Communication difficulties are a characteristic of autism spectrum disorder (ASD) (American Psychiatric Association, 2013). Children with ASD who are also nonverbal or minimally verbal have even more severe communication challenges (Tager-Flusberg & Kasari, 2013). The communicative ability of nonverbal and minimally verbal children ranges from nonverbal with no spoken language to vocalizations and atypical non-speech sounds to minimally verbal with limited expressive language consisting of a few words or echolalic speech (Tager-Flusberg & Kasari, 2013). An average of 25–30% of children with ASD are nonverbal or minimally verbal, defined as having fewer than thirty words in their vocabulary (Brignell et al., 2018). Furthermore, an estimated one-third of children with ASD who are nonverbal or minimally verbal will remain so as they grow older (Koegel et al., 2019) and may experience social isolation (Roux et al., 2015). Koegel et al. (2020) recommended that children over 18 months of age be classified as nonverbal if they have no consistent spoken words during standardized tests, from observations in different settings, and from parent reports. Koegel et al. (2020) also recommended that children 30 months or older with fewer than 50 spoken words, as determined by a credible word count procedure, be classified as minimally verbal.
Tager-Flusberg and Kasari (2013) noted that approximately 30% of children who are nonverbal or minimally verbal have concurrent cognitive disabilities. Children are assessed against alternate achievement standards (NCSER, 2009) if they have significant cognitive disabilities and receive modified instruction or significant support for skills generalization under their individualized education programs (IEP). According to the Every Student Succeeds Act (GovTrack.us, 2022), students with significant cognitive disabilities who take an alternate assessment account for approximately 1% of the student population. Furthermore, approximately 30% of students who take an alternate assessment have no spoken language or use fewer than three words (Erickson & Geist, 2016).
Initiations are vital to overall communication for children on the autism spectrum who are nonverbal or minimally verbal because they are a precursor to conversation skills (Tager-Flusberg & Kasari, 2013). The skills required for conversation include initiating communication with others and responding to initiations from others (Doggett et al., 2013). An initiation is a communication attempt and may involve gestures, asking for an item, expressing an opinion or a need, or protesting (Tager-Flusberg & Kasari, 2013). Initiations may also be pivotal to the performance of behaviors unrelated to communication (Doggett et al., 2013). Interventions focused on enhancing initiation and response rates have proven to be the most effective in boosting the communication abilities of minimally verbal children (Tager-Flusberg & Kasari, 2013).
Self-monitoring, a component of self-management, is an intervention that research has found to be effective for multiple behaviors of children with ASD, including initiations (Doggett et al., 2013; Koegel et al., 2014; Loftin et al., 2008; Newman & Ten Eyck, 2005). However, a review of the literature from the year 2000 revealed only limited research on self-management or self-monitoring of initiations for individuals who may fit the recommended definitions of nonverbal and minimally verbal by Koegel et al. (2020). Most of the studies for self-management or self-monitoring of initiations have been conducted with individuals who had severe social deficits but could speak in sentences of five or more words, spontaneously mand, tact, emit various intraverbals, and occasionally converse with peers or adults. For example, Koegel et al. (2014) used self-management for reciprocal conversations between four children aged 4–14 years who spoke in sentences of five or more words to request, protest, or discuss restricted interests, but had difficulty sustaining conversation. Koegel et al. (2014) observed that the intervention resulted in more expansive on-topic social conversations maintained in generalized settings. Doggett et al. (2013) used self-management to address question-asking in social conversations between two children aged 8–9 years on the autism spectrum. Both children spoke in sentences of five or more words and could engage in multiple verbal exchanges. Similarly, Loftin et al. (2008) implemented self-monitoring of social interactions and repetitive behavior in conversations between three student peers aged 9–10 years. All students could make statements and participate in conversation, although they rarely did so. Reynolds et al. (2014) studied social initiations in conversations with peers by four students aged 5–6 years. All students had the verbal ability to communicate in sentences and phrases with peers, although they had difficulty with pragmatic language, such as conversational turn-taking and question-asking (Reynolds et al., 2014).
