Journal of Fluency Disorders · Journal of Fluency Disorders 55 (2018) 6–45 Contents lists available at ScienceDirect Journal of Fluency Disorders Review Article A systematic literature

Journal of Fluency Disorders 55 (2018) 6–45

Contents lists available at ScienceDirect

Journal of Fluency Disorders

Review Article

A systematic literature review of neuroimaging research ondevelopmental stuttering between 1995 and 2016

Andrew C. Etchell a,e,∗, Oren Civierb,c,1, Kirrie J. Ballardd, Paul F. Sowmane

a Department of Psychiatry, University of Michigan, MI, United Statesb The Gonda Multidisciplinary Brain Research Center, Bar-Ilan University, Ramat-Gan, Israelc Department of Computer Science and Applied Mathematics, Weizmann Institute of Science, Rehovot, Israeld Faculty of Health Sciences, University of Sydney, Sydney, Australiae Department of Cognitive Science, Macquarie University, Sydney, Australia

a r t i c l e i n f o

Article history:Received 23 July 2016Received in revised form 25 January 2017Accepted 6 March 2017Available online 12 March 2017

Keywords:stutteringreviewPETfMRIdiffusion MRIDTIMEGEEGTMSVBMmorphologytractography

a b s t r a c t

Purpose: Stuttering is a disorder that affects millions of people all over the world. Over thepast two decades, there has been a great deal of interest in investigating the neural basisof the disorder. This systematic literature review is intended to provide a comprehensivesummary of the neuroimaging literature on developmental stuttering. It is a resource forresearchers to quickly and easily identify relevant studies for their areas of interest andenable them to determine the most appropriate methodology to utilize in their work. Thereview also highlights gaps in the literature in terms of methodology and areas of research.Methods: We conducted a systematic literature review on neuroimaging studies on devel-opmental stuttering according to the PRISMA guidelines. We searched for articles in thepubmed database containing “stuttering” OR “stammering” AND either “MRI”, “PET”, “EEG”,“MEG”, “TMS”or “brain” that were published between 1995/ 01/ 01 and 2016/ 01/ 01.Results: The search returned a total of 359 items with an additional 26 identified from a man-ual search. Of these, there were a total of 111 full text articles that met criteria for inclusionin the systematic literature review. We also discuss neuroimaging studies on developmen-tal stuttering published throughout 2016. The discussion of the results is organized first bymethodology and second by population (i.e., adults or children) and includes tables thatcontain all items returned by the search.Conclusions: There are widespread abnormalities in the structural architecture and func-tional organization of the brains of adults and children who stutter. These are evidentnot only in speech tasks, but also non-speech tasks. Future research should make greateruse of functional neuroimaging and noninvasive brain stimulation, and employ structuralmethodologies that have greater sensitivity. Newly planned studies should also investi-gate sex differences, focus on augmenting treatment, examine moments of dysfluency and
longitudinally or cross-sectionally investigate developmental trajectories in stuttering.© 2017 The Authors. Published by Elsevier Inc. This is an open access article under the CC
BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

∗ Corresponding author at: Department of Psychiatry, University of Michigan, Rachel Upjohn Building Rm 2701, 4250 Plymouth Rd., Ann Arbor,MI 48109, United States.

E-mail address: [email protected] (A.C. Etchell).1 Current address: Florey Institute of Neuroscience and Mental Health, Melbourne Brain Centre, Melbourne, Australia.

https://doi.org/10.1016/j.jfludis.2017.03.0070094-730X/© 2017 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

https://doi.org/10.1016/j.jfludis.2017.03.007

http://www.sciencedirect.com/science/journal/0094730X

http://crossmark.crossref.org/dialog/?doi=10.1016/j.jfludis.2017.03.007&domain=pdf

http://creativecommons.org/licenses/by-nc-nd/4.0/










mailto:[email protected]

https://doi.org/10.1016/j.jfludis.2017.03.007











C

1

sboG(b

shc2Irwotp

A.C. Etchell et al. / Journal of Fluency Disorders 55 (2018) 6–45 7

ontents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.1. Protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82.2. Information sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82.3. Search parameters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .82.4. Study selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3. Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93.1. Positron emission tomography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3.1.1. AWS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93.2. Functional MRI and near infrared spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

3.2.1. AWS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123.2.2. CWS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

3.3. Electroencephalography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163.3.1. AWS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173.3.2. CWS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

3.4. Magnetoencephalography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193.4.1. AWS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203.4.2. CWS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

3.5. Transcranial magnetic stimulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213.5.1. AWS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3.6. Voxel-based morphometry: gray matter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223.6.1. AWS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223.6.2. CWS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

3.7. Morphology: gray matter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233.7.1. AWS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233.7.2. CWS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

3.8. Voxel-based morphometry: white matter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233.8.1. AWS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243.8.2. CWS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

3.9. Diffusion MRI − voxelwise analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243.9.1. AWS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243.9.2. CWS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

3.10. Diffusion MRI − tractography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253.10.1. AWS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263.10.2. CWS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

4. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275. Thoughts for future research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28Appendix A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

. Introduction

Developmental stuttering is characterized by involuntary prolongations, repetitions and pauses that disrupt the flow ofpeech. Stuttering affects 1% of the general population (Craig, Hancock, Tran, Craig, & Peters, 2002) and about twice as manyoys as it does girls (Howell, 2007). While negative effects of stuttering are not generally observed in the first year followingnset, those who continue to stutter into adulthood can experience marked disruptions to their quality of life (Boyle, 2015;unn et al., 2014; Iverach et al., 2009, 2010) and mental health (Iverach et al., 2009). Because many people who stutter

PWS) speak fluently the majority of the time, stuttering is thought to be a disorder of the structure and/ or function of therain that transiently disrupts speech production.

Accordingly, over the last 20 years there has been a vast increase in the number of published studies documentingtructural and functional differences in the brains of those who stutter compared to their fluent peers. Most of this researchas been conducted on adults who stutter (AWS). Despite relatively widespread acknowledgment of the need to studyhildren who stutter (CWS e.g., Busanet al., 2013; Cai et al., 2014; Chang, Erickson, Ambrose, Hasegawa-Johnson, & Ludlow,008; Choo et al., 2011; Cieslak, Ingham, Ingham, & Grafton, 2015; Connally, Ward, Howell, & Watkins, 2014; Cykowski, Fox,

ngham, Ingham, & Robin, 2010) there have been very few neuroimaging studies in this population (see Chang, 2014 for aeview). This is likely attributable to the practical difficulties and special considerations that must be taken into account
hen testing young children. Children need to remain still for extended periods of time, maintain attention for the duration
f a task and manage potential anxiety about the scanning environment. In brief, these issues can be remedied by shorteninghe duration of tasks, involving tasks that do not require overt responses, familiarizing the child with the scanner throughractice and mock sessions, and making the experience child friendly through the use of rewards and playful themes (e.g.,

8 A.C. Etchell et al. / Journal of Fluency Disorders 55 (2018) 6–45

likening the scanner to a spaceship). Despite obvious difficulties, studying children who stutter (CWS) is important becausethe brains of CWS have had far less time to change in response to stuttering; observation of differences in brain activity inCWS is, therefore, more likely to reflect causal mechanisms of the disorder.

The aim of this review is to provide a comprehensive account of the past 20 years of neuroimaging research on develop-mental stuttering. It is intended that it will be used as a resource for anyone seeking to conduct any form of neuroimagingstudy on developmental stuttering: it will provide a summary of the research across broad topics and methodologies. We donot intend to provide a coherent account of developmental stuttering or point to the probable etiology of the disorder. Webriefly mention consistent findings and some of the theories they support, as well as inconsistencies between reports to dateand approaches that might be employed in future in an effort to minimize discrepancies that arise due to methodologicalissues. One of our goals is to highlight the advantages and disadvantages of various methodologies, helping researchersidentify the techniques most suitable for their research question. These are briefly summarized here and elaborated onwithin each of the respective sections below. Techniques like positron emission tomography (PET), functional magneticresonance imaging (fMRI) and near infrared spectroscopy (NIRS) rely on differences in the flow or magnetization of bloodand are indirect measures of neural activity. They have good spatial resolution (fMRI having the highest resolution of thethree), being able to determine precisely which cortical (PET, fMRI, and NIRS) and subcortical (PET and fMRI) regions of thebrain are active during a particular task. They do, however, have poor temporal resolution due to the time it takes for bloodto flow to a given region of the brain. In contrast, techniques like magnetoencephalography and electroencephalography(MEG and EEG) are direct measures of neural activity with high temporal resolution. EEG measures the electrical dischargeof the firing of tens of thousands of neurons, at the scalp, while MEG relies on the magnetic field generated perpendicular tothis electric field, slightly above the scalp. Because magnetic signals are less affected by differences in scalp conductivities, itis easier to determine the source of the neural activity in MEG than EEG. Techniques like transcranial magnetic stimulation(TMS) are used for a variety of purposes: two of the most common are probing cortical excitability of motor areas at specificpointsin time, or creating transient and reversible “virtual lesions”. More pragmatically, the purchase and operating costs ofPET, fMRI and MEG are considerably greater than NIRS, EEG and TMS.

The review is organized according to experimental methodology and, where appropriate, by the type of experimentaltask employed. It includes studies on AWS (18 years and older) and CWS (less than 18 years), as well as studies whoseexperimental sample included participants from both groups. This is based on the legal definition of adults and children inAustralia, and on the terminology used by the individual studies. Special attention will be given to experimental paradigmsthat were used in both AWS and CWS that thus permit a direct comparison between age groups. Some studies also investigatechildren who initially stuttered but later recovered (recovered children), and one study includes cases of adult-onset recoveryfrom stuttering (Kell et al., 2009).

2. Methods

2.1. Protocol

This systematic literature review followed the PRISMA guidelines (www.prisma-statement.org) and included any studythat used neuroimaging to investigate developmental stuttering, reported more than one participant, and was publishedbetween 1995/ 01/ 01(YYYY/ MM/ DD) and 2015/ 12/ 31. The rationale for using this range of time was to include all studiesover the past two decades.

2.2. Information sources

Articles were identified using the pubmed database located at the following url: http://www.ncbi.nlm.nih.gov/pubmed.Since no single database is likely to contain every relevant publication, we also cross-checked the results with a manual searchof google scholar (http://scholar.google.com.au/ ) and via bibliographies. The last search was conducted on 01/ 01/ 2016.

2.3. Search parameters

We identified relevant studies in the pubmed database by searching for any items that contained the words “stuttering”OR “stammering” in either the title or the main body of the article AND terms referring to neuroimaging methodologies.These neuroimaging methodologies were “positron emission tomography” OR “PET”, “magnetic resonance imaging” OR“MRI”, “electroencephalography” OR “EEG”, “magnetoencephalography” OR “MEG” and “transcranial magnetic stimulation”OR “TMS” in any part of the article. In an attempt to pick up articles that did not use these methodologies, we also includedthe more general search word “brain”. This search returned a total of 359 items without any additional filters. When filteringthe results to include “full text articles” (i.e., excluding abstracts or conference proceedings) the search returned a total of305 items. We then used the filter on the pubmed database to identify article types that were not appropriate for inclusion inthe systematic review. Of these 305 full text articles, 163 were excluded after filtering out reviews or meta analyses (n = 38),
case studies or clinical trials (n = 68), items containing “acquired” in either the title or the abstract (n = 7), or non-humanstudies (n = 50). The remaining 142 items were then assessed for eligibility for inclusion in the systematic literature review,as per the criteria in 2.4. Please note that, upon further inspection, no reviews or meta-analyses identified in the searchreported using a systematic search method such as that recommended in the PRISMA statement.
http://www.prisma-statement.org

http://www.ncbi.nlm.nih.gov/pubmed

http://www.ncbi.nlm.nih.gov/pubmed

2

daata

3

iwiawamout

oftgvos

tl

3

pib2F&sT

33igsidsaatTit

n


.4. Study selection

The full texts of the 142 articles were examined and excluded if they (1) did not collect neuroimaging or neurophysiologicalata,(2) focused on acquired rather than developmental stuttering (3) were review or hypothesis papers or (4) were notvailable in English. Based on these criteria, another 57 papers were excluded leaving a total of 85 papers. A further 26rticles were then identified through a manual search of bibliographies in these 85 articles and google scholar. All 26 methe eligibility criteria and were included in the full review. The flowchart in Fig. 1 depicts the identification and selection ofrticles for this review.

. Results and discussion

There were a total of 111 studies included in the systematic review. Important details of the neuroimaging studiesncluding the methodology, number of participants, experimental task and main findings are summarized in Tables 1–7

hich can be found in the appendix. They are organized first by methodology and then by age of participants, with studiesncluding AWS (18 years and up) and CWS (younger than 18) appearing in separate tables. Studies that included both CWSnd AWS (mixed-age group, denoted people who stutter, or PWS) are listed after the CWS studies, in the same table studieshose experimental paradigm was used to study a sample of AWS and a separate sample of CWS (often in separate papers)

re marked with asterisk (*). Studies that used multiple neuroimaging methodologies on the same sample of subjects arearked with a cross (+), and are listed in all relevant tables. The characteristics of the experimental and control groups

f each study are given in the Sample column; mean age or age range in years is listed in brackets after each group. Anp-to-date version of the tables, which includes future neuroimaging studies of developmental stuttering, is maintained byhe authors online (http://www.neurostuttering.org).

For ease of reading of the results and discussion, we have separated the studies into ten categories based on methodf imaging and analysis: (1) positron emission tomography (PET), (2) functional magnetic resonance imaging (fMRI) andunctional near infrared spectroscopy (fNIRS), (3) electroencephalography (EEG), (4) magnetoencephalography (MEG), (5)ranscranial magnetic stimulation (TMS), (6) gray matter voxel based morphometry (VBM) on structural MRI (sMRI), (7)ray matter morphology on structural MRI, (8) white matter voxel based morphometry on structural MRI, (9) white matteroxel wise analysis on diffusion MRI (dMRI) and (10) white matter tractography on diffusion MRI. Every category has itswn section, and studies that used multiple methodologies are detailed in all relevant sections. At the beginning of eachection, we outline the advantages and disadvantages of the methodology and list all relevant studies.

Please note that we have also included reference to studies published between 2016/ 01/ 01 and 2016/ 31/ 12 (usinghe same search parameters as the systematic review) to provide a more comprehensive and up to date summary of theiterature. These studies are not listed in the tables, though.

.1. Positron emission tomography

PET is a neuroimaging method used to measure metabolic processes in the brain. It has an unrivaled capacity to track theharmacodynamics of the human brain in vivo using radio labeled biomarkers (Gambhir, 2002; Judenhofer et al., 2008). PET

s particularly useful for measuring uptake and reuptake of neurotransmitters such as dopamine and serotonin. There haveeen a total of14 PET studies of stuttering, all of which have focused on AWS (Braun et al., 1997; De Nil, Kroll, Kapur, & Houle,000; De Nil, Kroll, & Houle, 2001; De Nil, Kroll, Lafaille, & Houle, 2004; Fox et al., 1996, 2000; Ingham et al., 1996; Ingham,ox, Ingham, & Zamarripa, 2000; Ingham et al., 2004; Ingham, Grafton, Bothe, & Ingham, 2012; Ingham, Wang, Ingham, Bothe,

Grafton, 2013; Stager, Jeffries, & Braun, 2004; Wu et al., 1995, 1997). They represent a range of conditions including restingtate, activation during speech tasks or non speech tasks, and changes in activation resulting from behavioral intervention.he absence of studies on CWS is understandable, given that this methodology requires the injection of radioactive tracers.

.1.1. AWS

.1.1.1. Resting state. The first study to examine resting state activity was by Ingham et al. (1996) who measured differencesn at-restcerebral blood flow (CBF) between AWS and adults who do not stutter (AWDS). There was weak evidence forroup differences in laterality of blood flow between regions involved in speech motor control, based on the results of theirtatistical analysis. This result did not withstand correction for multiple comparisons. The authors concluded that stutterings not associated with a ‘lesion’ or abnormal asymmetry. Braun et al. (1997) also measured CBF between AWS and AWDSuring rest, as part of a baseline condition for speech tasks, and found no differences between groups. Later, in a moreophisticated study, Ingham et al. (2012) compared neural activity using PET during eyes-closed resting state, oral readingnd monolog speech conditions. They reported significantly increased activation in the left putamen, left post central gyrus,nd the left preSMA in AWS during eyes-closed rest, as compared tothe AWDS. A number of regions including left superioremporal gyrus, cuneus, and right post central gyrus were also more active during speech (i.e., oral reading and monologs).he observation of similar neural group differences during resting state and the performance of overt speaking tasks was
ntriguing because it implied that fundamental differences in neural activation between AWS and AWDS normally attributedo task-related activity may exist in the absence of speech.
Whereas most PET studies report on functional brain activations, one focused on the rate at which the body metabolizeseurotransmitters. In the only study of this kind, Wu et al. (1997) found that AWS have significantly higher uptake of

http://www.neurostuttering.org


Fig. 1. Visual representation of systematic literature review process.

dopamine in the caudate tail and auditory areas during an eyes open resting state than do AWDS. This result highlighted theimportance of dopamine in the etiology of stuttering and was highly influential in driving future research, despite its verysmall sample size (n = 3).

3saFpatwIrrsssr

ssaa

ebwon

cra

btarw

3ibcw

3inTafad

isDcs

mel


.1.1.2. Speech tasks. For some helpful meta-analyses that include results from PET and fMRI studies of speech tasks intuttering, see Belyk, Kraft and Brown (2015), Budde, Barron and Fox (2014), and the earlier Brown, Ingham, Ingham, Lairdnd Fox (2005). Early studies focusing on differences between fluent and dysfluent speech presented inconsistent findings.or example, Ingham et al. (1996) found no between group differences in regional CBF during rest, reading aloud to anothererson (solo), or reading intime with someone else (chorus). However, Wu et al. (1995) contradicted these findings despitelso using solo and chorus reading tasks. During both conditions, AWS had approximately 50% less CBF in the left caudatehan AWDS. In AWS, Broca’s and Wernicke’s areas, and the right cerebellum were under active during the solo condition,hich was associated with more stuttering, but exhibited normal levels of CBF during the induced fluency chorus condition.

n contrast, Fox et al. (1996) reported that solo reading was associated with overactivation of the SMA, insula and cerebellumelative to AWDS. These over activations were largely reduced bychorus reading. The fluency inducing condition (choruseading) also increased activation in the left inferior frontal and superior temporal gyri in AW Fox et al. (2000), using theame paradigm as their previous work (Fox et al., 1996), found that cerebral and cerebellar activations correlated withtuttering rate and syllable rate. They interpreted cerebellar dysfunction to be an essential feature of stuttering. Despiteome contradictory findings, these studies suggested that CBF and activation patterns in cortical, subcortical, and cerebellaregions are dependent on whether an AWS is speaking fluently or not.

Some studies indicate that the left and right hemispheres play different roles in the production and attenuation oftuttered speech. Braun et al. (1997) observed that, whereas left-sided activations were positively associated with dysfluencycores (i.e., stuttered syllables), right-sided activations were negatively associated with dysfluency scores. Interestingly, theseuthors also reported that AWS showed greater activation in the left inferior frontal gyrus (IFG) and the supplementary motorrea (SMA) during non-linguistic oral motor movements than AWDS.

The influence of sex on stuttering and brain activation patterns has also been considered. Using the same paradigm as Foxt al., Ingham et al. (2004) reported that activation in some brain regions was positively correlated with stuttering rate foroth sexes (e.g., the SMA and primary motor cortex [M1]), whilst activation in other brain regions was negatively correlatedith stuttering rate for men but not women (e.g., left basal ganglia, left occipital lobe, right midcingulate gyrus and right

ccipital lobe). This highlighted the importance of considering sex differences in stuttering studies, something that is ofteneglected.

Later work examined whether there were commonalities in brain activation under a broader range of fluency inducingonditions. Stager et al. (2004) found that both singing and metronome-timed speech increased activation in auditory regionselative toAWDS. The authors suggested some fluency inducing mechanisms may allow AWS to make more efficient use ofuditory information (for a similar fMRI study, see Toyomura, Fujii, & Kuriki, 2011).

De Nil et al. (2000) compared neural responses to reading single words aloud or silently with the aim of examiningrain lateralization in AWS and AWDS. Whereas AWS showed increased activation of right hemisphere structures relativeo AWDS when reading aloud, AWS showed decreased activation of right hemisphere structures when reading silently. Thenterior cingulate cortex was overactive in the stuttering group during silent speech and was thought to reflect anticipatoryeactions to stuttering (though see Ingham et al., 2004; who noted that this region was not activated during stuttering andas therefore not necessary for stuttering to occur).

.1.1.3. Non-speech tasks. Because stuttering can be rare in the laboratory setting, Ingham et al. (2000) investigated whethermagined stuttering elicited the same patterns of neural activation as real stuttering. Real stuttering and imagined stutteringoth resulted inactivation of the SMA, the bilateral insula and cerebellum and decreased activations in the right auditoryortex. Interestingly though, the right superior lateral premotor cortex was strongly active during overt stuttering, but onlyeakly active when stuttering was imagined, suggesting that it may play an important role in moments of stuttering.

.1.1.4. Intervention. Consistent with Fox et al. (2000) and De Nil et al. (2001) highlighted the importance of the cerebellumn the control and timing of voluntary movements. Given that stuttering is often conceived as a disorder of temporal coordi-ation, De Nil et al. examined how activation in the cerebellum responded to an intervention designed to reduce stuttering.hey found differences in cerebellar activity in AWS prior to treatment and AWDS during overt and silent reading and during

verb generation task. The AWS were rescanned immediately after fluency enhancing treatment and again at a one-yearollow-up. Over the course of the three scans, the AWS showed a decrease in cerebellar activity toward normal levels. Theuthors interpreted this as evidence that AWS have heightened levels of monitoring and decreased levels of automaticityuring the execution of speech movements, but that this can be altered by speech therapy.

In a subsequent study, De Nil et al. (2004) showed that speech therapy shifts the altered right lateralization of activationn the inferior frontal and the precentral gyrus to the left homologue. Conversely, the focus of activation in the cerebellumhifted the other way, from left to the right after treatment. However, in contrast to previous work (Fox et al., 1996, 2000),e Nil et al. noted that increased activation in the left superior temporal gyrus is likely to reflect increasing awareness andontrol of articulatory movements. This is supported by the observation that activation in the STG of AWDS increases aspeech is altered (also see Tourville et al., 2006).

Interestingly, it has been reported that a decreased level of activation in the left putamen during speech, non-speechovements and during rest, is a predictor of the likelihood of successful or unsuccessful recovery from stuttering (Ingham

t al., 2013). This result from Ingham et al. (2013) is consistent with the theory that global overactivation of the putameneads to a lower signal-to-noise ratio (SNR) in that structure. The theory contends that this reduced SNR means that the basal


ganglia thalamo-cortical loop fails in its selection of motor programs for execution, resulting in stuttering (Civier, Bullock,Max, & Guenther, 2013). The idea that dysfunction of the basal ganglia may cause problems with speech production is inagreement with its involvement in slower, controlled speech and the fact that it is involved in the initiation and executionof speech (Price, 2010, 2012).

3.2. Functional MRI and near infrared spectroscopy

Historically, fMRI became available after PET. The shift from PET studies to fMRI was primarily driven by the fact that fMRIis non-invasive, does not involve exposure to ionizing radiation, and has better spatial resolution than PET. The hemodynamicresponse measured by fMRI typically peaks several seconds after a behavioral response. The operation of fMRI is considerablylouder than PET and the sound emitted tends to be rhythmic. Both noise masking (Civier, Tasko, & Guenther, 2010) and rhythm(Toyomura, Fujii, & Kuriki, 2015) have fluency enhancing effects and may present unique challenges to the study of stuttering.Due to the non invasiveness nature of fMRI, some studies have used it to test CWS; however, the loud sound, together withfMRI’s sensitivity to movement-related artifacts present challenges (see Loucks, Kraft, Choo, Sharma, & Ambrose, 2011).Nevertheless, fMRI is arguably the most dominant neuroimaging methodology to have been used in the study of stuttering.Since the main strength of both fMRI and PET is a high degree of spatial resolution, the studies using these paradigms havelargely utilized similar experimental tasks. This also includes resting-state, although in fMRI the analysis of such data is notstraightforward. Because raw measurements acquired using standard fMRI protocols are not comparable across participants,one must rely on contrasting different conditions when comparing regional activations between participants. In a resting-state design there is only a single condition, and one must rely instead on functional connectivity analysis, which utilizescorrelations or model-fitting across multiple regions of interest. Here, we focus first on fMRI studies in AWS, followed bystudies in CWS.

