JOURNAL OF APPLIED BEHAVIOR ANALYSIS 2019, 52, 756–771 NUMBER 3 (SUMMER)

Teaching mands for information using speech generating devices:

A replication and extension
M. ALICE SHILLINGSBURG
MAY INSTITUTE, RANDOLPH, MA

VIDESHA MARYA, BRITTANY L. BARTLETT AND TAYLOR M. THOMPSON

MARCUS AUTISM CENTER, ATLANTA, GA

Approximately 30% of individuals diagnosed with autism spectrum disorder (ASD) fail to
develop vocal communication and, therefore, use some form of augmentative or alternative com-
munication system. The current study replicates and extends previous research on teaching
“Who?” and “Which?” mands for information to 3 young children diagnosed with ASD using a
speech generating device. Procedures were evaluated using a multiple baseline across participants
design. All participants learned to mand for information and, subsequently, used the informa-
tion to access preferred items.
Key words: augmentative and alternative communication, autism spectrum disorder, mands
for information, speech generating device, “wh” questions

Autism spectrum disorder (ASD) is charac- supports the effectiveness of these methods in
terized by persistent impairments in social improving communication skills with non-
communication and restricted, repetitive pat- vocal children.
terns of behavior, interests, or activities Across individuals with developmental dis-
(American Psychiatric Association, 2013). abilities and language delays, mand training,
Within ASD, communication is considered to also referred to as request training, has often
be a core deficit. Although a majority of chil- been the focus of early communication inter-
dren learn to communicate vocally, close to vention (Cowan & Allen, 2007). Current
30% of children with ASD remain minimally research on SGD interventions shows a similar
vocal, even after years of intervention (Rose, trend, with mands often targeted as the first
Trembath, Keen, & Paynter, 2016; Tager- type of communication (van der Meer &
Flusberg & Kasari, 2013). For these individ- Rispoli, 2010). During typical development
uals, augmentative and alternative communi- (Turnbull & Justice, 2017), early mands may
cation (AAC) systems such as manual signs, include requests for preferred toys, snacks, or
picture exchange communication system people (e.g., mama). This repertoire becomes
(PECS), and speech generating devices (SGD) increasingly more complex by including mands
are often used to either supplement for items that are not present, mands for future
(i.e., augment) or replace (i.e., serve as an events (e.g., going to the playground), and
alternative to) vocal speech (Ganz, 2014). A mands for information (e.g., “Where is my
review conducted by Baxter, Enderby, Evans, ball?”; Turnbull & Justice, 2017). Although
and Judge (2012) highlighted evidence which typically developing children appear to learn to
ask questions effortlessly (Brown, 1968), chil-
Correspondence concerning this article should be dren with ASD and other language delays often
addressed to: Alice Shillingsburg, May Institute, 41 Pacella require specialized intervention to learn to
Park Dr., Randolph, MA 02368. E-mail: ashillingsburg@
mayinstitute.org mand for information from others (Endicott &
doi: 10.1002/jaba.579 Higbee, 2007).
© 2019 Society for the Experimental Analysis of Behavior
756
 MANDS FOR INFORMATION SGD 757

Several studies have demonstrated effective pro- significant deficits in vocal communication and
cedures to teach children with ASD and language primarily communicated using a SGD. None
delays to mand for information (e.g., Betz, of the participants requested information prior
Higbee, & Pollard, 2010; Endicott & Higbee, to the start of the study. Justin was a 6-year-old
2007; Lechago, Carr, Grow, Love, & Almason, male scoring predominately within level 3 on
2010; Shillingsburg, Valentino, Bowen, Bradley & the Verbal Behavior Milestones Assessment and
Zavatkay, 2011; Sundberg, Loeb, Hale, & Placement Program (VB-MAPP; Sundberg,
Eigenheer, 2002), and include conditions in 2008). Justin’s communication consisted of
which information is needed (i.e., establishing typing and selecting word icons on a technol-
operation [EO] condition) and in which informa- ogy device using the TouchChat with Word-
tion is not needed (i.e., abolishing operation Power application. Justin demonstrated fluent
[AO] condition) to ensure that acquired mands mands for items and activities and learned
are under appropriate functional control of the novel mands for items without specific training.
motivating operation (Howlett, Sidener, Progar, & Justin followed multistep instructions and
Sidener, 2011; Ingvarsson & Hollobaugh, could type words and sentences, including
2010; Lechago, Howell, Caccavale, & Peter- punctuation. Emma was a 7-year-old female
son, 2013; Shillingsburg, Bowen, Valentino, & scoring predominately within level 3 on the
Pierce, 2014). Notably, the majority of VB-MAPP. Emma’s communication consisted
the previous research on teaching mands for of selecting word icons on a technology device
information to children with ASD and other using the Proloquo2Go application, though she
developmental disabilities focuses on vocal lan- occasionally typed words. Emma demonstrated
guage. In one exception, Carnett and Ingvarsson fluent mands for items and activities, learned
(2016) taught a nonvocal child who communi- novel mands for items without specific training,
cated primarily using a speech generating device and followed multistep instructions. Bruce was
(i.e., Proloquo2Go®) the mand, “I don’t know, a 3-year-old male scoring predominantly within
please tell me.” Even though this study included level 2 on the VB-MAPP. Bruce demonstrated
only one participant, these results suggest that the fluent mands for items and activities, learned
same strategies used to teach vocal mands novel mands for items without specific training,
(Ingvarsson & Hollobaugh, 2010) can be success- and primarily selected word icons on a technol-
fully applied to those using SGDs. However, ogy device with the Proloquo2Go application
there remains a need to evaluate procedures to to communicate, though he occasionally typed
teach additional variations of question-asking to words. Bruce also followed simple instructions.
those using SGDs (Schlosser, 2003). The primary Both TouchChat with WordPower and
purpose of the current study was to replicate and Proloquo2Go are assistive technology applica-
extend procedures used to teach mands for infor- tions that provide symbol and text options for
mation using “who?” and “which?” (Shillingsburg communication (example screenshots of the
et al., 2014; Shillingsburg, Gayman, & Walton, specific SGD response topographies [i.e., typing,
2016) to three children with ASD using SGDs. text selection, and icon selection] can be found
in Supporting Information). They also include
a voice output option such that the symbol or
METHOD
text communication that is selected is delivered
Participants, Setting, and Materials via a digitized prerecorded voice.
Three children diagnosed with ASD who All participants received comprehensive behav-
attended an intensive behavioral clinic partici- ior analytic one-on-one intervention approxi-
pated in this study. All three children exhibited mately 10 hr per week. As part of intervention,
758 M. ALICE SHILLINGSBURG et al.