Other studies addressing self-monitoring or self-management of initiations included compliment-giving (Apple et al., 2005), responding (Newman et al., 2000; Tereshko et al., 2021), and verbal interactions (Parker & Kamps, 2011). Apple et al. (2005) addressed compliment-giving initiations using self-management by boys aged 4–5 years. The boys had language abilities at normal levels, were able to identify a compliment, spoke in sentences, and were able to engage in reciprocal conversation. Tereshko et al. (2021) addressed the effects of self-management of motor stereotypy on the social initiations and responding of a 5-year-old boy who was a verbal communicator able to mand, tact, and emit intraverbals of various forms spontaneously. Newman et al. (2000) used self-management for variations in verbal responding for three students aged from preschool to six years who all spoke in complete sentences. Parker and Kamps (2011) used self-monitoring for the completion of task steps, activity engagement, and peer-directed verbalizations with two students with ASD aged nine years. The students were high functioning with 3–5-word spontaneous initiations and able to follow scripts of sentences.
The literature review found three studies with at least one individual who may fit the definition of nonverbal or minimally verbal by Koegel et al. (2020). One study on the self-monitoring of initiations included a 20-year-old male with no verbal language who used a communication book for wants, needs, and emotions (Ganz & Sigafoos, 2005). Another study by Newman and Ten Eyck (2005) used self-management of initiations by children with ASD aged 6–9 years. Two students were verbal and spoke with 3–4-word utterances, while the third student used the picture exchange communication system (PECS). Morrison et al. (2001) used peer mediation and self-monitoring to improve initiations in the form of requesting, commenting, and sharing by four students aged 10–13 years. The communication levels ranged from well-developed with appropriate skills overall, to limited use of words and a lack of spontaneous language, to short sentences. According to Morrison et al. (2001), two students were not as verbally competent as the others, with limited use of words or two- to three-word phrases and no spontaneous language. These two students may have met the criteria Koegel et al. (2020) recommended for minimally verbal communication, although Morrison et al. (2001) noted that they could respond verbally to peers. Morrison et al. (2001) stated that the acquired skills might not transfer to students who were not as verbally capable and that further studies of students with less functional language skills were needed. Brignell et al. (2018) also state that it is wrong to assume that interventions that have been effective for children with ASD will work for those children who are also nonverbal or minimally verbal.
None of the three previous studies involved technology for self-monitoring, highlighting a gap in the research for the effectiveness of technology-based self-monitoring of initiations for non-speaking or minimally speaking children on the autism spectrum who participate in an alternate assessment. Interest has increased in the use of technology for self-monitoring interventions for individuals with ASD (Chia et al., 2018). However, there are no studies in the literature on the use of technology for the self-monitoring of initiations by children on the autism spectrum who are also non-speaking or minimally speaking and take the state’s alternate assessment. I-Connect is a technology-based application designed by researchers at the University of Kansas for self-monitoring (Wills & Mason, 2014) and has potential for self-monitoring of communication initiations by this population of students. I-Connect for self-monitoring by students on the autism spectrum has been effective for on-task behavior and academics (Beckman et al., 2019; Clemons et al., 2016; Huffman et al., 2019; Romans et al., 2020; Rosenbloom et al., 2016), reducing stereotypical behavior (Crutchfield et al., 2015), and inappropriate vocalizations (Wills et al., 2019). One study used I-Connect to self-monitor question-asking initiations by young adults with ASD whose primary communication was verbal speech (Bross et al., 2022). The adults from Bross et al. (2022) were at Level 1, requiring support, or Level 2, requiring substantial support, as defined in DSM-V (American Psychiatric Association, 2013). When using I-Connect for self-monitoring, Bross et al. (2022) found that all three adults increased the number of questions they asked.
The present study sought to investigate the effectiveness of I-Connect for self-monitoring of initiations by participants on the autism spectrum who are also non-speaking or minimally speaking and take an alternate assessment. The participants in this present study are defined as on the autism spectrum with limited language skills. Newman and Ten Eyck (2005) served as a reference study because the subjects in that study most closely aligned with the criteria of nonverbal or minimally verbal by Koegel et al. (2020) and the results demonstrated success with self-management of initiations.
