See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/327383037

Phonological working memory in developmental stuttering: Potential

Insights from the neurobiology of language and cognition

Article  in  Journal of Fluency Disorders · September 2018

DOI: 10.1016/j.jfludis.2018.08.006

CITATIONS

2READS

358

4 authors, including:

Andrew Bowers

University of Arkansas

16 PUBLICATIONS   173 CITATIONS   

SEE PROFILE

Lisa M. Bowers

University of Arkansas

16 PUBLICATIONS   95 CITATIONS   

SEE PROFILE

Daniel Hudock

Idaho State University

16 PUBLICATIONS   59 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Andrew Bowers on 01 November 2018.

The user has requested enhancement of the downloaded file.

Contents lists available at ScienceDirect

Journal of Fluency Disorders

journal homepage: www.elsevier.com/locate/jfludis

Phonological working memory in developmental stuttering:Potential insights from the neurobiology of language and cognitionAndrew Bowersa,⁎, Lisa M. Bowersa, Daniel Hudockb, Heather L. Ramsdell-Hudockb

aUniversity of Arkansas, Epley Center for Health Professions, 606 N. Razorback Road, Fayetteville, AR 72701, United Statesb Idaho State University, 650 Memorial Dr. Bldg. 68, Pocatello, ID 83201, United States

A R T I C L E I N F O

Keywords:Phonological working memorySensorimotor integrationExecutive functionNonword repetitionDual-taskStuttering

A B S T R A C T

The current review examines how neurobiological models of language and cognition could shedlight on the role of phonological working memory (PWM) in developmental stuttering (DS).Toward that aim, we review Baddeley’s influential multicomponent model of PWM and evidencefor load-dependent differences between children and adults who stutter and typically fluentspeakers in nonword repetition and dual-task paradigms. We suggest that, while nonword re-petition and dual-task findings implicate processes related to PWM, it is unclear from behavioralstudies alone what mechanisms are involved. To address how PWM could be related to speechoutput in DS, a third section reviews neurobiological models of language proposing that PWM isan emergent property of cyclic sensory and motor buffers in the dorsal stream critical for speechproduction. We propose that anomalous sensorimotor timing could potentially interrupt bothfluent speech in DS and the emergent properties of PWM. To further address the role of attentionand executive function in PWM and DS, we also review neurobiological models proposing thatprefrontal cortex (PFC) and basal ganglia (BG) function to facilitate working memory underdistracting conditions and neuroimaging evidence implicating the PFC and BG in stuttering.Finally, we argue that cognitive-behavioral differences in nonword repetition and dual-tasks areconsistent with the involvement of neurocognitive networks related to executive function andsensorimotor integration in PWM. We suggest progress in understanding the relationship be-tween stuttering and PWM may be accomplished using high-temporal resolution electromagneticexperimental approaches.

https://doi.org/10.1016/j.jfludis.2018.08.006Received 29 September 2017; Received in revised form 30 July 2018; Accepted 27 August 2018

Abbreviations: DS, developmental stuttering; CWS, children who stutter; AWS, adults who stutter; PWM, phonological working memory; PFC,prefrontal cortex; BG, basal ganglia; LTM, long term memory; PL, phonological loop; VS, visuospatial sketchpad; CE, central executive; EB, episodicbuffer; TFS, typically fluent speakers; PWS, people who stutter; CWNS, children who do not stutter; ERP, event-related potential; BGTC, basalganglia thalamo cortical; IFG, inferior frontal gyrus; PMC, premotor cortex; PT, planum temporale; Spt, sylvian parieto-temporal junction; STG,superior temporal gyrus; STS, superior temporal sulcus; MTG, middle temporal gyrus; SFC, state feedback control; CA, conduction aphasia; BA,Brodmann’s Area; IPL, inferior parietal lobule; SPL, superior parietal lobule; PBWM, prefrontal cortex basal ganglia working memory; SMA, sup-plemental motor area; ACC, anterior cingulate cortex; DIVA, directions in auditory space to velocities in articulator space; GODIVA, Gradient orderDIVA model; vMC, ventral motor cortex; fMRI, functional magnetic resonance imaging; SMG, supramarginal gyrus; DMN, default mode network;DAN, dorsal attention network; VAN, ventral attention network; FPN, fronto-parietal network; GMV, gray matter volume; MFG, medial frontalgyrus; FA, fractional anisotropy; PET, positron emission tomography; ALE, activation likelihood estimation; A1, auditory cortex; OFC, orbitofrontalcortex; SSI-4, Stuttering Severity Instrument – 4th Edition; EEG, electroencephalography; MEG, magnetoencephalagraphy; SMS, Speech Motor Skill

⁎ Corresponding author.E-mail addresses: albowers@uark.edu (A. Bowers), lmbowers@uark.edu (L.M. Bowers), hudock@isu.edu (D. Hudock),

ramsdell@isu.edu (H.L. Ramsdell-Hudock).

Journal of Fluency Disorders 58 (2018) 94–117

Available online 01 September 20180094-730X/ Published by Elsevier Inc.

T

1. Introduction

Developmental stuttering (DS) is a disorder of speech fluency associated with a high recovery rate in preschool-age children whostutter (CWS) (> 80%) and significant psychosocial and emotional consequences for those who persist into later childhood andadulthood. Multifactorial, dynamic models propose that DS is primarily a disorder affecting the sensorimotor control of speech that isinfluenced by other psycholinguistic and cognitive-emotional factors (Smith, 1999; Smith & Weber, 2017; Starkweather, 2002; vanLieshout, Hulstijn, & Peters, 2004). From this point of view, interacting general cognitive capacities such as working memory,attention, and executive function play some as yet unclear role in the moment-to-moment variability and phenotypic expression ofstuttering in both CWS (Chang et al., 2017; Pelczarski & Yaruss, 2016; Smith & Weber, 2017; Spencer & Weber-Fox, 2014) and adultswho stutter (AWS) (Alm, 2004, Alm & Risberg, 2007; Kell et al., 2017; van Leishout et al., 2004). In particular, cognitive capacitiesknown to support phonological working memory (PWM), including phonological storage and the articulatory rehearsal of speechsounds, have been implicated as a factor in preschool CWS that predicts subsequent recovery or persistence (Mohan & Weber, 2015;Spencer & Weber-Fox, 2014). In CWS, PWM capacity may also be related to disturbances in cognitive-linguistic processing and tovariability in the expression of stuttering in contexts that load other cognitive processes supporting it, including attentional controland executive function both in CWS (Bajaj, 2007; Eggers et al., 2012; Jones et al., 2017; Usler & Weber-Fox, 2015) and AWS (Alm,2004; Alm & Risberg, 2007; Bajaj, 2007; Bosshardt, 2006). As such, a better understanding of the cognitive processes underlyingPWM has the potential to address long held questions about the nature of stuttering.

Despite the potential significance of PWM, studies using tasks thought to load PWM in CWS and AWS have produced somewhatmixed results, and the underlying mechanisms mediating recently reported differences remain unclear (Anderson & Wagovich, 2010;Anderson et al., 2006; Byrd et al., 2015; Byrd et al., 2012; Smith et al., 2010). Perhaps as a consequence, while both CWS and AWSpresent with functional and structural anomalies in both sensorimotor (Buchsbaum & D’Esposito, 2008; Buchsbaum et al., 2011;Hickok et al., 2011; Jacquemot & Scott, 2006) and neurocognitive networks implicated in PWM (Chang et al., 2017; Kell et al., 2017;Sitek et al., 2016), there have been no attempts to integrate existing cognitive-behavioral and neuroimaging findings with neuro-biological models of language and cognition. As such, the aim of the current review is to examine how neurobiological modelsincorporating PWM in typical speech and language processing (Hickok & Poeppel, 2007; Jacquemot & Scott, 2006) and executivefunction (D’Esposito, 2007; Hazy et al., 2007) may shed light on underlying mechanisms mediating recent cognitive-behavioralstudies implicating lower PWM capacity in CWS (Anderson & Wagovich, 2010; Anderson et al., 2006; Spencer & Weber-Fox, 2014)and AWS (Byrd et al., 2015; Byrd et al., 2012; Smith et al., 2010). We argue that interpreting cognitive-behavioral findings in thecontext of neurobiological models may strengthen the relationship between underlying mechanisms affecting fluent speech pro-duction those underlying differences in PWM in both CWS (Spencer & Weber-Fox, 2014) and AWS (Byrd et al., 2015; Byrd et al.,2012).

Toward bridging the gap between cognitive-behavioral findings in DS and the neurobiology of PWM, Section 2 reviews therelationship between DS and PWM capacity in the context of Baddeley’s influential multicomponent working memory model andbehavioral findings from nonword repetition and dual-task paradigms that implicate aspects of the executive function and phono-logical storage in CWS and AWS. In Section 3, we review neurobiological models proposing that PWM is an emergent property ofcyclic input and output buffers in the dorsal sensorimotor stream (Buchsbaum & D’Esposito, 2008; Hickok et al., 2011; Jacquemot &Scott, 2006) and models proposing that interactions between the prefrontal cortex (PFC) and basal ganglia (BG) mediate workingmemory in attentionally demanding conditions (D’Esposito, 2007; Hazy et al., 2007). In support of differences in networks supportingPWM in DS, we review neuroimaging evidence implicating large-scale networks in both CWS and AWS, including PFC, BG, and dorsalstream sensorimotor regions. In Section 4, we posit that mistimed or ‘noisy’ dorsal stream sensorimotor integration previouslyproposed to cause stutter-typical disfluencies (i.e., part-word repetitions, prolongations, and blocks) may also account for load-dependent differences on nonword repetition tasks in CWS and AWS. We further suggest that increased executive control of speechmay play a modulatory role in regulating dorsal stream timing, as experience with stuttering accumulates over the course of speechand language development. Lastly, in Section 5, we propose that progress in understanding the cognitive processes underlying PWMin DS may be accomplished using noninvasive, high temporal resolution time-frequency methods in nonword repetition and dual-taskparadigms.

2. Phonological working memory in DS

Much of the most recent work investigating aspects of PWM in DS references Baddeley’s working memory model as an ex-planatory framework and thus a brief review of the model is worthwhile (Anderson & Wagovich, 2010; Anderson et al., 2006; Bajaj,2007; Bosshardt, 2006; Byrd et al., 2015; Spencer & Weber-Fox, 2014). Currently, the model consists of four primary componentsproposed to interact with long-term memory (LTM), which include the phonological loop (PL), visuospatial sketchpad (VS), centralexecutive (CE), and episodic buffer (EB) (Baddeley, 2003, 2012). The PL is arguably the component most relevant to the acquisitionof speech and language and is composed of two subcomponents, one component that provides temporary storage for phonologicalinput known as the phonological store and a second to refresh the first via subvocal articulatory rehearsal. Functionally, the PL isproposed to support new phonological learning, facilitate the acquisition of new vocabulary, support second-language acquisition,and potentially to influence action selection via subvocal rehearsal (Adams & Gathercole, 1995; Alloway et al., 2005; Baddeley,2012). The VS functions to integrate spatial, visual, and kinesthetic information into a single representation for temporary storage.The CE component provides access to the other components and is broadly tasked with the control of attention to items to be held inworking memory. The latest and least studied addition to the model is the EB, a process proposed to have the capacity to store

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

95

information in ‘chunks’ allowing it to exceed the capacity of the other components of working memory by interacting with LTM. Insection 2.1, we review evidence from nonword repetition tasks implicating aspects of Baddely’s PL in CWS and AWS.

2.1. A review of the evidence: nonword repetition in DS

A growing corpus of experimental and descriptive evidence implicates PWM differences in both CWS and AWS as compared totypically fluent speakers (TFS). Much of the evidence linking PWM with stuttering has come from what are known as nonwordrepetition paradigms. Nonword repetition involves repeating words that conform to phonological rules in the language but have nomeaning (i.e., semantic content). In Baddeley’s model, such a task is thought to tap the storage component of the PL without relianceon LTM for language (Gathercole et al., 1994). Following that reasoning, differences between people who stutter (PWS) and TFS onnonword repetition tasks in the absence of differences on verbal repetition tasks would provide evidence favoring differences in PWMas distinct from LTM for semantic content. Findings from nonword repetition tasks are of importance because nonword repetitioncapacity in preschool children, as distinct from verbal working memory tasks, has been reported to predict subsequent recovery orpersistence (Spencer & Weber-Fox, 2014). In particular, it has been suggested that articulatory rehearsal and phonological storagemay interact with other psycholinguistic capacities to influence persistence or recovery (Spencer & Weber-Fox, 2014). As such, in thesections that follow, we first review studies reporting both significant group differences and a lack of differences in nonword re-petition accuracy and subsequently discuss factors that may have led to nonsignificant findings in some studies. We suggest that theweight of the evidence implicates subtle, load-dependent differences in PWS from the preschool years up through adulthood.

2.1.1. Nonword repetition findings in preschool-age CWSEarly studies investigating nonword repetition in CWS provided somewhat mixed results with two studies in CWS (age range 2–8

years) reporting significantly lower behavioral performance on children’s test of nonword repetition (Gathercole et al., 1994), acommon measure of nonword repetition (Anderson et al., 2006; Hakim & Ratner, 2004). In a follow-up study, Anderson andWagovich (2010) reported that preschool CWS (3–5 years) who participated in both a picture naming task and a nonword repetitiontask were significantly lower on the nonword repetition task only compared to children who do not stutter (CWNS). Interestingly, thetwo groups did not differ on a common measure of attention via parental report. However, within the group of CWS a measure of‘attentional focusing’ was positively correlated with nonword repetition performance. More recently, Smith, Goffman, Sasisekaran,and Weber-Fox (2012) did not detect group differences on a nonword repetition task in a group of preschool-age CWS (4–6 years). Inthat study, differences were only apparent for CWS with concomitant speech-sound and/or language disorders, suggesting that otherarticulatory and cognitive-linguistic capacities may interact to influence performance on nonword repetition tasks. Further, althoughCWS without speech-sound/language disorders did not differ on a measure of nonword repetition accuracy, they did show highervariability on indices of oral motor coordination (i.e., lip aperture variability), suggesting that motor control of speech varied withnonword repetition load.

In contrast to Smith et al. (2012), Spencer and Weber-Fox (2014) did find significant differences in nonword repetition perfor-mance between a preschool group of CWS and CWNS (age range 3–5 years) with or without concomitant language and phonologicaldisorders. Additionally, findings revealed that nonword repetition performance was significantly lower for CWS who persisted re-lative to those who recovered. Importantly, neither measures of expressive/receptive language nor measures of working memory forwords with semantic content were associated with recovery or persistence, suggesting a disassociation between verbal workingmemory for words with semantic content and PWM. Most recently, Pelczarski and Yaruss (2016) reported statistically significant yetsubclinical differences on a norm-referenced test of nonword repetition in CWS (age range 5–6 years). Consistent with previousstudies, no differences were reported on a digit-span task, suggesting similar behavioral performance for words with semanticcontent. Taken together, the majority of nonword repetition studies in preschool CWS (5/6) suggest subtle, subclinical differencesthat may also be related to or interact with articulation, phonology, and attentional capacities. Further, studies that have alsoinvestigated aspects of verbal working memory via digit span or other memory tasks for words with semantic content also suggestthat reduced capacities for nonword sequences are not related verbal working memory capacity (Pelczarski & Yaruss, 2016) and donot appear to predict recovery or persistence (Spencer & Weber-Fox, 2014). Relevant details of the studies cited above, includingsample sizes, maximum syllable length of nonword stimuli, and effect sizes when reported are summarized in Appendix A (Table A1).

2.1.2. Nonword repetition findings in school-age CWSRelative to findings in preschool CWS, fewer published studies have investigated differences in school-age CWS and CWNS (age

range 5–15 years). However, unlike studies of preschool CWS, only 3/6 studies report significantly lower performance in the school-age CWS group when compared to CWNS. In the earliest report of nonword repetition in school-age CWS, Seery, Watkins, Ambrose,and Throneburg (2006) reported no significant main effect between CWS and CWNS but did report a significant interaction betweengroup and condition (i.e., nonword syllable length) at 5 syllable nonword length. Bakhtiar, Abadm, and Panahi (2007) found nosignificant differences in nonword repetition accuracy between 12 CWS and 12 CWNS (age range 5–7 years) using nonwords with amaximum length of 3 syllables. A retrospective study of 5 persistent CWS and 14 recovered CWS also reported no statisticallysignificant group differences on a nonword repetition task (Chon & Ambrose, 2007). Later, Weber-Fox, Spruill, Spencer, and Smith(2008) reported that nonword repetition accuracy in a group of 10 CWS was similar to a group of 30 CWNS in the same age range. Incontrast to previous studies, Oyoun, El Dessouky, Sahar, and Fawzy (2010) did find significant differences in nonword repetitionaccuracy between 30 CWS and 30 matched CWNS, suggesting that previous studies using smaller sample sizes may not have detecteda difference due to a lack of statistical power. A more recent study conducted by Sasisekaran and Byrd (2013) in CWS and CWNS ages

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

96

7–15, found no significant difference between CWS and CWNS in overall nonword repetition accuracy and further found no dif-ferences in a nonword repetition condition requiring the omission of a target phoneme (i.e., phoneme elision). However, furthersubdivision of the CWS and CWNS into younger (7–11) and older (11–15) groups revealed a difference between the younger andolder CWS that was not present in the CWNS, suggesting that age may be a particularly important factor in determining whether ornot between group differences are detectable in school-age children. In summary, while most studies of school-age CWS have notreported between subject differences, those studies also used nonword stimuli of shorter syllable length than studies of preschoolCWS. Further, the two studies that did detect overall lower performance in CWS, either employed a 5 syllable length condition (Seeryet al., 2006) or employed a larger sample size (Oyoun et al., 2010). As such, while it is as yet unclear whether school-age CWS presentwith robust differences in nonword repetition accuracy, nonword repetition load, age, and sample size appear to be important factorsin determining whether a difference is detectable. A table summarizing findings from the studies cited above is presented in AppendixA (Table A2).

