, or multiple sound changes). If the infants’ behavioral responses increase to the novel stimulus, it can be concluded that they have the ability to differentiate the habituated and novel stimuli. In this regard, habituation is fundamentally a method to index discrimination ability.

Fantz’s (1964) article in Science on visual habituation in the human infant broadly introduced using this task with very young participants to psychological researchers. However, it is important to note that previous studies had already used habituation with

Habituation Techniques 3

infants, including studies on auditory habituation. For example, Bartoshuk (1962) dem-onstrated that newborns habituate to tones and dishabituate to tones of a differing intensity, using heart rate as his dependent measure. Once it was determined that infants could habituate and dishabituate to auditory tones, it was a straight road for researchers to examine language sound (i.e., phoneme) discrimination using similar methods.

In one of the seminal works on infant language perception, Eimas, Siqueland, Jusczyk, and Vigorito (1971) used a habituation task with sucking as their measure to investigate 1‐ and 4‐month‐old infants’ discrimination of consonants, specifically a voicing con-trast. Consonants produced in the same place and manner can differ in the timing of the vibration of the vocal folds. For example, /b/ and /p/ are both produced from the lips and are stops, but they differ in voicing. Vocal cord vibrations occurring approximately 25 ms after the air burst from the mouth (or later) sound like a /p/to English speakers. Vibrations starting before that mark sound like a /b/. In Eimas et al., infants heard a repeated sound contingent on strong sucks. Once their sucking rate decreased by 20%, a novel stimulus was presented in two experimental conditions. In one condition, the novel sound was from the same phonological category (i.e., a new /b/ sound that dif-fered by 20 ms in voicing from the original stimulus) and in the other condition the novel sound came from a different category (i.e., a 20 ms voicing change that crossed the boundary from /b/ to /p/). In the control condition, the same sound was played after the 20% reduction in behavior. Only infants in the differing category condition had a dishabituation response—increased sucking when the sound change occurred.

The above experiment highlights some important aspects of infant habituation. First, habituation allows the researcher to test categorical perception in that we can determine if an acoustically different stimulus will engender a continued habituated response or a dishabituation response. As Thompson and Spencer (1966) highlight in their classic list of the characteristics of habituation, “habituation of response to a given stimulus exhibits stimulus generalization to other stimuli” (p. 19). Thus, we can assume that the lack of dishabituation to an acoustically different stimulus means that the infant considered it to fall into the same category, or the distinction is too weak to detect. This second explanation is unlikely if you include a condition where a similar magnitude difference elicits a dishabituation response due its crossing of a category boundary, as in the Eimas et al. work.

Second, the use of criterion equated the processing of the stimuli across infants. Based on Sokolov’s (1963) theory, the reduction in the target behavior is commensurate with the increasing robustness of the infant’s memory trace for the stimulus. But, different infants would potentially, and probably, have differing timing with respect to the building of the stimulus memory trace due to individual differences in cognitive skills, particu-larly attention. By requiring them to reach the same relative decrease, the researcher can assume that they have reached similar processing levels for the stimulus in question.

Apparatus

Testing infants’ language skills via habituation typically involves relatively little technology in comparison to some other methods present in the literature, like event‐related potentials (ERP), functional near infrared spectroscopy (fNIRS), or functional magnetic resonance imaging (fMRI). As such, the necessary apparatus can be implemented relatively quickly and inexpensively in the lab. Primarily, a

4 Research Methods in Psycholinguistics and the Neurobiology of Language

researcher requires devices to present stimuli in a controlled manner and to measure the target behavior. As the measurements of the latter need to feed back to control the former in habituation, we typically use the same device to do both: habituation software on a lab computer. A widely used free‐ware program is available for such purposes, called Habit 2 (Oakes, Sperka, & Cantrell, L., 2015). The program will control stimuli presentation, compute habituation criteria, and accumulate behavioral data. Stimuli are usually played from digitized files on the computer and are sent to the display and speaker in the testing room. The experimenter, who should be blind to the audio stimuli being presented and to whether a trial was a habituation or test trial, remotely monitors the infant’s behaviors via key presses.