There have been a total of 22 fMRI studies of AWS, some combined with adolescent CWS (Blomgren, Nagarajan, Lee, Li,& Alvord, 2004; Chang, Kenney, Loucks, & Ludlow, 2009; Chang, Horwitz, Ostuni, Reynolds, & Ludlow, 2011; De Nil et al.,2008; Giraud et al., 2008; Howell, Jiang, Peng, & Lu, 2012; Jiang, Lu, Peng, Zhu, & Howell, 2012; Kell et al., 2009; Loucks et al.,2011; Lu et al., 2009; Lu, Chen et al., 2010; Lu, Peng et al., 2010; Neumann et al., 2004, 2005; Preibisch et al., 2003; Sakai,Masuda, Shimotomai, & Mori, 2009; Toyomura et al., 2011; Toyomura et al., 2015; Van Borsel, Achten, Santens, Lahorte, &Voet, 2003; Watkins, Smith, Davis, & Howell, 2008; Wymbs, Ingham, Ingham, Paolini, & Grafton, 2013; Xuan et al., 2012),and only one published study using fNIRS with AWS (Tellis, Vitale, & Murgallis, 2015). Liu et al.’s (2014) fMRI study and Satoet al.’s (2011) fNIRS study examined brain activity in a mixed-age group spanning from young CWS to AWS. Only one studyhas used fMRI in CWS alone (Chang & Zhu, 2013).

Six other articles published since our search was completed (all in AWS) are also included in the discussion: Halag-Miloet al. (2016), Kell et al. (this issue), Lu et al., 2016, Neef et al., 2016; Sitek et al., 2016 and Yang, Jia, Siok and Tan (2016).One study has utilized fMRI to examine connectivity in CWS (Chang et al., 2016). Finally, one unpublished thesis (Kazenski,2015) used fNIRS to look at AWS and CWS, whilst another looked at AWS alone (Brown, 2015).

3.2.1. AWSfMRI was first used to investigate the neural substrates of stuttering around 2003 when PET was still the dominant

methodology. Most of the early fMRI studies examined differences in localized activations, whereas later studies haveincluded differences in functional connectivity between multiple regions of interest in the brain. Additionally, whilst themajority of studies have focused on speech production tasks, a small number have examined other contexts such as theresting state, speech perception, or changes induced by fluency enhancing treatments. All past meta-analyses of functionalimaging in stuttering included both fMRI and PET studies (Belyk et al., 2015; Brown et al., 2005; Budde et al., 2014; also seeNeef, Anwander, & Friederici, 2015; Neef, Hoang, Neef, Paulus, & Sommer, 2015 for a summary of these studies). There isas yet no meta-analysis of stuttering exclusively restricted to fMRI (or PET). This is unfortunate given the methodologicaldifferences between the techniques, which were discussed above.

3.2.1.1. Resting state. It has been proposed that any differences in regional brain activations between AWS and AWDS maybe present both during speaking and in the absence of a task, that is in the resting state. Xuan et al. (2012) found thatAWS had higher amplitude low frequency fluctuations (a measure of resting state brain activity) during rest than AWDSin several cortical regions associated with speech: left superior and middle temporal gyri and the triangular portion of leftIFG. Moreover, AWS had reduced amplitude low frequency fluctuations in bilateral SMA. Largely consistent with Xuan et al.,Yang et al. (2016) reported reduced connectivity between right SMA and basal ganglia as well as between bilateral superiortemporal gyrus and basal ganglia (cf. Chang & Zhu, 2013; Lu et al., 2012). These connectivity changes likely affect sensorimotorintegration during speech production. Yang et al. also detected increased connectivity between the cerebellum and right IFGin AWS compared to AWDS, which was positively correlated with stuttering severity and perhaps indicates a compensation
mechanism. Lu et al. (2016) reported resting state functional connectivity differed between the left IFG and Heschl’s gyrus.As this was related to their performance on a speech perception task, the authors suggested that anomalous connectivitycontributed to speech perception deficits in AWS. More recently, Desai et al. (2016) used pulse arterial spin labeling toacquire perfusion data in a group of AWS and CWS. They reported a reduced CBF in the left IFGof AWS as compared to AWDS

tb

3vdwFgwc

2thTst2v2

(wrlwss

(msaepta(mTlodMc

rtradvsAisfcst


hat was correlated with more severe symptoms. Notably, because this was also evident in CWS, the authors proposed it toe associated with the pathophysiology of stuttering.

.2.1.2. Speech production tasks. In 2003 Van Borsel et al. published the first fMRI study of AWS. This investigation used aariety of conditions including reading meaningful and nonsense words silently and aloud. Their analyses of AWS did notetect any activation of auditory cortices when contrasting silent reading and reading aloud. This was compared to AWDS,ho exhibited similar levels of activation in both conditions. This finding has since been replicated (e.g., De Nil et al., 2008).

urther, whereas AWDS demonstrated a left dominance for activation in regions such as Wernike’s area and middle temporalyrus, AWS activated right hemisphere structures to the same extent as left. Blomgren et al. (2004) obtained similar findingshen asking participants to “think” of a target word; although, the results were not significant when corrected for multiple

omparisons. Van Borsel et al. therefore proposed that stuttering arises because of abnormal cerebral dominance.The right inferior frontal gyrus (IFG) is a cortical structure often implicated in the etiology of stuttering (Brown et al.,

005; but see Budde et al., 2014). This region was shown to be overactive in AWS as compared to AWDS in a variety of speechasks (Lu et al., 2009; Neumann et al., 2005; Sakai et al., 2009 and see also Neef et al., 2016), with some studies showingigher right IFG activity associated with lower rates of dysfluency (Braun et al., 1997; Kell et al., 2009; Preibisch et al., 2003).his finding lead Preibisch and colleagues to propose that right IFG is involved in compensation for stuttering, possiblyubstituting for impaired networks in the left hemisphere. The precise nature of such compensation is unclear given that (a)he right IFG was also overactive in tasks that did not require overt or covert speech (Halag-Milo et al., 2016; Preibisch et al.,003), (b) it is not associated with symptom relief (Budde et al., 2014; also see Kell et al., 2009; Yang et al., 2016 for similariews), and (c) in some studies, higher right IFG activity is associated with higher, not lower, rates of fluency (Ingham et al.,000).

Theorizing that differences in brain activity may underlie alterations in motor behavior during stuttering, De Nil et al.2008) attempted to examine the neural correlates of simulated (i.e., pretended) stuttering. They compared neural activityhen participants were passively listening to words, when they were pretending to stutter the word, or when they fluently

epeated words aloud. AWS exhibited stronger activation of bilateral IFG, bilateral superior temporal gyrus, left insula andeft supramarginal gyrus when pretending to stutter than when simply repeating the stimulus word once. Notably however,

hen stuttering, AWS exhibited higher activation only in the right IFG. These findings were taken to suggest that the motorystems of AWS were more heavily taxed than those of AWDS, requiring the extra recruitment of the right IFG. They alsohow that increased activation of the superior temporal gyrus is not necessarily associated with an increase in fluency.

Inter-individual variation in neural activity associated with stuttering has also been a topic of interest. Wymbs et al.2013) noted that, for some individuals, there was bilateral overactivation of the superior temporal gyrus, insula and/ or

otor regions during stuttered speech, but little consistency in patterns of brain over- or under-activation during stutteredpeech between participants. There was not a great deal of overlap in brain activation associated with stuttered speechcross participants. The discrepancies between De Nil et al. (2008) who found differences at the group level and Wymbst al. who found differences at an individual level, may lie in how stuttering was elicited. In the De Nil et al. study, participantsretended to stutter, whereas in the Wymbs et al. study words were selected on an individual basis depending on how likelyhey were to elicit stuttering. Furthermore, neither study subdivided stuttered speech into different types of dysfluenciesnd it is possible that this contributed to the individual differences observed. This idea receives some support from Jiang et al.2012) who used pattern analysis of neural activity to classify dysfluencies in AWS during a sentence completion task as: (1)

ost typical (i.e., specific to stuttering speakers) or; (2) least typical (i.e., commonto both stuttering and fluent speakers).he left IFG and the bilateral precuneus showed higher brain activity for the most typical symptoms, whilst activity in theeft putamen and right cerebellum showed the strongest relationship to the least typical symptoms. Since different patternsf brain activation are associated with planning vs. execution of speech (Lu, Chen et al., 2010), as well as different types ofysfluency, it might be the case that stuttering could result from failure in either the planning or execution stage of speech.ore generally, this study points to the need to consider a wider range of behavioral variables to shed light on some apparent

ontradictory findings across studies.Whereas early imaging work tended to focus on localizing differences between AWS and AWDS in the cerebral blood flow

esponse, later studies have aimed to identify differences in functional connectivity between different brain regions. One ofhe first studies to employ connectivity analysis to the study of developmental stuttering was Lu, Peng et al. (2010), with theationale that differences in levels of neural activation during speech production may only be understood when consideredlongside functionally connected regions. Results from an overt picture-naming task showed that there were significantifferences between AWS and AWDS in the output of the basal ganglia to the SMA and to the middle temporal gyrus, but notice versa. Based on this finding, Lu, Peng et al. (2010) suggested that these regions were unable to receive proper timingignals from the basal ganglia. In contrast, the SMA was largely intact and was able to send signals to the basal ganglia.n earlier study by the same group (Lu et al., 2009) performed connectivity analysis on a much larger network of regions

nvolved in speech production, this time using a covert rather than an overt picture-naming task. The output of the leftuperior temporal gyrus to the left IFG was the same between AWS and AWDS groups, supporting the idea that processes
or retrieving the phonological code for an intended utterance are intact in AWS. However, the AWS lacked a functionalonnection from the left IFG to the precentral gyrus; a connection that was present in AWDS. Connections between theuperior frontal gyrus and the SMA and the precentral gyrus exhibited opposite patterns in AWS and AWDS, which wasaken to reflect AWS having difficulties in the initiation and inhibition of speech. There were also stronger connections


between the right cerebellum and the precentral gyrus in AWS than AWDS. Since the cerebellum has been implicated inerror monitoring and timing of complex sequences of movement, Lu et al. (2009) suggested that AWS require more effort tomonitor speech. This may occur either due to less automated programming, greater reliance on feedback (Cai et al., 2014)or high error-proneness in the speech programming system of AWS. Areas of the cerebellum are implicated in the temporalcontrol of speech (Mathiak, Hertrich, Grodd, & Ackermann, 2002) and temporal processing more generally (see Petter et al.,2016, for a review).

Since these studies point to deficits at potentially multiple levels of speech production, Lu, Chen et al. (2010) attemptedto separate the neural activity associated with planning vs. execution of speech. The task required participants to eithersay a one-syllable word once (non-repeated), repeat a one-syllable word three times (repeat condition), or to say a threesyllable word (three syllable). The contrast of repeat condition vs. the three syllable condition allowed an examination of theneural substrates associated with planning, and the contrast between the non-repeated and repeated conditions allowedan examination of the neural substrates associated with execution. Their analysis revealed atypical planning was associatedwith connectivity between the bilateral IFG and right putamen, and that the left angular gyrus and premotor areas, and theright insula and cerebellum, contributed most to atypical execution. These findings provide support for the dual premotorsystems theory (Alm, 2007; see also Etchell, Johnson, & Sowman, 2014).This theory posits motor control is made up of twosystems − the lateral system comprised of the cerebellum and the premotor areaand the medial system comprised of thebasal ganglia and the SMA. Lu, Chen et al. (2010) also drew particular attention to the angular gyrus as the interface betweenplanning and execution because AWS showed stronger connections from the bilateral IFG and weaker connections from thepremotor areas to the angular gyrus relative to AWDS. Evidence for the idea that the cerebellum can compensate for deficitsin the basal ganglia circuity is provided by Sitek et al. (2016) who found that AWS with decreased connectivity between thecerebellum and frontal regions exhibited the most severe stuttering.

Although not focusing on connectivity, Chang et al. (2009) also separated perception, planning and execution of speechand non-speech movements (examining changes at activation level rather than connectivity). Consistent with previouswork (e.g., Braun et al., 1997; Fox et al., 1996), the authors found that perception of speech elicited reduced activations inmotor and auditory areas of AWS, but that during production, AWS had greater activation of motor and auditory areas. Laterwork by the same group extended their previous study to examine functional connectivity. Chang et al. (2011) reportedthat compared to AWDS, AWS exhibited weaker functional connectivity between the left IFG and the premotor cortex andsuperior temporal gyrus (cf. Lu et al., 2009). AWS also showed greater connectivity between left Brodmann Area 44 and theright postcentral gyrus, and left anterior cingulate, but there were no differences in connectivity between BA44 and auditoryregions. Chang et al. observed similar deficits in connectivity during speech and non-speech tasks, again indicating that suchdeficits are likely related to the pathogenesis of stuttering.

One other study has documented abnormal connectivity in AWS relative to AWDS during control of rising and fallingtones in Mandarin (Howell et al., 2012). It was found that AWDS exhibited significant functional connections between theleft laryngeal motor cortex and the left insula. In contrast to this, AWS showed a connection from the left laryngeal motorcortex to the putamen, a region identified in the Belyk et al. (2015) meta-analysis as being commonly activated across severalstudies of speech in stuttering.

It is noteworthy that there are similar differences in the neural activity between AWS and AWDS during speech and non-speech movements. Specifically, Chang et al. (2009) showed that AWS show less activation in the left superior temporal gyrusand premotorareas but greater activation in the right auditory areas, putamen and precentral gyrus for both speech and non-speech oral movements, compared with AWDS. This highlights the possibility that while stuttering typically manifests inthe domain of speech, it affects other oral movements as well. Similar hypotheses have been forwarded for other conditionssuch as apraxia which are traditionally considered to be specific to speech (e.g., Ballard, Robin, & Folkins, 2003).

Other work has also examined activity in subdivisions of left and right frontal areas in AWS and AWDS to attain greateranatomical specificity with respect to the location of the deficits (Neef et al., 2016). These authors found reduced activityin a cluster typically associated with speech production. Notably, Kell et al. (2009) highlight the importance of consideringboth structural and functional deficits. They assert that whereas the left IFG exhibits structural and functional deficits andis associated with the pathophysiological basis of stuttering other areas are likely a downstream effect.

3.2.1.3. Speech perception tasks. In addition to examining neural activation differences during speech production, a smallnumber of fMRI studies have looked at differences in brain activity during speech perception. Chang et al. (2009) examinedneural activity associated with both speech perception and production. They reported that during perception AWS hadreduced activation in the SMA, insula and angular gyrus and less activation in superior temporal regions, but that duringproduction the reverse held true. This finding is consistent with stuttering being a disorder of speech production, whereinproduction places greater demands on brain regions as compared to speech perception in AWS. De Nil et al. (2008) andHalag-Milo et al. (2016) on the other hand, reported increased activation in bilateral auditory regions and right frontalregions during passive listening in AWS relative to AWDS. These discrepancies may arise due to differences in task demands(passive vs. active listening) or the complexity of stimuli. Indeed, as the magnitude of task complexity increases, so too does
the concentration of hemoglobin in the blood − the primary determinant of activation levels in fMRI experiments (Telliset al., 2015). In the Chang et al. study, the stimuli consisted of meaningless consonant vowel strings whereas in the De Nilet al. and Halag-Milo et al. studies, the stimuli were single syllable words and sentences, respectively. In Halag-Milo et al.,auditory over-activations were limited to left superior temporal gyrus, which was thought to be an attempt to overcome

dAs

pift

3stotfe

LiwbtraacrttTr

ssiciah

nfceps(bgA

3

b

3abtbb


ecreased temporal sensitivity in stuttering (see Beal et al., 2010, 2011). Recently, Lu et al. (2016) also reported that AWS andWDS exhibited different patterns of activation in the left IFG and anterior insula for both the perception and production ofpeech.

It is possible to identify differences between AWS and AWDS in areas classically associated with speech production anderception in approximately 3 min (Loucks et al., 2011). This is important because it provides a means by which to rapidly

dentify neural correlates of speech production in children and who have difficulty remaining still and maintaining attentionor extended period of time. The tasks in this experiment involved picture naming, but this rapid scanning sequence has yeto be applied to studies of CWS and children who do not stutter (CWDS).

.2.1.4. Intervention. Several studies have demonstrated that the functional abnormalities observed in speech regions duringpeech and non-speech tasks can be altered when AWS undergo speech therapy. For example, precision fluency speechherapy consisting of an inpatient program for 3 weeks and maintenance for 1–2 years appears to attenuate or eliminatever-activation in the right IFG (Neumann et al., 2004). In accordance with PET studies (De Nil et al., 2001; De Nil et al., 2004),herapy can also reduce cerebellar over-activation in AWS (Lu et al., 2012; Toyomura et al., 2015). This may be importantor remediating temporal control of speech because the cerebellum is hypothesized to be involved in the timing of externalvents (that is responding in a temporally dependent way to external cues).

Lu et al. (2012) reported that functional connectivity in the left IFG did not change in AWDS following speech treatment.u et al. (2012) however did not provide details about what other regions were functionally connected to the IFG. Thesenconsistencies between studies could partly be attributed to the former measuring differences in resting state connectivity

ith the latter measuring activation during a speech production task. In an alternative view, Kell et al. (2009) assert thatecause they observed both structural and functional deficits in the IFG, but only functional deficits in other areas such ashe basal ganglia, IFG likely has a critical role in the pathological basis of stuttering; moreover, changes in activations in otheregions are likely to be sequelae of stuttering. The same group (Kell et al., this issue) performed a reanalysis of their 2009 datand interestingly, showed that the increased activation previously associated with recovered adults who stutter was notssociated with increased connectivity between speech motor regions like the SMA. Instead, it was associated with reducedonnectivity, suggesting that perhaps activation of Brodmann Area 47/ 12 uncouples an abnormal Broca’s area from theest of the speech production system thereby allowing it to regain normal function and enabling long term recovery. Takenogether, these differences of opinion regarding the regions causally linked to stuttering, and their amenability or resistanceo change with treatment, highlight the need to carefully consider how studies are designed and how data are interpreted.hey further underscore the need to consider not only levels of activation, but the degree of connectivity between differentegions as well.

Although transient, fluency inducing techniques can have an effect on dysfluent speech similar to that conferred bypeech therapy. Such techniques include delayed and frequency shifted auditory feedback, metronomic pacing of speech,hadowed or chorals peech, and so on. Some of these appear to induce similar neural effects to traditional speech therapy,ncreasing activity in the bilateral superior temporal cortex in both AWS and AWDS (Watkins et al., 2008). Under theseonditions however, AWS still exhibit overactivity in the bilateral insula, cerebellum and midbrain as well as under-activityn the bilateral ventral premotor and sensorimotor cortex relative to AWDS. There is evidence that the differences in neuralctivity between AWS and AWDS tend to be confined to the left hemisphere rather than occurring bilaterally or in the rightemisphere alone (Toyomura et al., 2011).

The differences between fluency inducing conditions are important, having the potential to shed light on the range ofeural mechanisms by which fluency can be induced. Delayed auditory feedback (Watkins et al., 2008) and frequency altered

eedback (Sakai et al., 2009) lead to a greater activation in the right IFG than normal feedback. Metronomic pacing, but nothoral speech, and so on, increases activity in the basal ganglia of AWS to a level comparable to that seen in AWDS (Toyomurat al., 2011). This is thought to result from the metronome providing an external cue with which to pace speech, and theacing being integrated into the speech signal. Yet, the interpretation of results for some fluency inducing conditions can beomewhat difficult. For example, in the Watkins et al. study, both groups stuttered more during delayed auditory feedbackwhich is meant to induce fluency). Similarly, an unpublished thesis (Kazenski, 2015) found that laterality did not differetween AWS and AWDS during either normal speech or in fluency enhancing conditions. The AWS did however exhibitreater “effort” during both speech types as evidenced by greater bilateral change in hemoglobin concentration relative toWDS.

.2.2. CWSSince 2011, there have been a small number of fMRI studies examining CWS, with findings showing some similarities

ut also differences when compared with results from adult studies.

.2.2.1. Resting state. Chang and Zhu (2013) focused on the basal ganglia thalamo-cortical circuit and auditory-motor regionsnd found various differences between CWS and CWDS. Specifically, CWS had reduced resting state functional connectivity
etween the putamen and left SMA, superior temporal gyrus, insula/ rolandic operculum, and stronger connectivity betweenhe right superior frontal gyrus and the putamen, compared with CWDS. CWS also had reduced functional connectivityetween the left SMA and the left putamen, cerebellum and right superior temporal gyrus and increased connectivityetween the SMA and the left cerebellum and right paracentral lobule compared with CWDS. The authors suggest that


differences in these resting state networks likely relate to differences in the organization of self-timed movements in children.Much like the PET data for AWS (Ingham et al., 2013), which highlighted the role of the basal ganglia in stuttering, theseresults deserve further investigation. Further support for this contention comes from another study by the same group. Changet al. (2016) showed that resting connectivity from the putamen to the SMA and other regions of the brain is significantlyreduced in CWS relative to CWDS as it relates to a rhythm discrimination task

Most recently, Chang and colleagues reported a larger study of aberrant patterns of resting state connectivity betweenpersistent and recovered CWS and their fluent peers (Chang et al., this issue). Interestingly, there were significant groupdifferences in large scale intrinsic connectivity networks associated with attention and default mode networks that couldfurther differentiate between persistent and recovered individuals who stutter. The study by Chang et al. highlights the needfor researchers to consider regions and networks in stuttering that are not normally associated with speech production. Ofnote, Xuan et al. (2012) found decreased connectivity within the default mode network and increased connectivity withinthe sensorimotor networks of AWS. O’Neill et al. (2017) reported differences in neurometabolism in regions associated withattention such as the bilateral superior temporal gyrusand left putamen in CWS and AWS. Likewise, Desai et al. (2016)highlighted the role of the left IFG in the pathophysiology of stuttering as they observed that reduced cerebral blood flow tothe left IFG in both CWS and AWS was associated with greater stuttering severity.

Functional near infrared spectroscopy (fNIRS measures hemodynamic activity from the scalp) has shown promise as amore accessible and feasible method of studying brain activation during speech production in CWS. Using this method,Brown (2015) examined speech production in both CWS and AWS and reported both AWS and CWS showing increasedactivation over the right IFG and decreased activation in the left IFG during speech. The latter were interpreted to reflectinefficient feedforward mechanisms (Max et al., 2004); while others have argued that abnormal feedback systems, ratherthan feedforward systems explain stuttering behaviors, these studies have typically included only AWS (Cai et al., 2014).Because the average age of the children was around 11, the right hemisphere activation might still be the result of havingstuttered for a number of years. Apart from the novel methodology, this study also stands out in being an important cross-sectional investigation of preschool children, school-aged children, and adults as it is one of a few to directly compare AWSand CWS. Such studies are important because they can help disentangle patterns of activation that change over time inresponse to stuttering from those that are more causally related to it.

3.2.2.2. Speech perception tasks. Using the fNIRS method, Sato et al. (2011) found atypical laterality in CWS in a speechperception task. Specifically, they compared the responses in auditory cortex to phonetic and phonemic contrasts. Whilstthe AWDS and the CWDS exhibited a left lateralized response for the phonemic contrast relative to the phonetic contrast,there was no difference between the conditions for AWS and CWS. This was true even on an individual basis, suggestingthat functional lateralization of speech was disorganized in children who stutter.

3.2.2.3. Other. Based on the assumption that prenatal exposure to toxoplasma gondi could affect brain development and inparticular the basal ganglia, C elik et al. (2015) attempted to establish whether there was a relationship between this bacteriaand stuttering. They recruited a large sample of young CWS (n = 30), adolescents CWS (n = 35) and their fluent peers. Theseauthors found that the presence of the virus was significantly higher in the stuttering group but no differences were detectedin either MRI or the EEG signal between the groups. Unfortunately, the precise nature of any experimental tasks and themethods of statistical analysis were not described, thereby making it difficult to interpret the null imaging findings.