2.1.3. Nonword repetition findings in AWSAlthough most studies of school-age children have not reported significant differences in nonword repetition accuracy, most (5/7)

studies in AWS have reported lower behavioral performance. An early preliminary study reported significant differences in nonwordrepetition performance that was consistent with significantly lower capacity in AWS relative to TFS (Ludlow et al., 1997). In akinematic study of nonword repetition, Namasivayam and van Lieshout (2008) reported that AWS showed reduced practice effectsreflected in higher movement variability over several days when compared to controls. In another study of AWS who were cate-gorized with mild stuttering severity, Smith et al. (2010) found no differences in accuracy on nonword repetition but did report groupdifferences on a measure of inter-articulator coordination (i.e., labial aperture variability) with increasing length (1–4 syllables) andphonological complexity of nonwords. While earlier studies had used shorter nonword lengths, Byrd et al. (2012) investigated groupdifferences at higher nonword loads (i.e., 7 syllables). A significant group by syllable interaction was reported at the 7 syllable lengthnonwords only. In a subsequent study using a group of 9 AWS and matched controls, Sasisekaran (2013) found no behavioraldifferences on the Nonword Repetition Test (Dollaghan & Campbell, 1998) for 1–4 syllables but did find significant group differencesusing a measure of lip aperture variability on a nonword reading task in which word lengths were varied between 6 and 11 syllables.Subsequently, Sasisekaran and Weisberg (2014) reported that AWS showed an increasing likelihood of nonword repetition errorswith increasing length and phonological complexity of nonwords (i.e., 6 syllables), suggesting that the load induced by nonwordstimuli may be a critical factor in detecting between group differences in AWS.

Building on findings from Byrd et al. (2012), Byrd et al. (2015) reported that nonword repetition performance differed betweenAWS and TFS on 7 syllable nonwords but no differences were reported on 4 syllable nonwords. Additionally, decreases in perfor-mance were observed on a phoneme elision condition for 4 and 7 syllable nonwords in which participants silently identified thenonword with a target phoneme omitted, strongly implicating phonological memory traces in the observed differences. Finally, arecent report also demonstrated that AWS were less accurate at repeating low-frequency iambic stress patterns in a 2-syllablenonword repetition task relative to higher-frequency trochaic stress patterns in the absence of auditory-orthographic cues, suggestingthat even at shorter syllable lengths phonological load manipulations may allow for the detection of group differences in behavioralaccuracy (Coalson & Byrd, 2017). Thus, collectively findings manipulating nonword repetition load suggest that AWS present withlower performance on nonword repetition tasks when PWM is heavily loaded even into adulthood. A table summarizing findings fromthe studies cited above are presented in Appendix A (Table A3).

2.1.4. Implications of nonword repetition findings in CWS and AWSIn summary, the weight of the evidence implicates nonword repetition differences from the onset of stuttering in the preschool

years up through adulthood. First, while the majority of studies in preschool CWS have detected a group difference, some studies thatdid not find significance may not have controlled for a number of influential factors. In the only study of preschool CWS reporting nooverall group difference in accuracy, subgroups with language disorders and phonological disorders did present with significantlylower behavioral performance (Smith et al., 2012). By contrast, two other subsequent studies found group differences in preschoolCWS even without clinically significant language or phonological disorders (Pelczarski & Yaruss, 2016; Spencer & Weber-Fox, 2014).As noted by Pelczarski and Yaruss (2016), age, language ability, and nonword stimulus factors such as length and complexity may nothave been adequately examined in previous studies. Further, most of the studies have included preschool children that may sub-sequently recover, a factor that is likely to lead to higher overall behavioral performance in CWS groups (Spencer & Weber-Fox,2014). Given the weightof the evidence and factors that may have affected findings in previous studies reporting no group differ-ences, nonword repetition capacity is strongly implicated in preschool-age CWS (Table A4).

In school-age children who stutter, while only 3 of 6 studies detected group differences, the studies that did employed largersample sizes (Oyoun et al., 2010), separated CWS into younger and older groups (Sasisekaran & Byrd, 2013), or detected differencesonly at longer nonword lengths (Seery et al., 2006). These findings suggest that school-age CWS may present with higher PWMcapacities than preschool CWS and show differences primarily under high nonword repetition loads. That interpretation is consistentwith available evidence in AWS suggesting that the likelihood of detecting differences in behavioral performance is heavily de-pendent on factors affecting nonword repetition load (Byrd et al., 2012; Byrd et al., 2015; Coalson & Byrd, 2017). Further, kinematicdifferences, when measured, often show robust differences even at lower loads in CWS and AWS (Sasisekaran, 2013; Sasisekaran &Weisberg, 2014; Smith et al., 2010). Differences in kinematics in fluent nonword repetition have been taken as evidence of inefficient,immature, or compromised motor-control processes (Smith et al., 2010) or as compensatory motor control processes in AWS that arerelated to compromised speech motor skill acquisition (Namasivayam & van Lieshout, 2008). It is also important to consider that,

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

97

because differences in speech kinematics are present during perceptually fluent nonword repetition in AWS, differences coordination,amplitude, and slower adaptation over trials may also be interpreted as momentary compensation for instabilities in speech motorcontrol as a function of nonword repetition load.

Despite variability in nonword repetition findings across studies, there are a number of common themes. Interestingly, onecommon finding across recent studies is that preschool CWS tend to present with lower accuracy on nonword repetition tasks but donot present with differences on verbal repetition tasks with semantic content, suggesting a disassociation between verbal workingmemory and PWM early in DS (Pelczarski & Yaruss, 2016; Spencer & Weber-Fox, 2014). One possible interpretation is that thedisassociation may be related to compensatory processing in Baddeley’s EB for words with semantic content that is not available fornonwords, suggesting that LTM for language may play a compensatory role (Pelczarski & Yaruss, 2016). As noted in Spencer andWeber-Fox (2014), successful nonword repetition requires at least intact auditory-phonological processing, phonological analysis andstorage, speech planning, and ultimately execution. Those authors suggest that the amalgam of those processing demands may indexa fundamental process predictive of recovery and persistence. Finally, given the results of Anderson and Wagovich (2010), it alsoseems likely that some aspects of attention also interact with those processes in preschool CWS, suggesting that aspects of the CE mayalso be involved in observed differences. Further, differences in school-age CWS and AWS have only been observed for highernonword repetition loads, suggesting that aspects of executive function/attention may play an important role in observed differences.Differences in each process, including the processing of speech sounds (Neef et al., 2012; Saltuklaroglu et al., 2017), phonologicalencoding (Kolk & Postma, 1997), executive function/attention (Alm, 2004; Alm & Risberg, 2007; Eggers et al., 2012), and differencesin linguistic planning and execution (Howell & Au-Yeung, 2002; Kleinow & Smith, 2000), have been proposed to play some role in DS(Bloodstein & Bernstein-Ratner, 2008). As such, while nonword repetition appears to index an important process related to persis-tence/recovery, the underlying mechanisms relating nonword repetition to stuttering are unclear.

2.2. A review of the evidence: dual-task performance in DS

Given that the processes of attention and executive function likely interact with the function of the PL, consideration of thosecapacities is critical when discussing the potential role of PWM in DS. Behavioral evidence for the function of the CE in Baddeley’smodel is often examined through the lens of dual-task experimental paradigms (Baddeley, 2012). In these dual-task experiments,participants complete mental operations such as counting backwards while their performance is evaluated on some target task.Baddeley and colleagues suggest that such tasks interrupt the automaticity of the target task, placing load on aspects of cognitionsubserved by the CE component (Baddeley, 2012). Following that reasoning, dual-task experiments could potentially provide awindow into the link between PWM and aspects of cognition relegated to the CE in PWS. Dual-task studies in AWS have investigatedaspects of target and secondary task performance as a measure of processing load along with effects on measures of stuttering. Assuch, in the sections below we review evidence for differences in behavioral performance in dual-tasks along with evidence sug-gesting effects on stutter-typical disfluencies.

2.2.1. Dual-task performance in school-age CWS and AWSWhile a number of early dual-task studies investigated the performance of distracting tasks on aspects of speech and language

production in school-age CWS and AWS, those studies are difficult to interpret because it is unclear what cognitive processes wereengaged by secondary tasks. Early studies employed secondary tasks such as flickering lights, tracking a moving dot pattern, fingertapping, gross motor activity, and tracking line patterns while CWS and AWS were engaged in various speech production tasks(Arends et al., 1988; Brutten & Trotter, 1985; Brutten & Trotter, 1986). Two of the studies were conducted with school-age CWS andteenage CWS (age range 6–17 years) and reported somewhat lower finger-tapping rates as verbal task demands increased (Brutten &Trotter, 1985; Brutten & Trotter, 1986). Subsequent studies in AWS have employed secondary tasks that were more readily inter-pretable. In one early investigation, Caruso and Chodzko-Zajko (1994) reported differences in the accuracy of a speaking (speechresponse latency and rate) while AWS and matched TFS performed the Stroop color word task. More recently, Bosshardt and col-leagues conducted a series of studies in which AWS and matched TFS performed various speaking tasks along with a number ofsecondary tasks including mental calculation, rhyme and semantic category decisions, and memorizing phonologically similar anddissimilar words (Bosshardt, 2002; De Nil & Bosshardt, 2000). In the speaking tasks, group differences or interactions between taskand group showed differences in measures of inhalation rate or word duration. In sentence level speaking tasks, decreases in thenumber of propositions were also found in dual-task conditions relative to single task conditions (Bosshardt, 2002), suggesting thatAWS responded to task demands by reducing speech/linguistic output relative to TFS. In a review of those studies, Bosshardt (2006)concluded that AWS compensate for the effects of processing load in dual-tasks by reducing the ‘conceptual work’ of speech andlanguage production in an attempt to decrease stuttering.

More recently, Smits-Bandstra and De Nil (2009) compared a group of AWS and TFS on repeated practice of ten syllable nonsensesequences while performing a color recognition task. AWS demonstrated slowed speech sequence initiation over practice compared tothe TFS group and decreases in syllable sequence reading accuracy. Jones, Fox, and Jacewicz (2012) used a dual-task in whichparticipants made rhyme judgements while performing a letter recall task. While rhyme judgements did not show group differences,reaction time, and letter recall decreases in accuracy for higher loads (i.e., five letters) were observed. The differences in responsetime as opposed to accuracy in rhyme judgements suggest that AWS may need more time to process phonological information.Consistent with nonword repetition findings, apparent behavioral differences were only detected when phonological processing washeavily loaded. In accord with behavioral findings, a recent electrophysiological study also reported group differences in event-related potentials (ERPs) in a dual-task wherein participants performed a tone discrimination task while at the same time naming

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

98

pictures (Maxfield et al., 2016). Whereas no differences in ERPs to tones were reported in the simple tone judgement task, ERPs werereduced in the AWS group during the dual-tasks (i.e., phonological competition and lexico-semantic conditions). Most recently,Eichorn, Marton, Schwartz, and Melara (2016) reported no group differences in behavioral performance on a speaking task (narrativeretell) performed along with a range of tasks designed to load working memory/attention, including a digit span and a spatialworking memory task. While no differences in behavioral performance were found, the AWS group did reduce the number of syllablesproduced in the dual-tasks, suggesting that AWS adjusted speech and language to accommodate processing load. Taken together, themajority of findings support the notion that AWS make subtle adjustments to speech and language production to accommodateprocessing limitations in dual-task compared to single task conditions.

2.2.2. Dual-task effects on stutteringWhile many dual-task studies have reported subtle adjustments in speech or language, a smaller subset have also reported effects

on stuttering in dual-tasks relative to speaking tasks alone. Early studies employing a range of secondary tasks largely unrelated tospeech production or linguistic processing, found no effect on fluency or more often decreases in stuttering under dual-task conditions(Bajaj, 2007). By contrast, in one early study Caruso and Chodzko-Zajko (1994) reported increases in measures of stuttering in theStroop color-word task. Bosshardt and colleagues reported a complex relationship between dual-task conditions and measures ofstuttered speech in which secondary tasks that have minimal overlap with speech and language (e.g., mental calculation) result inreductions in stuttering (Bosshardt, 1999), while tasks that have ostensibly higher overlap with speech and language (e.g., mem-orizing phonologically similar or dissimilar words) are associated with increases in stuttering (Bosshardt, 2002). Other conditions inwhich AWS reduce speech and language output in a sentence generation task along with a secondary task (i.e., rhyme and categorydecisions) were associated with no significant change in measures of stuttering in dual relative to single task conditions, presumablybecause language output was reduced (Bosshardt, 2002). In contrast with those findings, recent evidence has also demonstrateddecreases in measures of stuttering in dual-task relative to single task conditions. Vasic and Wijnen (2005) reported that stutteredspeech reliably decreased in a dual-task in which AWS played video games while speaking. Similarly, Eichorn et al. (2016) alsoreported decreases in measures of stuttering under dual-task digit-span and spatial-working memory tasks. Taken as a whole, theavailable evidence suggests that the cognitive processes loaded in dual-task conditions may tap into a cognitive resource that isrelated to both increases (Bosshardt, 2002; Caruso & Chodzko-Zajko, 1994) and decreases in stuttering (Eichorn et al., 2016; Vasic &Wijnen, 2005) depending on the tasks or even compensatory strategies employed by AWS to reduce task demands (Bosshardt, 2002).What is as yet unclear from dual-task studies is what specific conditions or mechanisms are required to exacerbate or amelioratestuttering.

2.2.3. Implications of dual-task findingsCollectively, findings in AWS implicate behavioral adjustments to speech and language output in dual relative to single task

conditions. Despite findings suggesting dual-task adjustments to speech there are a number of unresolved issues. First, although theevidence implicates differences in behavioral performance between AWS and controls on a range of dual-tasks, there is much lessevidence for differences in CWS (Anderson & Wagovich, 2010; Brutten & Trotter, 1986). Although not a dual-task study, Eichorn,Martin, and Pirutinsky (2017) recently reported that school-age CWS have more difficulty with attentional control in a cognitive taskmeasuring cognitive flexibility, suggesting that aspects of executive function may indeed be ‘weak’ in CWS. Second, due in part to thewide array of secondary tasks employed in dual-task studies, it is difficult to tease apart underlying mechanisms mediating differencesin behavioral performance or stuttering. While the mechanisms are unclear, a number of explanations have been offered. Smits-Bandstra and DeNil (2007) suggested that reduced dual-task performance is reflective of low levels of automaticity mediated bybasal-ganglia-thalamo-cortical (BGTC) loops. In that case, deficits or differences in the BGTC loops of PWS cause decreases in newskill learning that do not become automatic with practice at the same rate as TFS. Under dual-task conditions, attentional load furtherdegrades the process of automation over a number of trials, suggesting a neurophysiological link between attentional resources,automation, and adaptive sensorimotor learning.

In more recent discussions, Maxfield et al. (2016) further propose that attentional resources allocated to atypical linguisticprocessing related to covert behavior in AWS (e.g., word avoidance) may have significant effects on word retrieval, speech-motorplanning, and subsequent execution. Studies reporting greater fluency also make reference to attentional and working memorymechanisms (Eichorn et al., 2016; Vasic & Wijnen, 2005). More specifically, Eichorn et al. (2016) interpret findings of decreasedstuttering in dual-task conditions as consistent with the Matched Filter Hypothesis (Chrysikou et al., 2014). According to that fra-mework, top-down attentional control can be detrimental for tasks that require relatively automatic motor performance (i.e., pro-cedural tasks). According to Eichorn et al. (2016), when working memory/attentional resources are otherwise engaged and top-downcontrol of speech is reduced, more automatic mechanisms (e.g., subcortical mechanisms) facilitate speech production. Thus, theeffects of top-down attentional control on speech in dual-tasks is consistent with mechanisms mediating enhanced fluency in somecontexts (Eichorn et al., 2016) and deleterious effects in others (Bosshardt, 2006; and see Metten et al., 2011), perhaps depending onthe direction of attention or the extent to which dual-tasks share cognitive resources for speech and language.

2.2.4. Implications of nonword repetition and dual-task findingsCollectively, nonword repetition and dual-task findings support the notion that limitations on the cognitive capacities underlying

PWM may be related to stuttering from the preschool years into adulthood. First, the majority of studies in preschool CWS suggest aload dependent vulnerability to cognitive factors also known to load PWM, including phonological analysis, temporary storage, andspeech motor planning prior to nonword repetition. While differences in behavioral accuracy are likely to be subclinical (Pelczarski &

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

99

Yaruss, 2016) and may interact with other cognitive-linguistic capacities (Anderson & Wagovich, 2010; Smith et al., 2012), they arealso related to subsequent recovery and persistence (Spencer & Weber-Fox, 2014). Further, load-dependent differences are presenteven in older school-age CWS and AWS. As such, the behavioral evidence from nonword repetition paradigms suggest subtle, loaddependent limitations on what Baddeley described as the PL in both CWS and AWS relative to TFS. We suggest that this difference inPWM capacity may be of importance because the processing demands required in nonword repetition appear to be related to per-sistence and recovery (Spencer & Weber-Fox, 2014; Mohan & Weber, 2015) and may interact with other psycholinguistic processes(e.g., syntax) over the course of language development to influence the expression of stuttering (Usler & Weber-Fox, 2015),

Second, while it is unclear from nonword repetition studies how PWM may be related to overt stuttered moments, dual-taskstudies suggest that loading aspects of attention in speech production is associated with effects on stuttered speech, especially underconditions in which PWM is loaded (Bosshardt, 2002). However, it also appears that effects on stuttering sensitively depend on studyconditions, with some studies providing evidence of increases in stuttering (Bosshardt, 2006; Caruso & Chodzko-Zajko, 1994) andothers paradoxically reporting decreases in stuttering (Eichorn et al., 2016; Vasic & Winjen, 2005). Stutter-typical disfluencies mayvary based on the internal attention to speech and language production or the extent to which secondary tasks load overlappingprocesses required for speech and language processing (Bosshardt, 2006). Thus, while the underlying mechanisms mediating dual-task findings are unclear, they do implicate aspects of what Baddeley described as the CE and PL in the modulation of languageproduction and stuttering under high cognitive-load conditions. We suggest that those findings are important because it is likely thatAWS also experience greater cognitive-linguistic loads in more naturalistic communicative exchanges when compared to TFS.

