Access to this full-text is provided by De Gruyter.
Content available from Paladyn, Journal of Behavioral Robotics
This content is subject to copyright. Terms and conditions apply.
Research Article
Shruti Chandra*, Garima Gupta, Torrey Loucks, and Kerstin Dautenhahn
Opportunities for social robots in the stuttering
clinic: A review and proposed scenarios
https://doi.org/10.1515/pjbr-2022-0001
received February 3, 2022; accepted April 15, 2022
Abstract: The inclusion of technologies such as teleprac-
tice, and virtual reality in the field of communication
disorders has transformed the approach to providing
healthcare. This research article proposes the employ-
ment of similar advanced technology –social robots, by
providing a context and scenarios for potential imple-
mentation of social robots as supplements to stuttering
intervention. The use of social robots has shown potential
benefits for all the age group in the field of healthcare.
However, such robots have not yet been leveraged to aid
people with stuttering. We offer eight scenarios involving
social robots that can be adapted for stuttering interven-
tion with children and adults. The scenarios in this article
were designed by human–robot interaction (HRI)and stut-
tering researchers and revised according to feedback from
speech-language pathologists (SLPs).Thescenariosspecify
extensive details that are amenable to clinical research. A
general overview of stuttering, technologies used in stut-
tering therapy, and social robotsinhealthcareisprovided
as context for treatment scenarios supported by social
robots. We propose that existing stuttering interventions
can be enhanced by placing state-of-the-art social robots
as tools in the hands of practitioners, caregivers, and clin-
ical scientists.
Keywords: stuttering, stammering, communication dis-
order, technology, social robots, assistive technology,
human–robot interaction, child–robot interaction, robot-
assisted therapy
1 Introduction
Innovative technical solutions have been proposed and
incorporated in treatments for several conditions and
have positively impacted the lives of people. The field
of communication disorders has also benefited from the
inclusion of technology [1]. This article is offered as a
position paper to illustrate the potential and utility of
similar advanced technologies, namely, social robots,
for stuttering interventions. In this article, we propose
scenarios that have been designed with input from human–
robot interaction (HRI)and stuttering researchers. Feed-
back of speech-language pathologists (SLPs)was included
to place the scenarios within a clinician-centric framework
involving three phases of treatment. This position article is
intended for the varied stakeholders in the field of stut-
tering including SLPs, researchers, and clients, as well as
stakeholders in the field of HRI who design and adapt
robotics for clinical and healthcare applications. Overall,
this article provides a starting point for productive colla-
borations among these stakeholders. It should be noted
that, to the author’s knowledge, no empirical research
has been conducted in this area, and, thus, this article initi-
ates research in a new direction. This is the first article
proposing clinician-centric HRI scenarios in which social
robots are presented as tools to aid SLPs and people with
stuttering.
The present article is structured as follows: Section 2
presents a brief review of stuttering. Section 3 presents a
review of technologies used in stuttering intervention. In
Sections 4 and 5, we introduce social robots and review
the available literature on social robotics in healthcare
and communication disorders. In Section 6, we describe
* Corresponding author: Shruti Chandra, Department of Electrical
and Computer Engineering, Social and Intelligent Robotics Research
Laboratory (SIRRL), University of Waterloo, Waterloo, Ontario,
Canada, e-mail: shruti.chandra@uwaterloo.ca
Garima Gupta: Department of Psychology, Faculty of Arts, University
of Waterloo, Waterloo, Ontario, Canada,
e-mail: g24gupta@uwaterloo.ca
Torrey Loucks: Department of Communication Sciences and
Disorders, Institute of Stuttering Treatment and Research (ISTAR),
Faculty of Rehabilitation Medicine, University of Alberta, Canada,
e-mail: loucks@ualberta.ca
Kerstin Dautenhahn: Department of Electrical and Computer
Engineering, Department of Systems Design Engineering, University
of Waterloo, Waterloo, Ontario, Canada,
e-mail: kerstin.dautenhahn@uwaterloo.ca
Paladyn, Journal of Behavioral Robotics 2022; 13: 23–44
eight play scenarios to illustrate the potential applica-
tions of social robots in the treatment of children and
adults who stutter. The article concludes with a summary
and discussion of the contributions, limitations, and
future steps for stuttering research using social robots
as tools.
2 Overview of stuttering
Stuttering, a developmental speech disorder prevalent
worldwide, has been defined as “disorders in the rhythm
of speech, in which the individual knows precisely what
[they]wish to say, but at the time is unable to say it
because of an involuntary, repetitive prolongation or ces-
sation of a sound”[2](World Health Organization, 1977).
The primary symptoms of stuttering are repetitions, pro-
longations, pauses and/or blocks that disrupt the rhythmic
flow of speech [3]. These primary symptoms may be
accompanied by physical (e.g., involuntary eye-blinking,
jaw jerking)and/or affective (e.g., avoidance behaviors,
negative emotions)secondary behaviors [2]. Researchers
have pointed to multiple possible causes for stuttering,
including sensorimotor, neurological, linguistic, and/or
genetic differences. The most common form of stuttering
is a developmental disorder that begins in the early child-
hood and follows a recovery trajectory or a chronic trajec-
tory [3]. A rare, acquired form of stuttering is also recog-
nized, which has an onset during adulthood, usually tied
to neurological disease [4]. The majority (65–75%)of chil-
dren diagnosed with developmental stuttering recover
naturally by the age of 8 years over a duration of 1 to 4
years, whereas 25–35% of children do not recover [5].
A review of 44 international studies of school-aged
children reported 1% prevalence rate for stuttering [6].
The incidence rate of stuttering is approximately 5%,
although a more recent large-scale survey suggests 5–8%
[7], with the onset occurring primarily during the preschool
years [8]. Stuttering is more prevalent in biological males
than biological females, with an approximate ratio of 4
males to 1 female [9]. The ratio is less pronounced in young
children, reportedly being 2:1 males-to-females [10].
The quality of life is negatively impacted by stut-
tering with coping difficulties extending beyond speech
into the social, emotional, and psychological domains,
eventually impacting major life choices [11,12]. These dif-
ficulties are exacerbated from childhood into adolescence
and adulthood. While a range of evidence-based treatment
programs (e.g., Comprehensive Stuttering Program [13])
address the behavioral, social, and emotional components
of stuttering, inclusion of social robotics could enhance
these treatment options. A comprehensive survey covering
a variety of stuttering treatment approaches including
formal programs, fluency induction techniques, and adjunct
therapy approaches that have been reported in the past
20 years (2000–2020)can be found in ref. [14].
3 Technology used in stuttering
interventions
Technological advancement has prompted the explora-
tion and application of technology in speech and lan-
guage therapy. The following is a brief description of
the technological tools used in stuttering therapy, which
provide contrast and context for considering possibilities
of social robots.
–Altered auditory feedback (AAF):The most common
AAF device is the SpeechEasy,¹ which is a self-con-
tained, in-the-ear fluency aid that alters the frequency
and delay of the user’s speech [15]. AAF devices can be
used in any language, and improvements in fluency
from these devices are generally stable over time. How-
ever, these devices are not curative and gains in flu-
ency are contingent on using the device [15].
–Metronomes: Metronome-paced speech, or speaking
to a beat, is a fluency induction technique based on
controlling speech rate [16]. Metronomes have been
used to establish target speech rates in the early stage
of therapy [17]. Metronomes have also been miniatur-
ized into electronic metronomes, which can be worn
behind the ear like hearing-aids.
–EMG biofeedback: A technique, which is not widely
used but has potential for fluency induction, uses sur-
face electromyography (EMG)to reduce excessive mus-
cular tension in the speech articulators (e.g., larynx,
tongue). Electrodesare placed overmuscles where there
is presumed to be excessive tension. The client and
clinician monitor muscular tension by a visual or audi-
tory signal and employ therapy techniques to reduce
tension to reach a subjective threshold [18].