3.3. Electroencephalography

In contrast to both fMRI and PET methods, which have relatively poor temporal resolution relative to the timescalesof neural activation, EEG has excellent temporal resolution. EEG is able to resolve brain activity occurring at millisecondtimescales rather than over several seconds. This permits, for example, temporal differentiation between speech preparationand production and the measurement of oscillatory brain activity. EEG is also considerably cheaper and quieter than fMRI.Because the recording electrodes are affixed to an individual’s scalp rather than in a set position, EEG is considerably moretolerant of head movement, providing a major advantage for studies of CWS. However, EEG has poor spatial resolutionand is therefore not able to reliably localize sources of neural activity in brain space. Because EEG is also very sensitive tomuscle artifacts, recordings during continuous speech are difficult; consequently, the majority of EEG studies have focusedon auditory and/ or linguistic processing whilst others have focused on preparatory speech activity.

There have been 25 EEG studies relating to speech production and auditory processing in AWS, some combined withadolescent CWS (Achim, Braun, & Collin, 2008; Andrade, Sassi, Matas, Neves, & Martins, 2007; Angrisani, Matas, Neves,Sassi, & Andrade, 2009; Arnstein, Lakey, Compton, & Kleinow, 2011; Corbera, Corral, Escera, & Idiazábal, 2005; Cuadrado &Weber-Fox, 2003; Daliri & Max, 2015; Dietrich, Barry, & Parker, 1995; Hampton & Weber-Fox, 2008; Joos, De Ridder, Boey,& Vanneste, 2014; Khedr, El-Nasser, Abdel Haleem, Bakr, & Trakhan, 2000; Liotti et al., 2010; Maxfield, Huffman, Frisch, &Hinckley, 2010; Maxfield, Pizon-Moore, Frisch, & Constantine, 2012; Maxfield, Morris, Frisch, Morphew, & Constantine, 2015;
Mock, Foundas, & Golob, 2015; Morgan, Cranford, & Burk, 1997; Murase, Kawashima, Satake, & Era, 2016; Rastatter, Stuart, &Kalinowski, 1998;Sassi, Matas, de Mendonc a, & de Andrade, 2011; Tahaei, Ashayeri, Pourbakht, & Kamali, 2014; Vanhoutteet al., 2015; Weber-Fox & Hampton, 2008; Weber-Fox, Spencer, Spruill, & Smith, 2004; Weber-Fox, 2001). Additionally, therehave been 9 EEG studies in CWS (Arnold, Conture, Key, & Walden, 2011; Jansson-Verkasalo et al., 2014; Kaganovich, Wray,

&2r

da

33e(prbwo

3pmpwr

tmssaeswmwc

rmpipi

owcabt1Ctrm

irais


Weber-Fox, 2010; Mohan & Weber, 2015; Özcan, Altınayar, Özcan, Ünal, & Karlıdag, 2009; Özge, Toros, & C ömelekoglu,004; Usler & Weber-Fox, 2015; Weber-Fox &Hampton, 2008; Weber-Fox, Wray, & Arnold, 2013). One single study hasecruited a mixed-aged group spanning from young CWS to AWS (Khedr et al., 2000).

Six papers have been published since the search was completed and, due to their relevance, they are considered in theiscussion: Maxfield et al. (2016), Mock, Foundas and Golob (2016), Piispala et al. (2016), Prestes et al. (2016), Sengupta etl. (2016) and Vanhoutte et al. (2016).

.3.1. AWS

.3.1.1. Resting state. One study has examined resting state EEG in AWS. Much like fMRI and PET studies, there is somevidence that AWS and AWDS exhibit different patterns of brain activation in the absence of any task. Recently, Joos et al.2014) found reduced oscillatory connectivity in the beta band (12.5–18 Hz) in AWS, relative to AWDS, between the leftars triangularis (BA45) and right primary motor cortex (BA4), as well as between the left pars opercularis (BA44) and theight premotor cortex (BA6) and primary motor cortex (BA4). The sources were localized using standardized low resolutionrain electromagnetic tomography (sLORETA), a method of distributed source analysis (Pascual-Marqui, 2002). It is unclearhether such differences are evident in AWS or CWS during or prior to speech, because no other study has investigated

scillatory connectivity in stuttering.

.3.1.2. Speech tasks. In task-based paradigms, there is growing evidence that AWS have abnormalities in not only the neuralrocesses associated with speech movement, but also the neural processes associated with the preparation to move. Oneeasure of movement preparation is the contingent negative variation (CNV): a slow wave measured in either EEG or MEG

receding the onset of movement that occurs between two defined stimuli. Achim et al. (2008) found that this componentas smaller in AWS than AWDS preceding both fluent and stuttered speech. The authors observed a larger CNV over the

ight hemisphere for AWS only when speech was dysfluent.The CNV has also been investigated in a more recent EEG study by Vanhoutte et al. (2015). Because the amplitude of

he CNV is dependent on baseline measures which can make it hard to detect between-group differences, Vanhoutte et al.easured the slope of the CNV, which is not baseline-dependent. These authors reported that during the fluent production of

ingle words, the slope of the CNV was steeper in AWS than in AWDS and the slope was positively correlated with stutteringeverity. Since the CNV is generated by structures in the basal ganglia-thalamo-cortical loop (Bares & Rektor, 2001), theuthors suggested AWS have a dysfunction in that network. This explanation is consistent with fMRI studies (e.g., Lu, Chent al., 2010; Lu, Peng et al., 2010). However, it is also possible that the increased slope reflects greater effort to achieve theame degree of motor readiness or potentially the strength of impulse control applied during the preparation period. In otherords, AWS might actually be less prepared to move and require a greater degree of preparation to execute the plannedovement. A more recent study by Vanhoutte et al. (2016) demonstrated that the CNV was smaller preceding stutteredords, compared to fluent words. This further correlated with measures of stuttering severity perhaps indicating it is a

ompensatory response that promotes fluency.Later, Ning et al. (2017) reported that a premature shift from the early to late CNV reflects an aberrant timing of speech

esponse. Interestingly, there are correlations between the beta band and the CNV, and both are influenced by drugs thatodulate dopamine (see Kononowicz & Penney, 2016). Mock et al. (2016) showed that beta desynchronization during

reparation for speech correlated with stuttering rate. Further, Mersov et al. (this issue) showed that beta desynchronisations reduced in AWS prior to stuttered as compared to fluent speech thereby providing further evidence that without sufficientreparatory activity, speech may be disfluent. More broadly, these indirectly confirm the suggestion that oscillatory function

n the beta band −associated with motor activity − is linked with stuttering (Etchell et al., 2014).Arnstein et al. (2011) used EEG to investigate error-monitoring processes during a task where AWS had to judge whether

r not a “target” was orthographically similar or dissimilar and rhymed or did not rhyme with a “test” word. Responsesere made via a button press and errors were indexed by the error-related negativity (i.e., N100, an electrophysiological

omponent that peaks about 100 ms after the initiation of an incorrect response) and the error related positivity (P300,nother component that peaks 200–400 ms after an incorrect response). For example, the N100 and P300 components coulde observed when judging that a target word rhymed with a test word when, in fact, it did not. The authors hypothesizedhat excessive error monitoring could lead to dysfluencies by disrupting speech planning (see also Postma, Kolk, & Povel,990). Speech is disrupted because, once an error is detected, a new correct speech plan must be activated or formed (seeivier et al., 2010). AWS showed a larger amplitude N100 regardless of the accuracy of their response. One interpretation ofhis result is that AWS perceive their speech plan as being incorrect, even when there is nothing wrong with it, and that theesulting attempts to repair speech cause stuttering. Indeed, even when speech rate is held constant, AWS make and detectore errors than AWDS (Brocklehurst & Corley, 2011).Abnormal electrophysiological activity associated with the preparation to speak is not limited to motoric processes;

t also extends to sensory processes. Daliri and Max (2015) investigated speech planning by examining auditory evoked
esponses to tones presented either prior to speech or prior to silent reading. Whereas the AWDS showed a reduction in themplitude of the N100 response on the tone trials relative to the control condition (tones presented before a fixation cross)n the speech condition, AWS did not. This was taken to suggest that AWS have difficulty in predictively modulating sensoryystems prior to speech. Mock et al. (2015) also presented participants with a tone while they were preparing to speak,


but they first received a cue that either allowed them to prepare to name a target or did not allow them to prepare. Theelectrophysiological response was thought to be an index of the strength of efference copy (i.e., motor cortex sending copiesof motor commands to sensory regions). Mock et al. (2015) found a significantly reduced amplitude of electrophysiologicalresponses at ∼130 ms in the bilateral ventral premotor cortex in AWS compared to AWDS suggesting that the generationof efference copies of motor commands might be impaired in AWS. A recent study also reported that, whereas AWDSdemonstrated adaptations to perturbations of auditory feedback, no such response was evident in AWS (Sengupta, Shah,Gore, Loucks, & Nasir, 2016). Additionally, behavioral responses were associated with corresponding differences in neuralactivity which may suggest a motor programming disruption whereby AWS are less able than AWDS to incorporate errorsin sensory feedback into internal models for motor control (Max et al., 2004) or feedforward motor commands (see Cai et al.,2012), resulting in restarts and initial syllable repetitions (Civier et al., 2010).

The deficits in sensorimotor processing above may explain a variety of the abnormal neural responses to auditory stimulithat have been described for AWS relative to AWDS. These include shorter middle latency auditory responses (Dietrichet al., 1995), left lateralized mismatch negativity responses to “oddball” stimuli representing phonetic contrasts (Corberaet al., 2005), and increased right hemisphere amplitude of P300 to infrequent pure tone stimuli (Morgan et al., 1997). Thesestudies support abnormal auditory processing and the hypothesis of altered cerebral dominance in stuttering. Hamptonand Weber-Fox (2008) cautioned that these abnormal responses may be modulated by skill level. They found that only theAWS individuals performing well below normal levels on an oddball detection task showed atypical N100, P200 and P300components.

There has been some controversy surrounding electrophysiological measures of AWS. Some studies have not founddifferences in the amplitude or latency of the P300 in response to pure tone stimuli for AWS (Andrade et al., 2007; Angrisaniet al., 2009; Sassi et al., 2011). It may be that neural responses to unexpected changes in tone frequency cannot reliablydifferentiate between AWS and AWDS. It might also depend on the task difficulty. In a single oddball task condition, Maxfieldet al. (2016) found no difference in P300 amplitude between AWS and AWDS. However, in a dual task condition whenperforming the oddball task and a picture naming task at the same time, the P300 amplitude was reduced in AWS comparedto AWDS. Maxfield et al. argued that demanding tasks might have a more detrimental effect on AWS than AWDS. Presteset al. (2016) also pointed out that some of the deficits in auditory processing in AWS may be due to deficits in temporalprocessing, such as detection of short gaps (0–40 ms) between sounds.

3.3.1.3. Other. Speech production can be affected not only by preparatory responses and perception, but also by psycholin-guistic factors. Accordingly, Weber-Fox and colleagues have run a series of studies examining differences between AWS andAWDS in electrophysiological indices of semantic, phonological, and syntactic violations. ERP peaks (N100, P200 and N400)in AWS were significantly reduced in amplitude, relative to cognitively-matched controls, to closed class words (i.e., wordsthat provide grammatical information), open class words (i.e., words that provide referential information) and to violationsof semantic expectation (Weber-Fox, 2001). However, ERP differences between AWS and AWDS are not confined to this earlytemporal window. AWS have also shown a reduced amplitude and duration of the P600 elicited in this particular case by aresponse to subject verb agreement violations (Cuadrado & Weber-Fox, 2003), perhaps reflecting a lack of syntactic revision.AWS have also shown a greater peak difference amplitude, in the right hemisphere only, between a task involving rhymejudgments and a baseline task (Weber-Fox et al., 2004). In comparison, the amplitude of the difference components wassimilar across hemispheres for the AWDS. Consequently, these authors concluded that the core neurophysiological deficitin stuttering did not relate to phonological deficits as there was a general similarity in evoked responses between AWS andAWDS. Instead, they suggest that differences between AWS and AWDS are the result of an interaction of increased cognitiveload and other factors such as language constraints.

In line with the reasoning that psycholinguistic factors may disrupt speech fluency, some authors have tested whetherspeech breakdown is due to difficulties associated with word retrieval. Maxfield and colleagues (Maxfield et al., 2010, 2012)conducted aseries of experiments to test if various types of priming would elicit typical or atypical electrophysiologicalcomponents prior to speech. This research largely focused on the N400 which indexes the spreading of activation betweenrelated and unrelated words (see Jescheniak, Schriefers, Garrett, & Friederici, 2002). Semantically or phonologically relatedwords reduce the amplitude of the N400 response in AWDS but increase the amplitude of the N400 response in AWS(Maxfield et al., 2010, 2012). This suggests that, while priming can facilitate speech production in AWDS, the same stimulusin AWS leads to a spreading of activation that makes access to the target more difficult. The authors reasoned that AWS maytherefore employ a center-surround inhibitory mechanism in order to select the appropriate target. Supporting this idea,Maxfield et al. (2015) reported that AWS, but not AWDS, showed a P280 component immediately before speech production,indicative of identity priming. This was taken to suggest that the AWS had more focused attention than AWDS immediatelyprior to speech production.

3.3.1.4. Intervention. While a number of fMRI and PET studies have investigated the effects of fluency inducing mechanismson neural activity of AWS, far fewer have used EEG to measure the effects of intervention or fluency inducing techniques.
Three previous studies (Andrade et al., 2007; Angrisani et al., 2009; Sassi et al., 2011), found no difference in electrophys-iological measures before or after speech treatment. In contrast to this, Rastatter et al. (1998) found that delayed auditoryfeedback and frequency-altered feedback reduced abnormally high beta band oscillations in AWS, particularly over the lefthemisphere.

33iic

3r1(rnbbe

3piT

sD(aa(

CotnAprDtfstwab(apd

c(N(Pcanpvl


.3.2. CWS

.3.2.1. Resting state. Oscillatory activity across a variety of frequency bands can be influenced by different patterns of breath-ng. Using this manipulation Özge et al. (2004) showed that, at rest and during hyperventilation, CWS exhibited markedlyncreased delta oscillations, related to cerebral maturation as well as decreased beta oscillations, related to movement,ompared to CWDS. These results with respect to the beta band are similar to those in AWS (see Joos et al., 2014).

.3.2.2. Non-speech tasks. Several EEG studies have focused on investigating responses to auditory tones in CWS. CWS showeduced auditory evoked brainstem responses to tones (Khedr et al., 2000; Tahaei et al., 2014). Additionally, while frequent

kHz tones elicit similar neural responses in CWDS and CWS, oddball 2 kHz tones elicit a P300 response only in CWDSKaganovich et al., 2010). Similarly, Jansson-Verkasalo et al. (2014) demonstrated that CWS fail to show a mismatch negativityesponse to changes in linguistic features such as intensity or frequency (the response to changes in duration of vowels wasormal, though). However, Özge et al. (2009), using a repetition suppression paradigm found no difference in P50 suppressionetween CWS and CWDS (though see Kikuchi et al., 2011 for a MEG study reporting differences in repetition suppressionetween AWS and AWDS). By and large, these results are consistent with findings in AWS (e.g., Corbera et al., 2005; Dietricht al., 1995; Morgan et al., 1997) and show that auditory deficits are present close to the onset of stuttering.

.3.2.3. Cognitive control. While there is considerable evidence that differences in auditory processing may affect speechroduction, there is less evidence that emotion can affect the speech of CWS. Arnold et al. (2011) measured the electrophys-

ological indices during speech production after CWS and CWDS listened to conversations that conveyed different emotions.hey reported no difference in EEG measurements between CWS and CWDS.

Others have examined attentional control and speed of reaction times in CWS. Although CWS and CWDS tend to haveimilar manual reaction times (Till, Reich, Dickey, & Seiber, 1983), this finding is not always consistent (Cross & Luper, 1983).espite this, the latency of the electrophysiological components (particularly the N200) is longer in CWS than in CWDS

Piispala, Kallio, Bloigu, & Jansson-Verkasalo, 2016). The authors suggested that, given that CWS exhibit difficulties shiftingttention (Eggers et al., 2012), the N200 delay was related to poor attentional control, leading to slower stimulus processingnd thereby slower reaction times. This finding is in agreement withthe idea that AWS are less prepared to move than AWDSAchim et al., 2008; Vanhoutte et al., 2015).

Weber-Fox, Spruill, Spencer and Smith (2008) replicated their 2004 rhyme judgment study on AWS in a group of CWS.onsistent with their previous study, both CWS and CWDS exhibited an N400 response to rhyming targets. The latencyf the N400 was earlier over the left hemisphere than the right hemisphere for CWDS than for CWS (for both prime andarget stimuli) and was taken to suggest that CWS have abnormal laterality in processing linguistic stimuli. The contingentegative variation (CNV), related tomemory and retention, was smaller in amplitude in the CWS group. Since the CNV ofWS and AWDS is no different (recorded, but unreported in Weber-Fox et al., 2004), this result suggests a developmentalhenomenon that resolves with maturation. It should be noted, though, that the results of Weber-Fox et al. may simplyeflect differences in task performance; CWS had lower overall accuracy, possibly due to delayed language development.elayed language development might be reflected in increased latencies of electrophysiological responses. Consistent with

his reasoning, Weber-Fox et al. (2013) showed that verb agreement violations in a passive listening task elicited an N400or both CWS and CWDS, but the latency of the component was delayed for the stuttering group. In another study by theame group that also considered grammatical violations when listening to sentences (e.g., ‘Every day the children pretendso be superheroes’), AWS but not AWDS showed an increase in the N400 amplitude (Weber-Fox & Hampton, 2008), whichas interpreted as reflecting greater processing difficulty and/ or less efficient lexical access. Furthermore, AWDS exhibited

P600 for sentences containing verb agreement violations but not semantic violations, while AWS exhibited a P600 foroth conditions. Similarly, indicating that AWS take more time to process the same stimuli than do AWDS. Murase et al.2015) reported that AWS exhibit a smaller N400 and larger late positive component when listening to sentences. Notably,ll of these studies were conducted in the absence of overt speech production. While they shed light on psycholinguisticrocesses that may influence speech production, they do not provide direct electrophysiological evidence that AWS or CWSiffer from their fluent peers in speech production processes.

Recently, Usler and Weber-Fox (2015) examined whether different brain responses to semantic and syntactic violationsould differentiate between children who persisted and who recovered from stuttering. Relative to canonical sentencese.g., ‘Pingu is building a castle on the floor’), semantic violations (e.g., ‘Pingu is building a music on the floor’) elicited an400 effect across CWDS and persistent and recovered CWS. Additionally, when compared to canonical phrase structures

e.g., ‘Mommy and Daddy look at their son’), syntactic violations (e.g., ‘Mommy and Daddy look at that their son’) elicited a600 component (often elicited by syntactic violations) that was equal in latency and amplitude across the three groups ofhildren. However, when listening to Jabberwocky sentences, the same phrase structure violations elicited a P600 in CWDSnd recovered CWS but an N400 effect in persistent CWS. Mohan and Weber (2015) also reported that, as compared to
on-rhyming targets, rhyming targets elicited an N400 over anterior sites for CWDS and recovered CWS, but was absent forersistent CWS group. Perhaps the N400 response has potential as a marker for persistence of stuttering in children. Thatarious psycholinguistic factors were associated with the persistence and recovery of stuttering highlight the influence ofanguage related factors in this disorder.


3.4. Magnetoencephalography

MEG, like EEG, has excellent temporal resolution and is quiet. MEG measures the strength of the magnetic fields outsidethe head associated with current flow in the brain. Because the recording of magnetic flux is less spatially affected bydifferences in tissue homogeneity across the head, it is better able to localize neural activity than EEG, though MEG is stillnot as spatially accurate as fMRI. Like MRI, however, MEG is less tolerant than EEG to head movement because the sensorsare in a fixed position. This may maketesting children difficult unless one uses techniques for head-movement correction.Because the sensors need to be as close as possible to the head, testing young children also requires a pediatric MEG withsmaller helmet size, though currently such infrastructure is not widely available. Furthermore, MEG is considerably moreexpensive than EEG due to the need for significant electromagnetic shielding and the operating costs of using liquid helium.

There have been six MEG studies focusing on AWS (Beal et al., 2010; Biermann-Ruben, Salmelin, & Schnitzler, 2005;Kikuchi et al., 2011; Salmelin et al., 1998; Salmelin, Schnitzler, Schmitz, & Freund, 2000;Walla, Mayer, Deecke, & Thurner,2004) and 3 MEG studies focusing on CWS (Beal et al., 2011; Etchell, Ryan, Martin, Johnson, & Sowman, 2016; Sowman, Crain,Harrison, & Johnson, 2014). The Kikuchi and colleagues group also recently reanalyzed their previous data in frequency space,i.e., in terms of brain oscillatory dynamics as opposed to evoked amplitudes in their previous analysis (Kikuchi et al., 2016).

3.4.1. AWS3.4.1.1. Speech production tasks. The first MEG study with AWS examined neural responses to tones presented while partic-ipants were speaking or reading, with the purpose of identifying whether there were difference in the temporal dynamicsof speech in AWS and AWDS (Salmelin et al., 1998). While the timings of evoked responses determined from dipoles in theauditory cortices were similar in both groups, the amplitude of the M100 response was larger in AWS relative to AWDS,perhaps suggesting AWS require greater “effort” to achieve the behavioral response. The amplitude of the M100 was alsolarger in the left than the right hemisphere for AWS, which the authors argued was indicative of an imbalance of hemi-spheric lateralization that could disturb speech production. A later study by the same group (Salmelin et al., 2000) testedbrain activation during single word reading. These authors found that, whereas neural activations in AWDS progressed fromthe left inferior frontal cortex (involved in articulatory planning) to the left fronto-parietal cortex and then to the motorcortex, AWS showed a reversed pattern, where activation progressed from the fronto-parietal cortex to the left inferiorfrontal cortex. These findings were thought to reflect disordered generation of motor programs before the completion ofarticulatory planning in stuttering.

Biermann-Ruben et al. (2005) also examined the temporal dynamics of cortical activation in AWS and AWDS but testedthe more complex task of sentence reading rather than isolated word reading. AWS, but not AWDS, exhibited activation inthe left inferior frontal cortex between 95 and 145 ms after stimulus onset. AWS also engaged the right frontal operculum.These results again were taken to indicate aberrant hemispheric dominance.

To investigate the preparatory stage of speech production in more detail, Walla et al. (2004) measured the Bereitschaftpotential which occurs 50 ms prior to the onset of speech. They found that, whilst AWDS exhibited a Bereitschaft potential,AWS did not (c.f. EEG study by Vanhoutte et al., 2015). They proposed that AWS lacked a state of ‘focused verbal anticipation’that is linked to gathering and preparing information required to produce a word. The lack of preparatory activity is perhapsalso linked to decreases in cortical excitability in AWS relative to AWDS (see TMS studies below). Reduced motor preparatoryactivity may be a cause of reduced excitability in the motor cortex, which in turn could affect the initiation of speechmovements.

Another important component of speech production is the phenomenon of speech-induced auditory suppression − thereduction in an auditory evoked response to self vs. externally initiated speech. Using MEG, Beal et al. (2010) found thatthe amount of suppression in AWS was remarkably similar to the amount of suppression observed in AWDS. There was,however, a delay in the peak of the M100 component in AWS relative to AWDS (see also Liotti et al., 2010; for a similar studywith high-density EEG), suggesting that AWS exhibit a temporal deficit in processing their own speech.

3.4.1.2. Auditory and speech perception tasks. One MEG study has focused on auditory processing abilities of AWS. Kikuchiet al. (2011) assessed the abilities of AWS and AWDS to gate (i.e., suppress) the second of two sequential pure tones.Whereas AWDS exhibited a suppression of the M100 response, AWS did not. This was taken to suggest AWS have difficultiesin suppressing irrelevant sensory input (see Özcan et al., 2009, for an EEG study of auditory gating in CWS). A reanalysis ofthe same MEG data (Kikuchi et al., 2016) revealed that AWS show increased interhemispheric synchronization in the betaband in the superior temporal gyri. They suggested that right hemisphere increases were compensating for left hemispheredeficits. Taken together, these results and the results of Beal et al. (2010) provide further support for abnormal auditorymotor integration in AWS.

3.4.2. CWS
The first to use MEG to investigate CWS were Beal et al. (2011) who reported a delay in the M50 component for CWS
relative to CWDS, suggesting that auditory processing was slower in the stuttering group. Notably, this was similar to theirprevious finding (Beal et al., 2010) of a delay in the peak of the M100 component in AWS relative to AWDS (see also Liottiet al., 2010). These findings are interesting, but it should be noted that the study was conducted on children aged 6–12 years,

wl

ls&

eif

3

neeP&eoSA

33iTchttftScismi

erHTodt

wfiiaifm(dose


ell past the age of onset for stuttering. Furthermore, they used an adult-sized MEG instead of a pediatric MEG which isikely to be non-optimal for detection of neuromagnetic flux in a significant proportion of the cohort studied.