we regularly conducted structured preference were conducted. A reduced number of trials

assessments to identify putative reinforcers for occurred due to difficulties scheduling availability
mand training and other skill acquisition pro- with extra known staff. Regular activities associ-
gramming. Preference assessments included a ated with treatment sessions, such as discrete-trial
multiple stimulus without replacement (MSWO; instruction, small group activities, and play were
DeLeon & Iwata, 1996) with modifications to interspersed throughout the day. Each participant
the number of repetitions to determine ranking had a history of manding for preferred items
(e.g., Carr, Nicolson, & Higbee, 2000). Once from adults in the center.
fluent mands for tangible items were established,
spontaneous mands were honored during instruc-
tion beginning on a fixed ratio 1 schedule and Dependent Measures and Experimental Design
gradually introducing wait intervals and periodic Trained therapists collected trial-by-trial data
denials. In general, participants emitted mands using paper and pen on the response “Which
for preferred items during instruction. cup?” or “Who has it?” via typing (Justin) or
We conducted training in a classroom with at selection of the text icon (Emma and Bruce).
least four known therapists present. Additional Responses were scored as either independent,
children and therapists, both known and prompted, irrelevant (non-target), or no response.
unknown, were present in the classroom but not Independent “Who?” and “Which?” responses
directly involved in the trials. The room con- occurred on EO and AO trials prior to the
tained tables, chairs, cabinets, bookshelves, and prompt. Specifically, participants using text icon
other materials typically found in a classroom set- selection (Emma and Bruce) navigated from the
ting. Additionally, when teaching the mand for “Home” page to the “Questions” folder and
information using “Which,” the therapist used selected the correct “wh-” question from an array
nine cups with different colors or pictures of five questions (i.e., “Who?”, “Which?”, and
attached to the side. When teaching the mand three distractors; see Supporting Information), at
for information using “Who,” preidentified thera- which point the speech output occurred. The
pists known to the participant were present. For icon arrangement on the question page did not
textual prompts, we used sheets of white com- vary. Typed responses consisted of typing the
puter paper with the phrase, “Which ___?” or “wh-” question on the digital keyboard and then
“Who has it?” printed on the paper. We selected touching the typed sentence to produce the
preferred items based on an interview with care- speech output. In all cases, the speech output
givers and direct observation of mands for the consisted of the full phrase for each question
items. However, the specific item used in each (i.e., “Who has it?” and “Which cup?”). “Who?”
trial was selected based on the mand spontane- and “Which?” responses were scored as prompted
ously emitted by a participant. The item that was if they occurred after a prompt (e.g., textual;
manded for on a daily basis was hidden under a described in Treatment section).
cup or was with a nearby therapist. Upon an We scored correct approaches when the partic-
instance of an unprompted mand by the partici- ipant approached the correct individual or
pant in the absence of the item, we initiated a selected and lifted the container within 5 s of
trial. Given that trials were initiated by the partic- the information being provided. When the par-
ipant’s mand, the time between trials could vary ticipant approached the correct person, navi-
from just a few seconds to more than 1 hr. Each gated to the preferred items folder, and selected
trial lasted approximately 1 min. We conducted the preferred item icon (Emma and Bruce) or
between 5 and 20 trials each day, with the excep- typed the name of the preferred item (Justin)
tion of 2 days for Justin, when only three trials within 5 s, a correct reissued mand was scored.
 MANDS FOR INFORMATION SGD 759