Newman and Ten Eyck (2005) implemented a multiple baseline design with non-contingent reinforcement, external reinforcement, and self-reinforcement phases. Each phase consisted of a minimum of six sessions, each with ten 90-second intervals. In each interval two interventionists interacted with the student to motivate, prompt, and reinforce initiations. A motivating condition was created by one interventionist who played with a toy in an opaque bag. The intent was to stimulate the student to make a play request (initiation) while the second interventionist prompted the student to initiate, if required. The student received a token for each successful initiation, moving towards self-managing their tokens and reinforcement. These tokens were exchangeable for a reinforcer of their choice. Data were collected on the number of intervals with initiations and the number of intervals when tokens were earned and taken. The study took place in a private room at a school. The results showed that all students increased their ability to initiate communication when they used self-management.
The present study modelled the procedures of the reference study with additions of: (a) I-Connect for self-monitoring, (b) a generalization phase, (c) the use of the Verbal Behavior Milestones Assessment and Placement Program (VB-MAPP) (Sundberg, 2014) to measure pre- and post-study verbal skills, (d) post-study social validity surveys, and (e) a classroom setting instead of a private room in a school.
The following research questions guided this study: (a) What is the effect of self-monitoring using I-Connect on the initiations of the participants? (b) To what degree do the participants accurately self-record their communication initiations on I-Connect? (c) To what degree do the participants who self-monitor their initiations using I-Connect in a classroom generalize the practice in a school setting? and (d) What is the social validity of any change in communication initiations of participants? The significance of this present study is that it addresses the gap in I-Connect for self-monitoring of initiations by elementary school children on the autism spectrum with limited language skills.
Method
Participants
The selection criteria for participants in the present study included elementary school students who: (a) were between the ages of 8–11 on the autism spectrum with limited language skills, (b) averaged ten or fewer communication initiations per day, (c) had already learned to initiate either verbally, with a picture icon, or with an augmentative and alternative speech-generating (AAC) device but were not consistently using the skill, and (d) completed a VB-MAPP test for verbal skills. In this present study an AAC device was an Apple iPad tablet with speech generating software that presented a series of images depicting customized wants and needs that when selected produced a voice output (Lorah et al., 2022).
Three students were initially selected as participants in this present study. However, one of the participants immediately increased their initiations in baseline prior to the commencement of the self-monitoring intervention. This participant was removed from the study and another participant joined the study to allow for three potential demonstrations of effect (Ledford et al., 2023). The impact of this action on internal validity of the present study is discussed in the Discussion section below.
Each participant had a vocabulary with an average of fewer than ten words from which they would use 1–3 words to request their basic needs such as bathroom, water, backpack, and highly preferred reinforcers. All participants were functioning at Level 3, requiring very substantial support, on the DSM-V (American Psychiatric Association, 2013), as determined by their parents and their IEPs.
The three participants were Ronnie, Vicky, and Andy. Ronnie was a 9-year-old Hispanic male. He communicated verbally with 1-3-word utterances, pointing, gesturing, or using his AAC device. Vicky was a 10-year-old White female who communicated by verbally requesting with 1-3-word utterances or by using her AAC device. Andy was a 9-year-old Hispanic male who communicated with limited vocalizations, gestures, picture icons, and speech but only initiated communication independently for highly preferred reinforcers. All the participants had strengths in receptive language skills. The Institutional Review Board approved this present study. The parents of the participants gave informed consent. The observer for the present study obtained assent from the participants using a visual consent form.
Setting and Interventionists
The present study was undertaken at an urban elementary school in the Southwestern United States. The setting was a self-contained classroom and general education art, library, and music classrooms on the school campus. There was a researcher and an observer. The researcher was the first author of this article with eight years of experience as a special education elementary school teacher for students on the autism spectrum.
The observer, a doctoral student in special education, collected interobserver and procedural fidelity data. The observer received information and an explanation of the self-monitoring procedures for I-Connect and data collection. Both the researcher and the observer were familiar with AAC devices.