3. PWM in neurobiological models of language and cognition

It is important to consider that, if CWS and AWS do present with differences in executive control interacting with a phonologicalloop, then the neurocognitive networks identified in CWS and AWS should be consistent with neurobiological models proposingmechanisms to account for those same cognitive capacities. Toward bridging the gap between cognitive-behavioral findings andneurobiological models, the sections below describe models of language proposing that PWM is an emergent property of sensory andmotor regions within the perisylvian network (Buchsbaum & D’Esposito, 2008; Hickok & Poeppel, 2007; Hickok et al., 2011; Majerus,2013). In addition, because DS is associated with differences in attentional control/executive function in both CWS and AWS, we alsodescribe a model proposing that interactions between the prefrontal cortex (PFC) and basal ganglia (BG) are also critical for workingmemory under distracting or attentionally demanding conditions (Hazy et al., 2007). Finally, we review brain imaging findings thatimplicate differences in large-scale networks supporting executive function/attentional control, PWM, and speech processing in thepathophysiology of both CWS and AWS. Taken as a whole, we suggest the networks implicated in the pathophysiology of stutteringare consistent with load dependent differences in behavioral accuracy in nonword repetition and dual-tasks.

3.1. PWM in the dual-stream model of language

While DS is most often associated with functional and structural differences within perisylvian regions (Etchell et al., 2018), it hasonly been in the last decade that neurobiological frameworks have proposed that perisylvian regions support PWM (Buchsbaum &D’Esposito, 2008; Herman et al., 2013; Jacquemot & Scott, 2006; Majerus, 2013). In an early model, Jacquemot and Scott (2006)proposed that auditory-sensory regions and premotor regions act as cyclic input and output buffers respectively allowing for pho-nological storage, suggesting a link between sensorimotor mechanisms in natural speech production and those involved in PWM.More recently, Hickok and Poeppel’s (2007) ‘dual-stream’ model of speech and language processing also proposed that interactionsbetween premotor, auditory-sensory, and audiomotor regions account for the emergent properties of PWM. According to their dual-stream model, a left hemisphere dominant dorsal stream, including aspects of the posterior inferior frontal gyrus (IFG), premotorcortex (PMC), and the planum temporale (PT) at the parieto-temporal boundary (Spt), mediate the conversion of input from auditory-phonological regions to articulatory output in the IFG/PMC regions. In the model, the primary auditory cortex along the superiortemporal gyrus (STG) is involved in spectrotemporal processing bilaterally and a left dominant region along the superior temporalsulcus (STS) is involved in retrieving phonological codes. A bilateral ventral stream in the anterior and more posterior aspects of themiddle temporal gyrus (MTG) and inferior temporal sulcus are more heavily involved in processing receptive language and mediatelexical access (see Hickok & Poeppel, 2007 for visual representation). Finally, although the dorsal and ventral streams are proposed tointeract in language formulation and processing, it is as yet unclear how and when the two streams interact to allow for speech andlanguage production in naturalistic contexts.

In the dual-stream model, short-term phonological memory (hereafter referred to as PWM) is proposed to emerge from inter-actions between auditory-phonological regions in the STS, audiomotor mapping in the PT, and premotor regions of the dorsal streamresponsible for motor programming and execution (Buchsbaum & D’Esposito, 2008; Hickok et al., 2003). In particular, a region of thePT at the sylvian parieto-temporal junction (i.e., area Spt) is active during tasks in which covert rehearsal is required for maintenanceof phonological information in working memory. The same region is active for auditory encoding, subvocal rehearsal, and overtproduction in a number of tasks, including multisyllabic nonword repetition (Buchsbaum et al., 2001), ‘jabberwocky’ phrase re-petition, and even covert rehearsal of piano melodies (Hickok et al., 2003). Finally, the dorsal stream network also mediates be-havioral performance on common clinical nonword auditory processing tasks, even in the absence of a requirement for production,strongly suggesting that the dorsal stream mediates PWM even when execution is not required (Perrachione et al., 2017). Thus, thedual-stream model suggests that the integration of sensory and motor representations are critical both for speech production and the

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

100

emergent properties of PWM. From this point of view, damage or disrupted input to the left hemisphere dorsal stream would beexpected to cause both effects on speech production and tasks that load PWM.

3.2. Disorders of dorsal stream sensorimotor integration and PWM

Working from a dual-stream model of language and a state feedback motor control (SFC) model, Hickok et al. (2011) proposedthat both DS and an acquired language disorder known as conduction aphasia (CA) are consistent with dorsal stream disruptionsaffecting speech production. From that point of view, individuals with CA have particular difficulty with word repetition related toimpaired PWM caused by damage to are Spt (Buchsbaum et al., 2011; Hickok et al., 2011). For individuals with CA, phonologicalerrors in natural speech production occur relatively infrequently but are more likely when producing multisyllabic and low frequencywords, suggesting parallels between the loci of phonological errors in CA (Buchsbaum & D’Esposito 2008; Buchsbaum et al., 2011)and similar linguistic factors known to influence the likelihood of stutter-typical disfluencies in DS (Howell & Au-Yeung, 1995,Howell & Au-Yeung, 2002). Thus, in the same way that individuals with CA experience load-dependent phonological errors in speechproduction and poor word repetition, PWS experience load-dependent disfluencies and load-dependent effects in nonword repetitiontasks.).

According to an SFC model of speech production, for accurate speech production to occur, an inhibitory model of the intendedsensory consequences (i.e., a forward model) is generated in premotor and motor regions (i.e., a motor-phonological system), whilean excitatory, corrective model is generated in somatosensory and auditory regions (i.e., a sensory-phonological system) (Hickoket al., 2011; Houde & Nagarajan, 2011). An audiomotor translator in area Spt mediates the conversion of motor and sensory pho-nological information. When the inhibitory forward model is accurate, no correction is needed and the two models are considered tocancel out. However, in instances wherein an error in the forward model is detected, excitatory corrective models are sent back to thePMC and primary motor areas for correction prior to overt production (Hickok, 2012). In people with CA, damage to Spt is proposedto disrupt forward predictions sent from motor-phonological areas to the auditory-phonological regions, causing errors in the internalmonitoring process. However, because the auditory-phonological regions are intact, a person with CA is nonetheless able to detect theerror and engages in repeated and often unsuccessful attempts to correct it.

By contrast, in DS the motor-phonological and sensory-phonological systems are intact but the audio-motor translator provides‘noisy’ or high variance input to the auditory-phonological system (Hickok et al., 2011). Due to this noisy input, forward sensorypredictions generate estimates of vocal-tract states and auditory-phonological targets that cause an invalid error signal to be sent backto the motor-phonological system. The behavioral consequence of this noisy translation is a predict-correct loop that, under con-ditions of high sensorimotor ‘load,’ results in overt disfluencies (e.g., part-word repetitions). Further, Hickok et al. (2011) predictedthat errors in sensorimotor mapping should be proportional to the load on the network. From this point of view, higher sensorimotor‘load’ would be expected on low frequency words, multisyllabic words, or potentially in situations taxing other systems related tospeech such as when speaking under time-pressure. Because nonword repetition tasks rely heavily on the same network, one pre-diction of this approach would be that high variance sensorimotor integration would also result in load dependent errors in repetitionas reported in both preschool CWS at lower loads (Pelczarski & Yaruss, 2016; Spencer & Weber-Fox, 2014) and older CWS and AWS athigher loads (Byrd et al., 2012; Byrd et al., 2015; Coalson & Byrd, 2017; Sasisekaran, 2013). As such, the above described mechanismsprovide a potential neurophysiological link between internal models of motor control proposed to explain the emergence stutter-typical disfluencies (Brown et al., 2005; Guenther et al., 2006; Hickok et al., 2011; Max et al., 2004; Tian & Poeppel, 2012) and thedorsal sensorimotor stream proposed to support PWM for learning new words, new phonological representations, and errors innatural, connected speech production (Buchsbaum & D'Esposito, 2008; Hickok et al., 2011; Jacquemot & Scott, 2006; and see Pageet al., 2007).

3.3. PWM in neurobiological models of executive control

While a dorsal stream deficit in DS is plausible, it is unclear from a dual-stream model alone how attention and executive functionunder high load conditions might interact with the dorsal stream. Whereas Baddeley’s cognitive model proposes interactions betweenthe CE and PL, other models suggest that task relevant information is not maintained in dedicated storage areas but as a subset ofprocesses that are in the focus of attention (Cowan, 1999; Cowan et al., 2005). A number of neurobiological accounts have implicatedthe dorsolateral prefrontal cortex (BA 46/BA 9) and ventrolateral PFC regions (BA 47/45) in attentional capacities along with theinferior parietal lobule (IPL) and superior parietal lobule (SPL) (e.g., Baddeley, 2003; Buchsbaum & D’Esposito, 2008; Majerus, 2013;Nee et al., 2013). Evidence from neuroimaging studies examining working memory in the contexts of visual, auditory, and affectivestimuli also suggest that PFC regions interact with other networks depending on the modality of the stimulus to be held in workingmemory (e.g., visual vs. auditory). In this way, the same networks involved in sensorimotor representations of the object to berecalled maintain it via interactions with attentional and executive control regions in the PFC (D’Esposito, 2007).

Other neurocomputational approaches emphasize interactions between the PFC and BG in working memory performance underattentionally demanding or distracting conditions (Baier et al., 2010; McNab & Klingberg, 2008; Voytek & Knight, 2010). Accordingto the prefrontal cortex and basal ganglia model of working memory (PBWM), the BG provides a ‘go’ or ‘nogo’ signal to the PFC tomodulate information relevance while more primitive BG to premotor connections (e.g, supplemental motor area and anteriorcingulate cortex) modulate action selection and execution (Hazy et al., 2007; Mink, 1996). In the PBWM computational model,dynamic gating via midbrain structures is learned via reinforcement mechanisms thought to rely on dopaminergic systems, while PFCrepresentations are updated using Hebbian and error-driven learning mechanisms. Bidirectional excitatory connections from the

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

101

thalamus to the frontal cortex are inhibited by tonically active neurons in the substantia nigra pars reticulata. In turn, the substantianigra is disinhibited by ‘go’ neurons in the dorsal striatum, allowing for disinhibition of the frontal cortex and gating of workingmemory in the PFC under distracting conditions. This process allows the PFC to maintain only relevant information in workingmemory under distracting conditions while the dopaminergic systems involved in learning and motivation (including the bilateralamygdala) evaluate the potential reward value of actions and goals maintained in the PFC, thereby modulating future actions.Finally, the model proposes an output-gating mechanism that can determine which of a set of actively maintained representationsshould be selected for motoric planning and output.

Working from the PBWM model, it is reasonable to propose that PFC-BG connectivity could modulate action selection and thusdorsal stream sensorimotor control networks in PWS as a lifetime of experience with stuttering accumulates. The model also providesa potential mechanism by which neural learning could affect the plasticity of networks involved in speech production, leading tovariability in the expression of stuttering as new experiences accumulate. In such cases, learned behaviors such as fluency enhancingmotor control strategies, anticipation of difficult social situations, and covert attempts to avoid or substitute words would bemodulated by ever evolving connections between the PFC, BG, and the dorsal sensorimotor stream. As such, in the sections following,we discuss theoretical frameworks implicating sensorimotor integration and speech-motor timing early in the development ofstuttering and neuroimaging evidence supporting a role for PFC and BG structures in connection with dorsal stream networks in CWSand AWS.

3.4. PFC, BG, and dorsal stream sensorimotor timing

While interactions between executive control in the PFC, BG, and dorsal regions involved in phonological ‘storage’ appear to playa role in DS, one potential problem is that the mechanism underlying ‘noisy’ sensorimotor integration is somewhat underspecified.For example, it is unclear how a difference in dorsal stream processing might lead to stutter-typical disfluencies in PWS as opposed tothe phonological errors observed in CA. There is not strong support for the notion that phonological errors of the kind observed in CAcovary with stuttering (Nippold, 2002; Nippold, 2012), suggesting that the sensorimotor errors in PWS are likely to involve a differentprocess. One possibility is that, rather than phonological errors caused by damage to the Spt region as in CA, disrupted sensorimotorintegration is instead a matter of inaccurate coordinative timing mediated by the dorsal sensorimotor stream and connections withsubcortical structures also implicated in speech-motor timing (e.g., the BG and cerebellum). While findings in AWS regarding timingin non-speech motor control have been variable, the notion that stuttering is primarily a disorder of timing or coordination for speechproduction has long been the subject of theoretical proposals (Alm, 2004; Etchell et al., 2014; Kent, 1984; Perkins et al., 1991; Smith& Weber, 2017; Van Riper, 1982). As such, subtle timing differences in sensorimotor integration could potentially account for theprocess disrupting both fluent speech production in PWS and differences in tasks loading PWM. In the sections below, we describeneurobiological frameworks and findings implicating speech-motor timing networks in sensorimotor integration and internal modelsof motor control.

3.4.1. Neurobiological frameworks supporting sensorimotor timing differences in DSSimilar to the dorsal stream SFC model described in Section 3.2, connections between the BG and frontotemporal-parietal net-

works involved in sensorimotor timing for speech have also been used to explain a wide range of phenomena in stuttering (Alm,2004; Alm & Risberg, 2007; Civier et al., 2013; Etchell et al., 2018; Wu, Maguire, Riley, Lee, Keator, & Tang, 1997). In an influentialproposal, Alm (2004) posited that differences in dopamine receptor concentrations in the BG early in speech motor developmentcause sensorimotor instabilities in CWS. According to Alm (2004), unstable timing cues provided by the BG are read out to thesupplemental motor area (SMA) for subsequent movement sequencing, eventually resulting in repetitions or prolongations ofmovement to compensate for the deficit. Drawing on Goldberg's dual-premotor hypothesis, Alm (2004) suggested a medial premotorsystem (SMA and BG) is more heavily involved in overlearned, automatic, and internally initiated movements whereas a lateralpremotor system (lateral PMC and cerebellum) is more heavily involved in the control of externally cued, less automatic, and novelmovement patterns. The increased fluency well-known to result from altered auditory feedback, choral speech, and rhythmic speech(e.g., speaking to a metronome) were proposed to be the result of a shift from a compromised medial premotor system to that of anuncompromised (or less compromised) lateral premotor system. In addition to fluency enhancing effects, it was also proposed thatattention to compensation for sensorimotor instability may shift internal timing cues from the medial to lateral premotor systemwhen speech production is less automatic or overlearned (e.g., when using fluency shaping techniques). More recently, Alm (2014)also proposed that situational variability related to social cognition mediated by the polar medial PFC (BA10), anterior cingulatecortex (ACC), and SMA could be related to situational variability in stuttering during social interactions. From that point of view, themedial PFC through connections with the BG and premotor regions may modulate motor control depending on social-cognitivedemands. Thus, connections between the PFC, BG, and frontotemporal-parietal networks are implicated in modulating sensorimotorcontrol and speech fluency depending upon context.

3.4.2. Neurocomputational modeling evidence supporting a link between BG timing networks and internal modelsIn addition to PFC and BG timing networks proposed by Alm (2004), elevated dopamine levels are also implicated in interaction

with an overreliance on sensory feedback mechanisms derived from internal models of motor control (Civier et al., 2010; Civier et al.,2013; Max et al., 2004). In a computational modeling study, Civier et al. (2010) used the directions in auditory space to velocities inarticulator space (DIVA) computational model to simulate an inefficient feedforward model, forcing the model to rely on time-laggedauditory feedback control. The results suggested that when accumulating errors between the feedforward and feedback models were

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

102

large (indexed by formant transitions for the auditory targets), self-repair was initiated in the form of a ‘reset’ or syllable repetition. Asecond modeling study used the gradient order DIVA model (GODIVA) to simulate two proposed brain differences in AWS, a white-matter anomaly in the ventral motor cortex (vMC) and elevated dopamine levels (Civier et al., 2013. Those authors reported thatwhereas elevated dopamine levels affecting the BG could account for syllable initial disfluencies, the white-matter anomaly couldonly account for disfluencies on syllables in the middle or later in an utterance, suggesting that the two potential causes of disfluencylikely interact. Computational modeling findings using the GODIVA model have recently been extended to nonword syllable re-petition (Markiewicz & Bohland, 2016) with more specific descriptions of the functional role of various input and output buffers insyllable production and their interaction with the BG (see Bohland, Bullock, and Geunther, 2010). Taken together, computationalmodeling findings do suggest a link between the functional role of the BG in speech production and internal models of motor control.

The potential link between timing cues for speech mediated by the BG and internal model proposals is of interest for explainingthe relationship between DS and PWM because timing in internal models of motor-control have also been implicated in PWM(Herman et al., 2013). Similar to neurobiological models of speech motor control, the maintenance of phonological traces betweenmotor-phonological and sensory-phonological regions would be expected to require timely cyclic interactions (Jacquemot & Scott,2006; Hickok et al., 2011; Herman et al., 2013). That is, cyclic activation and suppression between motor and sensory buffers must bemaintained prior to execution both in repetition tasks requiring maintenance of phonological traces and in natural speech production(Herman et al., 2013; Jacquemot & Scott, 2006; and see Page et al., 2007). As such, one likely source of variance in sensorimotorinput to the left PT region in PWS is generally slowed or mistimed information for phonological/lexical targets with subsequenteffects on execution. Further, it is likely that areas of the PFC modulate sensorimotor timing and dorsal stream coordination de-pending on context (Alm & Risberg, 2007). To explore that possibility, in sections following we review functional and structural brainimaging evidence implicating connectivity in large-scale neurocognitive networks in CWS and AWS consistent with differences in theconnectivity of PFC, BG and dorsal sensorimotor regions implicated both in speech production and PWM.