The cumulative number of mands for infor- (Which, Who, Reissued Mand, and Alternating
mation “Which cup?” or “Who has it?” was Scenarios; see Procedures). An agreement was
recorded. Mastery criteria consisted of 9 of 10 tri- defined as both the therapist and the secondary
als with correct independent mands for informa- observer recording a “Who?” or “Which?”
tion on EO trials and no mands for information response as independent, prompted, or that no
on the three most recent AO trials. For Bruce response or an irrelevant response occurred
and Justin, the mand for information “Which?” within a trial. Interobserver agreement (IOA)
was taught and mastered prior to teaching the was calculated by dividing the number of agree-
mand for information “Who?”. For Emma, the ments by the number of agreements and dis-
two mands were targeted simultaneously albeit in agreements and converting the ratio to a
separate blocks of trials at different times during percentage.
the day. These differences occurred due to logis- For “Which?” trials, IOA for Justin, Emma,
tical and scheduling issues idiosyncratic to the and Bruce was 100%, 97% (range, 75%-
participants. Following mastery of the mand 100%), and 97% (range, 75%-100%), respec-
“Who has it?”, data were collected on whether tively and was scored for 10%, 73%, and 70%
the participant reissued the mand for the item of sessions for Justin, Emma, and Bruce,
once he or she approached the correct person. respectively. For “Who” trials, IOA for Justin,
Mastery criteria consisted of four out of five cor- Emma, and Bruce was 92% (range, 75%-
rect reissued mands during the initial probes or 100%), 99% (range, 75%-100%), and 98%
9 out of 10 correct responses in treatment if (range, 50%-100%), respectively and was
treatment was necessary. scored for 98%, 80%, and 78% of sessions for
Following acquisition of both mands Justin, Emma, and Bruce, respectively. IOA for
“Which cup?” and “Who has it?” separately, Reissued Mand trials was 93% (range, 75%-
responding was evaluated when the scenarios 100%), 85% (range, 67%-100%), and 99%
were rapidly alternated to ensure discrimination (range, 75%-100%) for Justin, Emma, and
between the mand forms. During this evalua- Bruce, respectively and was scored for 67%,
tion, the therapist alternated between “Which” 38%, and 69% of sessions for Justin, Emma,
and “Who” trial types including both EO and and Bruce, respectively. IOA during the Alter-
AO trials as described below. Mastery criteria nating Scenarios for Justin, Emma, and Bruce
consisted of four out of five correct indepen- was 100%, 100%, and 99% (range, 75%-
dent responses (i.e., using the appropriate 100%) respectively and was scored for 94%,
mand under the correct conditions) with no 60%, and 34% of sessions for Justin, Emma,
additional intervention or 9 out of 10 trials in and Bruce, respectively.
treatment, if treatment was necessary. A second observer scored treatment integrity
We used a nonconcurrent multiple-baseline data in vivo during sessions for baseline, train-
across participants design (Watson & Work- ing, and posttraining. The second observer
man, 1981) to assess the effects of mand train- scored data on the therapist’s implementation
ing across conditions. of protocol procedures. Treatment integrity
during baseline, treatment, and post measures
was calculated via a six to eight item checklist
Interobserver Agreement and Treatment of the therapist behaviors required to complete
Integrity a trial. Generally, the therapist’s behaviors
A second independent observer scored data included providing appropriate antecedent con-
in vivo during trials in baseline, training, and ditions, implementing appropriate prompts,
posttraining for each phase of the study and delivering appropriate consequences (see
760 M. ALICE SHILLINGSBURG et al.

Supporting Information for example). Treat- row. This was adhered to in all cases with the
ment integrity was calculated by dividing the exception of Justin’s baseline for the “Which?”
number of correct therapist behaviors by the condition, in which case a procedural deviation
number of applicable therapist behaviors and occurred. For both conditions, a set of nine
converting the ratio to a percentage. cups was present and in sight (e.g., 0.3 to
Treatment integrity during trials for 0.9 m away on a table). A preferred item was
“Which?’ was 100% for all participants and placed under a predetermined cup out of sight
was scored during 6%, 35%, and 41% of ses- of the participants (e.g., when they were on a
sions for Justin, Emma, and Bruce, respectively. break away from the table). Once back at the
During trials for “Who?”, treatment integrity table, a trial began when the participant man-
for Justin, Emma, and Bruce was 92% (range, ded for the preferred item. In the AO Condi-
50%-100%), 100%, and 100% and was scored tion the therapist responded to the mand for
during 46%, 41%, and 45% of sessions, the item by providing the information needed
respectively. During trials for “Reissued to gain access to their preferred item (e.g., “It’s
Mands,” treatment integrity for Emma and under the yellow cup”). The participant was
Bruce was 100% and 98% (range, 80%-100%) then allowed to select and lift one cup. If the
and was scored during 23% and 31% of ses- participant selected the correct cup (i.e., the
sions, respectively. Treatment integrity data cup corresponding to the information pro-
were not collected for Justin due to unforeseen vided), the participant was allowed to pick up
scheduling conflicts. During trials for “Alternat- the cup and access the preferred item under the
ing Scenarios,” treatment integrity for Justin, cup. If the participant selected an incorrect
Emma, and Bruce was 98% (range, 80%- cup, the participant was again allowed to pick
100%), 100%, and 99% (range, 80%-100%) up the cup and see that the preferred item was
and was scored during 71%, 30%, and 26% of not there, and the trial ended. During the AO
sessions, respectively. trials mands for information did not produce
any programmed consequences. In the EO
Condition, following the participant’s mand for
Procedures a preferred item, the therapist did not provide
Preassessment. A preassessment was con- the information needed to gain access and
ducted to ensure participants could respond to instead provided a general statement (e.g., “It’s
instructions to go to a named person located under one of the cups”). Mands for informa-
within the classroom or to approach and select tion (e.g., “Which cup?”) were reinforced by
a named container. Following three consecutive providing information (e.g., “It’s under the yel-
correct approaches to the adult or container, low cup”) to access the preferred item. The par-
the response to the preassessment trial was ticipant was allowed to select one cup. If the
deemed known. For each participant, four participant selected the correct cup, they were
known adults and nine known containers were allowed access to the preferred item. If the par-
identified. No teaching was required for any ticipant selected an incorrect cup, the trial
participant. For a detailed description of these ended, and the participant was engaged in the
preassessment procedures, see Shillingsburg next activity or task as indicated on their indi-
et al. (2014). vidual schedules.
Baseline/Posttreatment. For the “Which” pro- For the “Who” protocol, trials were the same
tocol, trials were randomized between EO and as described above, with the exception that the
AO conditions such that no more than two tri- preferred item was in the possession of one of
als of the same condition were conducted in a the four known adults who were present in the
 MANDS FOR INFORMATION SGD 761