A later study by Sowman et al. (2014) pioneered the use of pediatric MEG in stuttering, and examined the hemisphericaterality of CWS and CWDS aged 3–6 years during a picture-naming task. These authors found no difference between groupsuggesting that abnormal laterality in stuttering develops only later in life (see also Choo, Chang, Zengin-Bolatkale, Ambrose,

Loucks, 2012).In a more recent study, the same group (Etchell, Ryan et al., 2016; Etchell, Johnson, & Sowman, 2016) showed that CWS

xhibit an out-of-phase pattern of beta band activity in response to isochronous sounds with a 450 ms interval. The authorsnterpreted this as evidence for a deficit in timing rather than in the processing of sounds (see also Alm, 2004), in line withMRI and diffusion tensor imaging (DTI) studies of CWS (e.g., Chang & Zhu, 2013).

.5. Transcranial magnetic stimulation

While the spatial and temporal dynamics of speech production can be readily examined with fMRI, PET, MEG and EEG,europhysiological techniques like transcranial magnetic stimulation (TMS) are unique because they allow probing of thexcitation and inhibition of cortical regions. There have been a total of seven studies that have used TMS to compare corticalxcitability and inhibition between AWS and AWDS (Alm, Karlsson, Sundberg, & Axelson, 2013; Busan et al., 2013; Neef,aulus, Neef, von Gudenberg, & Sommer, 2011; Neef, Jung et al., 2011; Neef, Hoang et al., 2015; Sommer, Wischer, Tergau,

Paulus, 2003;Sommer et al., 2009), and one additional study that was published after our search was completed (Busant al., 2016). Our search returned no TMS studies ofCWS. Notably, at the time of submission there were no published studiesf transcranial alternating current stimulation (tACS) or transcranial direct current stimulation (tDCS) in either AWS or CWS.ubsequently, Chesters, Watkins and Möttönen (2017) have published a study on the effects of tDCS on speech therapy inWS.

.5.1. AWS

.5.1.1. Nonspeech and speech tasks. The earliest TMS study on stuttering (Sommer et al., 2003) compared intra-corticalnhibition and intra-cortical facilitation between the hand area of the motor cortex of AWS and AWDS. By using a conditioningMS pulse to the hand area of the motor cortex followed by a test pulse, inhibitory and excitatory connections within motorortex can be measured. Whilst both of these measures were found to be normal in AWS, the authors incidentally reportedigher motor thresholds, suggesting reduced motor cortical excitability in AWS. Subsequently, the same group used TMSo examine the hypothesis that AWS might have aberrant inter-hemispheric communications. They used a TMS techniquehat measures inter-hemispheric inhibition; applying a single conditioning pulse to the motor cortex of one hemisphereollowed by a second test pulse approximately 10 ms later to the opposite hemisphere (Kujirai et al., 1993), and measuringhe resulting motor evoked potential (MEP). On recording activity from the abductus digitis minimi muscle in the hand,ommer et al. (2009) found that there was no significant difference between groups with respect to the amplitude of theonditioned motor evoked potentials (MEPs). This suggested that inter-hemispheric inhibition was normal in AWS. However,t may be that this measure was not sensitive enough. Accordingly, Busan et al. (2016) reported a higher threshold for theilent period (i.e., temporary suppression of muscular activity following a TMS pulse) in the left hemisphere of the tongueotor cortex in AWS compared to AWS. This finding likely reflects reduced intra-cortical inhibition in speech motor regions

n stuttering, even in the absence of stuttered speech.Other studies have found changes in motor cortex excitability for AWS (e.g., Alm et al., 2013; Busan et al., 2013; Sommer

t al., 2009). Alm et al. showed that motor thresholds in AWS were elevated in the left hemisphere relative to (a) their ownight hemisphere and (b) the left hemisphere of AWDS. In contrast, Busan et al. found no difference in resting thresholds.owever, they produced more evidence to suggest a reduced motor excitability in AWS, particularly in the left hemisphere.hey showed that, for stimulation at a given fraction above motor threshold, smaller MEPs were evoked in the right handf AWS compared to AWDS. Together these studies suggest that stuttering may be a symptom of a broader motor controlisorder because there is abnormal excitability, with elevated threshold and/ or reduced amplitude of motor excitability inhe hand representation of the motor cortex.

Until the last few years, most TMS studies of stuttering focused on the hand motor representation of the brain because itas deemed methodologically difficult to record from orofacial muscles like the tongue. Neef, Paulus et al. (2011) were therst to apply TMS in the oral mechanism in AWS. They showed that AWS have reduced short intra-cortical inhibition and

ntra-cortical facilitation in the tongue representation of the motor cortex. This highlighted that abnormalities exist botht rest in AWS relative to AWDS, as well as in speech (i.e., generating a verb; Neef, Hoang et al., 2015), in an area cruciallynvolved in articulation. Measuring the active motor threshold of AWS and AWDS during speech, Neef, Hoang et al. (2015)ound that AWS exhibited less excitability (i.e., higher motor thresholds) than AWDS in the tongue representation of the

otor cortex in both the left and right hemispheres of the brain. Additionally, AWS failed to show an increase in excitabilityMEP amplitude) prior to movement that was seen in AWDS up to 160 ms before speech onset. Importantly, this study
irectly linked observations of decreased cortical excitability to speech production. By demonstrating reduced excitabilityf the motor cortex during speech, Neef, Hoang et al. (2015) showed that alterations in excitability could directly impactpeech production. This parallels observations of reduced functional activation in the left motor regions, and might alsoxplain EEG and MEG results in AWS showing abnormal Bereitschaft potential (Walla et al., 2004) or contingent negative


variation (Vanhoutte et al., 2015) prior to normal or stuttered speech (Vanhoutte et al., 2016). No study has yet attemptedto examine differences in either motor thresholds or motor evoked potentials between CWS and CWDS.

Another study by Neef and colleagues explored the idea that stuttering is related to deficits in timing (Neef, Jung et al.,2011). These authors applied disruptive, repetitive TMS to the left and right dorsal premotor cortex of AWS and AWDS andexamined the effectson behavioral performance during a paced finger-tapping task. In AWDS, repetitive TMS of the leftdorsal premotor cortex impaired tap to tone asynchrony in the left but not the right hand. Conversely, in AWS, repetitiveTMS of the right dorsal premotor cortex impaired the tap to tone asynchrony when tapping with the left hand. This suggeststhat timing control in AWDS is mediated by the left hemisphere but that it is shifted to the right hemisphere in AWS. Thesefindings are generally consistent with rightward shifts inmotor and premotor activations in AWS (e.g., Braun et al., 1997;Fox et al., 2000) but also highlight the relevance of possible impaired motor predictions in AWS (see Pollok, Krause, Butz, &Schnitzler, 2009). To our knowledge, Neef, Jung et al. (2011) have published the only study to apply repetitive TMS in AWS.

One recent study examined the effects of tDCS on speech therapy in AWS (Chesters et al., 2017). It was a valiant attemptto use TDCS to investigate the brain basis of stuttering, but the results were not encouraging given the lack of significantdifferences between active and sham stimulation. The results may have been attributable to widespread differences in theseverity of stuttering, the manner of speech therapy, site of stimulation or using standard high definition tDCS systems. Theeffects of varying these parameters could be investigated in future studies.

3.6. Voxel-based morphometry: gray matter

Whereas PET, fMRI, EEG, MEG and TMS are used to investigate the function of the brain, they provide no informationabout the structure of the brain. The structure of the brain is important to examine because differences in structure couldunderlie differences in function. Voxel-based morphometry (VBM) is a fully automated technique to detect group differencesin the volume of gray matter in different parts of the brain. In VBM, the intensity of each brain voxel on the T1 image isused to estimate whether it contains gray matter, white matter, CSF, or combination of them. When used for gray matteranalysis, the estimations of gray matter density (or gray matter volume, GMV) in each vowel (expressed as the probabilitythat the voxel consists of gray matter only) is compared between groups in search of significant differences. To comparebrains across subjects, VBM uses both nonlinear spatial normalization and a smoothing kernel. A possible consequence ofthe former is that a brain with an anomaly will be erroneously wrapped to over-fit a healthy brain; fortunately, VBM solvesthis problem by taking into account the extent of wrapping before performing statistics. A report of VBM results usuallyincludes the location of clusters of affected voxels as inferred from an anatomical atlas. A total of six studies have used VBMto study structural differences between AWS and their fluent peers (Beal, Gracco, Lafaille, & Luc, 2007; Choo et al., 2011;Jäncke, Hanggi, & Steinmetz, 2004; Kell et al., 2009; Kikuchi et al., 2011; Lu, Peng et al., 2010) and a further two have usedVBM in school-aged CWS (Beal, Gracco, Brettschneider, Kroll, & Luc, 2013; Chang et al., 2008). In addition, one unpublishedthesis (Zengin, 2011) and one non-English paper (Song et al., 2007), which are not included in this systematic review, reporton differences in GMV as well.

3.6.1. AWSMost VBM studies of AWS have reported significant group differences in GMV, although these have been uncorrected for

multiple comparisons over the whole brain (usually p < 0.001 uncorrected). Exceptions are the study by Lu, Peng et al. (2010)that reported several corrected results, and the study by Kell et al. (2009) that reported on one region, the left IFG, whereAWS had reduced GMVcompared to controls (p < 0.05 corrected). Although Kell et al. also found a correlation between GMVin IFG and stuttering severity, these IFG effects have not been replicated by other studies.

Despite the localization of group differences in VBM studies varying considerably between studies, two brain regions standout: the right precentral gyrus and the right superior temporal gyrus (STG). In both these regions, multiple publicationshave shown that AWS have higher GMV than controls (Beal et al., 2007; Kikuchi et al., 2011; Lu, Chen et al., 2010; Lu,Peng et al., 2010; Song et al., 2007; Zengin, 2011). It has been asserted that the abnormal structure/ function in the righthemisphere in stuttering is a direct result of abnormal structure and functioning of the left hemisphere (Chang et al., 2011;Choo et al., 2011). For example, deficits in the left STG may result in increased reliance on the right hemisphere homologue(Kikuchi et al., 2011). These right hemispheric differences might be the result of adaptive or maladaptive compensation (seeCivier, Kronfeld-Duenias, Amir, Ezrati-Vinacour, & Ben-Shachar, 2015). Recently, Sowman et al. (2017) revealed significant(corrected) GMV differences in basal ganglia between AWS and AWDS, highlighting the importance of the basal ganglia instuttering.

3.6.2. CWSWhile most group differences in adults are higher GMV in AWS than AWDS, in children the opposite pattern is more

common: reduced GMV in CWS compared to control children. For example, the putamen has less GMV in CWS, perhapsreflecting the fact that it is under-utilized, not utilized efficiently, or did not develop properly in the first place (Beal et al.,
2013; Chang et al., 2008). Nevertheless, some regions do show the same effect in both age groups: the bilateral medial frontalgyrus exhibits decreased GMV (Chang et al., 2008; Lu, Peng et al., 2010) and, as with AWS, the right superior temporal gyrusexhibits increased GMV (Beal et al., 2007, 2013). Interestingly, children who persist at stuttering exhibit more GMV inthe right (and left) STG than those children who have recovered from stuttering (Chang et al., 2008). A baffling difference

bcem

Tsd

3

ieobhii(

3

tcisr

fsLrCb

3

2itm

(wavr

3

viVt

2u2t


etween CWS and AWS concerns the correlations with behavior detected in the IFG: in CWS, stuttering severity is negativelyorrelated with right IFG GMV (Beal et al., 2013) whereas in AWS, thesame behavioral correlation shows up in left IFG (Kellt al., 2009). Such findings raise questions regarding the hypothesis that right hemisphere group differences in stutteringerely reflect plasticity due to left hemisphere impairment.Zengin (2011) provide an approach that may shed light on GMV changes throughout a lifelong struggle with stuttering.

hese authors used VBM to study AWS, CWS and recovered children separately, with all groups being investigated using theame experimental paradigm. Unfortunately, with only 7 children per group in that study, definitive conclusions cannot berawn as yet.

.7. Morphology: gray matter

Although VBM is automatic, it has the drawback that the large number of statistical comparisons (all brain voxels)ncreases the probability of false positives and necessitates correction for multiple comparisons. To detect large scale differ-nces in the size of brain regions, an alternative approach is often applied instead − brain regions are manually segmentedn the T1 image and their size, in terms of voxels, is measured. A total of three MRI studies measured the size of gray matterrain regions in AWS (Foundas, Bollich, Corey, Hurley, & Heilman, 2001; Foundas et al., 2003, 2004) while a further twoave used this method in CWS (Foundas, Cindass, Mock, & Corey, 2013; Mock et al., 2012). One recent similar study on AWS

s not included in the systematic review (Watkins et al., 2008 Watkins et al. this issue). In addition, cortical thickness wasnvestigated by two studies on AWS (Cai et al., 2014; Lu et al., 2012), and one study on a combined group of CWS and AWSBeal et al., 2015). Lastly, one study investigated cortical folding in AWS only (Cykowski et al., 2008).

.7.1. AWSMost of the morphological studies were conducted by Foundas and colleagues (Foundas et al., 2001, 2003, 2004), with

he main finding being atypical laterality − AWS have larger structures on the right hemisphere relative to the left whenompared with controls. The left planum temporale, specifically, is larger than the right planum temporale in AWDS but notn AWS where the two regions are generally the same size (Foundas et al., 2001, 2004). It is of note though that a recenttudy, which utilized Foundas and colleagues’ methodology on a larger sample of PWS, could not replicate their results withelation to the planum temporale (Watkins et al., this issue).

A well-known characteristic of gray matter is that it changes in response to experience. Repeated use of brain regionsor a given task often results in expansion of that area whereas under-utilizing a brain region can lead to a decrease in itsize. For example, musicians have increased hand motor representation in the brain (see for review Herholz & Zatorre, 2012;appe, Herholz, Trainor, & Pantev, 2008). For this reason, it has been suggested that the atypical bias of gray matter to theight is likely a result of increased right hemisphere utilization, presumably in an attempt to compensate for stuttering (e.g.,ivier et al., 2015). Such inter-hemispheric reorganization might also account for the increased number of sulci and gyralanks detected in right perisylvian regions in AWS (Cykowski et al., 2008).

.7.2. CWSStudies on gray matter morphology in CWS have mostly replicated the findings in AWS (Foundas et al., 2013; Mock et al.,

012) suggesting that, if such reorganization does indeed happen, it is occurring soon after disorder onset. It is possible thatn stuttering, gray matter develops differently to begin with, but this hypothesis would be difficult to test empirically givenhe disorder’s early-childhood onset. Such efforts may require a study of children at risk of stuttering and collecting various

easures of brain structure throughout childhood.A recent study reported that the left IFG is the only brain region that exhibits an abnormal developmental trajectory

absent cortical thickening) over a lifetime of stuttering, from childhood to adulthood (Beal et al., 2015). This is consistentith the aforementioned finding of atypical GMV in left IFG in AWS (Kell et al., 2009). In that study, stuttering was also

ssociated with elevated fractional anisotropy values in left IFG (i.e., a measure of water diffusivity along axons, with atypicalalues, either too high or too low, indicative of microstructural anomaly in the white matter), with adult-onset unassistedecovery appearing to normalize them (Kell et al., 2009).

.8. Voxel-based morphometry: white matter

The VBM analysis of white matter is identical to that of gray matter other that it works on the estimates of white matterolume (WMV) in each voxel rather than gray matter volume. VBM being limited to detecting macrostructural differencesn white matter is inferior to diffusion MRI that can also detect microstructural differences. Moreover, the T1 scans used forBM does not give information on fiber directionality which is required for tractography. That said, T1 scans are short in

ime, and sometimes preferred for studying children (e.g., Beal et al., 2013).Past VBM studies of white matter include a total of four studies on AWS (Beal et al., 2007; Choo et al., 2011; Jäncke et al.,

004; Lu, Peng et al., 2010) and two more on school-aged CWS (Beal et al., 2013; Choo et al., 2012). Two of the above studiessed an identical experimental paradigm to permit fair, though descriptive, comparison between CWS and AWS (Choo et al.,011; Choo et al., 2012). These two studies also performed standard morphological analysis. In addition, one unpublishedhesis (Zengin, 2011) which is not included in the systematic review reports on differences in WMV as well.


3.8.1. AWSChoo et al. (2011) detected increased WMV in the anterior corpus callosum in AWS. In the same sample, they also showed

that the corpus callosum was larger on the average in AWS (as measured on the midsagittal plane). Jäncke et al. (2004)performed a whole-brain VBM analysis, as well as an analysis limited to predefined regions of interest. They identified aleftward bias in WMV of the auditory cortex in AWDS but not in AWS. Instead, AWS had increased WMV in the right superiortemporal gyrus, including planum temporale. Jäncke et al. suggested that their findings relate to atypical intra-hemisphericcommunication. Additional right hemisphere regions with increased WMV in AWS were reported by both Jäncke et al. (2004)and Lu, Peng et al. (2010); however, there is virtually no overlap between the studies in terms of the exact regions affected.

3.8.2. CWSThe Beal et al. (2013) study is the only published whole-brain white matter VBM investigation in CWS (but see an

unpublished thesis by Zengin, 2011). Its positive findings, which include group differences in the corpus callosum, shouldawait replication, as one paper that targeted callosal white matter in CWS reported null results (Choo et al., 2012). An earlierstudy by the same group did detect callosal white matter anomalies in AWS (Choo et al., 2011), which led the authors toconclude that callosal anomaly probably appears later in life in stuttering.

3.9. Diffusion MRI − voxelwise analysis

Diffusion MRI (dMRI) measures relate to underlying white matter microstructure. Although a certain biological inter-pretation of detected anomalies is usually not possible, the ability to detect anomalies and localize them to certain tractsadvances our knowledge considerably. dMRI studies can be divided into those using whole-brain voxelwise analysis, andthose employing tractography.

The major limitation of voxelwise dMRI studies is in aligning white matter structures between participants for groupanalysis. Even non-linear alignment techniques, usually used in gray matter studies, perform poorly in aligning the thinand elongated white matter structures, and statistical analysis requires substantial smoothing. The earliest dMRI study ofstuttering did rely on non-linear alignment and smoothing (Sommer, Koch, Paulus, Weiller, & Büchel, 2002) and identifiedreduced fractional anisotropy (FA) in a region in proximity to the speech motor representations of the tongue, larynx andpharynx as well as the arcuate fascicle. This result has been replicated by a number of other studies (see Neef, Anwander et al.,2015, for review). Most subsequent dMRI studies of stuttering employed the newer tract-based spatial statistics (TBSS; Smithet al., 2006) fully-automated technique that aligns lines or sheets (the centers of tracts) rather than blobs (the whole tracts)thus performing much better than non-linear alignment approaches. Not only that smoothing is not necessary anymore, butalso the FA measures extracted from the center of tracts are less prone to partial-voluming effects. To detect the centers oftracts (i.e., the white matter skeleton), TBSS searches the FA maps for elongated regional peaks, exploiting the fact that FAis modulated by fiber density which is higher at tract center. Despite its name, tract-based spatial statistics does not permitexamining of microstructural properties along specific tracts; instead, microstructural alterations are examined across thewhole white matter skeleton, and it is the scientist’s task to identify the affected tracts, usually using anatomical atlases.

A total of four dMRI studies on AWS have utilized TBSS (Cai et al., 2014; Civier et al., 2015; Cykowski et al., 2010; Kellet al., 2009). Kell et al. also studied adults with late-onset recovery. Additionally, two studies have focused on school-agedand young CWS (Chang et al., 2008, 2015, respectively), with the earlier study also examining recovered children. The latterstudy is particularly notable for its use of large samples of CWS. A further two studies have used mixed groups of AWS andadolescent CWS (Connally et al., 2014; Watkins et al., 2008). Because TBSS cannot properly align certain white matter tracts,two studies also measured average FA in specific ROIs in subject space (Cai et al., 2014; Connally et al., 2014). One studyinvestigated both developmetal and acquired stuttering, and therefore is not included in the systematic review (Chang et al.,2010).

3.9.1. AWSMost group differences detected in TBSS studies of stuttering indicate reduced FA in the left hemisphere (see Cai et al.,

2014; Cykowski et al., 2010). Indeed, in a recent meta-analysis, two of the three resulting clusters are left hemispheric (Neef,Anwander et al., 2015). Rather than consulting an atlas for tract identification, the authors used the clusters as seed regionsfor tractography, and identified the left superior longitudinal and arcuate fasciculus, and the callosal fibers connectingbilateral Rolandic regions. However, because most points in the brain are traversed by at least two fiber tracts crossing eachother (Jeurissen, Leemans, Tournier, Jones, & Sijbers, 2013), identification of affected pathways cannot always be certain.For example, Kronfeld-Duenias, Amir, Ezrati-Vinacour, Civier and Ben-Shachar (2016a) demonstrated that there are severalpathways passing through a cluster of voxels in the left Rolandic operculum, which is a region that came up over and overin diffusion MRI studies of stuttering (see Cykowski et al., 2010). For this reason, investigation of longitudinal tracts is betterconducted using tractography.

One of the clusters identified by Neef, Anwander et al. (2015), Neef, Hoang et al. (2015) is in the posterior mid body of
the corpus callosum, a region which is only traversed by inter-hemispheric fibers, thus permitting certain identification ofthe affected tract. Callosal anomaly in stuttering might suggest an inter-hemispheric reorganization, possibly leading to theaforementioned group differences in right-hemispheric gray matter morphology and volume, and to alterations in whitematter tracts such as the right superior longitudinal fasciculus (Cai et al., 2014; Kronfeld-Duenias et al., 2016a).

ftcBBtocthi

2cfirmaAoF

3

peWgasm

wraasnae

bsadc

3

dasafmvKo


Interestingly, in TBSS studies of stuttering, the corpus callosum is also where most significant group differences correctedor multiple comparisons show up (Civier et al., 2015; Cykowski et al., 2010; Kell et al., 2009). However, it is not yet clear whyhese TBSS corrected effects are in the forceps minor (also see Cai et al., 2014; for uncorrected effects), a part of the corpusallosum which is more anterior than the callosal section identified by Neef, Anwander et al. (2015), Neef, Hoang et al. (2015).ecause forceps minor anomalies were detected in adults but not in children who stutter (Choo et al., 2011, 2012; but seeeal et al., 2013), it is possible that late plasticity leads to the observed structural and micro-structural changes. The findinghat forceps minor FA reductions in AWS are correlated with decreased fluency suggests that such plasticity is maladaptiver epiphenomenal (Civier et al., 2015). If the affected forceps minor fibers are left-to-right projections inhibiting right frontalortex (see Civier et al., 2015), this would also explain why the activity of this region is associated with stuttering ratherhan with induced fluency (Budde et al., 2014). Regarding biological interpretation, the forceps minor anomaly is driven byigh radial diffusivity (i.e., the diffusion of water perpendicular to the main direction of axon bundles), which might suggest

mpaired myelin (Cykowski et al., 2010; but also see Civier et al., 2015).Voxelwise analysis was performed without TBSS in three studies (Cai et al., 2014; Connally et al., 2014; Sommer et al.,

002). Cai et al. performed an analysis in subject space, calculating average FA in the white matter regions directly beneathortical regions that were automatically defined. This may be most useful for detecting anomalies in local U-fiber (i.e.,bers that connect adjacent cortical gyri; Schmahmann & Pandya, 2009). They found three left hemisphere speech networkegions with lower average FA in AWS compared to AWDS (p < 0.05, not corrected): dorsal inferior frontal operculum, middleotor cortex and posterior dorsal superior temporal sulcus. Connally et al. defined cerebellar ROIs using manual and semi-

utomatic methods, and similarly, compared their average FA between groups. For all six ROIs in the cerebellar peduncles,WS had lower FA than their fluent peers. For ROIs that also included gray matter, white matter voxels were isolated basedn an automatic white/ gray matter segmentation performed on the T1 scan. Nevertheless, no group differences in averageA were detected in these mixed tissue regions.