–Virtual reality (VR):VR provides advanced human–
computer interfaces where a user is immersed in a simu-
lated version of 3D, real-world situations virtually. VR
immersion interface typically includes a standalone
visual interface, a VR headset for interacting with
1https://speecheasy.com/
24 Shruti Chandra et al.
virtual environments through vision and sometimes
audition. Researchers have reported that individuals with
stuttering display similar levels of affective, behavioral,
and cognitive symptoms in virtual and real environments
[19].VRhasbeeneffectively used for desensitization
therapy (e.g., reducing fear of public speaking [20]) and
the adaptation effect (i.e., gradually reducing dysfluen-
cies with repeated exposure to the same stimulus; [21]).
Such findings suggest that VR can function as a sys-
tematic, controlled, and confidential method to supple-
ment treatment.
–Telepractice: Telepractice is the delivery of profes-
sional healthcare services using tele-communications
technology, typically the Internet. Telepractice can be
synchronous (i.e., video or audio conferencing), asyn-
chronous (i.e.,e-mail)or hybrid [22]. A recent sys-
tematic review by McGill et al. [23]indicates that
synchronous telepractice is a promising service delivery
system that has been successfully incorporated in stut-
tering treatments, such as the Lidcombe and Camper-
down programs. Benefits of telepractice include cost
and time effectiveness and increased accessibility to
healthcare (e.g., decreased travel time, access from
remote areas, and during situations like the COVID-19
pandemic [23]). Limitations of telepractice include con-
cerns about patient or client privacy, technological dif-
ficulties (e.g., Internet speed), and poor patient or client
attendance [23].
–Mobile apps: Given the increased utilization of smart-
phones, mobile applications for stuttering management
are being developed [24,25]. Available applications include
different stuttering modification and fluency shaping
techniques (e.g., AAF). Popular mobile applications for
stuttering management include Speech4Good,² DAF
Professional,³ Dysfluency Index Counter,⁴Stamurai,⁵
MyLynel,⁶SPEECHTOOLS,⁷and Speechagain.⁸These
applications have been designed by professionals in
the fields of speech and language therapy, software
engineering, and artificial intelligence, and they show
promise in supporting users’therapy goals.
Considering the ongoing innovation in the field of
stuttering intervention, inclusion of advanced AI-based
technologies, such as social robots, is an important step
toward increasing the accessibility and potential effec-
tiveness of stuttering intervention.
4 Social robots
Social robots are one of the contemporary technologies
that have been investigated and used ineducation, health-
care, fitness, and entertainment [26–28]. According to
Breazeal et al. [26],“social robots are designed to interact
with people in human-centric terms and to operate in
human environments alongside people.”Social robots
“engage people in an interpersonal manner, [by]commu-
nicating and coordinating their behavior with humans
through verbal, nonverbal, and/or affective modalities”
[26]. Social robots interact with the environment using
different sensors and actuators such as cameras, micro-
phones, motors, and speakers with software algorithms
to support artificial speech recognition and generation.
As shown in Figure 1, the physical appearance of social
robots ranges from human-like (humanoids)to non-bio-
metric (animal-or toy-like)[29]. Such robots have the
capacity for social interactions using verbal and non-
verbal modalities, such as interpreting sounds and speech,
speaking, performing gestures using arms (e.g.,pointing),
face (emotional expressions, eye gaze, and joint attention
[30]) or full body (walking [31]). Social robots can be
operated using different modes of operation, including
fully autonomous, semi-autonomous, and wizard-of-oz
(WoZ [29,31]). Fully autonomous social robots sense their
environment, make decisions, and perform tasks without
human intervention. Semi-autonomous social robots have
some pre-programmed operations, while others are con-
trolled remotely by a human operator. In the WoZ model
of operation, social robots are teleoperated by a hidden
human operator. The design of HRI scenarios can include
one-to-one interaction (e.g., a social robot and a human
participant)or multiparty interaction (e.g., interaction
between one or more social robots with one or more
human participants). Further information regarding human–
robot interaction and social robots can be found in ref. [31].
There are numerous advantages to incorporating social
robots in the field of stuttering interventions.
4.1 Technology-focused advantages
–Social robots excel at repetitive tasks: Stuttering
treatments often require time-intensive practice, for
2https://innovationlabs.harvard.edu/current-team/speech4good/
3https://speechtools.co/daf-pro
4https://www.smartyearsapps.com/dysfluency-index-counter-2/
5https://stamurai.com/about
6https://mylynel.soft112.com/
7https://speechtools.co/
8https://speechagain.com/
Social robots in stuttering clinic 25
which the continual presence of the SLPs is not essen-
tial. Social robots could function as assistants for repe-
titive tasks because they maintain task consistency. In
addition, social robots’expansive ability to record ses-
sions and performance can provide clinicians with
more options to evaluate and chart progress. The
application of robots in this manner potentially allows
the clinicians to see more clients, focus on individual
needs, and offset wait lists.
–Social robots are programmable and adaptable:
Such robots can be tailored to accommodate individual
needs [32], which can address the variation presented
by people who stutter. These robots can be customized
according to speech patterns and stuttering severity,
as well as age, gender, and personality. They can play
different roles and exhibit varied behaviors in order to
make activities incorporating social robots appropriate
for children, adolescents, and adults. Moreover, they
could be tailored according to the particular intervention
used by a clinician.
–Social robots have a physical presence: Unlike other
technologies, such as mobile applications or virtual
reality, social robots are physically present during an
interaction. Studies in the field of education have
demonstrated that students learned more and faster in
the presence of a physically embodied social robot in
comparison to alternative technologies because phy-
sical presence offered a multimodal and richer pedago-
gical atmosphere [33,34]. Similarly, interactions with
social robots have been reported to be more motivating
and engaging than a virtual reality counterpart [35].
Furthermore, through their survey on people’spercep-
tion of physically present social robots versus virtual
agents, ref. [36]found that physically present social
robots were perceived more positively and considered
more persuasive.
–Social robots compared to other AI-driven technol-
ogies: Several studies have shown that people prefer
social robots over other technologies such as tablets or
smartphones. Westlund et al. [37]reported that com-
pared to learning from an iPad or a human teacher,
children preferred learning new words from a robot.
They also considered the social robot to be more like a
person than the iPad. Similarly, Zhexenova et al. [38]
compared children’s knowledge of Latin script in three
conditions: with a tablet, with a robot plus a tablet,
and a human teacher. The results showed that while
children gained the same amount of knowledge in all
three conditions, they reported higher likeability for
and positive mood change in the “robot plus tablet”
condition over the other two conditions. Further, Deu-
blein and Lugrin [39]conducted a user study with 84
students in a smart office environment where a tablet,
non-expressive social robot, or an expressive social
robot randomly requested the participants to perform
activities associated with physical well-being. The find-
ings suggested that compared to the tablet, both the
non-expressive and expressive robots were rated signif-
icantly higher in social presence and interaction within
the smart office environment.
Figure 1: Examples of social robots that have been used in the fields of therapy and healthcare. Left to Right: Pepper (2014-present, Source:
Softbank Robotics); Kaspar (2009-present, Source & credit: The Adaptive Systems research group, University of Hertfordshire ); Furhat
(2014-present, source: Furhat Robotics AB); Nao (2006-present, source: Softbank Robotics);QT(2018-present, LuxAI).
26 Shruti Chandra et al.
4.2 Human-focused advantages
–Social robots are enjoyable/engaging: Social robots
are novel contemporary technologies, and therefore,
they tend to inherently generate interest and/or exci-
tement regardless of whether they appear to be anthro-
pomorphic or toy-like [40]. Using them in conventional
interventions could introduce an element of fun, curi-
osity, and excitement, which engage clients and make
the therapy experiences and exercises more enjoyable
[35]. Social robots might be particularly attractive to
children, who find the interaction with a robot intrin-
sically rewarding, thereby assist in early interven-
tion [41].
–Social robots are non-judgemental: One of the key
advantages of using social robots for clinical popula-
tions is that people feel little to no judgement from
them, which could particularly facilitate therapeutic
outcomes in stuttering. Both the verbal and non-verbal
behaviors of a social robot can be consistently regulated
to ensure the atmosphere is comfortable, supportive,
and non-judgemental [42,43]. In certain therapeutic
contexts, social robots can be particularly useful for
mediating human–human interactions, as robots do
not show unconscious behaviors that could limit the
engagement and comfort of the clients. This has been
recognized in HRI research on children with special
needs, who prefer social robots for some therapy activ-
ities [44,45]. We expect that this non-judgemental
element, along with social robots’human-like com-
munication and interaction, could in turn, facilitates
practice of certain fluency skills [46].