room. Known adults held one hand behind and were identical to posttraining with the
their backs or otherwise out of sight while exception that trials for “Which” and “Who”
appearing occupied in an activity such as under EO and AO conditions were now inter-
watching another session or looking at a maga- spersed. The order of conditions was randomly
zine. For both EO and AO conditions, a pre- assigned such that all four trial types were
ferred item was given to a selected individual rotated through prior to repeating a trial type.
out of sight of the participant. A trial began Mand training (Which). During this condi-
when the participant manded for the preferred tion, nine cups were displayed in a visual array
item. In the AO condition, the therapist on the table in front of the participant (0.3 to
responded to the mand by providing the infor- 0.9 m away). The order of the cups varied from
mation needed to gain access to the preferred trial to trial. The therapist hid a preferred item
item (e.g., “Alfred has it”). If the participant under one of the cups out of sight of the partic-
approached the correct individual, he or she ipant. Trials were initiated once a participant
was given access to the preferred item. The par- emitted the previously acquired mand for a pre-
ticipant was only allowed to approach one indi- ferred item using his or her speech generating
vidual. During the AO trials mands for device. Subsequently, the therapist responded,
information did not produce any programmed “It’s under one of these cups” during EO trials
consequences. In the EO condition, the thera- or “It’s under the [named] cup” during AO tri-
pist did not provide the information needed to als. We used a constant time delay prompt
gain access to their preferred item (e.g., “One (Schuster, Gast, Wolery, & Guiltinan, 1988)
of your friends has it”). Mands for information which consisted of the therapist immediately
(e.g., “Who has it?”) were followed by specific presenting the textual prompt, “Which cup?”
information (e.g., “Sarah has it”) to access the on the table for the first eight trials. Following
preferred item. The participant was only eight prompted trials, with the last three trials
allowed to approach one individual. If the par- correct, the prompt was delayed by 2 s. Due to
ticipant approached the correct individual, he the lack of independent responding from
or she was given access to the preferred item. If Bruce, the delay was lengthened to 5 s, then
the incorrect adult was approached, no pre- 10 s, then 45 s, at which point we returned to
ferred item was provided, the participant was the 0-s delay for three trials and began to fade
redirected back to the desk, and the trial again. Gestural prompts consisted of pointing
ended. to the device or to a particular folder or key on
We noticed that some participants the keyboard to redirect attention or to keep
approached the correct adult, stood in front of going if the participant paused during
the adult, and waited for the item to be given. responding. Correct prompted and indepen-
In the present study, the adults knew why the dent responses were followed by information
participant was approaching them, but outside regarding the location of the preferred item.
of the study procedures, that may not always Mastery criteria consisted of 9 out of 10 consec-
be the case. Therefore, following mastery of the utive EO trials with independent responding
mand for information “Who?” we then assessed and no responding on the last three AO trials.
whether the participant reissued the mand for Mand training (Who). During this condition,
the initial item within 5 s of approaching the three additional known adults were positioned
named individual. with hands behind their backs, standing or sit-
Alternating scenarios assessment. Probes were ting in a natural posture, within the classroom
conducted following acquisition of the at a minimum of 1.5 m from the participant.
“Which” response and the “Who” response, The placement of the adults varied from trial
762 M. ALICE SHILLINGSBURG et al.