.9.2. CWSOne region that showed decreased FA only in studies on CWS (or studies that combined CWS and AWS) is the posterior

art of the left superior longitudinal fasciculus (SLF), adjacent to the posterior division of the supramarginal gyrus (Changt al., 2008; Chang, Zhu, Choo, & Angstadt, 2015; Connally et al., 2014; Neef, Anwander et al., 2015; Neef, Hoang et al., 2015;atkins et al., 2008). In the single AWS study, where a close-by region was implicated, FA was actually higher in the patient

roup (Kell et al., 2009). Although FA values in the posterior left SLF were never directly contrasted between CWS and AWS,n adjacent posterior callosal region did demonstrate large age-related FA differences (Civier et al., 2015). Future studieshould investigate whether FA values in these left posterior regions indeed change as the disorder develops, and what is theechanism that drives plasticity.Chang et al. (2015) also showed that young CWS exhibit increased FA in the cerebellum, which Chang et al. noted

as associated with the organization of sequential movements into chunks (e.g., Sakai, Hikosaka, & Nakamura, 2004). Theeason why such differences are evident as early as 3 years is unclear, but may suggest that some compensatory processesre occurring very early on. It is important to point out that many of these structural abnormalities observed in CWS aressociated with lower performance on arange of tasks, relative to CWDS (Choo, Burnham, Hicks, & Chang, 2016). There areome interesting structural differences between boys and girls who stutter. In particular, Chang et al. (2015) reported aegative correlation between FA in left IFG, cingulate and supramarginal gyrus for CWS, but this is only present in the boysnd not the girls who stutter. This may have significant implications for stuttering that more girls than boys recover (Changt al., this issue; see also Beal et al., 2015; Kell et al., 2009).

From a theoretical standpoint, the observation, of common abnormalities between AWS and CWS, is important to considerecause it could point to a biomarker of stuttering. For example, both AWS and CWS exhibit abnormal white matter micro-tructure in the left superior temporal gyrus (Chang et al., 2015; Lu, Peng et al., 2010), which might be related to theforementioned gray matter anomalies in bilateral auditory cortices in stuttering. However, because CWS and AWS haveifferent brain sizes, such commonalities are usually difficult to demonstrate using TBSS and VBM, which produce MNIoordinates as output. Tractography might be a better approach for this task.

.10. Diffusion MRI − tractography

In tractography, pathways of interest made of streamlines, are reconstructed from dMRI data, based on the diffusionirectionality and strength estimated for each voxel. Tractography can benefit from advanced acquisition methods, suchs high angular resolution diffusion imaging (HARDI), that permit estimating several diffusion directions per voxel − usingpecial tracking algorithms, crossing fibers may be accounted for (Farquharson et al., 2013). In principle, tractography-basednalysis can be performed entirely in subject space, with no need for inter-subject alignment, but see Chang and Zhu (2013)or an exception. Given the reconstructed pathways, one can perform morphological measurement such as tract volume, and

icro-structural measurement such as FA. The latter measurements can be performed by averaging properties across all tractoxels, or using more sophisticated methods that divide the tract into small sections which are then analyzed separately (e.g.,ronfeld-Duenias et al., 2016a). In probabilistic tractography, connectivity strength is defined as the number of streamlinesriginating in seed and reaching (or terminating in) target region, and it may represent either macro- or micro-structural


white matter properties. For a more detailed review of tractography studies in stuttering, please refer to Kronfeld-Dueniaset al. (this issue).

There have been three tract-of-interest tractography studies with AWS (Chang et al., 2011; Cieslak et al., 2015; Connallyet al., 2014) and only 1 on CWS (Chang & Zhu, 2013). Other two such studies not included in the systematic review areKronfeld-Duenias et al. (2016a) and Kronfeld-Duenias, Amir, Ezrati-Vinacour, Civier, & Ben-Shachar (2016b). However, itis also possible to investigate all major pathways of the brain without being constrained to tracts-of-interest dictated byprevious hypotheses. The challenge of such an approach is the large number of comparisons involved, and the fact thatmultiple-comparison correction methods based on the spatial extent (e.g., the method used in TBSS) are not applicable. Onlyone study utilized this network-level methodology to study stuttering, examining AWS (Cai et al., 2014). A follow up studyby the same group (Sitek et al., 2016) is not included in the systematic review.

3.10.1. AWSBecause TBSS studies often localized FA reductions in stuttering to the left dorsal language tracts − the left arcuate

fasciculus and the superior longitudinal fasciculus (SLF) (Chang et al., 2008; Civier et al., 2015; Cykowski et al., 2010; Kellet al., 2009; Watkins et al., 2008) − tractography studies tried to verify this finding by reconstructing these tracts, and directlymeasuring their mean FA (weighted or unweighted average of FA across all tract voxels). Surprisingly, Kronfeld-Duenias et al.(2016a) could not detect group difference in FA between AWS and AWDS in these tracts, and Connally et al. (2014) detecteda small group difference in FA only when left and right arcuate fasciculi were considered together. One explanation for thediscrepancy between TBSS and tractography studies is that the underlying anomalies are actually morphological, and TBSSwrongly identifies them as FA differences (see Kronfeld-Duenais et al., 2016a). Indeed, when the volume (number of voxels)of these left longitudinal tracts was examined, large group differences were detected, with AWS showing smaller volumescompared to AWDS (Kronfeld-Duenias et al., 2016a, 2016b; but see Connally et al., 2014, for null results). Cieslak et al.(2015), in particular, revealed that in AWS, the left arcuate exhibited reduced volume in specific tract sections. This is alsoconsistent with the report that connectivity strength from posterior superior temporal gyrus (pSTG) to IFG, the presumedendpoints of the arcuate fasciculus, is much weaker in AWS than in AWDS (Chang et al., 2011). It is of note, though, thatChang et al. could not detect parallel differences in functional connectivity in their study. Reduced left arcuate fasciculusvolume might compromise the pathway ability to carry interactions between motor and auditory regions, possibly affectingthe encoding of forward models important for speech production (Max et al., 2004). It is also possible that these motor-to-auditory models are intact, but their predictions cannot be read out from IFG to pSTG (Chang & Zhu, 2013), where thecomparison with speech’s auditory feedback is hypothesized to take place (Golfinopoulos et al., 2010).

The left dorsal language tracts are crossed by many other pathways smaller in size, and it is also possible that pastvoxelwise analyses confused an anomaly in one of these minor tracts with arcuate or SLF anomaly. One candidate pathwayis the left corticospinal tract (CST) whose inferior section’s volume is reduced, and mean diffusivity (MD) is increased, inPWS compared to PWDS (Connally et al., 2014; Kronfeld-Duenias et al., 2016a, 2016b). However, that these differences weredetected in inferior rather than superior CST makes confusion with the dorsal language tracts less likely. Smaller or impairedleft CST is consistent with findings of higher motor threshold for left hemisphere in AWS (see Alm et al., 2013), suggestingleft hemisphere motor impairment in stuttering.

Kronfeld-Duenias et al. (2016b) found that the anatomical connection between the SMA and the IFG − the frontal aslanttract (see Catani et al., 2013) − is abnormal in AWS as evidenced by increased levels of mean diffusivity (MD). That therewere differences in the level of MD not accompanied by differences in FA implies there are fewer constraints on diffusionthat are not specific to any direction. Interestingly, MD in this tract was correlated with decreased speech rate in AWS. Thistract has an important role in speech production as evidenced by the fact that electrically stimulating it leads to speecharrest (e.g., Vassal, Boutet, Lemaire, & Nuti, 2014) in much the same way that electrical stimulation or TMS of Broca’s areadoes (Devlin & Watkins, 2007). More interestingly, Kemerdere et al. (2016) demonstrated electrical stimulation of this tractinduces stuttering. The frontal aslant tract therefore is functionally relevant for speech, though little is understood about itssignificance in the etiology of stuttering.

It is important to note that white matter differences in any disorder, stuttering included, are likely not confined to a sin-gle tract but are rather widespread and involve multiple interconnected tracts or even multiple networks. Cai et al. (2014)reported the first investigation into network-level structural connectivity in AWS. To limit the number of comparisons, thenetwork was constrained to speech-related regions, but even then, most results did not survive correction for multiple com-parisons. Nevertheless, Cai et al. could identify a trend for AWS to show reduced connectivity of left hemisphere structures,primarily pre-Rolandic regions, consistent with many of the reports discussed above.

3.10.2. CWSChang and Zhu (2013) studied the basal ganglia thalamo-cortical circuit as well as the auditory-motor network in CWS.

They reported that CWS exhibit weaker connections than CWDS between the left putamen and several cortical regions: theleft insula, the left inferior and middle frontal gyri, and the SMA; findings that are consistent with the Beal et al. (2013)
report on reduced GMV in the left putamen in CWS. Chang and Zhu also found reduced connectivity in CWS from left IFGto pSTG, similar to their finding in AWS (Chang et al., 2011; and see also Chang et al., 2015). Interestingly, however, theaffected pathway in CWS does not seem to be the arcuate fasciculus but rather another pathway that interconnects thesetwo regions via a more inferior route (Chang & Zhu, 2013). This is important information, which is not usually available in

nti

4

flooi(ssranKs

spsitrocrb

(itCet2eer

5

assd

sSeasyila2g


etwork-level connectivity studies, such as Cai et al. (2014). Because the IFG-pSTG connections are the only ones reportedo be abnormal both in adults and children who stutter, there should be a concentrated effort to verify whether the samenferior tract is also the one affected in AWS.

. Summary

Overall, it appears that there are widespread functional and structural brain differences between AWS, CWS and theiruent peers. However, there is only limited consistency in the description of where or what these differences are. Method-logically, the majority of studies to date have used functional or structural MRI; a similar number have used EEG whilstther methodologies such as TMS, TES, NIRS and MEG are largely under-represented in stuttering research. Many neuroimag-ng studies in stuttering have been conducted using speech production (primarily fMRI and PET) and auditory perceptionprimarily EEG and MEG), and relatively few have focused on resting state activity. The functional neuroimaging literatureeems to be coming full circle. Early PET neuroimaging studies focused on resting state and differences between fluent andtuttered speech. Research then shifted somewhat to examining neural activation during speech and non-speech tasks. Cur-ently, the focus is shifting back to resting state and the effects of fluency inducing conditions, albeit with higher resolutionnd more sophisticated analyses. Structural neuroimaging, on the other hand, is still far from realizing its potential, andew approaches may yet detect novel anomalies in stuttering (e.g., Kronfeld-Duenias et al., 2016a, 2016b; Cai et al., 2014;ronfeld-Duenias et al., this issue). Below we highlight consistent themes in the literature following from the results of theystematic review. We then detail a series of recommendations and provide directions for future research.

One consistent theme in the TMS and M/ EEG literature is that AWS exhibit reduced excitability in motor areas prior topeech production. Notably this is also consistent with fMRI studies showing anomalies in the planning stages of speechroduction. Anomalies in neural activity before speech could lead to disfluencies during speech production. For example,ome studies directly link alack of increased excitability to moments of stuttering (For possible compensatory mechanismsn AWS that involve increased excitability see Mersov et al., this issue; Vanhoutte et al., 2016; and Section 3.3.1). Althoughhe number of stuttered syllables is small,these provide invaluable insights into stuttering. Examining preparatory speech-elated activity in young CWS, who presumably have not yet developed compensatory mechanisms, will shed further lightn abnormal motor development in stuttering. Collectively, the observation of increased motor thresholds and decreasedortical excitability suggests that the difficulty AWS have in initiating movements stems from reduced excitability in motoregions of the brain. These differences in cortical excitability are also particularly interesting in light of an inverse relationshipetween fractional anisotropy and motor threshold in AWDS (Klöppel et al., 2008).

A second consistent theme with respect to AWS is that they exhibit atypical activation in the left inferior frontal gyruse.g., Neef et al., 2016) and right auditory regions (Belyk et al., 2015). They also show structural abnormalities in the leftnferior frontal gyrus as measured by gray matter volume (Beal et al., 2015; Kell et al., 2009), and in the left IFG connectionso the SMA (Kronfeld-Duenias, Amir, Ezrati-Vinacour, Civier, & Ben-Shachar, 2016b), to temporal regions (Chang et al., 2011;ieslak et al., 2015; Connally et al., 2014; Kronfeld-Duenias et al., 2016a) and to the right IFG (Kell et al., 2009; Cykowskit al., 2010; Choo et al., 2011; Cai et al., 2014; Civier et al., 2015). Additionally, AWS have over-developed right superioremporal gyrus as indexed by higher gray matter volume in that region (Beal et al., 2007; Song et al., 2007; Kikuchi et al.,011; Sowman et al., 2017), as well as symmetric gray matter volume (Foundas et al., 2001) and white matter volume (Jäncket al., 2004) in auditory regions instead of the typical leftward bias (see for review Neef, Anwander et al., 2015; Neef, Hoangt al., 2015). To summarize, it seems that the left IFG and right auditory regions play an important role in the disorder, aseflected by functional, gray matter and white matter anomalies.

. Thoughts for future research

Research seems to be trending toward functional and structural connectivity across multiple regions rather than functionnd structure of individual regions. Currently, analysis methods are somewhat limited. For example, there are very fewtudies examining causal influences of one brain region on another. Likewise, while there are a growing number of DTItudies, many have difficulty resolving the crossing fibers in the brain. We suggest that dynamic causal modeling of functionalata and constrained spherical deconvolution of diffusion MRI can be used to address these gaps in the literature.

There are still other gaps in the literature, most notably the lack of longitudinal data in CWS. It should also be noted thattudies on CWS are becoming widespread, with several recent studies examining children aged 6–12 (Beal et al., 2011, 2013;ato et al., 2011) and others examining children aged 3–6 (Arnold et al., 2011; Etchell, Ryan et al., 2016; Etchell, Johnsont al., 2016; Sowman et al., 2014; Weber-Fox et al., 2013). Interestingly, the results of the former studies raise questionsbout the assumption that little or no neural reorganization takes place in the childhood years of PWS. Many studies ontuttering use the term children to refer to participants that would be more appropriately described as adolescents or evenoung adults. For example, in the Desai et al. (2016) study, children were defined as being under the age of 20. In the future,t will be necessary to make clearer distinctions in the literature to more easily determine whether neural reorganization is
ikely to have occurred. This will also allow a more nuanced perspective on how stuttering affects development at differentges. Longitudinal studies or direct comparisons of adults and children are rare (though see Beal et al., 2015; Choo et al.,012; Desai et al., 2016; O’Neill et al., 2017; and see also Choo et al., 2011), but will be necessary to elucidate which regionso through plastic changes that adaptively or maladaptively compensate for stuttering, and which have anomalies that are


part of the etiology of the disorder. Such work will also allow us to develop more definitive biomarkers for the likelihood ofpersistence or recovery from stuttering.

Accordingly, there is likely to be a proliferation in the use of less invasive methods like near infra-red spectroscopy (Satoet al., 2011) and more recently proton magnetic resonance spectroscopy (O’Neill et al., 2017) to image the brains of youngerchildren who stutter. Given sufficient time and money, it would be particularly interesting to focus on the neural markersthat contribute to a risk of stuttering, and recovery and persistence of stuttering as compared to controls (see Chang et al.,2008, 2015; and Chang et al., this issue). Encouragingly, such work is already underway in some labs. Notably, there has beenvery little investigation of sex differences in stuttering which is strange considering the high degree of incidence in malesas compared to females who stutter. Along asimilar line of reasoning, most previous work has focused on the perception orproduction of speech. There has been very little focus on attentional components of the disorder or on neural oscillations.This is unfortunate, as the few efforts in these directions did yield some interesting results that warrant further investigation(Etchell, Ryan et al., 2016; Etchell, Johnson et al., 2016; Kikuchi et al., 2016; Mersov et al., 2016; Mersov et al., this issue;Sengupta et al., 2016).

Future research should focus on using non-invasive brain stimulation. One line of research is to examine whethernon-invasive brain stimulation is an effective tool for augmenting therapy and whether or not such stimulation createsmeasureable structural and functional changes in the brain (e.g., Chesters et al., 2017). A second line of research is to tryand reproduce transient symptoms of stuttering in control participants, in order to elucidate the underlying neural mech-anisms of the disorder. Since there are many potential deficits that could lead to stuttering, artificial stuttering induced inthis manner might not reflect natural stuttering. Nevertheless, such artificial stuttering could still provide valuable insightsinto the disorder. This idea could then be combined with other forms of neuroimaging to examine precisely what hap-pens in the brain when stuttering is induced. Alternatively, researchers may even attempt to induce stuttering in AWS viabrain stimulation to establish a causal role of areas involved in stuttering. Along a similar line of reasoning, with improvedneuroimaging technologies, especially those concerned with handling head-movement artifacts, there may well be a trendtoward examining the neural correlates differentiating fluent and stuttered speech (e.g., Mersov et al., this issue; Sowmanet al., 2012; Vanhoutte et al., 2016) and how it is altered with fluency inducing mechanisms.

There is an extensive neuroimaging literature on stuttering. However, a clear and consistent picture of the brain basis ofstuttering is far from complete. While meta-analytic studies highlight the robust findings of anomalous brain structure (Neef,Anwander et al., 2015; Neef, Hoang et al., 2015), and function (Belyk et al., 2015; Brown et al., 2005; Budde, 2015) in AWS, thefunctional significance of these results is far from clear. Further, recent work has pointed to errors in the software commonlyfor ALE meta-analysis (Eickhoff, Laird, Fox, Lancaster, & Fox, 2017) thereby casting doubt on the results of the aforementionedstudies. Moreover, meta analytic studies, though comprehensive and useful, likely present an oversimplified picture ofthe brain basis of stuttering. They may miss areas that are not reported across multiple studies that may nevertheless beimportant. It is noteworthy that the meta-analyses to date have largely neglected M/ EEG data (the different data output fromsuch studies makes their integration with the MRI studies difficult) thereby omitting a large proportion of the literature.However, making sense of the data from a narrative perspective is perhaps an even more difficult task; one made morecomplex by the disparate theoretical frameworks which drive, and inherently bias the analytic direction of many of theinvestigations to date.

One means by which to clarify discrepant findings is multi-center studies. Multi-center studies would increase collabo-rative efforts and could produce higher quality studies. Rather than competing and investigating the same lines of research,institutions could cooperate and attempt to employ the identical experimental paradigms, scanning parameters and analy-ses pipelines to concurrently verify and independently replicate findings. The main benefit of such an enterprise is twofold.Firstly, it allows researchers to determine the reproducibility of results and pool data into much larger sample sizes, whichwould enable greater confidence in results and resolve discrepant findings. While this suggestion is valid for many areas ofneuroscience, the problem is more acute in stuttering research since many centers have very different ways of measuringstuttering severity. It would be advisable for different labs to attempt to standardize such clinical measures across studies.Secondly, multi-center collaborations would, due the involvement of different groups, facilitate researchers being morediverse in their thinking instead of designing and interpreting studies only within their own conceptual frameworks.

Stuttering is only one of many speech disorders, and must be understood in a wider context. Future work may wish toform hypotheses about how structural and functional abnormalities differentiate and cause one speech disorder rather thanthe other. Computational modeling approaches can be particularly valuable here by showing that a common frameworkfor speech production can generate different disorders, depending on the impairment introduced to the model (e.g., Civieret al., 2010; Terband et al., 2009). Since stuttering differs from many other speech disorders in that it can be temporarilyalleviated – even during a single experimental session – it is the ideal place to start a revolution in our understanding ofboth normal and disordered speech.

Acknowledgements

Ballard is supported by Australian Research Council Future Fellowship FT120100355. Sowman is supportedby the National Health and Medical Research Council, Australia (#1003760), the Australian Research Council(DE130100868 & DP170103148) and the Australian Research Council Centre of Excellence for Cognition and its Disorders(http://www.ccd.edu.au); CE110001021.

http://www.ccd.edu.au

A.C.

Etchell et

al. /

Journal of

Fluency D

isorders 55

(2018) 6–45

29

Appendix A.

Table 1PET studies of adults who stutter (18+).

Study Method Sample Task Main finding

Wu et al. (1995) PET 3M 1F AWDS (22–41)3M 1F AWS (20–54)

Reading task Caudate was 50% less active in AWS than duringstuttering and fluency

Fox et al. (1996) PET 10M AWDS (21–55)10M AWS (21–46)

Chorus vs. Solo reading AWS had overactivation in motor areas that wasreduced by choral reading

Ingham et al. (1996) PET 19M AWDS (21–55)10M AWS (22–46)

Resting state AWS and AWDS did not differ in cerebral blood flow atrest

Braun et al. (1997) PET 12M 8F AWDS (23–50)10M 8F AWS (23–51)

Fluency vs. Dysfluency Activation of LH and RH was associated with stutteredand fluent speech, respectively

Wu et al. (1997) PET 6M AWDS Resting state AWS had higher uptake of FDOPA than AWDSthroughout cortex3M AWS

De Nil et al. (2000) PET 10M AWDS (20–25) Silent vs. Oral word reading AWS showed greater RH activation and AWDS showedgreater LH activation when comparing oral vs. silentreading.

10M AWS (24–44)

Fox et al. (2000) PET 10M AWDS (32) Solo and choral reading vs. rest. Brain correlates of stuttering was non dominant lefthemisphere10M AWS (32)

Ingham et al. (2000) PET 4M AWDS (28–50) Solo vs. choral paragraph reading vs. Rest Imagined stuttering elicited brain activation similar tothat associated with overt stuttering4M AWS (30–46)

De Nil et al. (2001) PET 10M AWDS (20–45) Silent and oral reading, verb generation vs. Passiveviewing

Fluency Treatment increased cerebellar activation13M AWS (20–45)

De Nil et al. (2004) PET 10M AWDS (19–34) Oral and silent reading vs. viewing string of X’sbefore and after fluency treatment

.Prior to treatment AWS had right lateralized activity.Following treatment this activation became leftlateralized

13M AWS (20–40)

Ingham et al. (2004) PET 10F AWDS (42) Solo, choral reading and rest Female AWS showed bilateral neural activation that wasassociated with stuttering10F AWS (44)

Stager et al. (2004) PET 9M 8F AWDS (23–50) Same as Braun et al. (1997) Increased activation in auditory areas in AWS in fluencyinducing conditions enabled more efficient use ofauditory information

10M 7F AWS (23–50)

Ingham et al. (2012) PET 12 AWDS (20–65) Eyes closed rest, oral reading aloud (reading textaloud), monologs (continuous self formulatedspeech)

AWS and AWDS showed increased basal ganglia bloodflow to the basal ganglia during rest, but decreasedblood flow to the basal ganglia during speech. Stutteringfrequency was associated with activity in thecortico-striatal-thalamic network

18 AWS (20–67)

Ingham et al. (2013) PET 8 AWDS (20–64) PET scan before and after treatment. During thePET scan, SSX performed oral reading, monologand eyes closed rest (baseline)

Decreased regional cortical blood flow in the leftputamen was predictive of successfully completing thetreatment program

17M 5F AWS (20–64)

PET = Positron Emission Tomography; AWS = adults who stutter; AWDS = adults who do not stutter; CWS = children who stutter (<18); CWDS = children who do not stutter (<18); PWS = persons who stutter (mixedage); PWDS = persons who do not stutter (mixed age); Sample related acronyms preceded by ‘r’ denote “recovered” and those preceded by ‘p’ denote “persistent”; * denotes that the study belongs to a pair ofstudies that used the same paradigm, one of them to examine adults, and the other to examine children; + denotes that the study used multiple imaging methodologies.

30

A.C.

Etchell et

al. /

Journal of

Fluency D

isorders 55

(2018) 6–45

Table 2(A) fMRI studies of adults who stutter (18+). (B) t fMRI studies of children who stutter (<18) and persons who stutter (mixed-age).