As alluded to in the section above, many of the early studies in infant habituation used sucking or heart rate as the dependent measure. Measuring infant sucking strength and rates requires the experimenter to have a pressure transducer within a pacifier, and the corresponding connected equipment to measure the output from the transducer. Heart rate measures usually require three electrodes to be placed on the infants’ chest and abdomen, with the electrodes again connected to equipment to measure their output. While these measures are still used, the typical behavior being measured in modern habituation research is looking time (LT) to a visual display, even if one is testing habituation/dishabituation of auditory language stimuli. There is a positive relationship between attention to an auditory stimulus and visual fixation (Horowitz, 1975). Unlike the measures above, LT requires nothing to be in physical contact with the infants, which is an advantage. The researcher—appropriately blinded to the condition—simply needs to record, by pressing buttons on a keyboard connected to the same software, where the infant is looking, typically through watch-ing the infant via a closed‐circuit video camera. The use of a video camera also allows for a record that can confirm the real‐time measurements when coded post‐experiment. The ease of measuring looking behavior has led to its wide application.

Nature of the Stimuli

Due to the nature of the task, many habituation studies investigating language development have involved basic auditory stimuli, such as changes to the acoustic form of simple syllables. For example, tracking infants’ phonetic and phonological development has been the focus of many infant habituation studies, starting with Eimas et al. (1971). Using an example from Polka and Werker (1994), infants in such studies are habituated to one syllable (e.g., /dYt/) and then given a syllable involving a single phoneme change for the novel stimulus at test (e.g., /dut/). If infants dishabitu-ate to the novel syllable, they are able to distinguish the target phonemic contrast. See Figure 1.1 for a visualization of such a study. Such studies contributed to the finding that infants are initially universal listeners, able to distinguish sounds from both their native and from non‐native languages, but then become language‐specific listeners over the first year—failing to dishabituate to non‐native contrasts.

Habituation studies examining language development are not limited to simple sylla-bles, however. For example, Mehler, Jusczyk, Lambertz, Halsted, Bertoncini, and Amiel‐Tison (1988) used a habituation method where the target stimuli were narrative auditory passages from rhythmically similar and dissimilar languages (recorded from fluent bilinguals so that the voice did not differ). Using sucking rate as the dependent variable,

Habituation Techniques 5

they showed that infants of 2 months can distinguish their native language from a non‐native language based on their rhythmic class, but not two non‐native languages.

One can even use visual stimuli to demonstrate language discrimination. In a novel twist, Weikum, Vouloumanos, Navarra, Soto‐Faraco, and Sebastián‐Gallés (2007) presented infants with silent video clips of fluent French‐English bilinguals reciting passages in each language. Infants of 4 and 6, but not 8, months dishabituated to French clips after being habituated to English ones, and vice versa. This shows that infants have an early ability to visually discriminate non‐native from native lan-guages before perceptually narrowing to their native language in the visual domain. Interestingly, French‐English bilingual infants, for whom both languages were native, were able to discriminate the languages at the older age.

Finally, rather than focusing on audio or visual stimuli, some habituation e xperiments explore the connection between the two by pairing objects and word forms during habituation, and then test infants on novel word‐object associations (see Figure 1.1). As such, these studies contribute to a major area of early language development: early word learning. For example, a simple way to invoke word learning is to replace a visual pattern typically used in discrimination studies with an object that affords naming. But infants may succeed by ignoring the object and simply focus on the change in label during the novel trial in the test phase. Werker, Cohen, Lloyd, Casasola, and Stager (1998) corrected for this by creating an audio‐visual variation of the habitua-tion procedure called the Switch task (see Figure 1.1). Infants are habituated to two word‐object associations (e.g., Object A – Word A; Object B – Word B) and then tested on two trials: a Same trial comprising of one of the habituated pairings (e.g., Object A  –  Word A) and a Switch trial where an incorrect pairing is presented (e.g., Object A – Word B). Importantly, the Switch trial consists of a habituated object and a habituated word, but linked in a novel way. In this manner, infants should only disha-bituate if they have appropriately linked the word and object.