to trial. Out of the participant’s sight, one adult for “Which?” (left panel) and the approach
was given a preferred item to hold. Trials began behavior (right panel) under EO and AO con-
following a mand for the preferred item using ditions for Bruce (top panel), Emma (middle
the SGD. Following a mand, the therapist panel), and Justin (bottom panel). Each data
responded identically to baseline according to point represents one trial. The closed data path
the EO (e.g., “one of your teachers has it”) or represents EO trials and the open data path
AO (e.g., “Sarah has it”) condition. We represents AO trials. Each occurrence of correct
implemented a constant time delay prompt responding is depicted by an increase in the
(Schuster et al., 1988) in which the textual data path. A cumulative data display was
prompt “Who has it?” was used to teach the selected because each trial was initiated by a
mand similarly to the “Which” procedures with mand for an item by the participant, so sched-
the exception that no additional prompt delays uling sessions with a specific number of trials
past 2 s were needed for any participants. was not possible. In baseline, none of the par-
Reissued mands were probed once the ticipants manded for information under either
“Who?” mand for information was mastered. If condition. All three participants approached the
no correct responses occurred during the initial correct container only in the AO condition
probe trials, treatment was initiated (Emma when the information on the location of the
and Bruce only). Following the mand for infor- preferred item was already provided. In con-
mation “Who?”, the therapist immediately trast, participants did not approach the correct
prompted the participants to take their SGD as container in the EO condition during baseline,
they approached the named adult. Prompts as the information on the location was withheld
consisted of gestures (e.g., pointing to the and mands for information did not occur. Fol-
device) and/or physical guidance to pick up the lowing mastery of the “Which?” mand, all three
SGD. No additional prompts to use the SGD participants emitted this mand under EO con-
with the adult were needed. Following eight tri- ditions and never under AO conditions. All
als at a 0-s prompt delay, with the last three tri- three participants continued to approach the
als correct, the therapist provided a 2-s delay correct container under AO conditions, but
for subsequent trials. Mastery criterion con- also approached the correct container under
sisted of 9 out of 10 consecutive trials of inde- EO conditions after emitting the mand for
pendently reissuing the mand. information and being provided with the
Alternating scenarios training. If we observed requested information.
more than one error during the Alternating Figure 2 shows the cumulative number of
Scenarios assessment, intervention to promote mands for information during training for
discriminated manding was initiated (Bruce “Which?”. Emma and Justin both met criterion
only). Intervention was identical to that to fade to the 2-s delay quickly. Justin emitted
described in the “Which” and “Who” proce- two correct mands during the immediate
dures above; however, trials were interspersed prompt phase simultaneous to the therapist’s
among all conditions (i.e., EO and AO and prompt. Therefore, those responses were coun-
“Which” and “Who”). The order of conditions ted as independent. Emma met mastery crite-
was randomly assigned, as described above. rion within 44 trials of alternating EO and AO
conditions and consistently selected the correct
container when the information was provided.
RESULTS Justin met mastery criterion within 39 trials of
Figure 1 displays the cumulative number of alternating EO and AO conditions and consis-
mands during baseline and posttraining probes tently selected the correct container. Bruce did
 MANDS FOR INFORMATION SGD 763

Figure 1. Cumulative number of independent mands “Which” (left) and cumulative number of approaches (right)
across EO and AO conditions during pre- and postprobes for Bruce, Emma, and Justin.

not respond independently during the 2-s delay baseline none of the participants manded for
phase and so the delay was increased to 5 s and information under either condition. All three
then to 10 s. Independent responding began to participants approached the correct adult in the
occur at the 10-s delay but responding did not AO condition when the information on which
persist. A 45-s delay was implemented and lack therapist had the preferred item was already
of responding and errors occurred. Thus, provided. Justin and Bruce approached the cor-
immediate prompting to prevent errors was rect therapist by chance in one of the EO con-
reinstated. Correct prompted responding dition trials in baseline. Following mastery of
occurred consistently and independent the “Who?” mand (see Figure 4), all three par-
responding occurred with the re-initiation of ticipants emitted this mand under EO condi-
the 2-s delay. Bruce met mastery criteria within tions and did not emit the mand under AO
99 trials and consistently selected the correct conditions. All three participants approached
container when the information was provided. the correct adult under AO and EO conditions
Figure 3 displays the cumulative number of once the information regarding which therapist
mands during baseline and posttraining probes had the preferred item was provided (right
for the response “Who?” (left panel) and the panel, Figure 3).
approach behavior (right panel) under EO and Figure 4 depicts the cumulative number of
AO conditions for Justin (top panel), Emma mands during training for “Who?”. All three
(middle panel), and Bruce (bottom panel). In participants met criteria to fade to the 2-s delay
764 M. ALICE SHILLINGSBURG et al.

Figure 2. Cumulative number of independent and prompted mands “Which” (left) and cumulative number of
approaches (right) across EO and AO conditions during training for Bruce, Emma, and Justin.

following eight immediately prompted EO tri- EO and AO “Who?” and “Which?” trials were
als. Justin met mastery criteria within 44 total alternated for Bruce only. Following eight total
trials, Emma met mastery criteria within EO trials in which an immediate prompt was
68 total trials, and Bruce met mastery criteria provided, a 2-s delay was introduced. Bruce
within 29 total trials. All three participants con- responded correctly and independently on all
sistently approached the named adult when the subsequent trials and met mastery criteria
information on who had the preferred item was within 34 total EO and AO trials. Posttraining
provided. during alternating scenarios was then con-
Figure 5 depicts the cumulative number of ducted (see Figure 5) and Bruce differentially
mands during EO and AO conditions for both emitted mands for information “Which?”
the “Who?” and “Which?” trial types when and “Who?”
interspersed in the alternating scenario. Justin Figure 7 shows the results of the reissued
and Emma emitted mands for information mands for items once the named adult was
“Which?” and “Who?” differentially according approached during “Who?” trials. Justin
to the correct condition and only during EO reissued the original mand for the preferred
trials (left panel). In contrast, Bruce emitted item after he approached the correct adult on
the incorrect mand for information on 3 of all but the first trial (left panel). Thus, no train-
10 EO trials. Despite emitting the wrong mand ing was conducted. Neither Emma nor Bruce
type, he never manded for information on any reissued the original mand after they
AO trials. Figure 6 shows training trials when approached the named adult. The right panel
 MANDS FOR INFORMATION SGD 765