(A)


Preibisch et al. (2003) fMRI 16M AWDS (19–51) Overt reading vs. Viewing sentences AWS showed increased activation in the rightfrontal gyrus not present in AWDS and waspositively correlated with stuttering

16M AWS (18–48)

Van Borsel et al. (2003) fMRI 6M AWDS (19–37) Overt and covert reading of meaningful andnonsense paragraphs

AWS did not show activation across any conditions6M AWS (19–37)

Blomgren et al. (2004) fMRI 7M 2F AWDS (20–40) Listen to descriptions of verbs or nouns Responses were bilateral in AWS but leftlateralized in AWDS5M 2F AWS (19–38)

Foundas et al. (2004) fMRI 11M 3F AWDS (19–47) Reading under delayed and non altered auditoryfeedback

Delayed auditory feedback was associated withatypically increased activation in right planumtemporale

11M 3F AWS (22–47)

Neumann et al. (2004) fMRI 16 AWDS (19–51) Overt reading and semantic decision before duringand after therapy

Brain activation was right lateralized beforetherapy, left lateralized after and tended to be rightlateralised at followup

5M AWS (26–41)

Neumann et al. (2005) fMRI 9M AWS (24–41) Sentence reading Therapy increased activation in the left rolandicoperculum AWS

De Nil et al. (2008) fMRI 15 AWDS (24–48) Listen, vocalize, simulate stuttering and prolong aword

AWS showed significantly more activation inbilateral cortical regions during normal speech.AWS showed significantly more activation in theright inferior frontal gyrus than AWDS whensimulating stuttering

15 AWS (21-47)

Giraud et al. (2008) fMRI 16 AWS (18–48) Reading sentences vs. Silent viewing ofmeaningless letter like signs

Activity in basal ganglia (caudate) positivelycorrelated with stuttering severity before speechtherapy but not after (except for a small cluster)

Chang et al. (2009) fMRI 9M 11F AWDS (36.35) Perceive, plan and produce speech AWS exhibited reduced activation of motor regionsduring perception and greater activation in motorand auditory regions during production.Differences were similar for speech and nonspeech stimuli

11M 9F AWS (36.35)

Kell et al. (2009)+ fMRI 13M AWDS (23–44) Overt sentence reading vs. covert sentence reading(baseline) before and after therapy

Persistent stuttering was associated withmobilization of regions in right hemisphereOptimal repair associated with engagement of leftBA47/12

13M pAWS (18–39)13M rAWS (16–65)

Lu et al. (2009) fMRI (SEM) 5M 4F AWDS (22–29) Covert picture naming vs. passive viewing(baseline)

AWS exhibited large scale dysfunctional neuralinteractions across widely distributed brainregions, particularly those involved in auditory andmotor processing

9M 1F AWS (20–29)

Sakai et al. (2009) fMRI 3M 7F AWDS (24.1) Sentence reading during delayed and normalauditory feedback vs. Passive viewing of fixationcross

AWS showed greater activation of the right inferiorfrontal gyrus during normal speech. AWS had lessactivation of the right SMA and STG in bothconditions

3M 5F AWS (20–53)

A.C.

Etchell et

al. /

Journal of

Fluency D

isorders 55

(2018) 6–45

31

Lu, Peng et al. (2010)+ fMRI (SEM) 8M 4F AWDS (22-29) Picture naming task vs. Passive viewing task(baseline); baseline task prior to the naming task

AWS had weaker connections throughout BGTCnetwork. AWS had greater activation in the rightinferior frontal gyrus and insula

10M 2F AWS (19-31)

Lu, Chen et al. (2010) fMRI (SEM) 7M 5F AWDS (22–29) Picture naming a one or three syllable word orrepeating the one syllable word 3x

AWS exhibited atypical planning in the bilateralinferior frontal gyrus and right putamen. AWSexhibited atypical production in the rightcerebellum, insula and left premotor area

10M 2F AWS (20–29)

Chang et al. (2011)+ fMRI 12M 11F AWDS (33) See Chang et al. (2009) AWS had atypical functional connectivity from theleft inferior frontal gyrus to premotor regions andin the basal ganglia-thalamo-cortical pathway

13M 10F AWS (35)

Loucks et al. (2011) fMRI 9M 1F AWDS (25.2) Oral picture naming, Silent auditory phonememonitoring vs. Silent rest (baseline)

During picture naming, AWS showed higheractivity in the right inferior frontal gyrus, temporallobe and sensorimotor cortices relative to AWDS

10M 1F AWS (25.9)

Toyomura et al. (2011) fMRI 11M 1F AWDS (22-44) Sentence reading (solo, metronome or chorus) vs.Rest (baseline)

AWS exhibit reduced activity in the caudate duringsolo reading. This was increased to the levels ofAWS in the metronome condition but not thechorus condition. Basal ganglia were negativelycorrelated with stuttering severity

11M 1F AWS (18-55)

Howell et al. (2012) fMRI (SEM) 5M 4F AWDS (22–29) Picture naming (in mandarin). Pictures could benamed by characters with the following tones:high-flat, rising, falling-rising, and falling vs.passive viewing unnamable pictures (baseline)

AWS lacked functional connections between theinsula and left laryngeal motor cortex. AWS alsohad connections between the putamen and the leftlaryngeal motor cortex which were not seen inAWDS

7M 2F AWS (20–29)

Jiang et al. (2012) fMRI 20 AWS (26.8) Sentence completion task with same grammaticalstructure vs. resting

Whereas the left inferior frontal gyrus and thebilateral precuneus showed higher brain activity tomost typical than least typical symptoms, the leftputamen and the right cerebellum showed themost activation for the least typical symptoms

Lu et al. (2012) + fMRI 13 AWDS (24) Resting state connectivity compared before andafter speech therapy (eyes closed)

Resting state functional connectivity in thecerebellum was reduced following therapy butfunctional connectivity in the left inferior frontalgyrus was unchanged

15 AWS (24)

Wymbs et al. (2013) fMRI 4M AWS (19–25) Overt and covert production of (stutter prone andnon stutter prone words)

AWS exhibited consistent activations whenscanned on two different occasions. AWS did notshow strong overlap between different individualsduring overt and covert stuttering

Toyomura et al. (2015) fMRI 10 AWDS (22–23) Self vs. externally paced sentence reading, to thesound of the metronome, and rest (baseline).Tested before speech therapy (1 scanning session)and after speech therapy (another scanning

Before treatment AWS showed significantly lessactivation than AWDS for both the self and externalcondition in the putamen, caudate and globuspallidus. After treatment, there was no differencein the activation between AWS and AWDS

10 AWS (20–31)

32

A.C.

Etchell et

al. /

Journal of

Fluency D

isorders 55

(2018) 6–45

Table 2 (Continued)

(B)


Chang and Zhu (2013) + MRI 29 CWDS (3–9) Resting state functional connectivity CWS had reduced connectivity in networks thatsupport self paced movement27 CWS (3–9)

Watkins et al. (2008) + fMRI 8M 5F PWDS (14–27) Read sentences aloud (presented visually) vs.viewing a row of X’s with normal, delayed orfrequency shifted feedback

irrespective of feedback condition, PWS showedunderactivity in the ventral premotor regions12M 5F PWS (14–27)

Sato et al. (2011) * NIRS 10M AWS (18–44) Listen to phonemic contrast (/itta/vs./itte/orprosodic contrast/itta/vs./itta?/)

AWDS and CWDS showed LH advantage forphonemic vs. prosodic contrast; AWS and CWS didnot show laterality difference between conditions;Lateralization for speech processing abnormal inpreschool CWS

5M 2F CWS (6–12)5M 1F CWS (3–5)

Xuan et al. (2012) fMRI 46M PWDS (17–37) Resting state functional connectivity (eyes closed) PWS showed increased ALFF in left brain areasrelated to speech motor and auditory functions.PWS showed decreased ALFF in the bilateral nonspeech motor areas

44M PWS (17–37)

Liu et al. (2014) fMRI 33M 19F PWDS (6–50) Simon spatial incompatibility task On incongruent trials, PWS showed strongeractivation in the cingulate cortex, left prefrontalcortex, right medial frontal cortex. Activation inanterior cingulate cortex correlated with stutteringseverity. Inadequate readiness to execute motorresponse

29M 17F PWS (5–51)

fMRI = functional Magnetic Resonance Imaging; SEM = structural equation modeling. For definition of acronyms and symbols, see bottom of table 1

A.C.

Etchell et

al. /

Journal of

Fluency D

isorders 55

(2018) 6–45

33

Table 3(A) EEG studies of adults who stutter (18+). (B) EEG studies of children who stutter (<18) and persons who stutter (mixed-age).

(A)


Dietrich et al. (1995) EEG 10 M AWS (21–42) Listen to tones Latency of P3b wave was shorter in AWS

Cuadrado and Weber-Fox(2003)

EEG 8M 1F AWDS (22.1) Listen to sentences with and without verb agreementviolations

P600 to verb agreement violations was reduced in AWS

8M 1F AWS (22.1)

Corbera et al. (2005) EEG 10M 2F AWDS (19–30) Mismatch negativity paradigm AWS exhibited a larger mismatch negativity amplitudefor phonetic contrasts over left mastoid as comparedto AWDS and positively correlated with dysfluency

10M 2F AWS (19–30)

Andrade et al. (2007) EEG 3 M AWS (20–31) Oddball paradigm There were no significant differences in latency oramplitude after treatment

Achim et al. (2008) EEG 5 AWDS (21.2) Speech production AWDS showed left lateralized CNV and AWS showedright lateralized CNV5 AWS (28.4)

Hampton and Weber-Fox(2008) *

EEG 8M 3F AWDS (19–60) Oddball paradigm No statistical difference between amplitude or latencyof P300. Individual differences showed AWS P300 todeviant tones tended (though not significantly) to bereduced in in comparison with AWDS

8M 3F AWS (18–62)

Weber-Fox and Hampton(2008)

EEG 8M 2F AWDS (9–13) Listen to canonical sentences containing a verbagreement/violation and a semantically unexpectedverb

AWDS exhibited an N400 for reduced semanticexpectations and a P600 for verb agreement violations.AWS exhibited an N400 and a P600 for both conditions8M 2F AWS (9–13)

Angrisani et al. (2009) EEG 8M 10F AWDS (18–30) Oddball paradigm Middle latency response and P300 changed aftertreatment8M 8F AWS (18–30)

Sassi et al. (2011) EEG 6 AWDS (21.6) Oddball paradigm: 1000 Hz standard tone (80%),1500 Hz deviant tone (20%). Participants to raise fingerupon hearing deviant tone

There was no significant difference in amplitude orlatency of the P300 before or after treatment6 AWS (24.5)

Maxfield et al. (2012) EEG 2M 12F AWDS (23.6) Picture naming AWS exhibited reduced semantic priming andreversed phonological priming (enhancement of N400amplitude) for phono-logically related as opposed tophono-logically unrelated probes

11M 3F AWS (32.4)

Joos et al. (2014) EEG 11 AWDS (28) Resting state EEG (eyes closed) No significant differences between AWS and AWDSresting state functional activity. However, AWSexhibited decreased functional connectivity in betaband between left BA44/45 and contralateral premotorand primary motor areas

11 AWS (27.8)

Tahaei et al. (2014) EEG 21M 4F AWDS (16–25) Auditory brainstem responses to speech sounds(synthesized syllable/da/with a duration of 40 ms)

AWS had significantly increased latencies for the onsetand offset waves of V, A and O to speech ABR. AWSshowed deficient timing in early neural response tospeech sounds consistent with temporal processingdeficits

21M 4F AWS (16–35)

34

A.C.

Etchell et

al. /

Journal of

Fluency D

isorders 55

(2018) 6–45

Table 3 (Continued)

(A)


Daliri and Max (2015) EEG 11M 1F AWDS (19–45) Overt and covert reading and viewing ¨ + + + +Presenteda tone 400 ms after word onset on 1/3 of the trials ineach condition

AWS did not show a reduction in N1 amplitude tosounds prior to speech whereas AWDS did11M 1F AWS (18–46)

Maxfield et al. (2015) EEG 14M 5F AWDS (24) Fixation (500 ms), pattern mask “#####”(200 ms),prime word (identity or control), 70 ms, backwardmask (8 consonants, 50 ms), picture (remained untilresponse)

Priming improved reaction times in both AWS andAWDS. P1 amplitude correlated with expressivevocabulary in AWS but receptive vocabulary in AWDS.Identity priming reduced P280 amplitude in AWS

13M 6F AWS (26)

Mock et al. (2015) EEG 12 AWDS (23–55) 12 Cued or uncued picture naming with an early or lateprobe

Negative slow wave was associated with efferencecopy and was smaller in AWS relative to AWDSM AWS (23–54)

Vanhoutte et al. (2015) EEG 24M 11F AWDS (18–58) Picture naming task: Warning stimulus (S1 Picturepresented for 1000 ms), then a 1000 ms blank interval,and a imperative stimulus (S2, black line presented for2000 ms) and then a black screen (presented for afurther 2000 ms.) Name picture at (S2)

AWS showed an increased CNV slope as compared toAWDS. Slope of CNV positively correlates withstuttering severity and frequency. Cortical basalganglia loop is overactive during speech motorpreparation in AWS

19M 6F AWS (18–57)

Murase et al. (2015) EEG 9M 4F AWDS (30.9) Listen to sentences with a semantically correct orincorrect verb at the end

AWS showed an attenuated N400 late LPC only evidentin AWS12M 1F AWS (34.4)

(B)


Özge et al. (2004) EEG 21 CWDS (3–12) Hyperventilation vs. resting Amplitude of delta and theta wave was increased inCWS relative to CWDS. Amplitude of alpha band wasdecreased relative to CWDS

26 CWS (3–12)

Weber-Fox et al. (2008) * EEG 8M 2F CWDS (9–14) Judge whether or not a target word rhymes with aorthographically similar/dissimilar prime word

Latency of N400 earlier in left hemisphere than righthemisphere for CWDS, but similar latency between theleft and right hemisphere for CWS

8M 2F CWS (9–13)

Özcan et al. (2009) EEG 16M 4F CWDS 7-18) Listen to 2 successive ‘click’sounds of 100 ms duration CWS and CWDS were not significantly different interms of the amplitude or latency of the P50suppression

16M 4F CWS 7–18)

Arnold et al. (2011) EEG 6M 3F CWDS (3–6) Listen to emotionally charged conversations There were no significant differences between CWSand CWDS on EEG measures6M 3F CWS (3–6)

Kaganovich et al. (2010) * EEG 12M 6F CWDS (4–6) Oddball paradigm 1000 Hz (standard) and 2000 Hztones (deviant) presented monaurally and binaurally

CWDS but not CWS exhibited significant P3 responsesto deviant tones13M 5F CWS (4–6)

A.C.

Etchell et

al. /

Journal of

Fluency D

isorders 55

(2018) 6–45

35

Weber-Fox et al. (2013) EEG 18M 9F CWDS (3–6) Listen to grammatically correct sentences andsentences that contained either a semantic or syntacticviolation

There were longer peak latencies for the N400component; syntactic violations generate negativitybetween 150 and 350 ms; CWNS exhibited significantP600 in LH but CWS exhibited significant P600 in RH

20M 7F CWS (3–6)

Jansson-Verkasalo et al.(2014)

EEG 7M 5F CWDS (7–9) Multi-feature mismatch negativity MMN response was smaller in the CWS and CWDS onEEG measures. Only duration change elicited MMN inCWS10M CWS (6–9)

Mohan and Weber (2015) EEG 15M 7F CWDS (7.5) Listen to pairs of pseudo words that either did or didnot rhyme

N400 was greater for CWDS bilaterally, greater overright hemisphere for rCWS and absent in pCWS10M 2F pCWS (7.5)

10M 2F rCWS (7.7)

Usler and Weber-Fox(2015)

EEG 8M 1F CWDS (6–8) Listen to English or Jabberwocky sentences that wereeither normal or contained a semantic violation

Whereas both CWDS and recovered CWS exhibited ap600 to phrase structure violations in jabberwockysentences, pCWS did not8M 3F pCWS (6–8)

9M 2F rCWS (6–8)

Morgan et al. (1997) EEG 8M PWDS (17–36) Oddball task 5 of 8 PWS had higher P300 in LH and PWDS hadhigher P300 in RH8M PWS 17–36)

Rastatter et al. (1998) EEG 6M PWDS (16–44) Delayed vs. Non altered auditory feedback PWS showed a decrease in beta activity under delayedauditory feedback6M PWS (16–45)

Weber-Fox (2001) EEG 7M 2F PWDS (17–34) Covert sentence reading PWS had reduced amplitude to all conditions between200 and 450 ms relative to PWDS7M 2F PWS (17–34)

Weber-Fox et al. (2004) * EEG 7M 4F PWDS (17–44) Listen to rhyming and non rhyming word pairs PWS and PWDS showed similar EEG components. PWSbut not PWDS showed greater RH asymmetry7M 4F PWS (17–44)

Tahaei et al. (2014) EEG 21M 4F PWDS (16–35) Auditory brainstem responses to speech sounds(synthesized syllable/da/with a duration of 40 ms)

PWS have increased latencies for the onset and offsetwaves of V, A and O to speech ABR. AWS showeddeficient timing in early neural response to speechsounds consistent with temporal processing deficits

21M 4F PWS (16–35)

EEG = electroencephalography. For definition of acronyms and symbols, see bottom of Table 1.

36

A.C.

Etchell et

al. /

Journal of

Fluency D

isorders 55

(2018) 6–45

Table 4(A) MEG studies of adults who stutter (18+) . (B) MEG studies of children who stutter (<18) .

(A)

Study Method Sample Task Main finding height

Salmelin et al. (1998) MEG 8M 2F AWDS (25–52) Neural responses to tones during overt, cover andchoral reading

AWS had larger M100 responses to tone during choralreading7M 2F AWS (22–53)

Salmelin et al. (2000) MEG 8M 2F AWDS (25–52) Word reading and finger movements Sequence of brain activation was normal in AWDS butreversed in AWS7M 2F AWS (22–53)

Walla et al. (2004) MEG 8 AWDS (34) Overt, covert and delayed deading AWS failed to show preparatory motor activity whenreading aloud8 AWS (33.1)

Biermann-Ruben et al.(2005)

MEG 10M AWDS (26–42) Listening to nouns AWS activated left frontal cortex 95–145 ms afterstimulus onset but AWDS did not

10M AWS (26–40)

Beal et al. (2010)* MEG 12M AWDS (24–49) Listen to tones, listen to vowels/i/, listen to words(pre-recorded), produce vowels/i/, produce words

AWS and AWDS exhibited a similar amplitude ofspeech induced suppression. AWS had earlier rightthan left hemisphere M100 whereas AWDS did not

12M AWS (21–45)

Kikuchi et al. (2011) + MEG 18M AWDS (22-43) Auditory sensory gating (listen to 2 tones presented inquick succession)

Whereas AWDS exhibited a significant reduction inamplitude of the response to the second stimulus,AWS did not. This indicates they have abnormal leftsensory gating.

17M AWS (21-41)

(B)


Beal et al. (2011) * MEG 11M CWDS (6–12) Produce vowel sound ‘a’ Both CWDS and CWS showed suppression of auditoryM50; CWS had a delayed M50 latency relative to CWDS11M CWS (6–12)

Sowman et al. (2014) MEG 10M 2F CWDS (2–6) Picture naming There was no difference in laterality between CWS andCWDS10M 2F CWS (3–5)

MEG = magnetoencephalography. For definition of acronyms and symbols, see bottom of Table 1.

A.C.

Etchell et

al. /

Journal of

Fluency D

isorders 55

(2018) 6–45

37

Table 5TMS studies of adults who stutter (18+).


Sommer et al. (2003) TMS 18 AWDS (43.6) Active and resting motor threshold AWS had increased active and resting motor thresholdcompared to AWDS. No significant difference inintracortical inhibition

18 AWS (39.5)

Sommer et al. (2009) TMS (MEP, iSP) 15 AWDS (26.7) Paired pulse stimulation Recorded MEP’s from theabductor digiti minimi

There was no difference between groups forinterhemispheric inhibition or the ipsilateral silentperiod. AWS have normal inter-hemispheric inhibition

15 AWS (28.7)

Neef et al. (2011a) TMS (MEP) 9M 3F AWDS (29.5) Experiment 1: Active motor threshold paired pulsestimulation Experiment 2: MEP’s during active motorthreshold

AWS exhibited a reduced and delayed shortintracortical inhibition in the right hemisphere andreduced intracortical facilitation in both hemispheresrelative to AWDS. AWS also show a steeper MEPrecruitment curve than AWDS

9M 3F AWS (29.9)

Neef et al. (2011b) Repetitive(inhibitory) TMS

14M 1F AWDS (28.1) Repetitive TMS applied over the left or rightdorsolateral premotor cortex before paced fingertapping (in different blocks)

In AWS inhibition of the right dorsal premotor corteximpaired synchronization accuracy of the left handwhereas in AWDS inhibition of the left dorsal premotorcortex impaired synchronization of the left hand

13M 1F AWS (30.3)

Alm et al. (2013) TMS (MT) 14M 1F AWDS (20-52) TMS induced motor threshold of the abductor digitiminimi; compared motor threshold betweenindividuals

AWS had a higher MT in the left hemisphere ascompared to their own right hemisphere MT. AWS alsohad a higher MT in the left hemisphere compared tothe AWDS left hemisphere MT

14M 1F AWS (20–52)

Busan et al. (2013) TMS (MEP) 17M 6F AWDS (20-43) Resting and active motor threshold silent period of thefirst dorsal interosseous muscle

AWS showed reduced resting MEP amplitudecompared to AWDS in the left hemisphere whenaveraged across all stimulus intensities

11M 6F AWS (19–46)

Neef et al. (2015b) TMS (MEP) 9M 4F AWDS (23–44) Measured MEPs during speech AWS lacked an increase in cortical excitability in theleft hemisphere that was seen in AWDS. Negativecorrelation between facilitation and stuttering severityindicates pathophysiological role in the disorder

9M 4F AWS (21–55)

TMS = transcranial magnetic stimulation; MEP = motor evoked potential; MT = motor threshold. For definition of acronyms and symbols, see bottom of Table 1.

38

A.C.

Etchell et

al. /

Journal of

Fluency D

isorders 55

(2018) 6–45

Table 6(A) Gray matter studies of adults who stutter (18+). (B) Gray matter studies of children who stutter (<18) and persons who stutter (mixed-age).

(A)


Foundas et al. (2001) MRI 13M 3F AWDS (31.72) N/A AWS had a larger planum temporale than AWDS. Extra right perisylvian gyri in AWS13M 3F AWS (31.72)

Foundas et al. (2003) MRI 13M 3F AWDS (29.63) N/A AWS did not have a larger right than left prefrontal lobe volume or larger left thanright occipital lobe volume normally seen in AWDS13M 3F AWS (33.81)

Foundas et al. (2004) MRI 11M 3F AWDS (19–47) N/A Rightward asymmetry planum temporalein AWS was associated with more severestuttering and enhanced fluency under DAF11M 3F AWS (22–47)

Jäncke et al. (2004)+ MRI (VBM) 8M 2F AWDS (N/A) N/A No difference in GMV between AWS and AWDS.8M 2F AWS (N/A)

Beal et al. (2007)* + MRI (VBM) 28 AWDS (30.53) N/A AWS showed increased gray matter in bilateral STG and left inferior frontal gyrusrelative to AWDS26 AWS (30.29)

Cykowski et al. (2008) MRI 16 AWDS (21–63) N/A AWS show a significant increase in the number of sulci and the number of suprasylvian gyral banks along the right sylvian fissure. There was a positive correlationbetween the structural findings and stuttering severity

19 AWS (19-58)

Kell et al. (2009)+ MRI (VBM) 13M AWDS (23–44) N/A AWS had a decrease in GMV in the left IFG that negatively correlated with stutteringseverity13M pAWS (18–39)

13M rAWS (16–65)

Lu, Peng et al. (2010)+ MRI (VBM) 8M 4F AWDS (22-29) N/A AWS showed more GMV in the left putamen, bilateral precentral, occipital temporal,cingulate gyrus and less GMV in the left superior and medial frontal gyrus9M 1F AWS (19-31)

Lu et al. (2012)+ MRI 13 AWDS (24) N/A AWS had reduced thickness in left IFG and increased thickness in the SMA andcerebellum15 AWS (24)

Cai et al. (2014)+ MRI 14M 4F AWDS (19–43) N/A Thickness of the left frontal operculum was reduced in AWS relative to AWDS. Surfacearea of this region was increased in AWS relative to AWDS15M 5F AWS (18–47)

(B)


Chang et al. (2008) + MRI (VBM) 7M CWDS (9–12) N/A Risk of lifetime stuttering was associated with reduced GMV in left hemisphere8M pCWS (9–12)7M rCWS (9–12)

Mock et al. (2012) + MRI 14M CWDS (8-13) N/A Stuttering was associated with atypical brain torque in prefrontal areas. CWS had lessright prefrontal GMV than CWDS14M CWS (8–13)

Beal et al. (2013) * + MRI (VBM) 11M CWDS (6–12) N/A CWS had reduced GMV in left inferior frontal gyrus and putamen and increased GMVin right inferior frontal gyrus and superior temporal gyrus as compared to CWDS11M CWS (6–12)

Foundas et al. (2013) MRI 13M CWDS (8–13) N/A There was atypical caudate anatomy in 9/14 CWS. CWS have deficit incortico-striato-thlamo-cortical network14M CWS (8–13)

Beal et al. (2015) MRI 55M PWS (7–47) N/A In PWS only the left pars opercularis exhibited a different developmental trajectory ingray matter with age (None of the 30 other brain regions did)61M PWDS (6–48)

MRI = magnetic resonance imaging; GM/V = gray matter/volume; VBM = voxel based morphometry. For definition of acronyms and symbols, see bottom of Table 1.