–Social robots as a companion: Socially assistive robots
(SARs)have shown promising results in providing
companionship to children and adults in healthcare
facilities, where one might experience loneliness
and/or psychological distress [47–49]. For example,
Alemi et al. [50]employed the NAO robot as a tool for
robot-assisted therapy in the hands of psychologists
to reduce levels of anxiety, anger, and depression
amongchildrenwithcancer.Duringtheinvasive
treatment procedures, the robot played different playful
roles (e.g., peer, nurse, doctor)to put the patient at ease.
The findings suggest that positive effects in the levels of
anxiety, anger, and depression among the participants.
Similarly, SARs have been used to help older adults
foster social connection with others, combat loneli-
ness and depression, and improve mood and quality
of life [51].
4.3 Limitations of social robots
–Social robots are machines: As a complex machine
consisting of hardware and software components, the
capabilities of a social robot are dependent on the
functionalities and limitations of these components.
By leveraging these components, a social robot can
perform certain human-like behaviors reliably; how-
ever, this is far from having human-level perception
and decision-making capabilities [52].
–Dependency on pre-programming: Social robots depend
heavily on pre-programming, regardless of whether the
robot is teleoperated or fully autonomous. Some of the
technical challenges associated with social robots include
natural-language understanding [53], speech recogni-
tion [53,54],social-signal processing, and action selec-
tion among others [31].
–Limited capabilities: Even though social robots can
reply to questions and communicate to participants,
they have limited capabilities to understand the seman-
tics and context of a situation and do not learn these
elements in real time [31]. Thus, a natural, human-like,
open-ended dialogue with a robot is not possible at
present.
–Cost: The cost of social robots varies based on their
level of functionality [28]. While social robots have
demonstrated effectiveness and potential, the targeted
users, including researchers, teachers, doctors, thera-
pists, parents, may not have the funds to acquire a
social robot [55]. However, efforts are being continu-
ously made to develop cost-effective and affordable
robots [56]. Particularly promising are initiatives pro-
moting open-source robot development,⁹which could
allow others to replicate the robot hardware and use the
previously developed software. An example is the recently
developed MyJay robot, to support robot-assisted play for
children with physical disabilities [57].
–Maintenance: Like any other computer-based tech-
nology, robots may also require periodic maintenance.
Common maintenance issues include hardware or soft-
ware malfunction, software updates, and charging depen-
dency. Further, a trained human technician is usually
needed to complete the technical set-up required for the
robot. Social robots are not humans: Social robots are not
suitable for sensitive situations, where human-level per-
ceptual abilities, emotions, empathy, and general human-
9https://github.com/hamzaMahdi/myjay-bot
Social robots in stuttering clinic 27
level intelligence and expertise are essential [58].Ethics-
related concerns should be tackled meticulously to make
such robots safe and secure for use, so that social robots
can be employed effectively in different contexts with
human supervision and intervention.
However, despite these limitations, targeted program-
ming and designs allow social robots to interact success-
fully with humans. Overall, social robots present a unique
avenue for research and intervention for serving clinical
populations, such as people who stutter. In the context of
the proposed scenarios presented later, a social robot
cannot replace the experience, training, judgement, or
flexibility of the SLPs; however, their functionality comple-
ments therapy activities within the scope of the aforemen-
tioned benefits.
5 Use of social robots in healthcare
In the field of healthcare, social robots have taken on the
roles of assistants, companions, guides, and trainers to
boost performance and provide encouragement while cli-
ents learning a certain task [59](see Figure 2). Social
robots have also been used with clients of all ages in
therapeutic and assistive contexts, such as therapy for
children with autism spectrum disorder (ASD), people
with dementia, or individuals with diabetes or cancer
[59]. In addition, robots have provided companionship
to children during extended stays at hospitals [60,61]
and act as a distraction tool during medication proce-
dures [62,63]. Social robots have been used in clinical
studies to train literacy, self-management, and aware-
ness [48,64,65], alongside recent proposals to use social
robots for social anxiety interventions [66].Different
appearances of social robots have been used (e.g., huma-
noid, animal-like, cartoon-like, machine-like). Examples
include Kaspar,¹⁰Zeno,¹¹ Nao, Aibo,¹² Paro,¹³ Pleo,¹⁴and
Keepon¹⁵robots.
Regarding the ASD, social robots have fostered verbal
and non-verbal communication skills [67–69], enhanced joint
attention [70], collaborative skills [71], visual-perspective
taking [72], and social/emotional engagement [73],whichare
specificdifficulties for children with ASD [74]. Social robots
have also been used for long-term interventions outside con-
trolled lab settings and without extensive technical supervision
[46]. For example, Scassellati et al. [75]used social robots to
provide home-based social communication skill practice for
children with special needs. In this study, the children engaged
in a triadic interaction with the robot and their caregiver for 30
minutes per day for a month that lead to improvement in joint
attention skills with caregivers. For older adults and adults with
special needs and medical conditions, such as dementia,
social robots have been used as companions and service
robots. Companion robots are defined as robots that fulfil
certain tasks, e.g., provide cognitive, physical or social assis-
tance in activities of daily living, in a socially acceptable
manner, to enhance the overall health and well-being of
the individual [76]. Some of the most common companion
robots are PARO, Aibo, Nao, Furhat,¹⁶Pepper,¹⁷Zorabots,¹⁸
Buddy,¹⁹and AIDO.²⁰PARO robots, which have been
widely accepted, demonstrated positive outcomes, espe-
cially in dementia care [77]. Older adults with dementia
showed improved activity levels [78], strengthened social
connections, and reduced stress levels [79]. Service robots,
such as Care-O-Bot,²¹ Hobbit robot,²² and Pearl [80],have
supported independent living, i.e., to provide medication
reminders; monitor activity levels; provide cleaning ser-
vices, basic daily activities [81], exercise, and navigation
[82]; and maintaining safety [83]. The success of these
verbal and nonverbal interaction activities mediated by
social robots holds considerable potential for children
and adults who stutter by embedding fluency goals within
communicative contexts. For example, turn-taking games
combined with fluency goals could be implemented during
early therapy and during transfer and maintenance.
5.1 Social robots as tools in communication
disorders
Communication can be impacted by disorders of speech
(i.e., speech sound disorders, stuttering, cleft palate),lan-
guage (i.e., developmental language disorder, language
10 https://www.herts.ac.uk/kaspar/the-social-robot
11 https://www.hansonrobotics.com/zeno/?
utm_source=robots.ieee.org
12 https://us.aibo.com/
13 http://www.parorobots.com/
14 https://www.pleoworld.com/pleo_rb/eng/index.php
15 https://beatbots.net/my-keepon
16 https://furhatrobotics.com/furhat-robot/
17 https://www.softbankrobotics.com/emea/en/pepper
18 http://zorarobotics.be/
19 https://buddytherobot.com/en/buddy-the-emotional-robot/
20 http://aidorobot.com/
21 http://www.Care-O-Bot-4.de/
22 http://hobbit.acin.tuwien.ac.at/index.html
28 Shruti Chandra et al.
delay), and social communication [74].Whileresearch
on social robots in communication disorders is limited,
some promising results have been reported. In research
involving children with special needs, participants’verbal
production and engagement in the therapy sessions
increased when the session was conducted with the social
robots Nao and CommU [85]. Another study found that
when two social robots were placed in a disability unit
for adolescents with special needs for two years, improve-
ments in articulation, verbal participation and sponta-
neous conversations were noted [67]. Similarly, the robot
Kaspar, used for a long-term study by caregivers in a nur-
sery school for children with ASD, has shown beneficial
outcomes for the participants. This study also found good
acceptance among teachers and caregivers who used
the robot without direct supervision by researchers [46].