Type of task Pretest Habituation phase Test phase PosttestFamiliar Novel

Speech discrimination

“Neem.” “Gek.” “Gek.” “Gik.” “Neem.”

Word learning(Single object)

“Neem.” “Gek.” “Gek.” “Gik.” “Neem.”

Word learning(Two objects)

“Neem.” “Gek.” “Gik.” “Gek.” “Gik.” “Neem.”

Note: The two object version of the task is known as the Switch task.

Figure  1.1 Examples of various infant language habituation tasks. (See insert for color representation of the figure.)

6 Research Methods in Psycholinguistics and the Neurobiology of Language

Methodological Structure

Now that we have discussed the nature of the work using habituation tasks, we can turn to the typical structure of these tasks. Infant habituation studies in language research typically involve four phases: pretest, habituation, test, and posttest. Figure 1.1 outlines these four phases across three different studies. Each of these phases comprises discrete trials wherein a visual and an audio stimulus are concur-rently presented. Trials can be preceded by what is termed an attention‐getter in order to get the infant to orient to the screen. Various attention‐getters have been used in past research. Some examples include: a silent, flashing light; a silent, mor-phing, colourful shape; and the face of a baby with giggling as the accompanying audio track. Once the infant looks to the screen, the relevant trial commences. Trials can be of fixed length or infant‐controlled. The latter involves setting a criterion via which the trial will end if the infant disengages attention. For example, if an infant looks away from the stimulus for 2 seconds, the trial ends and the next trial commences.

Pretest

In the pretest phase, the infant experiences a stimulus that is different from the repeated one during the upcoming habituation phase. The reasoning behind this trial is that infants need to become accustomed to the presentation method. Thus, this phase serves as a warm‐up prior to presenting the stimulus of import in your study.

Habituation

The habituation phase follows and is, of course, key to the experiment. One impor-tant point to consider for this phase is the intensity of the audio stimuli. As Thompson and Spencer (1966) highlight in their list of habituation characteristics, “strong stimuli may yield no significant habituation” (p. 19). For example, it would be hard to habituate to a blaring, variable siren. Therefore, the audio stimulus is typically delivered at approximately 65 decibels to make it loud enough for infants to hear, but not too loud to invoke failure to habituate. Following similar logic, the visual display that is shown should also be only moderately engaging. Another setting the researcher must decide upon is the habituation criterion. A common criterion is a 50% reduction in LT (Ashmead & Davis, 1996), although some researchers advocate using more stringent criteria with younger infants (e.g., 70%) as they are more c ognitively immature and therefore may require more presentations to fully process stimuli (e.g., Flom & Pick, 2012). Another consideration is the window over which the decrease in response is based. If one stimulus is repeated, many researchers opt for a window of three trials. For example, if the infant looks for a total of 50 seconds over the first three trials, they need to fall below 25 seconds in a subsequent window of three back‐to‐back trials to reach habituation. Ashmead and Davis recommended the size of this window based on their computer modeling in that it was more stable than a window size of two.

Habituation Techniques 7

Two other important considerations are related to these windows. First, the researcher could opt for a fixed or a sliding window. A sliding window keeps a running total of LT to determine habituation (e.g., trials 2, 3, and 4 are first compared to trials 1, 2, and 3). A fixed window compares subsequent blocks of three trials to the criterion previous block (e.g., trials 4, 5, and 6 are first compared to trials 1, 2, and 3). Oakes (2010) recom-mends using the sliding window whenever possible, as it necessarily will lead to shorter experiments on average, and shorter habituation phases should result in less attrition. However, if the infant is being habituated to two types of stimuli (e.g., two word‐referent combinations), the fixed window is necessary despite the chance of increased attrition in order to ensure that the infants receive an equal number of examples from each stimulus type during habituation (and the window needs to be increased to four trials: two of each stimuli type per block). The second consideration is whether to base the habituation criterion on the first block of trials, which typically—but not always— has the highest infant behavioral response to the stimuli, or base it on the block of trials with the highest behavioral response, regardless of when it occurs in the experiment. Most researchers use the first block, as infants may have an increased response on a later block due to a factor unrelated to the habituation curve (e.g., a baby surprises him-self with a sneeze and reorients to the stimuli due to increased arousal).