Figure 3. Cumulative number of independent mands “Who” (left) and cumulative number of approaches (right)
across EO and AO conditions during pre- and postprobes for Justin, Emma, and Bruce.

of Figure 7 depicts training trials of reissued the provided information to access the pre-
mands. Emma met mastery criteria within ferred item. All participants responded differen-
26 trials and Bruce met mastery within 18 trials. tially between conditions in which the mand
Posttraining probes show that both Emma and frame “Who?” and the mand frame “Which?”
Bruce reissued the original mand on all or most were appropriate, and when information was
trials. needed versus when the information had
already been given. More specifically,
responding occurred during EO trials and never
DISCUSSION during AO trials when scenarios were alter-
The current study replicates and extends nated, suggesting that mands were under
procedures previously used to teach mands for appropriate functional control of the motivat-
information to vocal children with ASD ing operation.
(Shillingsburg et al., 2014, 2016) to those Some idiosyncratic results were observed
using SGDs. In the current investigation, three across the participants when “Who?” and
participants were successfully taught to mand “Which?” scenarios were alternated. Two out
for information using “Who?” and “Which?” of the three participants (Justin and Emma)
questions under conditions in which motiva- manded for information correctly using
tion to ask the question was present and to use “Who?” or “Which?” when the corresponding
766 M. ALICE SHILLINGSBURG et al.

Figure 4. Cumulative number of independent and prompted mands “Who” (left) and cumulative number of
approaches (right) across EO and AO conditions during training for Bruce, Emma, and Justin.

conditions were randomly alternated, without learned to correct those errors simply through
the need for additional training. Bruce, how- contact with differential reinforcement as
ever, made errors when the two scenarios were applied during the alternating scenarios assess-
alternated by occasionally responding “Who?” ment. For example, when he emitted the
on “Which?” EO trials and vice versa. This “Who?” mand in the which context, the
highlights the need to assess not only differen- response was ignored and no information was
tial responding between EO and AO condi- provided, prolonging delay to the preferred
tions, but also use of the correct mand frame. item given that he did not have the information
Potentially, for some individuals, teaching to efficiently access it. With repeated exposures
mands for information in isolation (e.g., only to trials in which the correct mand frame led to
one mand type), without plans to intersperse information to access the preferred item and
scenarios, could lead to errors. For example, the the incorrect mand frame did not, he may have
child may learn to ask a question when infor- learned to use the correct mand frames without
mation is needed, but may incorrectly use the being prompted to do so by the therapist. We
learned mand frame in novel contexts. It is pos- did not assess this possibility and, instead, pro-
sible that direct teaching to use the correct vided direct prompts to use the correct mand
mand frame in the correct context was not nec- frame in the correct context. In general, clini-
essary for Bruce. The initial assessment with cians should be prepared to assess correct mand
the alternating scenarios was very short, and frame use across EO and AO conditions, as
though he made a few errors, he may have well as across contexts, and to provide
 MANDS FOR INFORMATION SGD 767

Figure 5. Cumulative number of independent mands “Which” and “Who” (left) and cumulative number
approaches (right) during Alternating Scenarios probes for Justin, Emma, and Bruce.

Figure 6. Cumulative number of independent and prompted mands “Which” and “Who” (left) and cumulative
number of approaches (right) during training for Alternating Scenarios for Bruce.

additional training if needed. In previous stud- et al., 2014), and, as described in the present
ies, some participants have required additional study, to promote the correct mand frames by
procedures to promote responding on EO trials using “Who?” when referring to people and
only (Shillingsburg et al., 2014), some partici- using “Which?” when referring to the
pants have required additional training to emit container.
mands for information in the presence of other An additional component of the study was
materials (e.g., bags versus cups; Shillingsburg the assessment of a reissued mand (Shillingsburg
768 M. ALICE SHILLINGSBURG et al.

Figure 7. Cumulative number of independent reissued mands for item during pre- and postprobes for Justin,
Emma, and Bruce (left panel). Cumulative number of independent and prompted reissued mands for item during train-
ing for Emma and Bruce (right panel).

et al., 2016), which was not included in the approached adult. In the previous study with
original set of procedures (Shillingsburg et al., vocal participants (Shillingsburg et al., 2016),
2014). In the current study, once the mand those requiring intervention to reissue the mand
“Who has it?” was emitted and the information needed vocal prompts to emit the mand once
was provided, participants were required to carry the adult was approached. In the current study,
the SGD with them to the named adult and as long as the device was present, the reissued
mand again for the item. One participant mands occurred independent of prompts from
engaged in mands for the item from the named the clinician. These results further support that
adult, whereas two participants required training teaching children to keep their communication
to reissue the original mand to the approached devices with them, as in the Distance and Persis-
adult. More specifically, participants in the cur- tence training in Phase II of PECS (Bondy &
rent study approached the adult independently Frost, 2002), is quite important for functional
but did not bring the SGD with them and, use of acquired mands when a device is
therefore, could not reissue the mand. Thus, being used.
both participants needed to learn to carry the There are some important considerations
SGD with them. During training, physical and when attempting to teach mands for informa-
gestural prompts to take the SGD with them tion using SGDs. First, the decision to use
were faded until both participants independently AAC, whether high- or low-tech, should be
took the device and reissued the mand to the made by individuals experienced with the
 MANDS FOR INFORMATION SGD 769