Among children with pervasive developmental disorder,
the humanoid robot iRoboi was found to positively impact
their communication skills using augmentative and alter-
native communication strategies [86].Robles-Bykbaev et al.
[87]used a low-cost robot, named SPELTRA (Speech and
Language Therapy Robotic Assistant), to support therapy
sessions for children with neurodevelopmental disorders
such as cerebral palsy, intellectual disability, dysarthia,
among others. Specifically, the robot was programmed to
register the participants’information and results from the
therapy session, as well as supporting the client outside
theclinictoreinforcetheskillslearnedin-clinic. The
participants quickly adapted to SPELTRA and showed
improvements in phonological, morphosyntactical, and
semantic communication measures. Other researchers
have proposed different applications of social robots in
the context of communication disorders, which hints at
the multitude of potential ways in that social robots can
be leveraged to enhance interventions. For patients with
Figure 2: Social robots interacting with people in healthcare settings. (a)The social robot Kaspar which has been designed for interactions
with children with special needs engaging a child in an activity with educational and therapeutic objectives [84];(b)Nao robot playing a
self-management educational game with a child with diabetics [64];(c)the robot assisting an autism therapist in ASD diagnosis training
session [68];(d)the robot engaging in therapy tasks with a child with special needs [69];(e)the robot as a collaborator promoting
engagement and performance for gait rehabilitation of a patient with neurological disorder during a therapy session; (f)a training assistant
robot providing encouragement and motivation to a patient in a cardiac rehabilitation training session. Image credits.
23
23 Note: Figure 2. Figure 2(a): Credit-The Adaptive Systems
research group, University of Hertfordshire; with permission from
authors. Figure 2(b): Reprinted from International Journal of
Human-Computer Studies, Volume 106, Henkemans, O. A. B., B.
P. Bierman, J. Janssen, R. Looije, M. A. Neerincx, M. M. van
Dooren, J. L. de Vries, G. J. van der Burg, and S. D. Huisman,
Design and evaluation of a personal robot playing a self-manage-
ment education game with children with diabetes type 1., 63–76,
Copyright (2017), with permission from Elsevier. Figure 2(c)and
(d): Credit: Hospital Garcia de Orta & INESC-ID, with permission
from authors. Figure 2(e)and (f): Credit-Center for
Biomechatronics at the Colombian School of Engineering by Julio
Garavito, with permission with authors.
Social robots in stuttering clinic 29
aphasia, Pereira et al. [89]proposed the implementation
of the social robot Nao as a mediator in a memory game
between a speech therapist and a client with aphasia, to
promote understanding of imperative statements. Rama-
murthy and Li [88]developedanapplicationforchildren
with cleft lip and palate with the social robot Buddy to
practise articulation. Castillo et al. [90]developed an
application using a desktop social robot, called Mini, to
assist therapists with rehabilitation exercises for adults
with apraxia. In regard to stuttering, Kwaśniewicz et al.
[91]described an application of the social robot Nao to
provide “echo”(a combination of delayed auditory feed-
back and choral speech)as clients practised their fluency
skills. The authors predicted that Nao could potentially
enhance the echo effect because the robot also provides
visual feedback through arm movements and a sense of
company while the clients practice therapy tasks. Still,
the impact of the social robot is yet to be validated with
empirical evidence [91]. Although significant progress
has been made in utilizing social robots in the field of
healthcare and clinical practice, implementing these robots
in the field of stuttering interventions has not been explored
to date. In the next section, we introduce possible applica-
tions of social robots in stuttering intervention.
6 Proposed play scenarios with
social robots
In this section, we propose higher-level conceptualizations
of social robots in stuttering intervention. Formulating and
specifying human–robot interaction scenarios is viewed as
the first and fundamental step for introducing social robots
in a principled manner within established interventions
and clinical research. The scenarios are intended to guide
and inform future research. The exact details of the sce-
narios are likely to change during empirical research,
which will require co-design, system development and
evaluation, and feedback from clients, teachers, care-
givers, therapists, and other stakeholders, cf. [84].The
format of the following play scenarios is inspired by pre-
viously developed robot-assisted play scenarios [84,92]
and broadly conform to several current robot-assisted
interventions that have been used for children, in parti-
cular for children with ASD. The scenarios are presented
within a therapeutic framework common in stuttering
interventions that involve three general phases of establish-
ment (alternative term used for acquisition),transfer,and
maintenance. These phases are described by Kully [13]
in the context of the Comprehensive Stuttering Program.
According to Kully [13],establishmentisthefirst phase of
treatment during which clients learn strategies or skills to
support their fluency goals (acknowledging their skills will
differ depending on the stuttering treatment approach).
Transfer is the second phase in which clients apply their
skills and strategies in diverse contexts. These speaking
activities are finely sequenced to gradually build on factors
that influence the client’s ability to implement the strategies
in simulated and real-world situations. The clinician works
closely with the client to structure transfer activities that
provide functional practice and build the client’sconfi-
dence. Maintenance is the last phase, which lasts beyond
treatment, where the client continues to practice treatment
skills in everyday life and over the long term. The training
and experience of the SLPs is most necessary during the
establishment phase, when treatment skills relevant for
reducing dysfluency are initially taught and negative atti-
tudes are evaluated. Once the client has learned the relevant
treatment skills, the clinician can progress to the transfer
phase, where skills are elaborated in varying communica-
tion contexts and with diverse conversation partners. The
clinician must work closely with the client to appropriately
choose sequential transfer activities and adjust the com-
plexity of the task to support the client’sprogress.Since
social robots excel at repetition and are inherently non-jud-
gemental, they may have the most relevance for transfer
when multiple, intensive repetitions of new skills are
needed. The clinician may find that structured human–
robot transfer activities reduce the time requirements for
direct clinician interaction while still achieving transfer
goals. Maintenance is introduced asthe last phase of treat-
ment when the client is prepared to sustain learned treat-
ment skills and continue to work towards their personal
communication goals. As maintenance of treatment gains
often involves continued practice, a social robot could
support the client by providing a practice partner and
potentially certain forms of feedback related to the client’s
accuracy in producing treatment skills. Other treatment
approaches may vary in their approach to establishment,
transfer, and maintenance. Each of the following scenarios
is categorized according to a particular phase of treatment
with treatment objectives set by the clinician (oftenwithinput
from the client)and social robots programmed accordingly by
engineers and roboticists. We strive to propose scenarios from
aclinician’s perspective and based them on our established
experience of translating clinical activities into programs car-
ried out by robots. These scenarios include suggested activ-
ities only and describe possible applications of social robots
in stuttering therapy both in and outside the clinic. The
proposed scenarios would require some adjustment for
30 Shruti Chandra et al.
alternating implementation between clinicians, caregivers,
and teachers. As with other social robot experiences, the
client is often an active participant in determining the
extent and frequency of robot participation.
6.1 Components of play scenarios
The first item in each scenario is the objective of the
proposed scenario, which is based on the phase of treat-
ment (establishment, transfer, or maintenance)followed
by categories that list treatment domains (speech, social,
and emotional), treatment technique, type of play, and
interaction technique for each scenario. Other items in
the scenarios are as follows: (a)participant roles and
behaviors, (b)specification of robot configuration and
mode of operation, (c)setting and time-frame, (d)pos-
sible variations, and (e)benefits that social robots offer
to the clients and SLPs. The details are aimed at pro-
viding SLPs, roboticists, and researchers with starting
points for leveraging robots enhancing stuttering inter-
ventions. The following sections discuss the integration
of play and different interactions techniques within the
scenarios.
6.2 Integration of play
The proposed scenarios are play based because play is a
crucial aspect of life that promotes cognitive, physical,
social, and emotional well-being of children and youth
[93]. Play is “fun, educational, creative, stress-relieving
and encourages positive social interactions and commu-
nication”[94]. Play therapies are powerful modalities for
working with different populations [95]. With regard to
social robots, play can make human–robot interaction
more enjoyable and motivating. In the aforementioned
examples of social robots for children with ASD or people
with dementia, the interaction with the robot was inher-
ently rewarding, in the absence of any additional, explicit
reward [84]. The proposed scenarios incorporate social
and cognitive play, which is typically used with children
but can be extended to adolescents/adults. Social play
is further divided into the four categories of solitary,
parallel, associative, and cooperative [96], whereas cog-
nitive play, proposed by Piaget [97], includes three dis-
tinctions, which are practice, symbolic, and play with
rules. These types of play encourage participants to
enhance their skills in social, cognitive, psychological,
emotional, and communication domains, all of which
could benefit persons who stutter.