3.4.3. Functional imaging findings in CWSOne source of evidence for differences in large-scale neurocognitive networks in CWS comes from functional imaging studies.

Recent functional magnetic resonance imaging (fMRI) studies implicate differences in activation or connectivity of large-scale net-works related to attention/executive function and timing for the sensorimotor control of speech (Etchell et al., 2018). In resting-staterecordings of children 3–9 years of age, hypo-connectivity motor (IFG/ primary motor cortex/PMC), auditory (MTG/STG), andsomatosensory association regions (e.g., the supramarginal gyrus; SMG) have been reported, suggesting that a decreased capacity fortiming emerges early in intrinsic BGTC networks. Further, hyperconnectivity in the right superior frontal gyrus and putamen was alsofound in CWS relative to CWNS, implicating connections between the PFC and BG (Chang & Zhu, 2013). Another recent study ofcortical connectivity with the BG (i.e., the putamen) during rhythm perception also found decreased connectivity with the left PMC,left STG, SMA, and cerebellum that was also related to lower behavioral performance in CWS relative to CWNS (Chang, Chow,Weiland, & McAuley, 2016). In a study of resting-state cerebral blood flow of CWS and AWS, the CWS were reported to have lowerperfusion in Broca’s area and the superior frontal gyrus relative to TFS that inversely correlated with stuttering severity (Desai et al.,2016).

Most recently, differences in metabolite ratios between school-age and adolescent CWS and TFS (age range 5–17 years) withinwide-spread executive function/attentional networks, speech sensorimotor control networks, and networks involved in memory werealso reported using proton chemical shift imaging (O’Neill et al., 2017). Finally, a recent longitudinal study in CWS reportedanomalous connectivity in the resting-state between what is known as the default mode network (DMN), dorsal attention network(DAN), ventral attention network (VAN), fronto-parietal network (FPN) (including the PT region), and somatomotor network, furtherimplicating broader attentional, audiomotor, and somatomotor regions in the pathophysiology of stuttering early in speech andlanguage development (Chang et al., 2017). More importantly, anomalous connectivity between the DMN, DAN, VAN, and fronto-parietal network also predicted later persistence in CWS (Chang et al., 2017). Thus, taken together, functional neuroimaging findingsimplicate large-scale neurocognitive networks involved in, attention, speech-motor timing, and sensorimotor integration early in thepreschool and school-age years. Because the cognitive capacities implicated in those networks are also related to working memorybrain activation and behavioral performance (Zhu et al., 2013), those findings are consistent with differences in neurocognitivenetworks supporting PWM reported in the behavioral literature (Spencer & Weber-Fox, 2014).

3.4.4. Structural imaging findings in CWSIn addition to functional imaging studies, a relatively small number of studies have also investigated structural differences be-

tween CWS and CWNS. Those studies also implicate anomalies in regions considered to be important for attention/executivefunction, sensorimotor integration, and PWM. Studies of gray matter volume (GMV) have reported reduced GMV in the putamen andmedial frontal gyrus (MFG), whereas the right STG is reported to have higher GMV in CWS ages 9–12 years (Chang, Erikson,Ambrose, Hasegawa-Johnson, & Ludlow, 2008) and CWS ages 6–12 years (Beal et al., 2015). A comparison of recovered and per-sistent school-age CWS also showed more GMV the left and right STG in children who persist, suggesting that temporal lobe auditorysensory regions play an important role in persistence and recovery (Chang et al., 2008). Studies of gray matter morphology in CWShave also reported differences in the laterality of prefrontal regions in school-age children (8–13) (Mock et al., 2012) and the caudatenucleus of the BG (Foundas et al., 2013). In addition to gray matter differences, weaker white matter connections (as measured byfractional anisotropy; FA) between the left putamen and cortical networks associated with both cognition and motor control (MFG,SMA, left IFG and insula) have also been observed along with decreased connectivity between the left IFG and pSTG (Chang & Zhu,2013; Chang et al., 2015). Dorsal stream white matter connections, including the left IFG and pSTG are also correlated with

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

103

disassociations on a range of neuromotor, articulatory, and receptive/expressive language measures, suggesting that structural dif-ferences may underlie decreased neural efficiency for speech and language production (Choo et al., 2016). As such, structural findingsalso implicate networks thought to support cognitive capacities underlying working memory emerging early in the development ofstuttering (Zhu et al., 2013)

3.4.5. Functional neuroimaging findings in AWSWhile more than two decades of neuroimaging studies have shown widespread functional and structural differences in AWS relative to

TFS, findings have also been inconsistent depending in part on the tasks used, sample sizes, neuroimaging approach and analysis metho-dology (see Etchell et al., 2018 for comprehensive review; Ingham et al., 2012; Neef et al., 2015). Those caveats notwithstanding, there areseveral lines of evidence implicating differences in the connections between the BG and sensorimotor networks involved in speech pro-duction. Neuroimaging studies have reported differences in the BG and along with premotor, auditory, and cerebellar regions compared toTFS in the eyes-open resting state (Civier et al., 2013; Ingham et al., 2012; Ingham et al., 2004; Wu et al., 1995) during speech productiontasks (Giraud et al., 2008; Ingham et al., 2012; Ingham et al., 2004; Watkins et al., 2008; Wu et al., 1995), and even during non-speechmovements (Chang et al., 2009). Activity in the BG is also associated with measures of stuttering severity and is related to the fluencyenhancing effects of treatment (Giraud et al., 2008; Toyomura et al., 2015). Taken together, those individual reports implicate BG andsensorimotor networks in a range of conditions and suggest a relationship with treatment effects.

3.4.5.1. Evidence from meta-analyses. In addition to individual reports, meta-analyses of positron emission tomography (PET) and fMRIudies also implicate structures within the BG with sensorimotor regions involved in speech production (Belyk et al., 2015; Brown et al., 2005;Neef et al., 2015). An early activation likelihood estimation meta-analysis across speech production tasks found that AWS consistently presentwith increased activation in the right IFG/insula and vermal region of the cerebellum, with decreased activation in the bilateral STG duringspeech production (Brown et al., 2005). Those findings were interpreted as an error in the process of internal model formulation in whichsuppression of the auditory areas below baseline (i.e., inhibitory forward models) causes an error signal to be sent back to motor areas that, inturn, causes a predict-correct loop (e.g., part-word repetitions) (Brown et al., 2005). Two more recent functional imaging meta-analyses alsoreported patterns of activity related to trait (i.e., TFS compared to AWS) and state (i.e., disfluent speech relative to fluent speech) aspects ofstuttering (Belyk et al., 2015; Budde et al., 2014). In trait analyses, right-hemisphere over activations include the PMC, primary motor-cortex,rolandic operculum, insula, posterior IFG, orbital IFG, pre-SMA, MFG, IPL, and SPL. Left-hemisphere under activations include laryngealmotor cortex, MTG, STS, vermal region of the cerebellum, and red nucleus of the midbrain, suggesting widespread neurocognitive networksare involved in stuttering by adulthood. By contrast, state analyses comparing disfluent and fluent speech have found areas related todisfluency are primarily in the left-hemisphere (e.g, IFG, SMA, somatosensory cortex, and BG), while areas related to fluency are found inright-hemisphere regions (primary auditory cortex (A1), PT, pSTG, SMG, IPL, & IFG). In those meta analyses, the only fluency related regionin the left was the SMG, suggesting that left somatosensory association areas may also play a role in fluency or compensation for sensorimotorinstabilities in the audio-motor network (Belyk et al., 2015; Budde et al., 2014; Neef et al., 2015). Taken together, state analyses suggest thatregions consistent with typically left-hemisphere dominant frontotemporal-parietal network are under-active in AWS during stuttered speech.On the other hand, fluency related areas in the right hemisphere are hyperactive during fluent speech, suggesting that right-hemispherestructures may aid in compensation for sensorimotor instabilities on the left. Those findings are consistent with differences in dorsal streamregions proposed to mediate the emergent properties of PWM and sensorimotor integration in speech production.

3.4.5.2. Evidence from resting-state networks. Another line of evidence implicating broader neurocognitive networks in DS is recentevidence from the resting state. One distinct weakness of functional neuroimaging studies employing various speaking and reading tasks isthat it is difficult to compare activations across studies due to task differences (Chang et al., 2017; Etchell et al., 2018). While early findingssuggested a lack of consistent differences in the resting-state, more recent studies using larger sample sizes and more recent analyticapproaches have noted differences (Etchell et al., 2018). Xuan et al. (2012) reported that networks associated with cognitive controlincluding increased connectivity in the prefrontal cortex and DMN and increased connectivity in motor and auditory regions. Lu et al. (2012)reported decreases in functional connectivity during resting state in the left IFG and dorsolateral prefrontal cortex compared to TFS. In thatstudy, AWS who received treatment showed reduction in the left IFG and increases in the SMA and cerebellum, suggesting that treatmenteffects may have been related to compensatory speech timing networks. Yang, Fanlu, Siok, and Tan (2018) also reported reduced resting-state connectivity between the right SMA and BG, along with decreased connectivity between the STG and BG. More recent studies have alsofound differences in resting state connectivity between the left IFG and primary auditory cortex/Heschl's gyrus that is correlated withperformance on a speech perception task, further suggesting a relationship between intrinsic differences in sensorimotor networks andchallenging speech processing tasks (Halag-Milo et al., 2016; Lu et al., 2016). Thus, resting state findings imply differences in large-scaleneurocognitive networks in AWS, including aspects of the PFC, BG, and regions involved in dorsal stream sensorimotor integration in speechproduction and challenging speech perception tasks.

3.4.5.3. Evidence for the PFC in compensation and unassisted recovery. While resting state findings implicate attention and executivefunction in AWS, they do not explicitly link it to moment-to-moment variability in stutter-typical disfluencies during speech production.However, recent functional neuroimaging findings in AWS do implicate the left and right orbitofrontal cortex (OFC) in compensation forstuttering (Sitek et al., 2016) and in unassisted recovery from stuttering in adulthood (Kell et al., 2017). Using resting state and diffusion MRI,Sitek et al. (2016) reported that AWS with the least severe symptoms as determined by the Stuttering Severity Instrument – Fourth Edition(SSI-4; Riley, 2009) presented with increased functional and structural connectivity between the OFC and cerebellum compared to AWS withhigher SSI-4 severity ratings, suggesting that the OFC/cerebellum could be involved in compensatory mechanisms in the more fluent AWS.

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

104

Further, long-lasting recovery without treatment (i.e., recovered stuttering) in AWS has been associated with differences in the leftventrolateral prefrontal cortex (orbital IFG; BA47/12) (Kell et al., 2009). More recent functional connectivity analyses have reported‘hyperconnectivity’ between BA47, BA44, the SMA, and SMG prior to fluency shaping therapy. Following therapy, reductions in connectivitywere observed between BA44 (i.e., opercular IFG) and the SMG along with increased coupling between the anterior STG and ventral motorcortex. By contrast, in recovered stuttering, there was decreased connectivity from BA47 to the SMA, SMG, and superior cerebellum,suggesting that the OFC modulates somatomotor connectivity in recovered stuttering differently than in treated AWS. Thus, recent evidenceimplicates the PFC in both enhancing fluency and potentially in increasing stutter-typical disfluencies, similar to dual-task findings that loadaspects of attention and executive function.

3.4.6. Structural findings in AWSEvidence from recent structural neuroimaging studies comparing AWS and TFS implicates connectivity between PFC, BG, and dorsal

stream regions. One line of evidence in particular implicates structural differences in the PT with at least some AWS demonstrating greateroverall area and a more symmetric pattern in the PT (Foundas et al., 2001). Enhanced fluency induced by delayed auditory feedback in asubgroup of AWS is also associated with rightward PT asymmetry (Foundas et al., 2004). Other studies have reported increased white matterunderlying the right PT, further suggesting that compensatory processing in the right PT may be related to white matter plasticity andcompensation for anomalies in the left PT (Jäncke et al., 2004). Despite these early findings, the role of PT asymmetry in DS is uncertain, aslater studies with larger sample sizes and more stringent controls (e.g., inclusion of only right-handed adults) have not replicated gray matterPT asymmetry in AWS relative to TFS (Chang et al., 2008; Cykowski et al., 2008; Gough et al., 2018). One possible mediating factor isvariability in the age of study participants. For example, Gough et al. (2018) reported a positive correlation between PT asymmetry and agein TFS while no such relationship was observed for AWS. Another possibility is that there are subtypes of AWS in whom PT asymmetryinteracts with structural or functional anomalies in other cortical or subcortical structures (Foundas et al., 2013; Gough et al., 2018). Thoseuncertainties notwithstanding, PT asymmetry or PT connectivity with other dorsal stream structures implicates structural differences inregion thought to be critical for audiomotor integration and PWM in the dual-stream model.

A second line of evidence supporting dorsal stream pathophysiology comes from structural differences in regions thought to beconnected to the PT. Decreases in gray matter volume of AWS have also been reported in other structures consistent with PWMbuffers thought to be important for articulatory rehearsal. These areas include the IFG and aspects of the more anterior STG (Bealet al., 2007; Song et al., 2007). In particular, decreased gray matter volume in the left IFG has been found to negatively correlate withstuttering severity (Kell et al., 2009). Recent findings also suggest that the right BG and bilateral STG functionally and structurally aremore strongly connected than in TFS (Sitek et al., 2016). Sowman et al. (2017) have also reported differences in gray matter volumeof the caudate nucleus in AWS, while a number earlier studies reported differences in the right precentral gyrus and STG (Beal et al.,2007; Kikuchi et al., 2011; Song et al., 2007). In addition to gray matter anomalies, studies have also reported white matter dif-ferences in areas important for connecting premotor structures with somatosensory and audio-motor regions (Cai et al., 2014).Further, anomalies along white matter tracts connecting dorsal stream regions may also be present in the peripheral nervous systemalong the corticobulbar and corticospinal tracts, suggesting that dorsal stream anomalies may be related to decreased efficiency ofperipheral execution (Connally, Ward, Howell, & Watkins, 2014). A recent ALE meta-analysis in AWS also reported decreased FA inwhite matter tracts connecting left hemisphere frontal motor structures (IFG, vPMC, primary motor cortex) with temporal structures(pSTG, middle temporal gyrus; MTG) and parietal structures (SMG, primary somatosensory cortex) (Neef et al., 2015).

While methodological limitations make attributing white matter anomalies to the dorsal stream as opposed to the ventral streamuncertain, recent studies have reported a lack of white matter differences in the ventral stream with robust differences in the dorsalstream (see Kronfeld-Duenias et al., 2017 for debate). Finally, in the most recent study implicating the PFC in stuttering, Neef et al.(2017) reported that increased connection strength between the right frontal pole (BA10) and subthalamic nuclei is positivelycorrelated with stuttering severity, while the uncinate fasciculus connecting the right frontal pole to auditory and multisensoryregions was negatively correlated with stuttering severity, suggesting that aspects of the frontal pole may be related both to com-pensatory and inhibitory processes affecting fluent speech production. Taken as a whole, the structural imaging literature suggestsanomalous connectivity in large-scale neurocognitive networks including the PFC, BG, and dorsal sensorimotor stream implicated inPWM, while differences in the ventral stream important for lexical access remain uncertain.

3.4.7. Implications of neuroimaging findings in CWS and AWSAlthough a number of interpretations have been proposed to account for neuroimaging findings in CWS and AWS, a common theme is

that both CWS and AWS present with differences in networks thought to mediate timing in speech-motor control and sensorimotor in-tegration. Despite considerable spatial variability across studies, in a systematic review of the neuroimaging literature, Etchell et al. (2018)concluded that one common finding in both CWS and AWS is decreased activation and reduced gray matter volume in the left IFG andconnections with the SMA, temporal lobe regions, and right IFG. Additionally, in the only study to examine anomalous structure and functionin CWS and AWS, the IFG (pars opercularis) was also reported to be the only region to show an abnormal developmental trajectory (reducedgray matter thinning), suggesting that gray matter anomalies or underlying white matter anomalies are a strong candidate for the cause ofanomalous input to the sensorimotor network in the preschool years (Beal et al., 2015). Interestingly, the triangular and opercular portions ofthe IFG are also causally related to the transmission of information for timing and sequencing in speech production as opposed to articulatoryquality located in more dorsal premotor regions (Long et al., 2016). The commonly reported hyperactivity in the right IFG (Belyk et al., 2015;Brown et al., 2005; Budde et al., 2014), thought to compensate for deficits in the left IFG, is also considered a core structure active across arange of conditions requiring the processing of temporal information in perception and motor control (Weiner and Turkeltaub, 2010). Assuch, neuroimaging findings suggest that differences in the efficiency of timing networks related to both dorsal stream sensorimotor

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

105

integration and the emergent properties of PWM are different in both CWS and AWS. In this way, the available neuroimaging literaturesuggests that behavioral differences in accuracy on tasks heavily recruiting those same networks, such as nonword repetition or dual-tasks,are likely in both CWS and AWS.