Instrumentation
Instrumentation included the VB-MAPP assessment tool, I-Connect application, social validity survey form, online data collection forms, and interobserver agreement. These are described in more detail below. A web-based platform provided services for data collection, storage, and analytics.
VB-MAPP
VB-MAPP is an assessment tool with good to moderate reliability (Montallana et al., 2019) for measuring the language skills of children on the autism spectrum or other disabilities (Sundberg, 2014). The researcher administered the mand domain skill test for levels 1, 2, and 3 of the VB-MAPP Milestone Assessment in accordance with the VB-MAPP Guide (Sundberg, 2014) as a pre- and post-study test for each participant. Twenty items were used for each participant under Level 2 of the mand domain testing. Comparing the VB-MAPP pre- and post-study test results is a potential measure of changes in initiations for each participant.
I-Connect Portal
The present study utilized the I-Connect Portal (University of Kansas, n.d.) for administrative purposes. The researcher accessed the I-Connect Portal to create user accounts for the participants, establish the self-monitoring parameters for I-Connect, and monitor progress. Additional information related to the creation of user accounts and downloading I-Connect may be obtained from the University of Kansas website (University of Kansas, n.d.).
Social Validity Survey
An online social validity survey, completed by the parents and the teacher’s aide at the end of the study, contained questions about the use of I-Connect for the target behavior of initiations, the importance of initiations, and observed changes in initiations at school and home. The survey questions were consistent with the recommendations from Cooper et al. (2019), including the suitability of the intervention for the target behavior, the social significance of the intervention goals, and the social importance of the outcomes. An online visual survey form recorded participant satisfaction with I-Connect. This form comprised two questions the participants could read, or the researcher could read for them: (a) I liked using I-Connect, and (b) I want to use I-Connect again. The participants answered the questions with a “Yes” or “No” response aided by visual icons.
Online Data Collection Forms
The researcher and observer used a secure online data collection form to record initiations, I-connect responses, and the number of times the participants were prompted in each of the ten 90-second intervals during 15-minute sessions in each phase of this present study. The researcher also used a separate online form to collect pre-baseline data.
Interobserver Agreement
The interobserver agreement (IOA) calculation used the interval-by-interval formula from Cooper et al. (2019). The observer recorded IOA data for all participants for 33% of the sessions in each of the baseline, I-Connect training, and self-monitoring intervention phases.
Materials
The researcher downloaded the I-Connect application from the University of Kansas (University of Kansas, n.d.) to a Samsung A7 tablet used by the participants. I-Connect provided prompts, interval timing, goal setting to help establish self-monitoring behaviors, and progress reports for self-monitored activities. Each participant used a Samsung Galaxy A7 tablet for self-monitoring during individual I-Connect sessions.
A video camera recorded each session of the baseline, I-Connect training, and self-monitoring intervention phases to validate observations and the fidelity of each session. Picture icons, including “Can I play?” and a variety of other icons, were available for use by the participants to initiate communication. The AAC devices contained similar picture icons for initiations. A variety of toys were available for reinforcement.
Research Design
This present study initially intended to use a single-subject multiple baseline design across participants to provide internal and external validity to hypothesize a function relationship between the independent and dependent variables (Ledford & Gast, 2024). However, it became a single-subject non-concurrent multiple baseline design across participants with the withdrawal and replacement of one participant early in the present study. This is addressed in more detail in the Discussion section.
Experimental control required selecting participants with functionally independent behaviors but with sufficient functional similarity for replication of effect (Ledford & Gast, 2024). Functionally independent behaviors meant that behaviors were not highly correlated or the linear relationship between the behaviors was low (Ledford & Gast, 2024). A change in the behavior of one participant did not linearly translate into a change in the behavior of other participants, as explained by Ledford and Gast (2024). Functional similarity was simply the inability to initiate social communication or conversations consistently. The multiple baseline design in this present study included the four phases of baseline, I-Connect training, self-monitoring intervention, and generalization. The staggering of the self-monitoring intervention to each participant provided internal validation of functional independence, and the impact on the dependent variable for each participant confirmed the functional similarity.