A.C.

Etchell et

al. /

Journal of

Fluency D

isorders 55

(2018) 6–45

39

Table 7(A) White matter studies of adults who stutter (18+)t . (B) White matter studies of children who stutter (<18) and persons who stutter (mixed-age).

(A)


Sommer et al. (2002) dMRI (DTI) 11M 4F AWDS (23–43) N/A AWS had reduced WM integrity in the left tongue/laryngeal representation of themotor cortex relative to AWDS10M 5F AWS (18–44)

Jäncke et al. (2004) + MRI (VBM) 8M 2F AWDS (age N/A) N/A AWS had increased WM throughout RH of the brain8M 2F AWS (age N/A)

Beal et al. (2007) *+ MRI (VBM) 28 AWDS (30.53) N/A AWS showed increased WM in the right inferior frontal gyrus relative to AWDS26 AWS (30.29)

Kell et al. (2009) dMRI (DTI) 13M AWDS (23–44) N/A AWS had elevated FA in the left anterior insula and frontal regions. There were nodifferences between persistent and recovered adults who stutter (pAWS and rAWS,respectively)

13M pAWS (18–39)13M rAWS (16–65)

Cykowski et al. (2010) dMRI (DTI) 14 AWDS (30.4) N/A AWS showed a reduction in fractional anisotropy in left perisylvian regions. Thesewere driven by an increase in radial diffusivity (perpendicular to fiber tracts)13 AWS (31)

Lu, Peng et al. (2010) + MRI (VBM) 8M 4F AWDS (22–29) N/A AWS had less WM in the left superior temporal gyrus, left lingual gyrus and bilateralcerebellum. AWS had more WMV in the right superior and inferior frontal gyrus andsuperior temporal gyrus likely related to compensation.

9M 1F AWS (19-31)

Chang et al. (2011) + dMRI (DTI) 7M 7F AWDS (33) N/A AWS had atypical structural connections measured from the left inferior frontal gyrusto premotor regions and auditory regions9M 6F AWS (36)

Choo et al. (2011) MRI (VBM) 12M AWDS (20-35) N/A The rostrum and midbody of the corpus calllosum was larger in AWS than AWDSwhich reflects changes in the lateralization of language.11M AWS (20-35)

Kikuchi et al. (2011) + MRI (VBM) 15M AWDS (22–43) N/A AWS showed a significant increase in GMV in the right STG15M AWS (21–41)

Cai et al. (2014) + dMRI (DTI, networkbased statistics)

14M 4F AWDS (19–43) N/A AWS exhibited reduced FA in the left mid motor cortex and reduced connectivity withother cortical brain regions compared to AWDS. This reduction also positivelycorrelated with stuttering severity15M 5F AWS (18–47)

40

A.C.

Etchell et

al. /

Journal of

Fluency D

isorders 55

(2018) 6–45

Table 7 (Continued)

(A)


Cieslak et al. (2015) dMRI (diffusionspectrum imaging)

8 AWDS (20–31) N/A AWS were missing large portions of the bilateral arcuate fasciculus. AWS also had aconnection from the cortico-spinal tract to the temporal cortex that was not seen inAWDS8 AWS (20–39)

Civier et al. (2015) dMRI (DTI) 11M 3F AWDS (19–47) N/A AWS showed reduced FA in the anterior corpus callosum. This reduction is positivelycorrelated with a reduction in speech fluency. Likely represents weaker inhibition ofthe right frontal cortex

11M 3F AWS (19–52)

(B)


Chang et al. (2008) + dMRI (DTI) 7M CWDS (9–12) N/A Risk of persistent stuttering was associated with reduced WMV in left hemisphere8M pCWS (9–12)7M rCWS (9–12)

Choo et al. (2012) * MRI (VBM) 7M CWDS (9–12) N/A No difference in corpus callosum or WMV between pCWS, rCWS and CWDS8M pCWS (9–12)6M rCWS (9–12)

Mock et al. (2012) + MRI 14M CWDS (8–13) N/A Stuttering is associated with atypical brain torque in prefrontal areas; CWS had moreWMV in left and right hemispheres; positive correlation between left prefrontal WMVand dysfluency rate

14M CWS (8–13)

Beal et al. (2013) * + MRI (VBM) 11M CWS (6–12) N/A CWS had reduced WMV in the forceps minor of corpus callosum11M CWDS (6–12)

Chang and Zhu (2013) + dMRI (DTI) 29 CWDS (3–9) N/A Basal ganglia thalamo-cortical network develops abnormally in CWS27 CWS (3–9)

Chang et al. (2015) dMRI (DTI) 22M 20F CWDS (3-10) N/A CWS exhibit reduced FA in WM tracts connecting auditory and motor regions of thebrain and those that support timing control26M 19F CWS (3–10)

Watkins et al. (2008) + dMRI (DTI) 8M 5F PWDS (14–27) N/A PWS had reductions in WM in the ventral premotor regions relative to PWDS12M 5F PWS (14–27)

Connally et al. (2014) dMRI (DTI) 21M 8F PWS (14-42) N/A PWS had reduced fFA in the arcuate fasciculus. More severe dysfluency correlates withreduced white matter in angular gyrus23M 14F PWDS (14–45)

MRI = magnetic resonance imaging; dMRI = diffusion MRI; WMV = white matter volume; VBM = voxel based morphometry; FA = fractional anisotropy. For definition of acronyms and symbols, see bottom of Table 1.

R

AA

AA

A

A

A

BB

B

B

B

B

B

B

B

B

B

B

B

B

BB

B

B

C

C

C

C

C

C

C

CCC

C

C

C

C

C

C


eferences

chim, A., Braun, C. M., & Collin, I. (2008). Event-related potentials distinguish fluent and stuttered speech. Journal of Neurotherapy, 11(3), 15–23.lm, P. (2007). The dual premotor model of cluttering and stuttering: A neurological framework. In Proceedings of the first world conference on cluttering

(pp. 207–210).lm, P. A., Karlsson, R., Sundberg, M., & Axelson, H. W. (2013). Hemispheric lateralization of motor thresholds in relation to stuttering. PLOS ONE, 8(10)ndrade, C. R. F., Sassi, F. C., Matas, C. G., Neves, I. F., & Martins, V. O. (2007). P300 event-related potentials in stutterers pre and post treatment: A pilot

study. Pro Fono Revista de Atualizac ão Científica, 19(4), 401–405.ngrisani, R. M. G., Matas, C. G., Neves, I. F., Sassi, F. C., & Andrade, C. R. F. d. (2009). Electrophysiological auditory evaluation in stutterers pre and post

treatment. Pro Fono Revista de Atualizacão Científica, 21(2), 95–100.rnold, H. S., Conture, E. G., Key, A. P., & Walden, T. (2011). Emotional reactivity, regulation and childhood stuttering: A behavioral and

electrophysiological study. Journal of Communication Disorders, 44(3), 276–293.rnstein, D., Lakey, B., Compton, R. J., & Kleinow, J. (2011). Preverbal error monitoring in stutterers and fluent speakers. Brain and Language, 116(3),

105–115.allard, K. J., Robin, D. A., & Folkins, J. W. (2003). An integrative model of speech motor control: A response to Ziegler. Aphasiology, 17(1), 37–48.ares, M., & Rektor, I. (2001). Basal ganglia involvement in sensory and cognitive processing. A depth electrode CNV study in human participants. Clinical

Neurophysiology, 112(11), 2022–2030.eal, D. S., Cheyne, D. O., Gracco, V. L., Quraan, M. A., Taylor, M. J., & Luc, F. (2010). Auditory evoked fields to vocalization during passive listening and

active generation in adults who stutter. Neuroimage, 52(4), 1645–1653.eal, D. S., Gracco, V. L., Brettschneider, J., Kroll, R. M., & Luc, F. (2013). A voxel-based morphometry (VBM) analysis of regional gray and white matter

volume abnormalities within the speech production network of children who stutter. Cortex, 49(8), 2151–2161.eal, D. S., Gracco, V. L., Lafaille, S. J., & Luc, F. (2007). Voxel-based morphometry of auditory and speech-related cortex in stutterers. Neuroreport, 18(12),

1257–1260.eal, D. S., Lerch, J. P., Cameron, B., Henderson, R., Gracco, V. L., & De Nil, L. F. (2015). The trajectory of gray matter development in Broca’s area is abnormal

in people who stutter. Frontiers in Human Neuroscience, 9.eal, D. S., Quraan, M. A., Cheyne, D. O., Taylor, M. J., Gracco, V. L., & Luc, F. (2011). Speech-induced suppression of evoked auditory fields in children who

stutter. Neuroimage, 54(4), 2994–3003.elyk, M., Kraft, S. J., & Brown, S. (2015). Stuttering as a trait or state −An ALE meta-analysis of neuroimaging studies. European Journal of Neuroscience,

41(2), 275–284.iermann-Ruben, K., Salmelin, R., & Schnitzler, A. (2005). Right rolandic activation during speech perception in stutterers: A MEG study. Neuroimage,

25(3), 793–801.lomgren, M., Nagarajan, S. S., Lee, J. N., Li, T., & Alvord, L. (2004). Preliminary results of a functional MRI study of brain activation patterns in stuttering

and nonstuttering speakers during a lexical access task. Journal of Fluency Disorders, 28(4), 337–356.oyle, M. (2015). Relations between causal attributions for stuttering and psychological well-being in adults who stutter. International Journal of

Speech-Language Pathology, 1–10 (ahead-of-print).raun, A., Varga, M., Stager, S., Schulz, G., Selbie, S., Maisog, J., et al. (1997). Altered patterns of cerebral activity during speech and language production in

developmental stuttering. Brain, 120(5), 761–784.rocklehurst, P. H., & Corley, M. (2011). Investigating the inner speech of people who stutter: Evidence for (and against) the Covert Repair Hypothesis.

Journal of Communication Disorders, 44(2), 246–260.rown, S., Ingham, R. J., Ingham, J. C., Laird, A. R., & Fox, P. T. (2005). Stuttered and fluent speech production: An ALE meta-analysis of functional

neuroimaging studies. Human Brain Mapping, 25(1), 105–117.rown, B. T. (2015). Neurocorrelates of speech-motor planning and execution in adults and children who stutter. University of Iowa.udde, K. S., Barron, D. S., & Fox, P. T. (2014). Stuttering, induced fluency, and natural fluency: A hierarchical series of activation likelihood estimation

meta-analyses. Brain and Language, 139, 99–107.usan, P., D’Ausilio, A., Borelli, M., Monti, F., Pelamatti, G., Pizzolato, G., et al. (2013). Motor excitability evaluation in developmental stuttering: A

transcranial magnetic stimulation study. Cortex, 49(3), 781–792.usan, P., Del Ben, G., Bernardini, S., Natarelli, G., Bencich, M., Monti, F., et al. (2016). Altered modulation of silent period in tongue motor cortex of

persistent developmental stuttering in relation to stuttering severity. PLOS ONE, 11(10), e0163959.ai, S., Tourville, J. A., Beal, D. S., Perkell, J. S., Guenther, F. H., & Ghosh, S. S. (2014). Diffusion imaging of cerebral white matter in persons who stutter:

Evidence for network-level anomalies. Frontiers in Human Neuroscience, 8.atani, M., Mesulam, M. M., Jakobsen, E., Malik, F., Martersteck, A., Wieneke, C., et al. (2013). A novel frontal pathway underlies verbal fluency in primary

progressive aphasia. Brain, 136(8), 2619–2628.¸ elik, T., Gökc en, C., Aytas Özc elik, A., C elik, M., & C oban, N. (2015). The prevalence of anti-Toxoplasma gondii antibodies in stutterers is higher than in the

control group. Folia Parasitologica, 62, 066.hang, S. E. (2014). Research updates in neuroimaging studies of children who stutter. Seminars in speech and language, 35, 067–079. No. 02, Thieme

Medical Publishers.hang, S.-E., Erickson, K. I., Ambrose, N. G., Hasegawa-Johnson, M. A., & Ludlow, C. L. (2008). Brain anatomy differences in childhood stuttering.

Neuroimage, 39(3), 1333–1344.hang, S.-E., Horwitz, B., Ostuni, J., Reynolds, R., & Ludlow, C. L. (2011). Evidence of left inferior frontal–premotor structural and functional connectivity

deficits in adults who stutter. Cerebral Cortex, 21(11), 2507–2518.hang, S.-E., Kenney, M. K., Loucks, T. M., & Ludlow, C. L. (2009). Brain activation abnormalities during speech and non-speech in stuttering speakers.

Neuroimage, 46(1), 201–212.hang, S.-E., & Zhu, D. C. (2013). Neural network connectivity differences in children who stutter. Brain, 136(12), 3709–3726.hang, S.-E., Zhu, D. C., Choo, A. L., & Angstadt, M. (2015). White matter neuroanatomical differences in young children who stutter. Brain, 138(3), 694–711.hesters, J., Watkins, K. E., & Möttönen, R. (2017). Investigating the feasibility of using transcranial direct current stimulation to enhance fluency in people

who stutter. Brain and Language, 164, 68–76.hoo, A. L., Burnham, E., Hicks, K., & Chang, S.-E. (2016). Dissociations among linguistic, cognitive, and auditory-motor neuroanatomical domains in

children who stutter. Journal of Communication Disorders, 61, 29–47.hoo, A. L., Chang, S.-E., Zengin-Bolatkale, H., Ambrose, N. G., & Loucks, T. M. (2012). Corpus callosum morphology in children who stutter. Journal of

Communication Disorders, 45(4), 279–289.hoo, A. L., Kraft, S. J., Olivero, W., Ambrose, N. G., Sharma, H., Chang, S.-E., et al. (2011). Corpus callosum differences associated with persistent stuttering

in adults. Journal of Communication Disorders, 44(4), 470–477.ieslak, M., Ingham, R. J., Ingham, J. C., & Grafton, S. T. (2015). Anomalous white matter morphology in adults who stutter. Journal of Speech, Language, and

Hearing Research, 58(2), 268–277.
ivier, O., Tasko, S. M., & Guenther, F. H. (2010). Overreliance on auditory feedback may lead to sound/syllable repetitions: simulations of stuttering and
fluency-inducing conditions with a neural model of speech production. Journal of fluency disorders, 35(3), 246–279.ivier, O., Bullock, D., Max, L., & Guenther, F. H. (2013). Computational modeling of stuttering caused by impairments in a basal ganglia thalamo–cortical

circuit involved in syllable selection and initiation. Brain and Language, 126(3), 263–278.

http://refhub.elsevier.com/S0094-730X(16)30065-1/sbref0005


























































































































































































































































































































































































































































































































































































































































































































































































































































Civier, O., Kronfeld-Duenias, V., Amir, O., Ezrati-Vinacour, R., & Ben-Shachar, M. (2015). Reduced fractional anisotropy in the anterior corpus callosum isassociated with reduced speech fluency in persistent developmental stuttering. Brain and Language, 143, 20–31.

Connally, E. L., Ward, D., Howell, P., & Watkins, K. E. (2014). Disrupted white matter in language and motor tracts in developmental stuttering. Brain andLanguage, 131, 25–35.

Corbera, S., Corral, M.-J., Escera, C., & Idiazábal, M. A. (2005). Abnormal speech sound representation in persistent developmental stuttering. Neurology,65(8), 1246–1252.

Craig, A., Hancock, K., Tran, Y., Craig, M., & Peters, K. (2002). Epidemiology of stuttering in the community across the entire life span. Journal of Speech,Language, and Hearing Research, 45(6), 1097–1105.

Cross, D. E., & Luper, H. L. (1983). Relation between finger reaction time and voice reaction time in stuttering and nonstuttering children and adults.Journal of Speech, Language and Hearing Research, 26(3), 356–361.

Cuadrado, E. M., & Weber-Fox, C. M. (2003). Atypical syntactic processing in individuals who stutter evidence from event-related brain potentials andbehavioral measures. Journal of Speech, Language, and Hearing Research, 46(4), 960–976.

Cykowski, M. D., Fox, P. T., Ingham, R. J., Ingham, J. C., & Robin, D. A. (2010). A study of the reproducibility and etiology of diffusion anisotropy differencesin developmental stuttering: A potential role for impaired myelination. Neuroimage, 52(4), 1495–1504.

Cykowski, M. D., Kochunov, P. V., Ingham, R. J., Ingham, J. C., Mangin, J.-F., Rivi‘ere, D., et al. (2008). Perisylvian sulcal morphology and cerebral asymmetrypatterns in adults who stutter. Cerebral Cortex, 18(3), 571–583.

Daliri, A., & Max, L. (2015). Modulation of auditory processing during speech movement planning is limited in adults who stutter. Brain and Language, 143,59–68.

De Nil, L. F., Kroll, R. M., Lafaille, S. J., & Houle, S. (2004). A positron emission tomography study of short- and long-term treatment effects on functionalbrain activation in adults who stutter. Journal of Fluency Disorders, 28(4), 357–380.

De Nil, F., Beal, D. S., Lafaille, S. J., Kroll, R. M., Crawley, A. P., & Gracco, V. L. (2008). The effects of simulated stuttering and prolonged speech on the neuralactivation patterns of stuttering and nonstuttering adults. Brain and Language, 107(2), 114–123.

De Nil, L., Kroll, R. M., & Houle, S. (2001). Functional neuroimaging of cerebellar activation during single word reading and verb generation in stutteringand nonstuttering adults. Neuroscience Letters, 302(2), 77–80.

De Nil, L., Kroll, R. M., Kapur, S., & Houle, S. (2000). A positron emission tomography study of silent and oral single word reading in stuttering andnonstuttering adults. Journal of Speech, Language and Hearing Research, 43(4), 1038–1053.

Desai, J., Huo, Y., Wang, Z., Bansal, R., Williams, S. C., Lythgoe, D., et al. (2016). Reduced perfusion in Broca’s area in developmental stuttering. Human BrainMapping, 38(4), 1865–1874.

Devlin, J. T., & Watkins, K. E. (2007). Stimulating language: Insights from TMS. Brain, 130(3), 610–622.Dietrich, S., Barry, S. J., & Parker, D. E. (1995). Middle latency auditory responses in males who stutter. Journal of Speech, Language, and Hearing Research,

38(1), 5–17.Eickhoff, S. B., Laird, A. R., Fox, P. M., Lancaster, J. L., & Fox, P. T. (2017). Implementation errors in the Ginger ALE Software: Description and

recommendations. Human Brain Mapping, 38(1), 7–11.Etchell, A. C., Johnson, B. W., & Sowman, P. F. (2014). Behavioral and multimodal neuroimaging evidence for a deficit in brain timing networks in

stuttering: A hypothesis and theory. Frontiers in Human Neuroscience, 8, 467.Etchell, A. C., Johnson, B. W., & Sowman, P. F. (2016). Beta oscillations, timing, and stuttering. Frontiers in Human Neuroscience, 1036.Etchell, A. C., Ryan, M., Martin, E., Johnson, B. W., & Sowman, P. F. (2016). Abnormal time course of low beta modulation in non-fluent preschool children:

A magnetoencephalographic study of rhythm tracking. Neuroimage, 125, 953–963.Farquharson, S., Tournier, J. D., Calamante, F., Fabinyi, G., Schneider-Kolsky, M., Jackson, G. D., et al. (2013). White matter fiber tractography: Why we need

to move beyond DTI: Clinical article. Journal of Neurosurgery, 118(6), 1367–1377.Foundas, A., Corey, D., Angeles, V., Bollich, A., Crabtree-Hartman, E., & Heilman, K. (2003). Atypical cerebral laterality in adults with persistent

developmental stuttering. Neurology, 61(10), 1378–1385.Foundas, A. L., Bollich, A. M., Corey, D. M., Hurley, M., & Heilman, K. M. (2001). Anomalous anatomy of speech −Language areas in adults with persistent

developmental stuttering. Neurology, 57(2), 207–215.Foundas, A. L., Bollich, A. M., Feldman, J., Corey, D. M., Hurley, M., Lemen, L. C., et al. (2004). Aberrant auditory processing and atypical planum temporale

in developmental stuttering. Neurology, 63(9), 1640–1646.Foundas, A. L., Cindass, R., Jr., Mock, J. R., & Corey, D. M. (2013). Atypical caudate anatomy in children who stutter. Perceptual & Motor Skills, 116(2),

528–543.Fox, P. T., Ingham, R. J., Ingham, J. C., Hirsch, T. B., Downs, J. H., Martin, C., et al. (1996). A PET study of the neural systems of stuttering. Nature, 382, 158–162.Fox, P. T., Ingham, R. J., Ingham, J. C., Zamarripa, F., Xiong, J.-H., & Lancaster, J. L. (2000). Brain correlates of stuttering and syllable production. Brain,

123(10), 1985–2004.Gambhir, S. S. (2002). Molecular imaging of cancer with positron emission tomography. Nature Reviews Cancer, 2, 683–693.Giraud, A.-L., Neumann, K., Bachoud-Levi, A.-C., von Gudenberg, A. W., Euler, H. A., Lanfermann, H., et al. (2008). Severity of dysfluency correlates with

basal ganglia activity in persistent developmental stuttering. Brain and Language, 104(2), 190–199.Gunn, A., Menzies, R. G., O’Brian, S., Onslow, M., Packman, A., Lowe, R., et al. (2014). Axis I anxiety and mental health disorders among stuttering

adolescents. Journal of Fluency Disorders, 40, 58–68.Halag-Milo, T., Stoppelman, N., Kronfeld-Duenias, V., Civier, O., Amir, O., Ezrati-Vinacour, R., et al. (2016). Beyond production: Brain responses during

speech perception in adults who stutter. Neuroimage Clinical, 11, 328–338.Hampton, A., & Weber-Fox, C. (2008). Non-linguistic auditory processing in stuttering: Evidence from behavior and event-related brain potentials. Journal

of Fluency Disorders, 33(4), 253–273.Herholz, S. C., & Zatorre, R. J. (2012). Musical training as a framework for brain plasticity: Behavior, function, and structure. Neuron, 76(3), 486–502.Howell, P. (2007). Signs of developmental stuttering up to age eight and at 12 plus. Clinical Psychology Review, 27(3), 287–306.Howell, P., Jiang, J., Peng, D., & Lu, C. (2012). Neural control of rising and falling tones in Mandarin speakers who stutter. Brain and Language, 123(3),

211–221.Ingham, R. J., Fox, P. T., Ingham, J. C., Xiong, J., Zamarripa, F., Hardies, L. J., et al. (2004). Brain correlates of stuttering and syllable production gender

comparison and replication. Journal of Speech, Language, and Hearing Research, 47(2), 321–341.Ingham, R. J., Fox, P. T., Ingham, J. C., & Zamarripa, F. (2000). Is overt stuttered speech a prerequisite for the neural activations associated with chronic

developmental stuttering? Brain and Language, 75(2), 163–194.Ingham, R. J., Fox, P. T., Ingham, J. C., Zamarripa, F., Martin, C., Jerabek, P., et al. (1996). Functional-lesion investigation of developmental stuttering with

positron emission tomography. Journal of Speech, Language, and Hearing Research, 39(6), 1208–1227.Ingham, R. J., Grafton, S. T., Bothe, A. K., & Ingham, J. C. (2012). Brain activity in adults who stutter: Similarities across speaking tasks and correlations with

stuttering frequency and speaking rate. Brain and Language, 122(1), 11–24.Ingham, R. J., Wang, Y., Ingham, J. C., Bothe, A. K., & Grafton, S. T. (2013). Regional brain activity change predicts responsiveness to treatment for stuttering

in adults. Brain and Language, 127(3), 510–519.
Iverach, L., Jones, M., O’Brian, S., Block, S., Lincoln, M., Harrison, E., et al. (2010). Mood and substance use disorders among adults seeking speech treatment
for stuttering. Journal of Speech, Language, and Hearing Research, 53(5), 1178–1190.Iverach, L., O’Brian, S., Jones, M., Block, S., Lincoln, M., Harrison, E., et al. (2009). Prevalence of anxiety disorders among adults seeking speech therapy for

stuttering. Journal of Anxiety Disorders, 23(7), 928–934.Jäncke, L., Hanggi, J., & Steinmetz, H. (2004). Morphological brain differences between adult stutterers and non-stutterers. BMC Neurology, 4(1), 23.






















































































































































































































































































































































































































































































































































































































































































































































































































































































































J

J

J

JJ

J

K

K

KK

K

K

K

K

K

K

K

L

L

L

L

L

L

L

L

L

M

M

M

M

M

MM

M

M

M

M

N

N

N

N


ansson-Verkasalo, E., Eggers, K., Järvenpää, A., Suominen, K., Van den Bergh, B., De Nil, L., et al. (2014). Atypical central auditory speech-sounddiscrimination in children who stutter as indexed by the mismatch negativity. Journal of Fluency Disorders, 41, 1–11.

escheniak, J. D., Schriefers, H., Garrett, M. F., & Friederici, A. D. (2002). Exploring the activation of semantic and phonological codes during speechplanning with event-related brain potentials. Journal of Cognitive Neuroscience, 14(6), 951–964.

eurissen, B., Leemans, A., Tournier, J. D., Jones, D. K., & Sijbers, J. (2013). Investigating the prevalence of complex fiber configurations in white mattertissue with diffusion magnetic resonance imaging. Human Brain Mapping, 34(11), 2747–2766.

iang, J., Lu, C., Peng, D., Zhu, C., & Howell, P. (2012). Classification of types of stuttering symptoms based on brain activity. PLOS ONE, 7(6), e39747.oos, K., De Ridder, D., Boey, R. A., & Vanneste, S. (2014). Functional connectivity changes in adults with developmental stuttering: A preliminary study

using quantitative electro-encephalography. Frontiers in Human Neuroscience, 8, 783.udenhofer, M. S., Wehrl, H. F., Newport, D. F., Catana, C., Siegel, S. B., Becker, M., et al. (2008). Simultaneous PET-MRI: A new approach for functional and

morphological imaging. Nature Medicine, 14(4), 459–465.azenski, D. M. (2015). fNIRS measures of prefrontal cortex lateralization during stuttered and fluency-enhanced speech in adults who stutter. (Doctoral

dissertation). The University of Vermont and State Agricultural College.aganovich, N., Wray, A. H., & Weber-Fox, C. (2010). Non-linguistic auditory processing and working memory update in pre-school children who stutter:

An electrophysiological study. Developmental Neuropsychology, 35(6), 712–736.ell, C. A., Neumann, K., von Kriegstein, K., Posenenske, C., von Gudenberg, A. W., Euler, H., et al. (2009). How the brain repairs stuttering. Brain, 185.emerdere, R., de Champfleur, N. M., Deverdun, J., Cochereau, J., Moritz-Gasser, S., Herbet, G., et al. (2016). Role of the left frontal aslant tract in stuttering:

A brain stimulation and tractographic study. Journal of Neurology, 263(1), 157–167.hedr, E., El-Nasser, W. A., Abdel Haleem, E. K., Bakr, M. S., & Trakhan, M. N. (2000). Evoked potentials and electroencephalography in stuttering. Folia

Phoniatrica et Logopaedica, 52(4), 178–186.ikuchi, Y., Ogata, K., Umesaki, T., Yoshiura, T., Kenjo, M., Hirano, Y., et al. (2011). Spatiotemporal signatures of an abnormal auditory system in stuttering.