6.3 Integration of interaction techniques
The interaction techniques used in the proposed sce-
narios are to promote engagement and provide priming
to participants. These techniques are described as follows:
–Peer learning. This interaction technique has been
widely used in the field of education. Peer learning is
a reciprocal learning method, in which participants
play the roles of both teachers and students [98].
Further, it is a useful technique because participants
are required to take on the responsibility for their own
learning through communication, provision, and recep-
tion of feedback [99].Chandraet al. [100]used peer
learning in their study on human–robot interaction
during a writing activity where the children acted as
peers and the social robot (Nao)as instructor. Their
findings suggest that peer learning enhanced the parti-
cipants’learning gains [100,101].
–Learning-by-teaching. Also known as peer-tutoring,
in this interaction technique, a student takes on the
role of the teacher and teaches other learners, which
enhances their own learning [102]. Prior to teaching
the other learners, the student engages with the task
and specific content [103]. Teaching the material taps
into the three core aspects of learning interactions,
which are structuring, taking responsibility, and reflecting
[103]. There is substantial evidence suggesting that teaching
others is an effective method for personal learning [104].
Several studies on HRI also found that peer-tutoring
was an effective method for improving children’shand-
writing capabilities [101,105]. In these studies, children
taught a robot how to write, which led to improvements
in their own writing.
–Support groups. Support groups are gatherings of five
or more people with a common problem that may or
may not include a trained professional [106]. Support
groups are beneficial because they can provide a sense
of community, a safe environment for self-disclosure,
help foster new friendships, provide resources, recom-
mendations for coping with difficulties, facilitate improve-
ment in social skills, and reduce the level of distress [106].
Birmingham et al. [108]used a Nao robot as a mediator to
investigate trust dynamics in a support group of students
who were strangers. The results of this study validated
that the robot-mediated support group could improve
interpersonal trust among the group members.
–Model-rival method. Developed by Dietmar Todt, the
model-rival method involves a three-way interaction
between two researchers and one student [109]. One
of the researchers acts as an instructor, while the other
models the behaviors of interest and is the student’s
rival for the instructor’s attention. With this interaction
Social robots in stuttering clinic 31
technique, the goal is to indirectly teach the student
behaviors of interest. The student observes the response
the model gets from the instructor when they respond
correctly (i.e., reinforcement of student’sinterest)and
incorrectly to a prompt (i.e., punishment)[110]. The
student observes these interactions and learns the tar-
geted behavior in order to receive the reinforcement
of interest. Pepperberg successfully implemented this
technique for teaching her parrot, Alex, colours, shapes,
and numbers [110]. Inspired by this, Fishman [111]pro-
posed to use the Model/Rival method for preschool chil-
dren with complex communication needs where the rival
could be replaced with low-tech devices (e.g.,displays,
speech generative devices or a helping doll)to promote
greater communication partner involvement. However,
this technique has not been explored in the field of
HRI yet.
6.4 Integration of treatment techniques
The treatment techniques used in the proposed scenarios
are behavioral interventions typically used for stuttering.
These techniques fall into two categories: fluency shaping
and stuttering modification techniques.
–Fluency-shaping techniques: Also known as speech
restructuring/modification or prolonged speech treat-
ments, these techniques promote fluent speech by
teaching new speech production patterns to the client
[112–114].Examplesoffluency-shaping techniques
include prolonged speech, regulated breathing, and
syllable-timed speech [112,115].
–Prolonged speech: Slowed or prolonged speech is
an effective speech restructuring technique that focuses
on controlling speech rate. It typically involves learning
to produce elongated speech segments at a very slow
Table 1: Description of scenario 1
Scenario 1: Beats with peers
Objectives Acquisition and transfer of fluency skills
Treatment domain Speech domain
Treatment technique Syllable-timed speech
Play type (social ∣cognitive)Cooperative and practice play
Interaction technique Peer learning
Participants’role & behavior There are two participants in this scenario, a social robot and the individual who stutters.
Both participants alternate as instructors and learners
Activity description
This scenario entails a cooperative task between a social robot and an individual who
stutters. A metronome, downloaded on a tablet, will be adjusted to a target syllable rate. At
the beginning of the scenario, the robot and the client are learners who practice syllable-
timed speech. As a co-learner, the robot can make the establishment phase more engaging
by actively modelling the technique and interacting with the client as a peer. After
establishing the basic skill, the robot and the client can engage in a peer learning game,
where they alternate between tasks of modelling and learning syllable-timed speech. The
participant who is modelling will demonstrate how to pace their speech according to the
metronome, and the other party will repeat after the modelling. Then the participants will
switch roles
Robot configuration & mode of operation A social robot that has speech capability will be used, such as the Pepper robot [131]. This
social robot will also have a tablet embedded into its chest to display the words or phrases
and well as the metronome. This robot can be operated in the Wizard-of-Oz (WoZ)or semi-
autonomous manner
Setting & time This scenario can be carried out in a home, school, or clinic setting over multiple sessions.
Telepractice is also a feasible option for both clinic and school settings. The duration and
frequency of the sessions would be set according to the clinician’s treatment plan
Variation The level of difficulty can be adjusted (i.e., increasing the number of syllables per words
and/or sentence lengths)in order to adapt the scenario to other age groups and treatment
hierarchy. The activity can also include more participants to promote transfer to group
speaking situations
Benefits In this scenario, the individual who stutters will benefit from a social robot’s ability to offer
customized, repetitive, and non-judgemental practice in a play based format. If effective,
such practice could reduce time demands, allowing the SLP to see more clients or provide
more focused intervention. The social robot’s ability to record the client’s performance
could help in monitoring progress
32 Shruti Chandra et al.
rate, typically timed at a syllable-by-syllable level. The
overall goal is to gradually increase speech rate while
maintaining fluency and naturalness [115,116].
–Regulated breathing: Also known as habit reversal,
regulated breathing seeks to reduce stuttering by
teaching more effective speech-related respiratory beha-
viors that are incompatible with stuttering. [117,118].Itis
amulti-component approach that includes awareness,
relaxation, competing response time, motivation, and
generalization training [119].
–Syllable-timed speech: This treatment involves timing
each syllable to a rhythmic beat. For example, a client is
instructed to produce syllables of equal duration in time
to a metronome beat. This technique is believed to sta-
bilize the speech motor system by reducing the varia-
tions in linguistic stress [120,121].
–Stuttering modification techniques: These techni-
ques are anxiolytic (i.e.,anxiety-reducing)in nature
and aim to promote desensitization to stuttering, aware-
ness of stuttering moments, and acceptance of one’sstut-
tering [115]. The focus of stuttering modification techni-
ques is to reduce fear, anxiety, low self-esteem, shame,
and avoidance of stuttering moments, or speaking situa-
tions among individuals who stutter [112,122].Examples
of these techniques include pseudo-stuttering and self-
disclosure of stuttering [112].
–Voluntary stuttering: Also called pseudo-stuttering,
this technique entails deliberate production of overt dys-
fluencies that resemble stuttering by the client or clini-
cian [123,124]. It is used to provide desensitization to
stuttering and reduce fear, negative emotions associated
with stuttering, and feeling of loss of control [123,125].