Dependent Variable
The dependent variable was the number of unprompted initiations made by a participant during 15-minute observation periods, as observed by the researcher. An initiation in this study was a mand, operationally defined as a participant making a request either verbally, with an AAC device, or by placing a picture icon into the researcher’s hand. The data were categorical and limited to correct, incorrect, or no response. The observer recorded IOA data for correct response, incorrect response, or no response.
Independent Variable
The independent variable was the self-monitoring of initiations using I-Connect. I-Connect was installed on a Samsung Galaxy A7 tablet which was used by each participant for self-monitoring during individual I-Connect sessions.
Procedures
The self-monitoring intervention comprised six phases including: (a) a study initialization phase, (b) a pre-baseline phase for observations and reinforcer assessment, (c) a baseline phase during which the participants received reinforcement regardless of whether they reached a target number of initiations, (d) an I-Connect training with externally provided reinforcement phase, (e) a self-monitoring intervention with self-reinforcement phase, and (f) a generalization phase. This present study used self-reinforcement because it was a component of self-management used in the reference study. All phases had individualized reinforcers for each participant. The researcher provided external reinforcement during the I-Connect training phase. The self-monitoring intervention phase involved self-reinforcement by the participant for having reached their target number of ten initiations. All participants attended research sessions during scheduled classroom days over ten consecutive weeks.
Study Initialization
A reinforcer assessment identified toys and other items for each participant. Interviews with parents first identified general items for the assessment using the Reinforcer Assessment for Individuals with Severe Disabilities (RAISD) protocol described by Cooper et al. (2019). The researcher then conducted a contrived free operant observation involving observing and recording items the participant chose during a period of unrestricted access (Cooper et al., 2019). This process confirmed participants’ preferred items, including toy cars, spikey balls, slime, plush dolls, or bubbles. The researcher chose a selection of preferred items for use during all phases of the present study, and the remainder comprised a visual menu for self-reinforcement.
Pre-baseline Observations
The researcher made pre-baseline observations of the participants over three consecutive days during regular classroom activities and recorded their initiations using the online data collection form. The researcher conducted VB-MAPP testing of the participants.
Baseline
All participants began in the baseline phase. The baseline consisted of six or more 15-minute sessions each divided into 90-second intervals. These sessions involved the researcher, observer, and participant seated at a small table. The participant sat near the middle of the table. The researcher was opposite the participant, and the observer, when available, sat to the side. The researcher held a dark bag containing a selection of preferred reinforcers for the participant. The preferred reinforcers were different for each participant. The remaining reinforcers were out of sight in a separate bag. The participants did not initially know that the bag contained items of interest to them.
The researcher began by playing with an item in the dark bag to establish a motivating condition for the participant to request to play. If the participant did not attempt to discover the contents of the bag within one minute, the researcher modeled the correct response through verbal prompts such as “Can I play?”, guiding them to the correct response on their AAC device, or pointing to the picture icon “Can I play?“. Other forms of response guidance included positional prompts, such as placing the AAC device closer to the participant, or gestural prompts.
If a participant made an initiation to the researcher at any time during the session, whether prompted or unprompted, an item from the bag was handed to them to play with for 30 s. Immediately after the participant initiated, the researcher would say “Nice asking” while handing over the item from the bag. After 30 s of playtime, the researcher asked for the item back by saying “My turn.” The researcher and the observer collected baseline data on each initiation made by the participant using their preferred communication method. At the end of the session the participants could choose from the reinforcer menu regardless of whether they had made an initiation or not. When baseline data for each participant showed a consistent level of initiations with low variability, I-Connect training commenced with the first participant while the others stayed in baseline (Ledford & Gast, 2024).