Finally, in CWS regions within the PFC appear to be related to persistence and recovery in connection with sensorimotor networks(Chang et al., 2017). In AWS, subregions of the PFC in connection with sensorimotor networks also appear to be related to com-pensation for stuttering (Sitek et al., 2016), treatment effects (Kell et al., 2009), and unassisted recovery (Kell et al., 2017). The mostrecent study to date to interpret functions of the PFC also suggested that different right hemisphere connections may subserve bothfluency enhancing compensatory and detrimental inhibitory processes, suggesting that the right PFC via different pathways may beinvolved in processes that are both deleterious to and facilitate fluency in AWS (Neef et al., 2017). Those findings are broadlyconsistent with discrepant dual-task effects on stuttering depending on the experimental conditions (Bosshardt, 2006). Taken to-gether, those findings are consistent with a modulatory role for the PFC that may be related to both compensation for sensorimotorinstabilities and conditions that increase the likelihood of stuttering.

4. Linking brain and behavior

Differences in connectivity of the PFC, BG, and dorsal stream sensorimotor regions may provide explanations for common behavioralfindings in nonword repetition, auditory-perceptual, and dual-tasks in DS. First, disrupted timing cues from the forward model to the inversemodel affecting speech production could explain subtle behavioral differences commonly observed on nonword repetition tasks in preschoolCWS (Spencer & Weber-Fox, 2014). If differences in sensorimotor timing or ‘high variance’ input to internal models of motor control accountfor incorrect nonword repetition trials, the process would not always be expected to result an incorrect response. In other words, on sometrials CWS would be expected to compensate for increased variance or interbuffer timing differences much like they would in natural speechproduction (i.e., perceptually fluent speech) (c.f. Namasivayam & van Lieshout, 2011). Such a mechanism might account for statisticallysignificant differences in nonword repetition that are nonetheless within normal limits in preschool CWS (Pelczarski & Yaruss, 2016). In AWS,the proposal is consistent with differences in fine motor control during nonword repetition tasks at lower loads (e.g., 4 syllables), suggestingcompensation for sensorimotor instabilities (Smith et al., 2010; Smith et al., 2012), while behavioral differences emerge only when the PWMsystem is heavily loaded (Byrd et al., 2012; Byrd et al., 2015; Coalson & Byrd, 2017). Finally, while lower performance on nonword repetitiontasks is commonly observed in CWS, semantic repetition tasks typically do not result in significantly lower accuracy (Spencer & Weber-Fox,2014; Anderson & Wagovich, 2010; Pelczarski & Yaruss, 2016) and do not appear to be related to persistence and recovery (Spencer &Weber-Fox, 2014). Interestingly, syntactic phrase structure violations embedded in sentences with semantic content are not associated withprocessing differences between CWS who recover and those who persist, but similar violations are associated with processing differences (i.e.,N400-like response) between the two groups in nonsense ‘jabberwocky’ phrase repetition tasks. Those findings suggest limitations on PWMinteracting with syntax may also be related to persistence/recovery while semantic processes alone are not (Usler & Weber-Fox, 2015). Takentogether, findings are more consistent with dorsal stream pathophysiology than ventral stream pathophysiology proposed to mediate LTM forsemantic content or lexical access (Hickok et al., 2011; and see Majerus, 2013) and may explain why single word, verbal repetition taskstypically do not result in significantly lower behavioral performance or predict persistence/recovery.

Second, a deficit in dorsal stream sensorimotor integration is consistent with lines of evidence suggesting neurobiological and subtlebehavioral differences between PWS and controls in challenging auditory perception tasks in the absence of a functional deficit in languagecomprehension (Chang et al., 2009; Halag-Milo et al., 2016; Neef et al., 2012; Saltuklaroglu et al., 2017; Sato et al., 2011). According to thedual-stream model, the dorsal stream may aid perceptual tasks in which working memory and attention are focused on speech-soundanalysis, whereas the ventral stream mediates receptive processing wherein the goal is linguistic comprehension (i.e., a double dissociation)(Hickok & Poeppel, 2007). Forward models of speech production may be adapted in some instances to difficult perceptual tasks (e.g.,discrimination in noise), allowing for constraints on expected upcoming speech percepts that may later aid working memory for the sub-sequent response (e.g., via covert rehearsal) (Hickok et al., 2011). Thus, mistimed or ‘noisy’ sensorimotor integration would be expected tocause subtle differences in auditory-perceptual tasks without functional language comprehension deficits. Given the link between sensor-imotor integration in speech perception and production, PWM may also be related to a higher risk of phonological disorders in children whostutter (Ambrose et al., 2015). As such, the evidence implicates differences in the dorsal sensorimotor network that are related to auditoryspeech processing, phonological storage, and the sensorimotor control of speech production in the absence of functional language com-prehension deficits thought to be mediated by the ventral stream.

Third, there is some evidence from preschool-age CWS that aspects of attention interact with the process of phonological storage fornonword repetition (Anderson & Wagovich, 2010) and the neurocognitive networks supporting attention and executive function are im-plicated in persistence and recovery (Chang et al., 2017). A potential role for the PFC and BG in executive control of both working memoryunder distracting conditions and speech production under attentionally demanding conditions (e.g., social situations or under time-pressure)is consistent with those findings and recent brain imaging evidence implicating both intrinsic and disfluency related neurocognitive networksin CWS and AWS (Chang et al., 2017; Kell et al., 2017; Sitek et al., 2016). In such case, tasks that ensure attention is directed away fromspeech production, decoupling of the PFC-BG and dorsal sensorimotor/cerebellar network would be expected along with fluency en-hancement (e.g., Eichorn et al., 2017; Vasic & Winjen, 2005). However, in other tasks in which attention is directed internally to speechproduction (e.g., Bosshardt, 2002; Caruso & Chodzko-Zajko, 1994) or that load highly overlapping cognitive-linguistic processes (Bosshardt,2002; Metten et al., 2011), greater PFC-BG coupling with the dorsal stream would be expected along with instabilities in dorsal streamsensorimotor timing. That potential function of PFC-BG modulation of the dorsal stream during internal model formulation is broadlyconsistent with the OFCs role in compensation for sensorimotor instabilities and disconnection of the OFC from the audiomotor and so-matosensory regions in unassisted recovery (Kell et al., 2017). Thus, the PFC may be important in compensating for internal sensorimotor

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

106

errors in addition to increasing the probability of stuttering depending upon the circumstances. To investigate those possibilities, future workwould need to carefully manipulate the direction of attention in dual-tasks and the extent to which cognitive resources are shared betweentasks.

5. Future directions: neural oscillations in PWM

While there appear to be links between the neurobiology of stuttering and PWM, it is still unclear what precise mechanismsmediate that relationship. We suggest that mechanisms underlying differences in sensorimotor control and PWM in CWS and AWScould be further explored by investigating cyclic dorsal stream interbuffer timing differences in nonword repetition and dual-tasksguided by the neurobiological networks reviewed. Fig. 1 depicts cortical networks likely to be related to PWM according to the dual-stream model proposed by Hickok and Poeppel (2007) (blue, green, yellow) and more recent models and data proposing an executivenetwork involved in attentional control (red) (D’Esposito, 2007; Majerus, 2013; Nee et al., 2013). While it is difficult to examinecyclic interbuffer timing in the encoding, maintenance, and production phases of a nonword repetition task when using relatively lowtemporal resolution imaging approaches (e.g., fMRI), one way to investigate ongoing sensorimotor timing in each phase of the task isvia event-related changes in the ongoing time-frequency of the electroencephalography (EEG) and magnetoencephalography (MEG)known as ‘neural oscillations.’ According to a ‘dynamic’ model of oscillatory function, neural oscillations are critical for long-distanceinterregional communication and timely correlations between the phase of oscillations in prefrontal regions and sensorimotor regionsmay reflect top-down control of interregional communication in both perception and action-control. (Fries, 2005; Siegel et al., 2012).As such, the power and phase of neuronal oscillations has the potential to allow for the study of large scale neuronal dynamics in thevarious stages of nonword repetition related to stimulus encoding, trace maintenance, and repetition (i.e., execution).

Multifactorial models of stuttering have long predicted that even subtle differences in cognitive capacities affect motor control for speechin DS (Smith, 1999; Starkweather, 2002). In particular, the speech motor skill (SMS) framework predicts that cognitive-sensorimotor loadsmay destabilize central neural oscillators, resulting in decreased oscillatory entrainment with speech-motor effectors critical for feedbackcontrol (Namasivayam & van Lieshout, 2011; van Lieshout et al., 2014; van Lieshout et al., 2004). From that point of view, the repetition ofnovel phonological sequences and dual-tasks loading cognitive-sensorimotor oscillators may cause load-dependent breakdowns in entrainedsensorimotor-effector oscillations. One likely difference related to sensorimotor timing in AWS and CWS is neural oscillations in the beta band(15–25 Hz) (Etchell et al., 2014). The beta band is a known correlate of motor activation and more recently has been proposed to mediatecoordinative timing in audio-motor integration for speech along with lower frequency bands (Morillon & Schroeder, 2015). In CWS dif-ferences in the beta band have been found at rest (Özge, Torosm & Çömelekoğlu, 2004) and reversals in the timing of peak beta band powerreduction (i.e., suppression) and enhancement in a rhythmic tone perception task has been observed in CWS relative to CWNS (Etchell et al.,2016). In AWS, differences have been found beta band interhemispheric connectivity at rest (Joos et al., 2014), in speech production tasks(Mersov et al., 2017), and in speech perception tasks (Saltuklaroglu et al., 2017). As such, we suggest that the study of neural oscillations innonword repetition and dual-tasks may provide a means of investigating a much needed link between the sensorimotor mechanisms pro-posed to mediate stutter-typical disfluencies (Hickok et al., 2011; Tian & Poeppel, 2012) and sensorimotor mechanisms mediating PWM

Fig. 1. A depiction of networks likely to interact in nonword repetition and dual-tasks based on Hickok and Poeppel’s dual-stream model (Hickokand Poeppel, 2007) (blue, green, yellow) and an attentional/executive network in working memory (see Majerus, 2013 and Nee et al., 2013) (red).Regions in red include the dorsolateral prefrontal cortex (DLPFC), the ventrolateral prefrontal cortex (VLPFC), inferior parietal lobule (IPL) andsuperior parietal lobule (SPL). The dorsal stream includes the IFG, PMC, and PT/Spt regions (blue) along with the primary auditory cortex (green)and superior temporal sulcus (yellow). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version ofthis article).

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

107

(Herman et al., 2013). Finally, if underlying dorsal stream timing mechanisms can be identified in oscillatory bands, such methods mayinform frequency specific approaches to neurofeedback or neuromodulation such as transcranial alternating current stimulation (Ali et al.,2013) in tasks designed to manipulate attention/executive function and working memory (e.g., see Metten et al., 2011). In this way, the studyof the relationship between cognitive-load and sensorimotor oscillations may lead to more specific neuromodulatory targets to augmentcurrent behavioral treatment approaches.

6. Conclusions and cautions

The primary aim of the current review was to further explicate how the neurobiological networks implicated in stuttering couldbe related to reported differences in nonword repetition and dual-task performance. To that end, we highlighted a number of findingsfrom nonword repetition and dual-task paradigms that are broadly consistent with a role for sensorimotor and executive capacitiesunderlying PWM. Second, we proposed that current neurobiological models of language, executive function, and neuroimagingfindings in DS are consistent with overall lower capacities on tasks that place a heavy load on PWM and sensorimotor integration.Taken as a whole, there is accumulating evidence that the emergent properties of PWM are related to the dorsal stream and aremodulated by PFC-BG connectivity both of which have been strongly implicated in the pathophysiology and behavior of DS (Etchellet al., 2018). Because behavioral tasks thought to load those same networks may be related to the subtypes/phenotypes of recoveryand persistence in CWS (Spencer & Weber-Fox, 2014) and unassisted recovery in adulthood (Kell et al., 2017), we believe thatattempts to bridge the gap between cognitive-behavioral and brain findings may result in further insights into the cognitive op-erations underlying those important phenomena.

While the review suggests parallels between the capacities mediated by PWM and the neurobiological networks supporting them,some caution is also warranted. First, although we hypothesize that cyclic timing errors in sensorimotor mapping will be related toload-dependent processing in CWS and AWS (e.g. in a nonword repetition task), it is unclear how errors in sensorimotor mapping leadto overt, stutter-typical disfluencies in more naturalistic communicative interactions. The difference in sensorimotor integration forspeech production proposed to account for stuttering and here to mediate lower PWM capacity is also considered subtle and highlyload-dependent. Further, organization for language appears to share features with other cognitive domains, potentially allowing formultiple interactions between linguistic and other cognitive processes in naturalistic speaking contexts (Friederici and Singer, 2015).Overt stuttered moments, as opposed to sensorimotor errors in timing, may also be an emergent process requiring multiple, dynamicinteractions between cognitive-linguistic, cognitive-emotional, and sensorimotor processes (Namasivayam & van Lieshout, 2011;Smith, 1999; Smith & Weber, 2017; Starkweather, 2002; van Lieshout et al., 2004). In other areas of neuroscience, significantadvances in understanding the relationship between brain and behavior have relied on similar concepts of multiple realizabilitydriven by previous theoretical and experimental decomposition of behavior (Krakauer, Ghazanfar, Gomez-Marin, MacIverr, &Poeppel, 2017). As such, to understand the relationship between stuttering and cognitive-sensorimotor capacities, future work willneed to examine the relationship between internal errors in sensorimotor mapping, carefully defined behavioral measures of cog-nitive-sensorimotor load, and the pathway by which multiple interactions between those processes lead to what listeners perceive asovertly stuttered moments.

Another note of caution is that the neurobiological models discussed in this review are themselves the subject of ongoing con-troversy, including ongoing discussion about the functional segregation of the PT region in speech processing (Hickok & Poeppel,2007). There is also evidence that the dorsal stream is more bilaterally organized than the dual-stream model indicates, with differentfunctions in the right and left hemispheres (Ylinen et al., 2014). Similarly, controversies concerning the functional role of variousprefrontal regions in executive function and attentional control are also ongoing and it is likely that different PFC regions (e.g. medialvs. lateral) have different functional roles (Koechlin & Summerfield, 2007). Finally, neuroimaging findings in PWS vary considerablyacross studies (Etchell et al., 2018; Ingham et al., 2012; Neef et al., 2015) and structural imaging methodology is currently limited.For those reasons, differences in particular regions or tracts should be interpreted cautiously (Etchell et al., 2018). Finally, neuralnetworks that are stable within an individual AWS nonetheless differ significantly between AWS, suggesting significant inter-individual differences in neural networks depending on previous experiences with stuttering (Wymbs et al., 2013). Such a finding isnot surprising considering that individual PWS have variable phenomenological experiences (Plexico et al., 2005) and to some extentfunctional reorganization would be expected to reflect those experiences (Ingham et al., 2012; Wymbs et al., 2013).

Despite the above cautions, further investigation of the link between the neurobiological underpinnings of PWM and stutteringhas a number of advantages. First, the study of PWM has the advantage of a rich theoretical history, stable and replicable empiricalparadigms, and supporting neurobiological networks that have also been implicated in both CWS and AWS. Second, because much ofthe neuroimaging literature in stuttering has not been driven by current neurobiological models of language and cognition, theapplication of models outside the field could potentially facilitate collaboration across laboratories. Finally, if PWM is related tovariability in the expression of stuttering, it may aid in explaining interindividual differences in brain networks. The networksimplicated here predict the extent to which relevant information is stored and are associated with individual differences in workingmemory capacity (McNab & Klingberg, 2008). As such, a better understanding of the mechanisms underlying PWM has the potentialto explain how sensorimotor loads or capacities are related to individual differences in brain networks. In the future, investigationsmay continue to aid in identifying CWS who are more likely to persist or recover and AWS that would benefit from cognitive-behavioral or mindfulness training approaches focusing on cognitive-linguistic and cognitive-emotional processes (Boyle, 2011;Menzies et al., 2009).

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

108

TableA1

Sum

mar

yof

nonw

ord

repe

titio

nst

udie

sin

pres

choo

l-age

child

ren

who

stut

ter.

Stud

yA

ge(r

ange

)N

Mat

chin

gD

iffer

ence

inno

nwor

dre

petit

ion

accu

racy

Max

imum

sylla

ble

leng

thD

iffer

ence

inki

nem

atic

mea

sure

sEff

ect

Size

Hak

im&

Ratn

er(2

004)

4;1-

8;4

8CW

S;8

CWN

Sag

e,se

x,m

ater

nale

duca

tion

yes

5N

Md

=1.

41A

nder

son

etal

.(20

06)

3;0-

5;2

12CW

S;12

CWN

Sag

e,se

x,SE

Sye

s5

NM

2=

.29

And

erso

n&

Wag

avic

h(2

010)

3;6-

5;2

9CW

S;14

CWN

Sno

neye

s5

NM

2=

.42

Smith

etal

.(20

12)

4-6

31CW

S;22

CWN

SSE

Sno

*4

yes

NR

Spen

cer&

Web

er-F

ox(2

014)

3;9-

5;8

21CW

S-Re

c;9

CWS-

Per;

25CW

NS

age,

SES,

nonv

erba

lrea

soni

ngye

s4

NM

NR

Pelc

zars

ki&

Yaru

ss(2

016)

5;5-

5;8

11CW

S;11

CWN

Sag

e,la

ngua

geab

ility

,sex

,mat

erna

led

ucat

ion

yes

7N

MN

R

Note.

CWS

=ch

ildre

nw

host

utte

r;CW

NS

=ch

ildre

nw

hodo

not

stut

ter;

CWS-

Rec=

child

ren

who

stut

ter,

reco

vere

d;CW

S=

Per=

child

ren

who

stut

ter

who

pers

ist;

SES

=So

cioe

cono

mic

Stat

us;

NM

=no

tmea

sure

d;N

R=

not

repo

rted

.*

Smith

etal

.,(2

012)

did

notr

epor

tsig

nific

antd

iffer

ence

sin

over

alla

ccur

acy

ofno

nwor

dre

petit

ion

betw

een

CWS

and

CWN

Sgr

oups

,but

did

repo

rta

sign

ifica

ntdi

ffere

nce

forC

WS

with

artic

ulat

ion/

lang

uage

diso

rder

s.