Table 2: Description of scenario 2
Scenario 2: Musical modelling
Objectives Transfer of treatment techniques
Treatment domain Social and speech domains
Treatment technique Syllable-timed speech, regulated breathing, or voluntary stuttering along with positive
reinforcement
Play type (social ∣cognitive)Cooperative and symbolic play
Interaction technique Peer learning and support group [132–134]
Participants’role & behavior A speech-language pathologist, small group of individuals who stutter and one social
robot are involved. The social robot will model treatment techniques, while the speech
language pathologist mediates group interactions and activities
Activity description
In this scenario, the group of individuals who stutter will review a stuttering treatment
technique that they previously learned. During a review session, the social robot will model
the stuttering treatment technique for the participants as the clinician explains it. After this
review, the group of individuals who stutter will engage in a game similar to musical chairs
(see note below). The participants will pass a ball while the music plays, and when it stops,
the participant with the ball will select an activity under their name on a tablet. The social
robot will model the selected activity (as many times as required), and then, the participant
will follow the social robot’s lead to complete the activity. The SLP will monitor the
participants and mediate when required (e.g., providing an alternative model if participant
has difficulty with the task)
Robot configuration & mode of operation A social robot with speech capability, like Pepper, that can operate in a wizard of oz [133]
or semi-autonomous manner will be used
Setting & time This scenario can be conducted in a school or clinical setting over multiple sessions and
duration as prescribed by the SLP
Variation The words or phrases displayed on the tablet can be customized to different reading levels,
stuttering severity, and treatment goals
Benefits Through this scenario, individuals can transfer a therapy technique to a group setting,
which can build confidence and introduce variation in using a skill. The social robot can
record the progress of a participant, which the SLP can review at any time. The time
requirements for direct SLP interaction during practice sessions can be reduced
Note: A game where a set of chairs are arranged in a circular fashion and the number of chairs are fewer than the number of participants.
While music plays, players walk around the chairs and when the music stops abruptly, all the players must occupy a chair. The player who
fails to find a chair gets eliminated from the game. For the next round, a chair is removed, and the process repeats until one player remains
in the game and is declared a winner.
Social robots in stuttering clinic 33
Table 3: Description of scenario 3
Scenario 3: Bouncy bingo
Objectives Desensitisation to stuttering
Treatment domain Social and emotional domains
Treatment technique Voluntary stuttering
Play type (social ∣cognitive)Cooperative play and games with rules
Interaction technique Peer learning
Participants’role & behavior In this play scenario, a child who stutters, and a social robot operate as peers
Activity description
This scenario consists of a cooperative game with rules between a social robot, and a child
client. Before starting the game, the clinician or a caretaker will review voluntary stuttering
for the client and a social robot programmed to model voluntary stuttering. After the
review, the client and the social robot engage in a game, i.e., Bingo! (see note below)[136],
mediated by an application run by the robot. On each turn, a participant selects a card
presented on a tablet, which has words or phrases adjusted to individual treatment goals.
Then the client and social robot use voluntary stuttering to produce the stimulus, provide
feedback to each other, potentially assign a score and then continue the game
Robot configuration & mode of operation A social robot with the capability to speak and operate autonomously, such as Pepper,
QT, Nao
Setting & time This play scenario can be conducted in home or clinic setting over multiple sessions as
prescribed. Telepractice can also be incorporated in this scenario. Increase the number of
participants or increase difficulty of stimuli to facilitate transfer. The fluency technique can
also be varied, such as using cancellations or pull-outs
Benefits The client benefits from the social robot’s non-judgemental engagement that also helps to
foster a comfortable play environment and more communication
Note: Players are provided a card with random numbers in different arrangements. A host announces different numbers that the players
mark on their cards. The player who completes the card by marking 5 numbers in a row or column yells “Bingo!”to stop the game. If all the
marked numbers were actually announced and marked correctly, then the player is the winner and a new round is started.
Table 4: Description of scenario 4
Scenario 4: Talking robots
Objectives To promote practice of a fluency skill
Treatment domain Speech domain
Treatment technique Prolonged speech and operant conditioning
Play type (social ∣cognitive)Associative and practice play
Interaction technique Model-rival method [109,110,137]
Participants’role & behavior In this scenario, there will be a social robot as a model/rival, a clinician as an instructor,
and a client as a learner
Activity description
This scenario entails associative and practice play between an individual who stutters, a
clinician who indirectly teaches the client a fluency-enhancing technique, and a social
robot that will assume the role of model-rival for the client. The clinician will first train the
social robot on the fluency technique intended for the client. The clinician will then provide
verbal and non-verbal positive reinforcement each time the social robot produces the
target behavior. The client will first observe the target skill and the positive reinforcement
from the clinician. After the practice round with the social robot, the clinician will repeat
the task with the client in the same format to model the skill and provide appropriate
reinforcement
Robot configuration & mode of operation A social robot with speech capability and appropriate emotional expression, such as
Pepper and Nao, functioning in a WoZ or semi-autonomous manner
Setting & time This scenario can be conducted in a clinic or school setting over multiple sessions and
duration. Telepractice can be incorporated in both settings
Variation Different fluency skills and complexity can be used based on the treatment goals. The roles
and order of participation can also be varied. As an alternative, the robot can be controlled
by the SLP in a WoZ mode to question “bumpy”vs “smooth”speech as part of a modified
Lidcombe approach
Benefits The use of model-rival framework and positive reinforcement can encourage clients to
practise therapy skills modelled by the social robots in a more engaging routine
34 Shruti Chandra et al.
–Operant conditioning: Several stuttering interven-
tions incorporate the principles of operant conditioning,
which is described as “the process by which the fre-
quency of a response is changed as a result of the con-
sequences of that response”[126]. Positive reinforce-
ments are typically used in techniques with operant
conditioning to increase or decrease the frequency of a
specificresponse[127,128]. For individuals who stutter,
treatment techniques include parents’verbal contingen-
cies for stuttered and stutter-free speech. For example,
the clinician teaches parents to reinforce fluent utter-
ances and to correct disfluent utterances [129].
–Time-out: This operant conditioning technique,
which is also known as “response-contingent time-
out,”involves a deliberate cessation of speech after
a stuttering episode [130]. This technique can incor-
porate clinician-led time-outs or self-administered
time-outs by the client. This technique is typically
used for reducing dysfluency in teenagers and adults
[130].
6.5 Robot as a tool box –case study
In addition to the scenarios presented in Tables 1–8, a
social robot can function as an embodied toolbox,i.e.,a
fluency buddy, that provides different tools for enhancing
fluency in varied settings. The robot can be customized
by the SLP with different fluency-inducing applications
(e.g., delayed auditory feedback),fluency techniques
(e.g., stretched speech), speed/timing of applications,
variable feedback delivery, and language difficulty/com-
plexity. In addition, the fluency buddy could complement
clinical activities by allowing the client to practice targeted
skills in an engaging manner outside the clinic.
7 Conclusion
Stuttering impacts a large segment of the population,
approximately 1% of people stutter. Individuals with
Table 5: Description of scenario 5
Scenario 5: Musical jeopardy!
Objectives To promote transfer of a fluency skill
Treatment domain Speech and social domains
Treatment technique Variable treatment skills –e.g., prolonged speech
Play type (social ∣cognitive)Cooperative, parallel play and play with rules
Interaction technique Support group
Participants’role & behavior In this scenario, there will be four or more clients who stutter and two social robots, who
will model speech behaviors for the participants in a setting resembling a support group
Activity description
This cooperative, parallel and play with rules scenario involves multiple clients who stutter
and two social robots. The clients are divided into teams consisting of at least two clients
and a social robot to play a modified version of Jeopardy! (see note below). Using a pre-
programmed jeopardy technique in the robots, each group will take turns to select a
potential category (e.g., movies, geography, music). In each category, there will be an
option to select one of five speech tasks that have a hierarchy of increasing difficulty (e.g.,
200 points for the easiest speech task and 1000 points for the most difficult speech task).
On each turn, a speech task will be displayed on the robot’s tablet, who will model a
fluency skill and prompt the client to complete the task using the skill. The teams can
compete for the highest score under the discretion of the SLP
Robot configuration & mode of operation A social robot with speech capability, such as Pepper, Nao, or QT, operating in a semi-
autonomous manner
Setting & time School or clinic
Variation In an alternate version, the social robot can lead the first session as a model, followed by
the clients who alternate as leader. The complexity of the reading material can be varied
according to reading level and treatment hierarchy
Benefits The clients practice treatment skills in a play based transfer session where the speech
tasks can be customized for the treatment goals with modelling provided by social robots
and other group members
Note: Jeopardy is a game show where participants engage in a general knowledge quiz. Each participant selects a dollar value and a
category from the game board, and responds to a trivia prompt in the form of a question [138].