Despite the presence of a criterion, the researcher should cap the number of possible trials in an experiment. Without such a cap, the experiment would not end for some infants in a reasonable amount of time, as they would not reach criterion. Dannemiller (1984) recommended that the maximum amount of trials in an infant habituation study should be 15 trials. At that number, according to his Monte Carlo modelling, there would be a 5% risk that infants are habituating by chance. But the trade off of having a small maximum number of trials is that you will have fewer infants reaching criterion within that overall trial limit. But, increasing the maximum amount of trials increases both the chance of random habituation and of attrition. Oakes (2010) recommends piloting infants and/or examining similar studies in the literature to determine the optimal maximum number of trials for a particular study.

Test

The test phase of the experiment should include both the novel stimulus and a r epetition of the familiar stimulus from habituation, with the order counterbalanced across participants (e.g., Werker et al., 1998). Why not only present the novel stim-ulus and compare it to the last habituation block? The major issue in taking that approach is that behavioral responses in the final block of habituation trials are necessarily low, and may be artificially so (see Cohen, 2004). This could be due to an infant reducing attention to the stimuli for a reason unrelated to habituation, such as a distraction in the room like the parent shifting in their seat. Thus, this comparison may falsely indicate dishabituation. By presenting both the novel stimulus and a rep-etition of the familiar stimulus, the experimenter can determine whether the infants can detect the difference between the habituated stimuli and something new. Some researchers run this manipulation as a between‐subject design, but this is not recom-mended for both statistical and practical reasons. One would introduce more error into the design related to the individual differences between the groups, and it requires doubling the amount of participants.

8 Research Methods in Psycholinguistics and the Neurobiology of Language

Posttest

Finally, the posttest trial should be presented last and should be maximally different from the habituation and test trials. It is expected that if infants are still engaged in the experiment, looking time would recover to near pretest level during this final trial.

Collecting and Analysing Data

As mentioned earlier, the best method to collect the data is via habituation‐specific software, like Habit. The program is designed to continuously compare an infant’s behavioral response to the stimulus, in this case LT, in a block of trials to previous blocks to determine when the infant has reached the habituation criterion. Importantly, before delving into statistical analyses of the habituation and test phases, one has to establish the reliability of the experimenter’s coding of the target variable (e.g., manual key strokes in response to LT) if it is not a direct measure (e.g., eye tracker for LT, ECG for heart rate). One standard is for another coder to recode the target trials of 25% of the participants from the video records of those experi-ments. In this case, the original coding would be considered to be reliable if a Pearson product‐moment correlation of the two coders’ measures is equal to or greater than .95. A more exact method is to have two coders score all the video records using a frame‐by‐frame analysis using freeware such as SuperCoder (Hollich, 2005).

One should also report analyses of the habituation phase prior to testing for novelty and familiarity effects. To determine whether infants maintained interest throughout the experiment and recovered from habituation, one possibility is to run a series of planned orthogonal comparisons to first compare pretest to posttest and, if these two trials are found to be the same, to then compare these trials to the last habituation block. It is expected that if infants were still engaged in the experiment, looking time would recover to near pretest level during the posttest. Thus, there should be no significant difference between the pretest and posttest. However, the pretest and posttest should be significantly different from the last habituation block to demonstrate recovery. One can then compare the first habituation block to the last via a paired‐sample t‐test to confirm a significant drop in looking time across the habituation phase. Finally, a full descriptive analysis of the habituation phase should be reported (i.e., mean number of habituation trials, mean looking time during habituation). If there are multiple conditions or groups in the experiment, these habituation variables should be compared in a mixed ANOVA to ensure similar habituation across conditions or groups. For example, in Table 1.1, infants in

Table 1.1 Mock habituation data from four experiments with looking time as the dependent variable.

Experiment Habituation Trials Habituation Time Familiar Trial Novel Trial

1 8 112 s 11.5 s 12.2 s2 8 180 s 8.1 s 12.8 s3 12 200 s 7.9 s 12.4 s4 12 210 s 8.3 s 8.1 s

Note: All data represent the mean.