various systems and be informed by relevant depicting a keyboard. Although both behaviors
family goals and evidence for the effectiveness require conditional discriminations and the
of each system considered (see Ganz, 2014). emission of a chain of responses to complete
Once a device is selected, the specific commu- them, teaching the typing response was not
nication application and set-up of the system replicated.
must be determined. Devices may be set up to Another limitation is the lack of assessment
display a static array in which the icon or text of generalization or additional programming to
display stays the same until someone manually promote generality. Unlike Shillingsburg et al.
changes it. Devices may also be set up with (2014), we did not formally assess generaliza-
dynamic displays in which pages of icons or tion of mands by contriving additional novel
text are linked together and pushing one but- scenarios or including novel people. Thus, it is
ton automatically takes the user to another dis- not clear whether any of the participants in the
play. In this way, the user navigates to various current study would emit the mand “Which?”
displays through a series of button selections. to access information about a container other
The participants in this study were all fluent than a cup or would emit the mand “Who?”
navigators of their communication devices prior when referring to people other than their thera-
to the start of the study. No specific procedures pists (e.g., peers, family). Previous research has
were needed to teach them to navigate to the shown that generalization of mands for infor-
correct folder in their device. This is an impor- mation may occur without specific program-
tant consideration, as teaching navigation ming (Lechago et al., 2010) and when it does
simultaneously with teaching a mand for infor- not, additional procedures may be effective in
mation may complicate procedures. Future promoting it (Shillingsburg et al., 2014). How-
research could examine whether fluent folder ever, it is still unclear whether those findings
navigation facilitates learning mands for infor- extend to children with ASD who communi-
mation when folder navigation is part of the cate using a SGD. Future research should assess
response chain. generalization across discriminative stimuli
Another consideration is the decision to use (e.g., people, materials, across settings), as well
typing, text icon selection, or picture icon selec- as across MOs (Miguel, 2017).
tion. The participants in the current study had Overall, this study adds to the previous liter-
varying skills in terms of their use of SGDs. ature suggesting that children who use SGDs
For Bruce and Emma, responding consisted of to communicate can be taught more complex
selecting word icons; however, Justin’s communicative repertoires (van der Meer &
responding consisted of typing out the correct Rispoli, 2010). Although research on the use of
response. These response modalities were those SGDs has focused on mand training, the tar-
that were commonly used in each individual’s gets have predominately consisted of mands for
day-to-day communication and were, thus, specific items (e.g., food and snacks) and have
deemed most appropriate for the mands for typically involved selecting icons from a single
information. static display without navigation (Alzrayer,
A potential limitation of the current study is Banda, & Koul, 2014). This is the first study
that a replication was not completed with a to demonstrate acquisition of mands for infor-
participant who engaged in a typing response. mation using high-tech SGDs with dynamic
Although three participants learned to mand displays, specifically TouchChat with Word-
for information using an SGD, two were tau- Power and Proloquo2Go. With the ubiquitous
ght to navigate through folders and select icons, use of technology in schools and homes, and
and one was taught to type words on a screen the ease with which families and clinicians can
770 M. ALICE SHILLINGSBURG et al.