Social robots in stuttering clinic 35
the chronic form experience significant reductions in the
quality of their life, with many facing bullying or peer-
rejection, emotional embarrassment/frustration, and psy-
chological sequelae (e.g., self-stigma and social anxiety).
Given such reductions in quality of life, it is critical to
investigate technologically intensive interventions that
have the potential to enhance current treatment options.
Research studies using technologies such as virtual rea-
lity and telepractice have revealed the potential and
benefits of these technologies for people with stuttering.
Similarly, we propose social robots as state-of-the-art tech-
nology that could enhance stuttering interventions. In this
position paper, a context and scenarios for practical appli-
cations of social robots to supplement varied stuttering
interventions, which seek to promote fluency, offset any
negative impacts of stuttering and promote effective, low-
effort communication.
Social robots are designed to interact with people in
interpersonal, social, and cooperative manners. Social
robots offer a multitude of benefits, including the ability
to facilitate non-judgemental therapeutic environments, main-
tain task consistency, adaptability, and generate interest/
engagement while being physically present during an interac-
tion. Given these characteristics and the available literature
demonstrating the effectiveness of incorporating social robots
in interventions for children with special needs, and people
with dementia, diabetes, or cancer, we believe that social
robots hold tremendous potential for persons who stutter.
To exploit the benefits of social robots, this article proposes
robot-assisted novel play scenarios that leverage upcoming
technologies with stuttering intervention practices. Each play
scenario presented in Tables 1–8 consists of several compo-
nents, including play-types, treatment domain, and treatment
techniques. These components are presented in considerable
detail and are tangible contributions that we hope can
be applied in therapy and clinical research (see Table 9).
The proposed scenarios have been designed to partner
with SLPs, caregivers, and teachers, according to a ther-
apeutic framework that involves three phases –establish-
ment, transfer and maintenance –with social robots
mostly contributing to transfer activities (see Table 9).A
limitation of the proposed scenarios is the social robots’
presently limited ability to automatically recognize and
generate speech, but this is an active area of research
and great progress has been made [139,140], including
automated techniques to recognize stuttering [141,142].
Turn-taking, programmed speech production, and game
applications have already been instantiated in many
Table 6: Description of scenario 6
Scenario 6: Smooth skits
Objectives To promote transfer and maintenance of fluency skills through a skit
Treatment domain Speech and social domains
Treatment technique Fluency-shaping or stuttering modification techniques
Play type (social ∣cognitive)Cooperative, parallel, and practice play
Interaction technique Peer learning
Participants’role & behavior In this scenario, the client who stutters, and a social robot engage in a skit delivered by the
robot
Activity description
This scenario consists of an activity that entails elements of cooperative, parallel, and
practice play with the social robot as a peer. The social robot and client adopt one of the
roles in a two-actor skit, which could be a simulated job interview, television show, or
movie. Both the robot and client will read the lines of their assigned role using a target
fluency skill. After receiving feedback from the clinician, teacher, or a caretaker, they can
switch roles and repeat the skit. The robot can provide the client with positive verbal
feedback during and after the performance
Robot configuration & mode of operation A social robot with the capability to speak and operate semi-autonomously, such as
Pepper, QT, or Nao
Setting & time School, clinic, or home setting. In addition, telepractice can be incorporated in school or
clinic settings
Variation The skits can be adjusted to the age and interests of the clients. For example, an adult who
stutters might be interested in practicing skits rooted in real-life social interactions, such
as job interviews. The number of participants can be increased to promote social
interaction and generalization of learned skills
Benefits The client can apply fluency skills in different contexts by engaging in scripted simulations
of every day and imaginative situations to promote transfer and maintenance
36 Shruti Chandra et al.
Table 8: Description of scenario 8
Scenario 8: Your wish is my command!
Objectives To promote acquisition of fluency in imperative speech
Treatment domain Speech, emotional, and social domains
Treatment technique Fluency-shaping or stuttering modification techniques
Play type (social ∣cognitive)Cooperative and practice play
Interaction technique Peer learning
Participants’role & behavior In this scenario, the client who stutters acts as instructor and the robot as the assistive
agent
Activity description
This scenario consists of elements of cooperative and practice play. The premise of this
scenario is rooted in the difficulty clients who stutter may have with imperative statements
and requests for service. In this scenario, the client could take on the role of the instructor
or individual seeking a service in different everyday social scenarios (e.g., ordering food,
calling a store or playing a song), while the social robot complies with requests. The
imperative commands will need to be fluent in order to elicit the response from the social
robot
Robot configuration & mode of operation A social robot with the capability to speak and operate semi-autonomously, such as
Pepper, QT, or Nao
Setting & time A skit scenario can be used in a school or home setting
Variation The imperative commands can be scripted and presented to the individual who stutters on
a tablet in order to provide the client with a point of reference. The context in the scenario
can involve the most relevant social situations for the client
Benefits Clients can practice imperative statements and requests in an engaging yet non-
judgemental environment. For the individuals with stuttering, the social robot can function
as a stepping stone that assists with establishing skill and confidence needed outside the
clinic
Table 7: Description of scenario 7
Scenario 7: scavenger hunt
Objectives To promote acquisition, transfer, and maintenance of therapy techniques
Treatment domain Speech and social domains
Treatment technique Fluency-shaping or stuttering modification techniques
Play type (social ∣cognitive)Cooperative and symbolic play
Interaction technique Peer interaction
Participants’role & behavior In this scenario, the client who stutters and a social robot engage in a search and discover
activity
Activity description
This scenario contains elements of parallel and symbolic play. A client who stutters and the
social robot will engage in a search and recover activity, where both the client and the
social robot search for a hidden item containing a cut-out that specifies a speech task (i.e.,
words or phrases with different complexities). Once the participants find the cut-out, the
client will be prompted to complete the speech task using a targeted therapy technique. As
needed, the social robot could model the speech task for the participant and feedback
could be provided by the clinician. After completing the task successfully, the participants
can search for the next item
Robot configuration & mode of operation A social robot with the capability to speak, move, and operate semi-autonomously, such as
Pepper, QT, or Nao
Setting & time Clinic, school, or home setting
Variation The tasks can be adjusted to the age and reading level of the clients
Benefits This scenario allows clients to practice a variety of therapy skills while engaging the social
robot’s haptic abilities to make the practice session more interesting and entertaining
Social robots in stuttering clinic 37
Table 9: Summary of the proposed play scenarios
Scenario 1 2 3 4 5 6 7 8
Scenario name Beats with peers Musical modelling Bouncy Bingo Talking robots Musical Jeopardy! Smooth skits Scavenger hunt Your wish is my
command!
Play Type Cooperative,
practice
Cooperative, symbolic Cooperative,
games with rules
Associative,
practice
Cooperative,
parallel, game
with rules
Cooperative,
parallel, practice
Cooperative,
symbolic
Cooperative,
practice
Play medium Digital Hybrid Hybrid Digital Digital Digital Hybrid Digital
Treatment
domain
Speech Social, speech Social, emotional Speech Speech, social Speech, social Speech, social Speech, emotional
and social
Treatment
technique
Syllable-timed
speech
Syllable-timed speech,
regulated breathing,
voluntary stuttering,
with positive
reinforcement
Voluntary
stuttering
Prolonged
speech with
operant
conditioning
Variable treatment
skills, e.g.,
prolonged speech
Fluency-shaping or
stuttering
modification
techniques
Fluency-shaping or
stuttering
modification
techniques
Fluency-shaping or
stuttering
modification
techniques
Interaction
technique
Peer learning Peer learning, support
group
Peer learning Model-rival
method
Support group Peer learning Peer learning Peer learning
Participants’
role
R: Instructor/
learner C:
Instructor/learner
R: Model C: Player SLP:
Moderator
R: Peer C:Peer R: Model C: Rival
SLP: Instructor
R1: Model R2:
Model C: Player
R: Peer C: Peer R: Peer C: Peer R: Peer C: Peer
Number of
participants
R:1 C:1 R: 1 C: 2 or more R: 1 C:1 R: 1 C:1 SLP:1 R1, R2:1 C:4
or more
R: 1 C: 1 R: 1 C:1 R: 1 C:1
Setting Home, school, or
clinic (including
telepractice)
School or clinic Home, school, or
clinic (including
telepractice)
School or clinic
(including
telepractice)
Home, school, or
clinic
Home, school
(including
telepractice)or
clinic (including
telepractice)
Clinic, school,
or home
School or home
Note: 1. Digital: The interaction is mediated by an electronic tablet or another electronic device. 2. Hybrid: The interaction is mediated by an electronic tablet and other non-electronic play
material, such as a ball. 3. “R”stands for social robot, “C”stands for the client/human participant, and “SLP”stands for the clinician.