App

endixA

.

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

109

TableA2

Sum

mar

yof

nonw

ord

repe

titio

nst

udie

sin

scho

ol-a

gech

ildre

nw

host

utte

r.

Stud

yA

ge(r

ange

)N

Mat

chin

gD

iffer

ence

inno

nwor

dre

petit

ion

accu

racy

Max

imum

sylla

ble

leng

thD

iffer

ence

inki

nem

atic

mea

sure

sEff

ect

Size

Seer

yet

al.(

2006

)8;

6-12

;614

CWS;

11CW

NS

none

yes*

5N

MN

RBa

khtia

ret

al.(

2007

)5;

1-7;

1012

CWS;

12CW

NS

none

no3

NM

NR

Chon

&A

mbr

ose

(200

7)10

-13

14CW

S-Re

c;5

CWS-

Per

none

no4

NM

NR

Web

er-F

oxet

al.(

2008

)9;

4-13

;910

CWS;

10CW

NS

age,

sex

no**

3N

MN

RO

youn

etal

.(20

10)

5-13

30CW

S;30

CWS

age,

sex

yes**

*3

NM

NR

Sasi

seka

ran

&By

rd(2

013)

8-15

14CW

S;14

CWN

Sag

e,se

xno

****

4N

MN

R

Note.

CWS

=ch

ildre

nw

host

utte

r;CW

NS

=ch

ildre

nw

hodo

not

stut

ter;

CWS-

Rec=

child

ren

who

stut

ter,

reco

vere

d;CW

S=

Per=

child

ren

who

stut

ter

who

pers

ist;

NM

=no

tm

easu

red;

NR

=no

tre

port

ed.

*Se

ery

etal

.,(2

006)

repo

rted

asi

gnifi

cant

inte

ract

ion

betw

een

grou

pan

dsy

llabl

ele

ngth

and

repo

rted

sign

ifica

ntdi

ffere

nces

inpo

st-h

occo

mpa

riso

nsbe

twee

nCW

San

dCW

NS

at5-

sylla

ble

leng

th.

**W

eber

-Fox

etal

.(20

08)

repo

rted

sim

ilar

nonw

ord

repe

titio

nac

cura

cyin

both

grou

psth

atw

asco

mpa

rabl

eto

30CW

NS

inth

esa

me

age

rang

ein

apr

evio

usst

udy.

***

Oyo

unet

al.(

2010

)al

sore

port

eda

sign

ifica

ntco

rrel

atio

nbe

twee

nsu

bjec

tag

ean

dse

veri

tyw

ithno

nwor

dre

petit

ion

perf

orm

ance

.**

**Sa

sise

kara

n&

Byrd

(201

3)fu

rthe

rsub

divi

ded

CWS

into

youn

ger(

7–11

)and

olde

rgro

ups(

11–1

5)an

dre

port

edsi

gnifi

cant

diffe

renc

esin

aph

onem

eel

isio

nta

skbe

twee

nyo

unge

rand

olde

rgro

ups

and

nodi

ffere

nces

betw

een

age

mat

ched

CWN

Sgr

oups

.

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

110

TableA3

Sum

mar

yof

nonw

ord

repe

titio

nst

udie

sin

adul

tsw

host

utte

r.

Stud

yN

Mat

chin

gD

iffer

ence

inno

nwor

dre

petit

ion

accu

racy

Max

imum

sylla

ble

leng

thD

iffer

ence

inki

nem

atic

mea

sure

sEff

ect

Size

Ludl

owet

al.(

1997

)5

AW

S;7

TFS

none

yes

4N

MN

RN

amas

ivay

am&

van

Lies

hout

(200

8)5

AW

S;5

TFS

age,

sex,

educ

atio

nle

vel

NM

4*ye

sd

=1.

64Sm

ithet

al.(

2010

)17

AW

S;17

TFS

age,

sex,

educ

atio

nle

vel

no4

yes

NR

Byrd

etal

.(20

12)

14A

WS;

14TF

Sag

e,se

xye

s7**

NM

2=

.15

Sasi

seka

ran

(201

3)9

AW

S;9

TFS

age,

sex

no4**

*ye

s2

=.3

2Sa

sise

kara

n&

Wei

sber

g(2

014)

10A

WS;

10TF

Sag

e,se

xye

s6

yes

NR

Byrd

etal

.(20

15)

10A

WS;

10TF

Sag

e,se

x,ed

ucat

ion

leve

lye

s7

NM

2=

.38

Coal

son

&By

rd(2

017)

26A

WS;

26TF

Sno

neye

s2**

**N

Md=

.34

Note.

AW

S=

adul

tsw

host

utte

r;TF

S=

typi

calfl

uent

spea

kers

;NM

=no

tm

easu

red;

NR

=no

tre

port

ed.

*N

amas

iyam

&va

nLi

esho

ut(2

008)

repo

rted

prac

tice

effec

tson

two

bisy

llabl

e,bi

labi

alno

nwor

dpr

oduc

tions

.The

effec

tsiz

ere

port

edis

for

am

ain

effec

tof

mov

emen

tam

plitu

de.

**By

rdet

al.,

(201

2)re

port

eda

sign

ifica

ntin

tera

ctio

nbe

twee

ngr

oup

and

sylla

ble

leng

thin

ano

nwor

dco

nditi

onw

ith7

sylla

bles

.**

*Sa

sise

kara

n(2

013)

repo

rted

nodi

ffere

nces

ona

nonw

ord

repe

titio

nta

skbu

tdi

dre

port

eddi

ffere

nces

onki

nem

atic

mea

sure

sdu

ring

ano

nwor

dre

adin

gta

sk.T

heeff

ect

size

repo

rted

isfo

rlip

aper

ture

vari

abili

ty.

****

Coal

son

&By

rd(2

017)

used

only

two

sylla

ble

leng

thsb

utm

anip

ulat

edst

ress

patt

erns

and

repo

rted

asi

gnifi

cant

inte

ract

ion

ofta

lker

grou

p,st

ress

,and

task

indi

catin

gth

atgr

oup

diffe

renc

esw

ere

driv

enby

only

the

mos

tde

man

ding

cond

ition

s(e

.g.,

low

freq

uenc

yst

ress

patt

erns

).

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

111

TableA4

Sum

mar

yof

dual

-task

stud

ies

inad

ults

who

stut

ter.

Stud

yN

Mat

chin

gD

iffer

ence

sin

prim

ary

task

/mea

sure

sD

iffer

ence

sin

seco

ndar

yta

sk/m

easu

res

Effec

tson

stut

teri

ng

Caru

soet

al.(

1994

)9

AW

S;9

TFS

none

yes

(lat

ency

,spe

ech

rate

)no

(Str

oop

colo

rw

ord)

incr

ease

sBo

ssha

rdt

etal

.(19

99)

9A

WS;

10TF

Sm

enta

lcal

cula

tion

perf

orm

ance

yes

(inh

alat

ion

rate

,len

gth

ofbr

eath

grou

p,w

ord

dura

tion)

no(m

enta

lcal

cula

tion)

incr

ease

spr

imar

ilyin

asu

bgro

upof

AW

SD

eN

il&

Boss

hard

t(20

00)

12A

WS;

12TF

Sno

ne* (s

ente

nce

plan

ning

)*

NR

(rhy

me

and

cate

gory

deci

sion

s)*N

MBo

ssha

rdt

etal

.(20

02)

14A

WS;

16TF

Sag

e,ed

ucat

ion,

sex,

mem

ory

span

,vo

cabu

lary

no(p

ause

rate

,inh

alat

ion

rate

)**

yes

(rea

ding

and

mem

oriz

ing

phon

olog

ical

lysi

mila

ran

ddi

ssim

ilar

wor

ds)

incr

ease

s

Boss

hard

tet

al.(

2002

)14

AW

S;16

TFS

age,

educ

atio

n,se

x,di

git

span

,vo

cabu

lary

*** ye

s(#

ofpr

opos

ition

sin

sent

ence

prod

uctio

n)no

(rhy

me

and

cate

gory

deci

sion

s)no

chan

ge

Vasi

c&

Wijn

en(2

005)

22A

WS;

10TF

Sno

neno

(nar

rativ

ere

tell)

****

NM

decr

ease

sSm

its-B

ands

tra

&D

eN

il(2

009)

NM

Jone

set

al.(

2012

)9

AW

S;9

TFS

sex,

educ

atio

nalb

ackg

roun

dno

(rhy

me

judg

emen

t)ye

s(l

ette

rre

call)

NM

Max

field

(201

6)15

AW

S;15

TFS

none

yes

(nam

ing

accu

racy

)no

(ton

eju

dgem

enta

ccur

acy)

NM

Eich

orn

etal

.(20

16)

20A

WS;

20TF

Sno

ne**

*** no

(spe

ech

erro

rs)

no(s

patia

lwor

king

mem

ory

and

digi

tsp

anta

sks)

decr

ease

s

Note.

AW

S=

adul

tsw

host

utte

r;TF

S=

typi

calfl

uent

spea

kers

;NM

=no

tm

easu

red;

NR

=no

tre

port

ed.

*D

eN

il&

Boss

hard

t(20

00)u

sed

ase

nten

cepl

anni

ngta

skw

here

inno

beha

vior

alm

easu

resw

ere

avai

labl

e.N

om

easu

reso

flan

guag

epr

oduc

tion,

stut

teri

ng,o

rsta

tistic

alte

stso

frhy

min

gan

dca

tego

ryde

cisi

ons

wer

ere

port

ed.

**Bo

ssha

rdt

etal

.(20

02)

repo

rted

sign

ifica

ntly

low

erw

ord

reca

llac

cura

cyin

the

TFS

grou

pco

mpa

red

toth

eA

WS

grou

p.**

*Bo

ssha

rdte

tal.

(200

2)re

port

edsi

gnifi

cant

lya

high

erpe

rcen

tage

ofst

utte

red

sylla

bles

inth

eA

WS

grou

pre

lativ

eto

cont

rols

butd

ual-t

asks

wer

eno

tass

ocia

ted

with

mor

est

utte

ring

than

sent

ence

prod

uctio

n.**

**Va

sic

&W

ijnen

(200

5)us

eda

dist

ract

ing

vide

oga

me

insi

mpl

ean

dm

ore

chal

leng

ing

cond

ition

sbu

tdid

notr

epor

tbe

havi

oral

perf

orm

ance

onth

eta

sk.

****

*Ei

chor

net

al.(

2016

)di

dno

trep

ort

mor

esp

eech

erro

rsin

dual

-task

sco

nditi

ons

for

AW

Sre

lativ

eto

cont

rols

but

did

repo

rtlo

wer

spee

chra

tein

the

AW

Sgr

oup

com

pare

dto

cont

rols

.

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

112

References

D’Esposito, M. (2007). From cognitive to neural models of working memory. Philosophical Transactions Biological Sciences, 362(1481), 761–772. https://doi.org/10.1098/rstb.2007.2086.

Adams, A., & Gathercole, S. (1995). Phonological working memory and speech production in preschool children. Journal of Speech Language and Hearing Research, 38,403–414. https://doi.org/10.1044/jshr.3802.403.

Ali, M. M., Sellers, K. K., & Frohlich, M. (2013). Transcranial alternating current stimulation modulates large-scale cortical network activity by network resonance.Journal of Neuroscience, 33(27), 11262–11275. https://doi.org/10.1523/JNEUROSCI.5867-12.2013.

Alloway, T., Gathercole, S. E., Adams, A., Willis, C., Eaglen, R., & Lamont, E. (2005). Working memory and phonological awareness as predictors of progress towardsearly learning goals at school entry. The British Journal of Developmental Psychology, 23, 417–426. https://doi.org/10.1348/026151005X26804.

Alm, P. (2004). Stuttering and the basal ganglia circuits: A critical review of possible relations. Journal of Communication Disorders, 37, 325–369. https://doi.org/10.1016/j.jcomdis.2004.03.001.

Alm, P. (2014). Stuttering in relation to anxiety, temperament, and personality: Review and analysis with focus on causality. Journal of Fluency Disorders, 40, 5–21.https://doi.org/10.1016/j.jfludis.2014.01.004.

Alm, P., & Risberg, J. (2007). Stuttering in adults: The acoustic startle response, temperamental traits, and biological factors. Journal of Communication Disorders, 40,1–41. https://doi.org/10.1016/j.jcomdis.2006.04.001.

Ambrose, N. G., Yairi, E., Loucks, T. M., Seery, C. H., & Throneburg, R. (2015). Relation of motor, linguistic and temperament factors in epidemiologic subtypes ofpersistent and recovered stuttering: Initial findings. Journal of Fluency Disorders, 45, 12–26. https://doi.org/10.1016/j.jfludis.2015.05.004.

Anderson, J. D., & Wagovich, S. A. (2010). Relationships among linguistic processing speed, phonological working memory, and attention in children who stutter.Journal of Fluency Disorders, 35(3), 216–234. https://doi.org/10.1016/j.jfludis.2010.04.003.

Anderson, J., Wagovich, S., & Hall, N. (2006). Nonword repetition skills in young children who do and do not stutter. Journal of Fluency Disorders, 31, 177–199.https://doi.org/10.1016/j.jfludis.2006.05.001.

Arends, N., Povel, D. J., & Kolk, H. (1988). Stuttering as an attentional phenomenon. Journal of Fluency Disorders, 13, 141–151. Retrieved from: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.442.6419&rep=rep1&type=pd.

Baddeley, A. (2003). Working memory: Looking back and looking forward. Nature Reviews Neuroscience, 4, 829–839. https://doi.org/10.1038/nrn1201.Baddeley, A. (2012). Working memory: Theories, models, and controversies. Annual Review of Psychology, 63, 1–29. https://doi.org/10.1146/annurev-psych-120710-

100422.Baier, B., Karnath, H., Dieterich, M., & Birklein, F. (2010). Keeping memory clear and stable—The contribution of human basal ganglia and prefrontal cortex to

working memory. Journal of Neuroscience, 30, 9788–9792. https://doi.org/10.1523/JNEUROSCI.1513-10.2010.Bajaj, A. (2007). Working memory involvement in stuttering: Exploring the evidence and research implications. Journal of Fluency Disorders, 32, 218–238.Bakhtiar, M., Abad, A. D., & Panahi, M. S. (2007). Nonword repetition ability of children who do and do not stutter and covert repair hypothesis. Indian Journal of

Medical Science, 61(8), 462–470.Beal, D., Gracco, V., Lafaille, S., & Luc, F. (2007). Voxel-based morphometry of auditory and speech-related cortex in stutterers. Neuroreport, 18, 1257–1260. https://

doi.org/10.1097/WNR.0b013e3282202c4d.Beal, 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, 89. https://doi.org/10.3389/fnhum.2015.00089.Belyk, M., Kraft, S., & Brown, S. (2015). Stuttering as a trait or state – An ALE metaanalysis of neuroimaging studies. The European Journal of Neuroscience, 41, 275–284.

https://doi.org/10.1111/ejn.12765/full.Bloodstein, O., & Bernstein-Ratner, N. (2008). A handbook on stuttering. New York: Thomson Delmar Learning.Bohland, J. W., Bullock, D., & Guenther, F. H. (2010). Neural representations and mechanisms for the performance of simple speech sequences. Journal of Cognitive

Neuroscience, 22(7), 1504–1529. https://doi.org/10.1162/jocn.2009.21306.Bosshardt, H. (1999). Effects of concurrent mental calculation on stuttering, inhalation, and speech timing. Journal of Fluency Disorders, 24, 43–72. https://doi.org/10.

1016/S0094-730X(98)00022-9.Bosshardt, H. (2002). Effects of concurrent cognitive processing on the fluency of word repetition: Comparison between persons who do and do not stutter. Journal of

Fluency Disorders, 27(2), 93–114. https://doi.org/10.1016/S0094-730X(02)00113-4.Bosshardt, H. (2006). Cognitive processing load as a determinant of stuttering: Summary of a research programme. Clinical Linguistics & Phonetics, 20, 371–385.

https://doi.org/10.1080/02699200500074321.Boyle, M. (2011). Mindfulness training in stuttering therapy: A tutorial for speech-language pathologists. Journal of Fluency Disorders, 36, 122–129. https://doi.org/10.

1016/j.jfludis.2011.04.005.Brown, S., Ingham, R., & Ingham, J. (2005). Stuttered and fluent speech production: An ALE meta-analysis of functional neuroimaging studies. Human Brain Mapping,

25, 105–117. https://doi.org/10.1002/hbm.20140/full.Brutten, G. J., & Trotter, A. C. (1985). Hemispheric interference: A dual-task investigation of youngsters who stutter. Journal of Fluency Disorders, 10, 77–85. https://

doi.org/10.1016/0094-730X(85)90016-6.Brutten, G. J., & Trotter, A. C. (1986). A dual-task investigation of young stutterers and nonstutterers. Journal of Fluency Disorders, 11, 275–284. https://doi.org/10.

1016/0094-730X(86)90015-X.Buchsbaum, B., & D’Esposito, M. (2008). The search for the phonological store: From loop to convolution. Journal of Cognitive Neuroscience. https://doi.org/10.1162/

jocn.2008.20501.Buchsbaum, B. R., Hickok, G., & Humphries, C. (2001). Role of the posterior superior temporal gyrus in phonological processing for speech perception and production.

Cognitive Science, 25, 663–678. https://doi.org/10.1207/s15516709cog2505_2.Buchsbaum, B. R., Baldo, J., Okada, K., Berman, K. F., Dronkers, N., D’ Esposito, M., & Hickok, G. (2011). Conduction aphasia, sensory-motor integration, and

phonological short-term memory – An aggregate analysis of lesion and fMRI data. Brain and Language, 119(3), 119–128. https://doi.org/10.1016/j.bandl.2010.12.001.