now access applications geared toward enhanc- Endicott, K., & Higbee, T. (2007). Contriving motivat-
ing the communication of those with signifi- ing operations to evoke mands for information in
preschoolers with autism. Research in Autism Spectrum
cant spoken language deficits, more research is Disorders, 1, 210-217. https://doi.org/10.1016/j.rasd.
needed to evaluate procedures for promoting 2006.10.003
the functional use of complex communication Ganz, J. B. (2014). Aided augmentative communication for
individuals with autism spectrum disorders. Springer:
skills. New York.
Howlett, M. A., Sidener, T. M., Progar, P. R., &
Sidener, D. W. (2011). Manipulation of motivating
REFERENCES operations and use of a script-fading procedure to
teach mands for location to children with language
Alzrayer, N., Banda, D. R., & Koul, R. K. (2014). Use of delays. Journal of Applied Behavior Analysis, 44, 943-
iPad/iPods with individuals with autism and other 947. https://doi.org/10.1901/jaba.2011.44-943
developmental disabilities: A meta-analysis of com-
Ingvarsson, E. T., & Hollobaugh, T. (2010). Acquisition
munication interventions. Review Journal of Autism
of intraverbal behavior: Teaching children with
and Developmental Disorders, 1, 179-191. https://doi.
autism to mand for answers to questions. Journal of
org/10.1007/s40489-014-0018-5
Applied Behavior Analysis, 43, 1-17. https://doi.org/
American Psychiatric Association (2013). Diagnostic and 10.1901/jaba.2010.43-1
statistical manual of mental disorders (5th ed.). Arling-
ton, VA: American Psychiatric Association. Lechago, S., Carr, J. E., Grow, L. L., Love, J. R., &
Almason, S. M. (2010). Mands for information gen-
Baxter, S., Enderby, P., Evans, P., & Judge, S. (2012).
eralize across establishing operations. Journal of
Interventions using high-technology communication
Applied Behavior Analysis, 43, 381-395. https://doi.
devices: A state of the art review. Folia Phoniatrica et
org/10.1901/jaba.2010.43-381
Logopaedica, 64, 137-144. https://doi.org/10.1159/
000338250 Lechago, S., Howell, A., Caccavale, M. N., &
Betz, A. M., Higbee, T. S., & Pollard, J. S. (2010). Pro- Peterson, C. W. (2013). Teaching “how?” mand-for-
moting generalization of mands for information used information frames to children with autism. Journal
by young children with autism. Research in Autism of Applied Behavior Analysis, 46, 781-791. https://doi.
Spectrum Disorders, 4, 501-508. https://doi.org/10. org/10.1002/jaba.71
1016/j.rasd.2009.11.007 Miguel, C. F. (2017). The generalization of mands. The
Bondy, A., & Frost, L. (2002). A picture’s worth: PECS Analysis of Verbal Behavior, 33, 191-204. https://doi.
and other visual communication strategies in autism org/10.1007/s40616-017-0090-x
(2nd ed.). Bethesda, MD: Woodbine House. Rose, V., Trembath, D., Keen, D., & Paynter, J. (2016).
Brown, R. (1968). The development of wh questions in The proportion of minimally verbal children with
child speech. Journal of Verbal Learning and Verbal autism spectrum disorder in a community-based early
Behavior, 7, 279-290. https://doi.org/10.1016/ intervention programme. Journal of Intellectual Dis-
S0022-5371(68)80002-7 ability Research, 60, 464-477. https://doi.org/10.
Carnett, A., & Ingvarsson, E. T. (2016). Teaching a child 1111/jir.12284
with autism to mand for answers to questions using a Schlosser, R. (2003). Roles of speech output in augmentative
speech-generating device. The Analysis of Verbal and alternative communication: Narrative review. Aug-
Behavior, 32, 233-241. https://doi.org/10.1007/ mentative and Alternative Communication, 19, 5-27.
s40616-016-0070-6 https://doi.org/10.1080/0743461032000056450
Carr, J. E., Nicolson, A. C., & Higbee, T. S. (2000). Schuster, J. W., Gast, D. L., Wolery, M., & Guiltinan, S.
Evaluation of a brief multiple-stimulus preference (1988). The effectiveness of a constant time-delay
assessment in a naturalistic context. Journal of Applied procedure to teach chained responses to adolescents
Behavior Analysis, 33, 353-357. https://doi.org/10. with mental retardation. Journal of Applied Behavior
1901/jaba.2000.33-353 Analysis, 21, 169-178. https://doi.org/10.1901/jaba.
Cowan, R. J., & Allen, K. D. (2007). Using naturalistic 1988.21-169
procedures to enhance learning in individuals with Shillingsburg, M. A., Bowen, C. N., Valentino, A. L., &
autism: A focus on generalized teaching within the Pierce, L. E. (2014). Mands for information using
school setting. Psychology in the Schools, 44, 701-715. “who?” and “which?” in presence of establishing and
https://doi.org/10.1002/pits.20259 abolishing operations. Journal of Applied Behavior
DeLeon, I. G., & Iwata, B. A. (1996). Evaluation of a Analysis, 47, 136-150. https://doi.org/10.1002/
multiple-stimulus presentation format for assessing jaba.101
reinforcer preferences. Journal of Applied Behavior Shillingsburg, M. A., Gayman, C., & Walton, W. (2016).
Analysis, 29, 519-532. https://doi.org/10.1901/jaba. Using textual prompts to teach mands for informa-
1996.29-519 tion using “Who?” The Analysis of Verbal Behavior,
 MANDS FOR INFORMATION SGD 771

32, 1-14. https://doi.org/10.1007/s40616-016- van der Meer, L. A. J., & Rispoli, M. (2010). Communica-
0053-7 tion interventions involving speech-generating devices
Shillingsburg, M. A., Valentino, A. L., Bowen, C. N., for children with autism: A review of the literature.
Bradley, D., & Zavatkay, D. (2011). Teaching chil- Developmental Neurohabilitation, 13(4), 294-306.
dren with autism to request information. Research in https://doi.org/10.3109/17518421003671494
Autism Spectrum Disorders, 5, 670-679. https://doi. Watson, P. J., & Workman, E. A. (1981). The non-
org/10.1016/j.rasd.2010.08.004 concurrent multiple baseline across-individuals
Sundberg, M. L. (2008). Verbal behavior milestones assess- design: An extension of the traditional multiple base-
ment and placement program: The VB-MAPP. Con- line design. Journal of Behavior Therapy and Experi-
cord, CA: AVB Press. mental Psychiatry, 12, 257-259. https://doi.org/10.
1016/0005-7916(81)90055-0
Sundberg, M. L., Loeb, M., Hale, L., & Eigenheer, P.
(2002). Contriving establishing operations to teach
mands for information. The Analysis of Verbal Received June 26, 2017
Behavior, 18, 15-29. Final acceptance March 14, 2019
Action Editor, Caio Miguel
Tager-Flusberg, H., & Kasari, C. (2013). Minimally ver-
bal school-aged children with autism spectrum disor-
der: The neglected end of the spectrum. Autism
Research, 6, 468-478. https://doi.org/10.1002/aur. SUPPORTING INFORMATION
1329
Additional Supporting Information may be
Turnbull, K. L. P., & Justice, L. M. (2017). Language
development from theory to practice (3rd ed.). Upper found in the online version of this article at the
Saddle River, NJ: Pearson. publisher’s website.