I-Connect Training
The I-Connect training phase began upon completion of the baseline phase. For the I-Connect setup, the location was the self-contained classroom, the interval was every 90 s of a 15-minute session, and the goal was to record ten “Yes” responses to the “Did you ask?” prompt on I-Connect. Each participant learned to discriminate between an initiation attempt and a non-initiation attempt according to the procedures in Koegel et al. (1990). The participants made initiation attempts either verbally, using their AAC device, or by placing a picture icon into the researcher’s hand. The participants recorded their initiations on I-Connect by pressing “Yes” or “No” once for each interval. For example, at the 60-second mark of each 90-second interval, I-Connect sounded a chime, and the question “Did you ask?” appeared on the screen along with the options “Yes” with a green check mark or “No” with a red “X” mark. The participant pressed either “Yes” or “No” to indicate whether they initiated a request for an item. Table 1 describes the step-by-step procedure for training the participants in the use of I-Connect.
Table 1
Step-by-Step I-Connect Training Procedure
Step | Description |
|---|---|
1 | The researcher began by introducing I-Connect to the participants and told them that the goal was to make ten initiations and to record each of these on I-Connect when they heard the chime by pressing “Yes” if they asked to play or “No” if they did not. |
2 | The researcher told the participants that when I-Connect displayed the goal of “10” they could choose an item from the reinforcer menu. |
3 | The researcher then modelled the process on I-Connect by role-playing, asking “Can I play?” and then pressing “Yes” after the sound of the I-Connect chime. To model a no response, after not asking, the researcher pressed the “No” button after the chime and said “I did not ask.” |
4 | The researcher began by playing with an item in the dark bag. If the participant did not ask to play or ask what was in the bag within 60 s, the researcher prompted them to do so. |
5 | To teach the participant to self-record initiations, the researcher mirrored the question displayed by I-Connect by saying “Did you ask?” followed by “Yes, you asked,” and then prompting the participant to press the “Yes” button. |
6 | The participants received praise when they initiated and independently recorded a “Yes” response on I-Connect. If the participant did not record a “Yes” after making an initiation, they received a prompt from the researcher to press the “Yes” button using least-to-most prompts (Cooper et al., 2019; Libby et al., 2008). |
7 | When the participant reached the goal of 10 initiations, the researcher said “You made it to 10,” while pointing to the number 10 on I-Connect. The researcher then gave the participant an item from the reinforcer menu. |
8 | The process repeated, as per Koegel et al. (1990), until the participant independently recorded “Yes” or “No” on the device, depending upon whether they initiated using their preferred mode of communication. |
The researcher and the observer (when in attendance) each collected data on the participants’ recording of “Yes” or “No” on I-Connect, or a “No Response.” Prompts may have been required for the participants to record whether they made an initiation on I-Connect. Once the participants reached a stable level with low variability (Ledford et al., 2017) they began the self-monitoring phase.
Self-monitoring Intervention
The objective of the self-monitoring intervention phase was to guide the participants into self-monitoring and self-recording their initiations. This phase was a continuation of the I-Connect training phase, with the researcher fading the prompts over six or more sessions until the participant was initiating and then self-monitoring with I-Connect. Fading began in the first session of the self-monitoring intervention phase by reducing the frequency of verbal or gestural prompts.
The setting and placement of participants were the same as for the I-Connect training phase. The researcher began playing with an item in the dark bag. When required, the researcher verbally prompted the participant to initiate a “Can I play?” request to the researcher either verbally, using an AAC device, or an appropriate picture icon. The researcher praised the participant by saying “Nice asking.” The researcher then said “Did you ask?” as a cue for the participant to press “Yes” or “No” on I-Connect (rather than telling them to do so). The researcher then handed the participant the item from the bag, allowing them 30 s to play with it. The researcher gradually faded the prompts until the participant independently self-monitored their initiations by indicating “Yes” or “No” on I-Connect. I-Connect accumulated running totals of “Yes,” “No,” and “No Response” from the participants. When the session ended, I-Connect displayed the number of “Yes” responses, allowing participants to gauge when they had earned reinforcement. In this event, rather than telling the participant “You reached 10” and then directing them to choose an item from the reinforcer menu, the researcher asked “Did you reach 10?” to enable the participant to decide if they earned a reinforcer reward. The participant could respond by showing “10” displayed on I-Connect to the researcher to indicate they were aware that they had achieved their goal. The participant could then self-reinforce by choosing an item from the reinforcer menu.