Budde, 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. https://doi.org/10.1016/j.bandl.2014.10.002.

Byrd, C., McGill, M., & Usler, E. (2015). Nonword repetition and phoneme elision in adults who do and do not stutter: Vocal versus nonvocal performance differences.Journal of Fluency Disorders, 44, 17–31. https://doi.org/10.1016/j.jfludis.2015.01.004.

Byrd, C., Vallely, M., Anderson, J., & Sussman, H. (2012). Nonword repetition and phoneme elision in adults who do and do not stutter. Journal of Fluency Disorders,37(3), 188–201. https://doi.org/10.1016/j.jfludis.2012.03.003.

Cai, 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 fornetwork-level anomalies. Frontiers in Human Neuroscience, 8, 54. https://doi.org/10.3389/fnhum.2014.00054.

Caruso, A., & Chodzko-Zajko, W. (1994). Emotional arousal and stuttering: The impact of cognitive stress. Journal of Fluency Disorders, 19(3), 162–170.Chang, S. E., Angstadt, M., Chow, H. M., Etchell, A. C., Garnett, E. O., Choo, A. L., & Sripada, C. (2017). Anomalous network architecture of the resting brain in children

who stutter. Journal of Fluency Disorders. https://doi.org/10.1016/j.jfludis.2017.01.002.Chang, S., Chow, H. M., Wieland, E., & McAuley, D. (2016). Relation between functional connectivity and rhythm discrimination in children who do and do not stutter.

NeuroImage Clinical, 12, 442–450. https://doi.org/10.1016/j.nicl.2016.08.021.Chang, S., Erikson, K., Ambrose, N., Hasegawa-Johnson, M. A., & Ludlow, C. (2008). Brain anatomy differences in childhood stuttering. NeuroImage, 39, 1333–1344.

https://doi.org/10.1016/j.neuroimage.2007.09.067.Chang, S., Kenney, M., Loucks, T., & Ludlow, C. (2009). Brain activation abnormalities during speech and non-speech in stuttering speakers. NeuroImage, 46(1),

201–212. https://doi.org/10.1016/j.neuroimage.2009.01.066.

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

113

Chang, S., Zhu, D., Choo, A., & Angstadt, M. (2015). White matter neuroanatomical differences in young children who stutter. Brain, 138(3), 694–711. https://doi.org/10.1093/brain/awu400.

Chang, S., & Zhu, D. (2013). Neural network connectivity differences in children who stutter. Brain, 136(12), 3709–3726. https://doi.org/10.1093/brain/awt275.Chon, H., & Ambrose, N. (2007). Nonword repetition abilities of children with persistent and recovered stuttering: Preliminary study. Poster Presented at the American

Speech-Language-Hearing Association Convention. Retrieved from https://scholar.google.com/scholar?q=Chon+and+Ambrose+2007+stuttering&btnG=&hl=en&as_sdt=0%2C4.

Choo, A. L., Burnham, E., Hicks, K., & Chang, S. E. (2016). Dissociations among linguistic, cognitive, and auditory-motor neuroanatomical domains in children whostutter. Journal of Communication Disorders, 61, 29–47. https://doi.org/10.1016/j.jcomdis.2016.03.003.

Chrysikou, E. G., Weber, M. J., & Thompson-Schill, S. L. (2014). A matched filter hypothesis for cognitive control. Neuropsychologia, 62, 341–355. https://doi.org/10.1016/j.neuropsychologia.2013.10.021.

Civier, O., Bullock, D., Max, L., & Guenther, F. H. (2013). Computational modeling of stuttering caused by impairments in a basal ganglia thalamo-cortical circuitinvolved in syllable selection and initiation. Brain and Language, 126(3), 263–278. https://doi.org/10.1016/j.bandl.2013.05.016.

Civier, O., Tasko, S., & Guenther, F. (2010). Overreliance on auditory feedback may lead to sound/syllable repetitions: Simulations of stuttering and fluency-inducingconditions with a neural model of speech. Journal of Fluency Disorders, 35(3), 246–279. https://doi.org/10.1016/j.jfludis.2010.05.002.

Coalson, G. A., & Byrd, C. T. (2017). Nonword repetition in adults who stutter: The effects of stimuli stress and auditory-orthographic cues. PloS One, 12(11),e0188111. https://doi.org/10.1371/journal.pone.0188111.

Connally, E. L., Ward, D., Howell, H., & Watkins, K. E. (2014). Disrupted white matter in language and motor tracts in developmental stuttering. Brain and Language,131, 25–35. https://doi.org/10.1016/j.bandl.2013.05.013.

Cowan, N. (1999). An embedded-processes model of working memory. In A. Mikaye, & P. Shah (Eds.). Models of working memory: Mechanisms of active (pp. 62–101).Cambridge, UK: Cambridge University Press.

Cowan, N., Elliott, E., Saults, J., Morey, C., & Mattox, S. (2005). On the capacity of attention: Its estimation and its role in working memory and cognitive aptitudes.Cognitive Psychology, 51(1), 42–100. https://doi.org/10.1016/j.cogpsych.2004.12.001.

Cykowski, M. D., Kochunov, P. V., Ingham, R. J., Ingham, J. C., Mangin, J. F., Rivere, D., ... Fox, P. T. (2008). Perisylvian sulcual morphology and cerebral asymmetrypatterns in adults who stutter. Cerebral Cortex, 18, 571–583. https://doi.org/10.1093/cercor/bhm093.

De Nil, L., & Bosshardt, H. G. (2000). Studying stuttering from a neurological and cognitive information processing perspective. In H.-G. Bosshardt, J. S. Yaruss, & H. F.M. Peters (Eds.). fluency disorders: Theory, research, treatment and self-help. Proceedings of the Third world congress of fluency disorders in Nyborg, Denmark (pp. 53–58).Nijmegen: Nijmegen University Press.

Desai, J., Huo, Y., Wang, Z., Bansal, R., Williams, S., Lythgoe, D., Fernando, Z., & Peterson, B. (2016). Reduced perfusion in Broca’s area in developmental stuttering.Human Brain Mapping, 38, 1865–1874. https://doi.org/10.1002/hbm.23487.

Dollaghan, C., & Campbell, T. F. (1998). Nonword repetition and child language impairment. Journal of Speech Language and Hearing Research, 41, 1136–1146. https://doi.org/10.1044/jslhr.4105.1136.

Eggers, K., De Nil, L., & Bergh, B. (2012). The efficiency of attentional networks in children who stutter. Journal of Speech Language and Hearing Research, 55, 946–959.https://doi.org/10.1044/1092-4388(2011/10-0208).

Eichorn, N., Martin, K., & Pirutinsky, S. (2017). Cognitive flexibility in preschool children with and without stuttering disorders. Journal of Fluency Disorders. https://doi.org/10.1016/j.jfludis.2017.11.001.

Eichorn, N., Marton, K., Schwartz, R., & Melara, R. (2016). Does working memory enhance or interfere with speech fluency in adults who do and do not stutter?Evidence from a dual-task paradigm. Journal of Speech Language and Hearing Research, 59, 415–429. https://doi.org/10.1044/2015_JSLHR-S-15-0249.

Etchell, A., Civier, O., Ballard, K., & Sowman, P. (2018). A systematic literature review of neuroimaging research on developmental stuttering between 1995 and 2016.Journal of Fluency Disorders. https://doi.org/10.1016/j.jfludis.2017.03.007.

Etchell, A., Johnson, B., & Sowman, P. (2014). Behavioral and multimodal neuroimaging evidence for a deficit in brain timing networks in stuttering: A hypothesis and theory.https://doi.org/10.3389/fnhum.2014.00467.

Etchell, A., Ryan, M., Martin, E., & Johnson, B. (2016). Abnormal time course of low beta modulation in non-fluent preschool children: A magnetoencephalographicstudy of rhythm tracking. Neuroimage. Retrieved from http://www.sciencedirect.com/science/article/pii/S1053811915010174.

Foundas, A., Bollich, A., Corey, D., & Hurley, M. (2001). Anomalous anatomy of speech–language areas in adults with persistent developmental stuttering. Neurology.Retrieved from http://www.neurology.org/content/57/2/207.1.short.

Foundas, A., Bollich, A., Feldman, J., & Corey, D. (2004). Aberrant auditory processing and atypical planum temporale in developmental stuttering. Neurology.Retrieved from http://www.neurology.org/content/63/9/1640.short.

Foundas, A., Cindass, R., Mock, J., & Corey, D. (2013). Atypical caudate anatomy in children who stutter. Perceptual and Motor Skills: Physical Development andMeasurement, 116(2), 528–543. https://doi.org/10.2466/15.10.PMS.116.2.528-543.

Friederici, A., & Singer, W. (2015). Grounding language processing on basic neurophysiological principles. Trends in Cogntive Sciences, 19(6), 329–338.Fries, P. (2005). A mechanism for cognitive dynamics: Neuronal communication through neuronal coherence. Trends in Cognitive Sciences, 9(10), 474–480. https://doi.

org/10.1016/j.tics.2005.08.011.Gathercole, S., Willis, C., Baddeley, A., & Emslie, H. (1994). The children’s test of nonword repetition: A test of phonological working memory. Memory, 2(2), 103–127.

https://doi.org/10.1080/09658219408258940 Retrieved from.Giraud, A., Neumann, K., & Bachoud-Levi, A. (2008). Severity of disfluency correlates with basal ganglia activity in persistent developmental stuttering. Brain and

Language, 104(2), 190–199. Retrieved from http://www.sciencedirect.com/science/article/pii/S0093934X07000697.Gough, P. M., Connally, E. L., Howell, P., Ward, D., Chesters, J., & Watkins, K. E. (2018). Plaum temporale asymmetry in people who stutter. Journal of Fluency

Disorders, 55. https://doi.org/10.1016/j.jfludis.2017.06.003 94.105.Guenther, F., Ghosh, S., & Tourville, J. (2006). Neural modeling and imaging of the cortical interactions underlying syllable production. Brain and Language, 96(3),

280–301. https://doi.org/10.1016/j.bandl.2005.06.001.Hakim, H. B., & Ratner, N. B. (2004). Nonword repetition abilities of children who stutter: An exploratory study. Journal of Fluency Disorders, 29, 179–199. https://doi.

org/10.1016/j.jfludis.2004.06.001.Halag-Milo, T., Stoppelman, N., Kronfeld-Duenias, V., Civier, O., Amir, O., Ezrati-Vinacour, R., & Ben-Shachar, M. (2016). Beyond production: Brain responses during

speech perception in adults who stutter. NeuroImage Clinical, 11, 328–338. https://doi.org/10.1016/j.nicl.2016.02.017.Hazy, T., Frank, M., & O’Reilly, R. (2007). Towards an executive without a homunculus: Computational models of the prefrontal cortex/basal ganglia system.

Philosophical Transactions Biological Sciences, 362, 1601–1613. https://doi.org/10.1098/rstb.2007.2055.Herman, A. B., Houde, J. F., Vinogradov, S., & Nagarajan, S. S. (2013). Parsing the phonological loop: Activation timing in the dorsal speech stream determines

accuracy in speech reproduction. Journal of Neuroscience, 33(13), 5439–5453. https://doi.org/10.1523/JNEUROSCI.1472-12.2013.Hickok, G. (2012). Computational neuroanatomy of speech production. Nature Reviews Neuroscience, 13, 135–145. https://doi.org/10.1038/nrn3158.Hickok, G., & Poeppel, D. (2007). The cortical organization of speech processing. Nature Reviews Neuroscience, 8, 393–402. Retrieved from http://www.nature.com/

nrn/journal/v8/n5/abs/nrn2113.html.Hickok, G., Buchsbaum, B., Humphries, C., & Muftuler, T. (2003). Auditory–Motor interaction revealed by fMRI: Speech, music, and working memory in area Spt.

Journal of Cognitive Neuroscience, 15(5), 673–682. https://doi.org/10.1162/jocn.2003.15.5.673.Hickok, G., Houde, J., & Rong, F. (2011). Sensorimotor integration in speech processing: Computational basis and neural organization. Neuron, 69(3), 407–422.

https://doi.org/10.1016/j.neuron.2011.01.019.Houde, J. F., & Nagarajan, S. S. (2011). Speech production as state feedback control. Frontiers in Human Neuroscience, 5, 82. https://doi.org/10.3389/fnhum.2011.

00082.Howell, P., & Au-Yeung, J. (1995). The association between stuttering, Brown’s factors, and phonological categories in child stutterers ranging in age between 2 and 12

years. Journal of Fluency Disorders, 20(4), 331–344. https://doi.org/10.1016/0094-730X(94)00016-M.

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

114

Howell, P., & Au-Yeung, J. (2002). The EXPLAN theory of fluency control applied to the diagnosis of stuttering. In E. Fave (Ed.). Clinical linguistics: Theory andapplications in speech pathology and therapy (pp. 75–94). Philadelphia, PA: John Benjamins Publishing Company.

Ingham, R., Grafton, S., Bothe, A., & Ingham, J. (2012). Brain activity in adults who stutter: Similarities across speaking tasks and correlations with stutteringfrequency and speaking rate. Brain and Language, 122(1), 11–24. https://doi.org/10.1016/j.bandl.2012.04.002.

Ingham, R. J., Fox, P. T., Ingham, J. C., Xiong, J., Zamarripa, F., Hardies, & Lancaster, J. (2004). Brain correlates of stuttering and syllable production. Journal of SpeechLanguage and Hearing Research, 47(2), 321. https://doi.org/10.1044/1092-4388(2004/026).

Jacquemot, C., & Scott, S. K. (2006). What is the relationship between phonological short-term memory and speech processing? Trends in Cognitive Sciences, 10(11),480–486. https://doi.org/10.1016/j.tics.2006.09.002.

Jäncke, L., Hänggi, J., & Steinmetz, H. (2004). Morphological brain differences between adult stutterers and non-stutterers. BMC Neurology. Retrieved from https://bmcneurol.biomedcentral.com/articles/10.1186/1471-2377-4-23.

Jones, R., Fox, R., & Jacewicz, E. (2012). The effects of concurrent cognitive load on phonological processing in adults who stutter. Journal of Speech Language andHearing Research, 55, 1862–1875. https://doi.org/10.1044/1092-4388(2012/12-0014).

Jones, R. M., Walden, T. A., Conture, E., Erdemir, A., Lambert, W. E., & Porges, S. W. (2017). Executive functions impact the relation between respiratory sinusarrhythmia and frequency of stuttering in young children who do and do not stutter. Journal of Speech Language and Hearing Research, 60, 2133–2150. https://doi.org/10.1044/2017_JSLHR-S-16-0113.

Joos, K., De Ridder, D., Boey, R., & Vanneste, S. (2014). Functional connectivity changes in adults with developmental stuttering: A preliminary study using quan-titative electroencephalography. Frontiers in Human Neuroscience, 8, 1–9. https://doi.org/10.3389/fnhum.2014.00783.

Kell, C., Neumann, K., & Behrens, M. (2017). Speaking-related changes in cortical functional connectivity associated with assisted and spontaneous recovery fromdevelopmental stuttering. Journal of Fluency Disorders. Retrieved from http://www.sciencedirect.com/science/article/pii/S0094730X16300948.

Kell, C., Neumann, K., von Kriegstein, K., Posenenske, C., von Gudenberg, A. W., Euler, H., & Giraud, A. L. (2009). How the brain repairs stuttering. Brain, 132(10),2747–2760. https://doi.org/10.1093/brain/awp185.

Kent, R. (1984). Stuttering as a temporal programming disorder. In R. F. Curlee, & W. H. Perkins (Eds.). Nature and treatment of stuttering: New directions (pp. 283–301).San Diego, CA: College-Hill Press.

Kikuchi, Y., Ogata, K., Umesaki, T., Yoshiura, T., Kenjo, M., Hirano, Y., Okamoto, T., Komune, S., & Tobimatsu, S. (2011). Spatiotemporal signatures of an abnormalauditory system in stuttering. Neuroimage, 55, 891–899. https://doi.org/10.1016/j.neuroimage.2010.12.083.

Kleinow, J., & Smith, A. (2000). Influences of length and syntactic complexity on the speech motor stability of the fluent speech of adults who stutter. Journal of SpeechLanguage and Hearing Research, 43, 548–559. https://doi.org/10.1044/jslhr.4302.548.

Koechlin, E., & Summerfield, C. (2007). An information theoretical approach to prefrontal executive function. Trends in Cognitive Sciences, 11, 229–235. Retrieved fromhttp://www.sciencedirect.com/science/article/pii/S1364661307001052.

Kolk, H., & Postma, A. (1997). Stuttering as a covert repair phenomenon. Nature and Treatment of Stuttering: New Directions, 182–203. Retreived from: https://ci.nii.ac.jp/naid/110003297375.

Krakauer, J., Ghazanfar, A., Gomez-Marin, A., MacIver, M. A., & Poeppel, D. (2017). Neuroscience needs behavior: Correcting a reductionist bias. Neuron, 93, 480–490.Kronfeld-Duenias, V., Amir, O., & Ezrati-Vinacour, R. (2017). The dorsal language pathways in stuttering: Response to commentary. Cortex. https://doi.org/10.1016/j.

cortex.2017.01.026.Long, M., Katlowitz, K., Svirsky, M., & Clary, R. (2016). Functional segregation of cortical regions underlying speech timing and articulation. Neuron, 89(6),

1187–1193. https://doi.org/10.1016/j.neuron.2016.01.032.Lu, C., Chen, C., Peng, D., You, W., Zhang, X., Ding, G., ... Howell, P. (2012). Neural anomaly and reorganization in speakers who stutter: A short-term intervention

study. Neurology, 79(7), 625–632. https://doi.org/10.1212/WNL.0b013e31826356d2.Lu, C., Long, Y., Zheng, L., Shi, G., & Liu, L. (2016). Relationship between speech production and perception in people who stutter. Frontiers in Human Neuroscience, 10,

224. https://doi.org/10.3389/fnhum.2016.00224.Ludlow, C., Siren, K., & Zikria, M. (1997). Speech production learning in adults with chronic developmental stuttering. In W. Hulstijn, H. F. M. Peters, & P. van

Lieshout (Eds.). Speech production: Motor control, brain research and fluency disorders (pp. 221–230). NY: Elsevier.Majerus, S. (2013). Language repetition and short-term memory: An integrative framework. Frontiers in Human Neuroscience, 7, 357. https://doi.org/10.3389/fnhum.