38 Shruti Chandra et al.
commercial robots (e.g., NAO or Pepper),makingthedevel-
opment of social robots feasible with collaborations between
engineers, clients, clinicians, and researchers. These sce-
narios are proposed as the first steps towards productive
collaborations between stakeholders in the fields of stuttering
and HRI research that could further expand and enhance
stuttering treatment options.
To conclude, technological advancements in robotics
and HRI are unlikely to slow down and have the potential
to significantly benefit clinical populations such as people
who stutter. Hence, we argue for the need to explore
the potential of these new technologies through surveys
and pilot studies with relevant stakeholders. Exploring
contemporary technologies, such as social robots, for stut-
tering and other communication disorders, not only pro-
vides a platform to test these technologies in clinical and
healthcare domains but also potentially expands the scope
of treatments. We hope that the appropriate introduction
of social robots in the treatment of people who stutter can
benefitbothspeech-language pathologists and clients.
Presently, we are planning experiments and surveys in
collaboration with speech language pathologists to explore
and further refine the aforementioned scenarios and inves-
tigate how we can demonstrate a therapy context where
social robots are effective tools in the hands of clinicians.
Acknowledgments: We thank Holly Lomheim and Jessica
Harasym for relevant feedback on the article and HRI sce-
narios. it’s repeated in Funding Information.
Funding information: This research was funded, in part,
by the Canada 150 Research Chair Programme. Shruti
Chandra, Garima Gupta, Kerstin Dautenhahn: Canada
150 Research Chair Programme.
Author contributions: SC contributed to the conceptuali-
zation, methodology, writing, and editing of the original
draft preparation. GG worked on the conceptualization
and editing of the original draft. TL supervised, reviewed,
and edited the manuscript. KD supervised, reviewed,
edited the manuscript, and acquired the funding. The
authors applied the sequence of authors (SDC)approach
for the sequence of authors.
Conflict of interest: The authors state no conflict of
interest.
Informed Consent: The conducted research does not
include any study, as no informed consent obtained
during the research work.
Ethical approval: The conducted research is not related to
either human or animals use.
Data availability statement: Data sharing is not applic-
able to this article, as no datasets were generated during
the research work.
References
[1]A. Almudhi, “Evolution in technology and changes in the
perspective of stuttering therapy: A review study,”Saudi J.
Biol. Sci., vol. 28, no. 1, p. 623, 2021.
[2]W. H. Organization, Manual of the international statistical
classification of diseases, injuries, and causes of death,
Geneva (Switzerland), WHO, 1977, vol. 1.
[3]J. E. Prasse and G. E. Kikano, “Stuttering: an overview,”
American Family Physician, vol. 77, no. 9,
pp. 1271–1276, 2008.
[4]C. Theys, A. Van Wieringen, and F. Luc, “A clinician survey of
speech and non-speech characteristics of neurogenic stut-
tering,”J. Fluency Disorders, vol. 33, no. 1, pp. 1–232008.
[5]E. Yairi and N. G. Ambrose, Early Childhood Stuttering,
Pro Ed, Austin, USA, 2004.
[6]O. Bloodstein and N. Bernstein-Ratner, A Handbook
on Stuttering, 6th edn, New York, NY: Thomson-
delmar, 2008.
[7]S. Reilly, M. Onslow, A. Packman, M. Wake, E. L. Bavin,
M. Prior, et al., “Predicting stuttering onset by the age of 3
years: A prospective, community cohort study,”Pediatrics,
vol. 123, no. 1, pp. 270–277, 2009.
[8]H. Maaansson, “Childhood stuttering: Incidence and
development,”J. Fluency Disorders, vol. 25, no. 1,
pp. 47–57, 2000.
[9]A. Craig, K. Hancock, Y. Tran, M. Craig, and K. Peters,
“Epidemiology of stuttering in the community across the
entire life span,”J. Speech Language Hearing Res., vol. 45,
no. 6, pp. 1097–1105, 2002.
[10]E. Yairi and N. Ambrose, “Epidemiology of stuttering: 21st
century advances,”J. Fluency Disorders, vol. 38, no. 2,
pp. 66–87, 2013.
[11]O. Bloostein and N. Bernstein Ratner, A Handbook on
Stuttering, Clifton Park, NY: Thompson/Delmar, 2008.
[12]J. S. Yaruss, “Assessing quality of life in stuttering treatment
outcomes research,”J. Fluency Disorders, vol. 35, no. 3,
pp. 190–202, 2010.
[13]D. Kully, “Intensive treatment of stuttering in adolescents
and adults,”Stuttering Related Disorders Fluency,
pp. 213–232, 2007.
[14]G. Gupta, S. Chandra, T. Loucks, and K. Dautenhahn,
“Stuttering treatment approaches from the past two dec-
ades: Comprehensive survey and review,”J. Student Res.,
2022, in press.
[15]J. Kalinowski, V. K. Guntupalli, A. Stuart, and T.
Saltuklaroglu, “Self-reported efficacy of an ear-level pros-
thetic device that delivers altered auditory feedback for the
Social robots in stuttering clinic 39
management of stuttering,”Int. J. Rehabil. Res., vol. 27,
no. 2, pp. 167–170, 2004.
[16]J. H. Davidow, “Systematic studies of modified vocalization:
the effect of speech rate on speech production measures
during metronome-paced speech in persons who stutter,”
Int. J. Language Commun. Disorders, vol. 49, no. 1,
pp. 100–112, 2014.
[17]V. A. Coppola and E. Yairi, “Rhythmic speech training with
preschool stuttering children: An experimental study,”
J. Fluency Disorders, vol. 7, no. 4, pp. 447–457, 1982.
[18]S. Block, M. Onslow, R. Roberts, and S. White, “Control of
stuttering with EMG feedback,”Adv. Speech Language
Pathol., vol. 6, no. 2, pp. 100–106, 2004.
[19]S. B. Brundage and A. B. Hancock, “Real enough: Using vir-
tual public speaking environments to evoke feelings and
behaviors targeted in stuttering assessment and treatment,”
Am. J. Speech-Language Pathol., vol. 24, no. 2,
pp. 139–149, 2015.
[20]A. Moïse-Richard, L. Ménard, S. Bouchard, and A.-L. Leclercq,
“Real and virtual classrooms can trigger the same levels of
stuttering severity ratings and anxiety in school-age children
and adolescents who stutter,”J. Fluency Disorders, vol. 68,
p. 105830, 2021.
[21]A. Almudhi, “Evaluating adaptation effect in real versus vir-
tual reality environments with people who stutter,”Expert
Rev. Medical Devices, vol. 19, no. 1, pp. 1–7, 2021.
[22]E. Haynes and M. Langevin, “Telepractice at the institute for
stuttering treatment and research (istar),”in: Paper pre-
sented at the 13th International Stuttering Awareness Day
Online Conference for ISAD on-line Forum. October, 2010.
[23]M. McGill, N. Noureal, and J. Siegel, “Telepractice treatment
of stuttering: A systematic review,”Telemed. e-Health,
vol. 25, no. 5, pp. 359–368, 2019.
[24]R. N. Madeira, P. Macedo, P. Pita, Í. Bonança, and
H. Germano, “