To establish interobserver agreement during each session, the researcher and the observer collected data on the participants’ self-monitoring attempts for: (a) the frequency and accuracy of initiation attempts, (b) the number of prompts given to the participant by the researcher, (c) the correct responses, (d) the incorrect responses, and (e) no responses. The I-Connect Portal automatically recorded I-Connect responses. Upon completion of the in-classroom phases of the present study, the remainder of the study took place in the general education setting.
Generalization
The generalization phase for this present study occurred in the general education settings of the school during art, library, and music classes. There were two planned sessions at each location for up to six sessions. As in the baseline, I-Connect training, and self-monitoring intervention phases, the researcher recorded data for each 90-second interval during 15-minute sessions. The locations for the generalization sessions rotated so that each day had one 15-minute generalization session in a different location. The participants were not re-trained in I-Connect procedures. The participant sat at a table with I-Connect positioned in front of them. The researcher had a dark bag with items of interest to the participant. Location specific items that the participant enjoys were added to the bag. For example, in the library, favorite books were added to the bag. In the art room, glitter sticks or crayon rocks in the participants’ favorite colors were rotated into the bag. The participants self-monitored their initiations with I-Connect.
Post-Study Surveys and Tests
After the generalization phase, the parents and the teacher’s aide completed the social validity survey. The participants took the post-VB-MAPP test and completed the visual social validity survey.
Results
Figure 1 shows the number of unprompted communication initiations for each phase of the present study. All participants had an immediate demonstration of effect from the baseline to the self-monitoring intervention phase, illustrating a functional relationship between the independent variable of self-monitoring and the dependent variable of initiations (Ledford & Gast, 2024). Descriptive statistics of level, variability, percentage of non-overlapping data (PND), and trend for each phase and each participant assisted in the visual analysis (Ledford et al., 2017). Visual analysis of the results of the initiations made by participants in Fig. 1 shows that in the baseline phase, the percentage of intervals with an initiation was 30% or less for each participant. Although there are points of overlap in the data, each participant demonstrated an immediate effect with the introduction of self-monitoring in the I-Connect training phase, with increases in the percentage of intervals with an initiation. In the self-monitoring intervention phase, each participant reached a consistent level of initiations relative to the baseline with 0% overlapping data. Table 2 shows the Tau-U effect sizes for the participants and confirms three potential demonstrations of effect. Tau-U statistics take baseline trend and non-overlap between phases into account (Parker et al., 2011). The Tau-U results include the overall weighted effect size average supporting three potential demonstrations of effect.
Fig. 1
Number of Unprompted Communication Initiations
×
![]()
Table 2
Tau-U Effect Sizes for Participants
Includes I-Connect Training | Excludes I-Connect Training | |||
|---|---|---|---|---|
Participant | Tau-U | P-Value | Tau-U | P-Value |
Ronnie | 0.97 | 0.001 | 1 | 0.0039 |
Vicky | 0.98 | 0.00 | 1 | 0.0006 |
Andy* | 1.00 | 0.0001 | 0.8125 | 0.0118 |
Weighted Average | 0.98 | 0.00 | 0.9382 | 0.0000 |
Ronnie
In baseline, Ronnie began with a low level of initiations, moderate variability, and a slightly decreasing trend. Ronnie maintained initiation consistency over three data points before the introduction of I-Connect training. There was an immediate effect in I-Connect training as seen by the increasing trend in his initiations. However, there was significant variability and inconsistency of initiations during this phase. The percentage of nonoverlapping data between baseline and I-Connect training phase was 83%. There was another immediate effect from I-Connect training to the self-monitoring intervention phase, and he then maintained a consistent level of ten initiations over five data points. There were no overlapping data between baseline and the self-monitoring intervention phase. After Ronnie started the generalization phase his level of initiations initially decreased (he was absent for one session) but subsequently increased to a high level with some variability. His I-Connect “Yes” response accuracy improved from 95% in the self-monitoring intervention phase to 99% in the generalization phase. There were no overlapping data between baseline and the generalization phase. Referring to Table 2, a Tau-U = 1.00 and p-value = 0.0039 indicates a large effect size from baseline to the self-monitoring intervention phase and confirms the visual observations of effect.