2013.00357.Markiewicz, C. J., & Bohland, J. W. (2016). Mapping the cortical representation of speech sounds in a syllable repetition task. Neuroimage, 141, 174–190. https://doi.

org/10.1016/j.neuroimage.2016.07.023.Max, L., Guenther, F., & Gracco, V. (2004). Unstable or insufficiently activated internal models and feedback-biased motor control as sources of disfluency: A

theoretical model of stuttering. Contemporary Issues in Communication Sciences and Disorders, 31, 105–122. Retrieved from http://vivavoceinstitute.com/wp-content/uploads/2016/01/LMax04a.pdf.

Maxfield, N., Olsen, W., Kleinman, D., & Frisch, S. (2016). Attention demands of language production in adults who stutter. Clinical Neurophysiology, 127(4),1942–1960. https://doi.org/10.1016/j.clinph.2016.01.016.

McNab, F., & Klingberg, T. (2008). Prefrontal cortex and basal ganglia control access to working memory. Nature Neuroscience, 11(1), 103–107. https://doi.org/10.1038/nn2024.

Menzies, R. G., Onslow, M., Packman, A., & O’Brian, S. (2009). Cognitive behavior therapy for adults who stutter: A tutorial for speech-language pathologists. Journalof Fluency Disorders, 34, 187–200. https://doi.org/10.1016/j.jfludis.2009.09.002.

Mersov, A., Cheyne, D., Jobst, C., & De Nil, L. (2017). A preliminary study on the neural oscillatory characteristics of motor preparation prior to dysfluent and fluentutterances in adults who stutter. Journal of Fluency Disorders. https://doi.org/10.1016/j.jfludis.2017.05.003.

Metten, C., Bosshardt, H. G., Jones, M., Eisenhuth, J., Block, S., Carey, B., & …Menzies, R. (2011). Dual tasking and stuttering: From the laboratory to the clinic.Disability and Rehabilitation, 33, 933–944. https://doi.org/10.3109/09638288.2010.515701.

Mink, J. (1996). The basal ganglia: Focused selection and inhibition of competing motor programs. Progress in Neurobiology, 50, 381–425. Retrieved from http://www.sciencedirect.com/science/article/pii/S0301008296000421.

Mock, J., Zadina, J., Corey, D., Cohen, J., Lemon, L., & Foundas, A. (2012). Atypical brain torque in boys with developmental stuttering. DevelopmentalNeuropsychology, 37(5), 434–452. https://doi.org/10.1080/87565641.2012.661816.

Mohan, R., & Weber, C. (2015). Neural systems mediating processing of sound units of language distinguish recovery versus persistence in stuttering. Journal ofNeurodevelopmental Disorders, 7(28), https://doi.org/10.1186/s11689-015-9124-7.

Morillon, B., & Schroeder, C. (2015). Neuronal oscillations as a mechanistic substrate of auditory temporal prediction. Annals of the New York Academy of Sciences,1337(1), 26–31.

Namasivayam, A., & van Lieshout, P. (2008). Investigating speech motor practice a-nd learning in people who stutter. Journal of Fluency Disorders, 33, 32–51. https://doi.org/10.1016/j.jfludis.2007.11.005.

Namasivayam, A., & van Lieshout, P. (2011). Speech motor skill and stuttering. Journal of Motor Behavior, 43(6), 477–489. https://doi.org/10.1080/00222895.2011.628347 Retrieved from.

Nee, D., Brown, J., Askren, M., & Berman, M. (2013). A meta-analysis of executive components of working memory. Cerebral Cortex, 23(2), 264–282. https://doi.org/10.1093/cercor/bhs007.

Neef, N. E., Sommer, M., Neef, A., Paulaus, W., von Gudenburg, A. W., Jung, K., & Wustenberg, T. (2012). Reduced speech perceptual acuity for stop consonants inindividuals who stutter. Journal of Speech Language and Hearing Research, 55, 276–279.

Neef, N. E., Anwander, A., Butfering, C., Schmidt-Samoa, C., Friederici, A. D., & Paulus, W. (2017). Structural connectivity of right frontal hyperactive areas scales withstuttering severity. Brain, 141, 191–204. https://doi.org/10.1093/brain/awx316.

Neef, N. E., Anwander, A., & Friederici, A. D. (2015). The neurobiological grounding of persistent stuttering: From structure to function. Current Neurology andNeuroscience Reports, 15(9), 63. https://doi.org/10.1007/s11910-015-0579-4.

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

115

Nippold, M. (2002). ). Stuttering and phonology: Is there an interaction? American Journal of Speech-language Pathology, 11, 99–110. https://doi.org/10.1044/1058-0360(2002/011.

Nippold, M. (2012). Stuttering and language ability in children: Questioning the connection. American Journal of Speech-language Pathology, 21, 183–196. https://doi.org/10.1044/1058-0360(2012/11-0078.

O’Neill, J., Dong, Z., Bansal, R., Ivanov, I., Hao, X., Desai, J., Pozzi, E., & Peterson, B. (2017). Proton chemical shift imaging of the brain in pediatric and adultdevelopmental stuttering. JAMA Psychiatry, 74, 85–94. https://doi.org/10.1001/jamapsychiatry.2016.3199.

Oyoun, H. A., El Dessouky, H., Sahar, S., & Fawzy, A. (2010). Assessment of working memory in normal children and children who stutter. The Journal of AmericanScience, 6(11), 562–569.

Özge, A., Toros, F., & Çömelekoğlu, 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. Retrieved from http://www.springerlink.com/index/R0J7487713874809.pdf.

Page, M. P. A., Madge, A., Cumming, N., & Norris, D. G. (2007). Speech errors and the phonological similarity effect in short-term memory: Evidence suggesting acommon locus. Journal of Memory and Language, 56(1), 49–64. https://doi.org/10.1016/j.jml.2006.09.002.

Pelczarski, K. M., & Yaruss, S. J. (2016). Phonological memory in young children who stutter. Journal of Communication Disorders, 62, 54–66. https://doi.org/10.1016/j.jcomdis.2016.05.006.

Perkins, W., Kent, R. D., & Curlee, R. F. (1991). A theory of neuropsycholinguistic function in stuttering. Journal of Speech Language and Hearing Research, 34, 734–752.https://doi.org/10.1044/jshr.3404.734.

Perrachione, T. K., Ghosh, S. S., Ostrovskaya, I., Gabrieli, J. D. E., & Kovelman, I. (2017). Phonological working memory for words and nonwords in cerebral cortex.Journal of Speech Language and Hearing Research, 60, 1959–1979. https://doi.org/10.1044/2017_JSLHR-L-15-0446.

Plexico, L., Manning, W., & DiLollo, A. (2005). A phenomenological understanding of successful stuttering management. Journal of Fluency Disorders, 30, 1–22. https://doi.org/10.1016/j.jfludis.2004.12.001.

Riley, G. D. (2009). Stuttering severity instrument (fourth edition). Austin, TX: Pro-Ed.Saltuklaroglu, T., Harkrider, A., Thornton, D., & Jenson, D. (2017). EEG Mu (μ) rhythm spectra and oscillatory activity differentiate stuttering from non-stuttering

adults. NeuroImage. https://doi.org/10.1016/j.neuroimage.2017.04.022.Sasisekaran, J. (2013). Nonword repetition and nonword reading abilities in adults who do and do not stutter. Journal of Fluency Disorders, 38, 275–289. https://doi.

org/10.1016/j.jfludis.2013.06.001.Sasisekaran, J., & Byrd, C. T. (2013). Nonword repetition and phoneme elision skills in school-age children who do and do not stutter. International Journal of Language

& Communication Disorders, 48, 625–639. https://doi.org/10.1111/1460-6984.12035.Sasisekaran, J., & Weisberg, S. (2014). Practice and retention of nonwords in adults who stutter. Journal of Fluency Disorders, 41, 55–71. https://doi.org/10.1016/j.

jfludis.2014.02.004.Sato, Y., Mori, K., Koizumi, T., & Minagawa-Kawai, Y. (2011). Functional lateralization of speech processing in adults and children who stutter. Frontiers in Psychology,

2, 70. https://doi.org/10.3389/fpsyg.2011.00070.Seery, C., Watkins, R., Ambrose, N., & Throneburg (2006). Nonword repetition by school-age children who stutter. Poster Presented at the Annual Meeting of the American

Speech-Language and Hearing Association (ASHA) Convention.Siegel, M., Donner, T. H., & Engel, A. K. (2012). Spectral fingerprints of large-scale neuronal interactions. Nature Reviews Neuroscience. https://doi.org/10.1038/

nrn3137.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. https://doi.org/10.3389/fnhum.2016.00190.

Smith, A. (1999). Stuttering: A unified approach to a multifactorial, dynamic disorder. In N. Bernstein-Ratner, & E. C. Healey (Eds.). Stuttering research and practice:Bridging the gap (pp. 27–44). New York, NY: Psychology Press.

Smith, A., & Weber, C. (2017). How stuttering develops: The multifactorial dynamic pathways theory. Journal of Speech Language and Hearing Research, 60, 2483–2505.https://doi.org/10.1044/2017_JSLHR-S-16-0343.

Smith, A., Goffman, L., Sasisekaran, J., & Weber-Fox, C. (2012). Language and motor abilities of preschool children who stutter: Evidence from behavioral andkinematic indices of nonword repetition performance. Journal of Fluency Disorders, 37, 344–358. https://doi.org/10.1016/j.jfludis.2012.06.001.

Smith, A., Sadagopan, N., Walsh, B., & Weber-Fox, C. (2010). Increasing phonological complexity reveals heightened instability in inter-articulatory coordination inadults who stutter. Journal of Fluency Disorders, 35(1), 1–18. https://doi.org/10.1016/j.jfludis.2009.12.001.

Smits-Bandstra, S., & DeNil, L. (2007). Sequence skill learning in persons who stutter: Implications for cortico-striato-thalamo-cortical dysfunction. Journal of FluencyDisorders, 32, 251–278. https://doi.org/10.1016/j.jfludis.2007.06.001.

Smits-Bandstra, S., & DeNil, L. (2009). Speech skill learning of persons who stutter and fluent speakers under single and dual task conditions. Clinical Linguistics &Phonetics, 23(1), 38–57. https://doi.org/10.1080/02699200802394914.

Song, L., Peng, D., Jin, Z., Yao, L., & Ning, N. (2007). Gray matter abnormalities in developmental stuttering determined with voxel-based morphometry. Zhonghua YiXue Za, 87(41), 2884–2888.

Sowman, P., Ryan, M., Johnson, B., Savage, G., Crain, S., Harrison, E., Martin, E., & Burianova, H. (2017). Grey matter volume differences in the left caudate nucleus ofpeople who stutter. Brain and Language, 164, 9–15. https://doi.org/10.1016/j.bandl.2016.08.009.

Spencer, C., & Weber-Fox, C. (2014). Preschool speech articulation and nonword repetition abilities may help predict eventual recovery or persistence of stuttering.Journal of Fluency Disorders, 41, 32–46. https://doi.org/10.1016/j.jfludis.2014.06.001.

Starkweather, R. (2002). The epigenesis of stuttering. Journal of Fluency Disorders, 27, 269–288. https://doi.org/10.1016/S0094-730X(02)00144-4.Tian, X., & Poeppel, D. (2012). Mental imagery of speech: Linking motor and perceptual systems through internal simulation and estimation. Frontiers in Human

Neuroscience, 6, 314. https://doi.org/10.3389/fnhum.2012.00314.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(1), 458–468. https://doi.org/10.1016/j.neuroimage.2015.01.024.Usler, E., & Weber-Fox, C. (2015). Neurodevelopment for syntactic processing distinguishes childhood stuttering recovery versus persistence. Journal of

Neurodevelopmental Disorders, 7, 4. https://doi.org/10.1186/1866-1955-7-4.van Lieshout, P., Ben-David, B., & Lipski, M. (2014). The impact of threat and cognitive stress on speech motor control in people who stutter. Journal of Fluency

Disorders, 40, 93–109. https://doi.org/10.1016/j.jfludis.2014.02.003.van Lieshout, P., Hulstijn, W., & Peters, H. (2004). Searching for the weak link in the speech production chain of people who stutter: A motor skill approach. In B.

Maassen, R. Kent, H. F. M. Peters, P. van Lieshout, & W. Hulstijn (Eds.). Speech motor control in normal and disordered speech (pp. 313–356). Oxford, UnitedKingdom: Oxford University Press.

Van Riper, C. (1982). The nature of stuttering. Englewood Cliffs, NJ: Prentice-Hall.Vasic, N., & Winjen, F. (2005). Stuttering as a monitoring deficit. In R. Hartsuiker, R. Bastiannse, A. Postma, & F. Winjen (Eds.). Phonological encoding and monitoring in

normal and pathological speech (pp. 226–247). Hove, UK: Psychology Press.Voytek, B., & Knight, R. (2010). Prefrontal cortex and basal ganglia contributions to visual working memory. Proceedings of the National Academy of Sciences, 107(42),

18167–18172. https://doi.org/10.1073/pnas.1007277107.Watkins, K., Smith, S., Davis, S., & Howell, P. (2008). Structural and functional abnormalities of the motor system in developmental stuttering. Brain, 1(1), 50–59.

https://doi.org/10.1093/brain/awm241.Weber-Fox, C., Spruill, J., Spencer, R., & Smith, A. (2008). Atypical neural functions underlying phonological processing and silent rehearsal in children who stutter.

Developmental Science, 11, 321–337. https://doi.org/10.1111/j.1467-7687.2008.00678.x.Weiner, M., Turkeltaub, P., & Coslett, H. B. (2010). The image of time: A voxel-wise meta-analysis. NeuroImage, 49, 1728–1740. https://doi.org/10.1016/j.

neuroimage.2009.09.064.

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

116

Wu, J., Maguire, G., Riley, G., Fallon, J., & LaCasse, L. (1995). A positron emission tomography deoxyglucose study of developmental stuttering. NeuroReport, 6,501–505.

Wu, J., Maguire, G., Riley, G., Lee, A., Keator, D., & Tang, C. (1997). Increased dopamine activity associated with stuttering. Clinical Neuroscience and Neuropsychology,8, 767–770.

Wymbs, N., Ingham, R., Ingham, J., & Paolini, K. (2013). Individual differences in neural regions functionally related to real and imagined stuttering. Brain andLanguage, 124(2), 153–164. https://doi.org/10.1016/j.bandl.2012.11.013.

Xuan, Y., Meng, C., Yang, Y., Zhu, C., Wang, L., Yan, Q., & Lin, C. (2012). Resting-state brain activity in adult males who stutter. PloS One, 7(1), e30570. https://doi.org/10.1371/journal.pone.0030570.

Yang, Y., Fanlu, J., Siok, W. T., & Tan, L. H. (2018). Altered functional connectivity in persistent developmental stuttering. Scientific Reports, (6), 19128. https://doi.org/10.1038/srep19128.

Ylinen, S., Nora, A., Leminen, A., Hakala, T., Houotilainen, M., Shtyrov, Y., Makela, J., & Service, E. (2014). Two distinct auditory motor circuits for monitoring speechproduction as revealed by content-specific suppression of the auditory cortex. Cerebral Cortex, 1–11.

Zhu, Q., Ross, T. J., Gu, H., Geng, X., Zuo, X.-N., Hong, L. E., ... Yang, Y. (2013). Intrinsic resting-state activity predicts working memory brain activation andbehavioral performance. Human Brain Mapping, 34(12), 3204–3215. https://doi.org/10.1002/hbm.22136.

Andrew Bowers, PhD CCC-SLP is an Assistant Professor at the University of Arkansas where he directs the EEG Neurofeedback Laboratory. He conducts researchfocusing on sensorimotor integration in speech perception and production using high density electroencephalography (EEG), autonomic measures, and surfaceelectromyography (EMG). His research examines relationships between sensorimotor integration, cognition, and measures of task-performance toward non-invasivebrain-computer interface methods designed to augment treatment approaches in fluency disorders.

Lisa Bowers, PhD CCC-SLP is an Assistant Professor at the University of Arkansas. Her research focuses on the speech, language, and literacy abilities of children frominfancy to adolescence. She is a certified speech-language pathologist and has experience in Early Intervention and school settings working with children fromculturally and linguistically diverse backgrounds.

Daniel Hudock, PhD CCC-SLP, is an Associate Professor in Communication Sciences and Disorders at Idaho State University and Director of the Northwest Center forFluency Disorders. His research interests are psycho-social-emotional aspects of fluency disorders, interprofessional collaborations with mental health professionals forthe holistic assessment and treatment of fluency disorders, and the neuroscience of speech perception and production in people who stutter using high-densityelectroencephalography (EEG) time-course analysis.

Heather L. Ramsdell-Hudock, PhD CCC-SLP, University of Memphis, 2009, is an Associate Professor at Idaho State University. She leads the Infant VocalDevelopment Laboratory and her research and teaching interests span infant vocal development, caregiver-infant reciprocity, phonetics, phonology, the scholarship ofteaching and learning, and mentoring graduate and undergraduate students through the scientific process.

A. Bowers et al. Journal of Fluency Disorders 58 (2018) 94–117

117

View publication statsView publication stats