Neuroimage, 55(3), 891–899.ikuchi, Y., Okamoto, T., Ogata, K., Hagiwara, O., Umezaki, T., Kenjo, M., et al. (2016). Abnormal auditory synchronization in stuttering: A

magnetoencephalographic study. Hearing Research, 344, 82–89.löppel, S., Bäumer, T., Kroeger, J., Koch, M. A., Büchel, C., Münchau, A., et al. (2008). The cortical motor threshold reflects microstructural properties of

cerebral white matter. Neuroimage, 40(4), 1782–1791.ronfeld-Duenias, V., Amir, O., Ezrati-Vinacour, R., Civier, O., & Ben-Shachar, M. (2016a). The frontal aslant tract underlies speech fluency in persistent

developmental stuttering. Brain Structure and Function, 221(1), 365–381.ronfeld-Duenias, V., Amir, O., Ezrati-Vinacour, R., Civier, O., & Ben-Shachar, M. (2016b). Dorsal and ventral language pathways in persistent

developmental stuttering. Cortex, 81, 79–92.ujirai, T., Caramia, M., Rothwell, J. C., Day, B., Thompson, P., Ferbert, A., et al. (1993). Corticocortical inhibition in human motor cortex. The Journal of

Physiology, 471, 501.appe, C., Herholz, S. C., Trainor, L. J., & Pantev, C. (2008). Cortical plasticity induced by short-term unimodal and multimodal musical training. The Journal

of Neuroscience, 28(39), 9632–9639.iotti, M., Ingham, J. C., Takai, O., Paskos, D. K., Perez, R., & Ingham, R. J. (2010). Spatiotemporal dynamics of speech sound perception in chronic

developmental stuttering. Brain and Language, 115(2), 141–147.iu, J., Wang, Z., Huo, Y., Davidson, S. M., Klahr, K., Herder, C. L., et al. (2014). A functional imaging study of self-regulatory capacities in persons who

stutter. PLOS ONE, 9(2), e89891.oucks, T., Kraft, S. J., Choo, A. L., Sharma, H., & Ambrose, N. G. (2011). Functional brain activation differences in stuttering identified with a rapid fMRI

sequence. Journal of Fluency Disorders, 36(4), 302–307.u, C., Chen, C., Ning, N., Ding, G., Guo, T., Peng, D., et al. (2010). The neural substrates for atypical planning and execution of word production in stuttering.

Experimental Neurology, 221(1), 146–156.u, C., Peng, D., Chen, C., Ning, N., Ding, G., Li, K., et al. (2010). Altered effective connectivity and anomalous anatomy in the basal ganglia–thalamocortical

circuit of stuttering speakers. Cortex, 46(1), 49–67.u, C., Chen, C., Peng, D., You, W., Zhang, X., Ding, G., et al. (2012). Neural anomaly and reorganization in speakers who stutter a short-term intervention

study. Neurology, 79(7), 625–632.u, C., Ning, N., Peng, D., Ding, G., Li, K., Yang, Y., et al. (2009). The role of large-scale neural interactions for developmental stuttering. Neuroscience, 161(4),

1008–1026.u, C., Long, Y., Zheng, L., Shi, G., Liu, L., Ding, G., et al. (2016). Relationship between speech production and perception in people who stutter. Frontiers in

Human Neuroscience, 10.athiak, K., Hertrich, I., Grodd, W., & Ackermann, H. (2002). Cerebellum and speech perception: A functional magnetic resonance imaging study. Journal

of Cognitive Neuroscience, 14(6), 902–912.axfield, N. D., Huffman, J. L., Frisch, S. A., & Hinckley, J. J. (2010). Neural correlates of semantic activation spreading on the path to picture naming in

adults who stutter. Clinical Neurophysiology, 121(9), 1447–1463.axfield, N. D., Morris, K., Frisch, S. A., Morphew, K., & Constantine, J. L. (2015). Real-time processing in picture naming in adults who stutter: ERP

evidence. Clinical Neurophysiology, 126(2), 284–296.axfield, N. D., Olsen, W. L., Kleinman, D., Frisch, S. A., Ferreira, V. S., & Lister, J. J. (2016). Attention demands of language production in adults who stutter.

Clinical Neurophysiology, 127(4), 1942–1960.axfield, N. D., Pizon-Moore, A. A., Frisch, S. A., & Constantine, J. L. (2012). Exploring semantic and phonological picture – Word priming in adults who

stutter using event-related potentials. Clinical Neurophysiology, 123(6), 1131–1146.ock, J. R., Foundas, A. L., & Golob, E. J. (2015). Speech preparation in adults with persistent developmental stuttering. Brain and Language, 149, 97–105.ock, J. R., Zadina, J. N., Corey, D. M., Cohen, J. D., Lemen, L. C., & Foundas, A. L. (2012). Atypical brain torque in boys with developmental stuttering.

Developmental Neuropsychology, 37(5), 434–452.ock, J. R., Foundas, A. L., & Golob, E. J. (2016). Cortical activity during cued picture naming predicts individual differences in stuttering frequency. Clinical

Neurophysiology.ohan, R., & Weber, C. (2015). Neural systems mediating processing of sound units of language distinguish recovery versus persistence in stuttering.

Journal of Neurodevelopmental Disorders, 7(1).organ, M. D., Cranford, J. L., & Burk, K. (1997). P300 event-related potentials in stutterers and nonstutterers. Journal of Speech, Language, and Hearing

Research, 40(6), 1334–1340.urase, S., Kawashima, T., Satake, H., & Era, S. (2015). An event-related potential investigation of sentence processing in adults who stutter. Neuroscience

Research, 106, 29–37.eef, N. E., Anwander, A., & Friederici, A. D. (2015). The neurobiological grounding of persistent stuttering: From structure to function. Current Neurology

and Neuroscience Reports, 15(9), 1–11.eef, N. E., Hoang, T. L., Neef, A., Paulus, W., & Sommer, M. (2015). Speech dynamics are coded in the left motor cortex in fluent speakers but not in adults

who stutter. Brain, 138(3), 712–725.eef, N. E., Jung, K., Rothkegel, H., Pollok, B., von Gudenberg, A. W., Paulus, W., et al. (2011). Right-shift for non-speech motor processing in adults who

stutter. Cortex, 47(8), 945–954.eef, N., Paulus, W., Neef, A., von Gudenberg, A. W., & Sommer, M. (2011). Reduced intracortical inhibition and facilitation in the primary motor tongue

representationof adults who stutter. Clinical Neurophysiology, 122(9), 1802–1811.



















































































































































































































































































































































































































































































































































































































































































Neumann, K., Euler, H. A., von Gudenberg, A. W., Giraud, A.-L., Lanfermann, H., Gall, V., et al. (2004). The nature and treatment of stuttering as revealed byfMRI: A within-and between-group comparison. Journal of Fluency Disorders, 28(4), 381–410.

Neumann, K., Preibisch, C., Euler, H. A., Von Gudenberg, A. W., Lanfermann, H., Gall, V., et al. (2005). Cortical plasticity associated with stuttering therapy.Journal of Fluency Disorders, 30(1), 23–39.

Ning, N., Peng, D., Liu, X., & Yang, S. (2017). Speech timing deficit of stuttering: Evidence from contingent negative variations. PLOS ONE, 12(1), e0168836.O’Neill, J., Dong, Z., Bansal, R., Ivanov, I., Hao, X., Desai, J., et al. (2017). Proton chemical shift imaging of the brain in pediatric and adult developmental

stuttering. Jama Psychiatry, 74(1), 85–94.Özcan, S., Altınayar, S., Özcan, C., Ünal, S., & Karlıdag, R. (2009). P50 sensory gating in children and adolescents with developmental stuttering. Bulletin of

Clinical Psychopharmacology, 19(3), 241–246.Özge, A., Toros, F., & C ömelekoglu, Ü. (2004). The role of hemispheral asymmetry and regional activity of quantitative EEG in children with stuttering.

Child Psychiatryand Human Development, 34(4), 269–280.Pascual-Marqui, R. D. (2002). Standardized low-resolution brain electromagnetic tomography (sLORETA): Technical details, Methods & Findings in

Experimental &. Clinical Pharmacology, 24(Suppl. D), 5–12.Piispala, J., Kallio, M., Bloigu, R., & Jansson-Verkasalo, E. (2016). Delayed N2 response in Go condition in a visual Go/ Nogo ERP study in children who

stutter. Journalof Fluency Disorders, 48, 16–26.Pollok, B., Krause, V., Butz, M., & Schnitzler, A. (2009). Modality specific functional interaction in sensorimotor synchronization. Human Brain Mapping,

30(6), 1783–1790.Postma, A., Kolk, H., & Povel, D. J. (1990). Speech planning and execution in stutterers. Journal of Fluency Disorders, 15(1), 49–59.Preibisch, C., Neumann, K., Raab, P., Euler, H. A., von Gudenberg, A. W., Lanfer- mann, H., et al. (2003). Evidence for compensation for stuttering by the

right frontal operculum. Neuroimage, 20(2), 1356–1364.Prestes, R., de Andrade, A. N., Santos, R. B. F., Marangoni, A. T., Schiefer, A. M., & Gil, D. (2016). Temporal processing and long-latency auditory evoked

potential in stutterers. Brazilian Journal of Otorhinolaryngology, 83(2), 142–146.Price, C. J. (2010). The anatomy of language: A review of 100 fMRI studies published in 2009. Annals of the New York Academy of Sciences, 1191(1), 62–88.Price, C. J. (2012). A review and synthesis of the first 20 years of PET and fMRI studies of heard speech, spoken language and reading. Neuroimage, 62(2),

816–847.Rastatter, M. P., Stuart, A., & Kalinowski, J. (1998). Quantitative electroencephalogram of posterior cortical areas of fluent and stuttering participants

during reading with normal and altered auditory feedback. Perceptual and Motor Skills, 87(2), 623–633.Sakai, K., Hikosaka, O., & Nakamura, K. (2004). Emergence of rhythm during motor learning. Trends in Cognitive Sciences, 8(12), 547–553.Sakai, N., Masuda, S., Shimotomai, T., & Mori, K. (2009). Brain activation in adults who stutter under delayed auditory feedback: An fMRI study.

International Journal of Speech-Language Pathology, 11(1), 2–11.Salmelin, R., Schnitzler, A., Schmitz, F., & Freund, H.-J. (2000). Single word reading in developmental stutterers and fluent speakers. Brain, 123(6),

1184–1202.Salmelin, R., Schnitzler, A., Schmitz, F., Jäncke, L., Witte, O. W., & Freund, H. J. (1998). Functional organization of the auditory cortex is different in

stutterers and fluent speakers. Neuroreport, 9(10), 2225–2229.Sassi, F. C., Matas, C. G., de Mendonc a, L. I. Z., & de Andrade, C. R. F. (2011). Stuttering treatment control using p300 event-related potentials. Journal of

Fluency Disorders, 36(2), 130–138.Sato, Y., Mori, K., Koizumi, T., Minagawa-Kawai, Y., Tanaka, A., Ozawa, E., et al. (2011). Functional lateralization of speech processing in adults and children

who stutter. Frontiers in Psychology, 2, 70.Schmahmann, J. D., & Pandya, D. (2009). Fiber pathways of the brain. OUP USA.Sengupta, R., Shah, S., Gore, K., Loucks, T., & Nasir, S. M. (2016). Anomaly in neural phase coherence accompanies reduced sensorimotor integration in

adults who stutter. Neuropsychologia, 93, 242–250.Sitek, K. R., Cai, S., Beal, D. S., Perkell, J. S., Guenther, F. H., & Ghosh, S. S. (2016). Decreased cerebellar-orbitofrontal connectivity correlates with stuttering

severity: Whole-brain functional and structural connectivity associations with persistent developmental stuttering. Frontiers in Human Neuroscience,10, 190.

Smith, S. M., Jenkinson, M., Johansen-Berg, H., Rueckert, D., Nichols, T. E., Mackay, C. E., et al. (2006). Tract-based spatial statistics: Voxelwise analysis ofmulti-subject diffusion data. Neuroimage, 31(4), 1487–1505.

Song, L. P., Peng, D. L., Jin, Z., Yao, L., Ning, N., Guo, X. J., et al. (2007). Gray matter abnormalities in developmental stuttering determined with voxel-basedmorphometry. Zhonghua yi xue za zhi, 87(41), 2884–2888.

Sommer, M., Knappmeyer, K., Hunter, E. J., Gudenberg, A. W., Neef, N., & Paulus, W. (2009). Normal interhemispheric inhibition in persistentdevelopmental stuttering. Movement Disorders, 24(5), 769–773.

Sommer, M., Koch, M. A., Paulus, W., Weiller, C., & Büchel, C. (2002). Disconnection of speech-relevant brain areas in persistent developmental stuttering.The Lancet, 360(9330), 380–383.

Sommer, M., Wischer, S., Tergau, F., & Paulus, W. (2003). Normal intracortical excitability in developmental stuttering. Movement Disorders, 18(7),826–830.

Sowman, P. F., Crain, S., Harrison, E., & Johnson, B. W. (2014). Lateralization of brain activation in fluent and non-fluent preschool children: Amagnetoencephalographic study of picture-naming. Frontiers in Human Neuroscience, 8. 354.

Sowman, P. F., Ryan, M., Johnson, B. W., Savage, G., Crain, S., Harrison, E., et al. (2017). Gray matter volume differences in the left caudate nucleus of peoplewho stutter. Brain and Language, 164, 9–15.

Stager, S. V., Jeffries, K. J., & Braun, A. R. (2004). Common features of fluency-evoking conditions studied in stuttering participants and controls: An H215Opet study. Journal of Fluency Disorders, 28(4), 319–336.

Tahaei, A. A., Ashayeri, H., Pourbakht, A., & Kamali, M. (2014). Speech evoked auditory brainstem response in stuttering. Scientifica.Tellis, G. M., Vitale, C., & Murgallis, T. (2015). Near infrared spectroscopy (NIRS): A pilot study to measure hemoglobin concentration changes in the brains

of persons who stutter and typically fluent speakers. Procedia Social and Behavioral Sciences, 193, 261–265.Till, J. A., Reich, A., Dickey, S., & Seiber, J. (1983). Phonatory and manual reaction times of stuttering and nonstuttering children Journal of Speech.

Language and Hearing Research, 26(2), 171–180.Toyomura, A., Fujii, T., & Kuriki, S. (2011). Effect of external auditory pacing on the neural activity of stuttering speakers. Neuroimage, 57(4), 1507–1516.Toyomura, A., Fujii, T., & Kuriki, S. (2015). Effect of an 8-week practice of externally triggered speech on basal ganglia activity of stuttering and fluent

speakers. Neuroimage, 109, 458–468.Usler, E., & Weber-Fox, C. (2015). Neurodevelopment for syntactic processing distinguishes childhood stuttering recovery versus persistence. Journal of

Neurodevelopmental Disorders, 7(1), 1–22.Van Borsel, J., Achten, E., Santens, P., Lahorte, P., & Voet, T. (2003). fMRI of developmental stuttering: A pilot study. Brain and Language, 85(3), 369–376.Vanhoutte, S., Santens, P., Cosyns, M., van Mierlo, P., Batens, K., Corthals, P., et al. (2015). Increased motor preparation activity during fluent single word

production in DS: A correlate for stuttering frequency and severity. Neuropsychologia, 75, 1–10.Vanhoutte, S., Cosyns, M., van Mierlo, P., Batens, K., Corthals, P., De Letter, M., et al. (2016). When will a stuttering moment occur?: The determining role of

speech motor preparation. Neuropsychologia, 86, 93–102.Vassal, F., Boutet, C., Lemaire, J.-J., & Nuti, C. (2014). New insights into the functional significance of the frontal aslant tract: An anatomo-functional study

using intraoperative electrical stimulations combined with diffusion tensor imaging-based fiber tracking. British Journal of Neurosurgery, 28(5),685–687.


























































































































































































































































































































































































































































































































































































































































































































































































W

W

WW

W

W

W

W

WW

XYZ

APpta

OtUNr

KAmaa

P(Sco


alla, P., Mayer, D., Deecke, L., & Thurner, S. (2004). The lack of focused anticipation of verbal information in stutterers: A magnetoencephalographicstudy. Neuroimage, 22(3), 1321–1327.

atkins, K. E., Smith, S. M., Davis, S., & Howell, P. (2008). Structural and functional abnormalities of the motor system in developmental stuttering. Brain,131(1), 50–59.

eber-Fox, C. (2001). Neural systems for sentence processing in stuttering. Journal of Speech, Language, and Hearing Research, 44(4), 814–825.eber-Fox, C., & Hampton, A. (2008). Stuttering and natural speech processing of semantic and syntactic constraints on verbs. Journal of Speech, Language,

and Hearing Research, 51(5), 1058–1071.eber-Fox, C., Spencer, R. M., Spruill, J. E., & Smith, A. (2004). Phonologic processing in adults who stutter: Electrophysiological and behavioral evidence.

Journal of Speech, Language, and Hearing Research, 47(6), 1244–1258.eber-Fox, C., Spruill, J. E., Spencer, R., & Smith, A. (2008). Atypical neural functions underlying phonological processing and silent rehearsal in children

who stutter. Developmental Science, 11(2), 321–337.eber-Fox, C., Wray, A. H., & Arnold, H. (2013). Early childhood stuttering and electrophysiological indices of language processing. Journal of Fluency

Disorders, 38(2), 206–221.u, J. C., Maguire, G., Riley, G., Fallon, J., LaCasse, L., Chin, S., et al. (1995). A positron emission tomography [18F] deoxyglucose study of developmental

stuttering. Neuroreport, 6(3), 501–505.u, J. C., Maguire, G., Riley, G., Lee, A., Keator, D., Tang, C., et al. (1997). Increased dopamine activity associated with stuttering. Neuroreport, 8(3), 767–770.ymbs, N. F., Ingham, R. J., Ingham, J. C., Paolini, K. E., & Grafton, S. T. (2013). Individual differences in neural regions functionally related to real and

imagined stuttering. Brain and Language, 124(2), 153–164.uan, Y., Meng, C., Yang, Y., Zhu, C., Wang, L., Yan, Q., et al. (2012). Resting-state brain activity in adult males who stutter. PLOS ONE, 7(1), e30570.ang, Y., Jia, F., Siok, W. T., & Tan, L. H. (2016). Altered functional connectivity in persistent developmental stuttering. Scientific Reports, 6.engin, H. (2011). Neuroanatomical differences in adults and children who stutter: A voxel-based morphometry study. (Doctoral dissertation). University of

Illinois at Urbana-Champaign.

ndrew C. Etchell received his PhD in Human Cognition and Brain Science from Macquarie University in 2015 under the supervision of Associaterofessor Paul Sowman. In this position, he used magnetoencephalography to investigate rhythm processing in children who stutter. He is currently aostdoctoral fellow at the University of Michigan (Ann Arbor United States) working with Associate. Prof.essor Soo-Eun Chang and Dr. Ho Ming Chowo unravel predictors of persistence and recovery from stuttering based on longitudinal dataset collected over the last 5 years. He is also interested inpplying non invasive brain stimulation to the study of stuttering.

ren Civier received his PhD from Boston University. Under the supervision of Prof. Frank Guenther, he used computational modeling to investigatehe neural substrates of stuttering and induced fluency. In his postdoctoral fellowship in Dr. Michal Ben-Shachar’s Neurolinguistics lab at Bar-Ilanniversity, he studied white matter and functional anomalies in stuttering. Currently, Dr. Civier is a postdoctoral fellow in the Florey Institute ofeuroscience and Mental Health, where he develops tools to study structural and functional connectivity in the human brain. He has utilized multiple

esearch methodologies in the past, including dMRI, fMRI, MEG, Electromagnetometry and camera-based motion tracking.

irrie J. Ballard is Professor of Speech Motor Control and Disorders and Australian Research Council Future Fellow at the University of Sydney,ustralia, and holds an honorary appointment with Neuroscience Research Australia. She has published over 80 research articles and book chapters onechanisms, diagnosis and treatment of acquired and developmental speech − language disorders, and on development of speech-controlled mobile

pplications and games for enhancing engagement in speech therapy. She iscurrently Editor of the International Journal of Speech-Language Pathologynd Chair of the Academy of Neurologic Communication Disorders and Sciences committee on evidence-based practice guidelines for apraxia of speech.

aul Sowman graduated with a PhD in Physiology from the University of Adelaide in 2008. Since then he has held fellowships from both the NHMRCNational Health and Medical Research Council) and ARC (Australian Research Council) whilst in his current position within the Department of Cognitivecience at Macquarie University in Sydney. His research uses Magnetoencephalography and non-invasive brain stimulation methods to understandognitive processes underpinning normal and abnormal speech production. He has a particular interest in inhibitory control processes and in disordersf inhibitory control that affect speech such as stuttering and Tourette syndrome.








































































































































































































































Documents

Journal of Fluency Disorders · Journal of Fluency Disorders 55 (2018) 6–45 Contents lists available at ScienceDirect Journal of Fluency Disorders Review Article A systematic literature