Vicky
In baseline, Vicky began with a moderate level of initiations, high variability, and a slightly decreasing trend. Vicky maintained a consistent level of initiations over four data points before the introduction of I-Connect training. There was an immediate effect in I-Connect training with an increasing trend in her initiations. The percentage of non-overlapping data between baseline and I-Connect training was 67%. During the self-monitoring intervention phase Vicky maintained a high level of initiations with moderate variability and a slightly increasing trend. There were no overlapping data between baseline and the self-monitoring intervention phase. The trend began to decrease during the generalization phase and Vicky’s I-Connect “Yes” response accuracy declined from 95% in the self-monitoring intervention phase to 65% in the generalization phase. The percentage of nonoverlapping data between baseline and generalization was 86%. Referring to Table 2, a Tau-U = 1.00 and p-value = 0.0006 indicates a large effect size from baseline to the self-monitoring intervention phase and confirms the visual observations of effect.
Andy
In baseline, Andy began with a low level of initiations, low variability, and a slightly increasing trend. Andy began I-Connect training with a moderate level of initiations with high variability. This improved during I-Connect training to a high level of initiations with low variability. Andy maintained this level of initiations during the self-monitoring intervention phase. After an initial drop when generalization began his level of initiations improved, but with high variability. Andy’s I-Connect “Yes” response accuracy declined from 53% in the self-monitoring phase to 45% in the generalization phase. There was no overlapping data between baseline and any of the phases. Referring to Table 2, his Tau-U, after baseline correction, was 0.8125 with a p-value = 0.0118 indicating a strong effect size from baseline to the self-monitoring intervention phase confirming the visual observations of effect.
Social Validity Survey
The parents and the teacher’s aide took the social validity survey after the study to determine their opinion of the self-monitoring intervention. 62.5% of the respondents either strongly agreed or somewhat agreed that self-monitoring with I-Connect helped their child/student to improve their initiation attempts, 37.5% were neutral. 25% of the respondents strongly agreed that self-monitoring with the I-Connect application is acceptable for communication needs, 25% somewhat agreed, and 50% were neutral. 87.5% of the respondents strongly agreed that initiations are important for overall communication for their child/student, while 12.5% were neutral.
The participants took the visual survey to determine their level of satisfaction with the self-monitoring intervention and their interest in the future use of I-Connect. All participants indicated that they liked I-Connect and would use it again.
VB-MAPP Mand Domain Testing
The participants’ VB-MAPP pre- and post-study tests for levels 1, 2, and 3 of the mand domain indicate that Ronnie, Vicky, and Andy improved their test scores by 8%, 6%, and 3%, respectively.
Interobserver Agreement
IOA observations were a combination of in-person attendance by the observer at participant sessions and from viewing videos of sessions. The average IOA across all participants was 90%.
Procedural Fidelity
Procedural fidelity was assessed by the observer and researcher for 33% of sessions in each of the baseline, I-Connect training, and self-monitoring intervention phases. A 10-point checklist focused on the performance and participant use of I-Connect, consistency of procedures during the 90-second intervals, and recording of data.
For the sessions during which procedural data was collected, the procedural fidelity was 89.5% with a range of (85–94%). The IOA between the researcher and observer for procedural fidelity was 90.4%. IOA for procedural fidelity was determined using the recommendations of Essig et al. (2023) as guidelines.
Discussion
During pre-baseline observations, all participants initiated, on average, less than ten times per day. The study commenced with Ronnie, Vicky, and a third participant who was replaced by Andy when Ronnie was in the I-Connect training phase. This may have compromised internal validity. However, Ledford et al. (2023) state the need for flexibility in single-case design, especially in special education. Ledford et al. (2023) also state that three potential demonstrations of effect, rather than three demonstrations of effect, are adequate in special education interventions. Furthermore, Slocum et al. (2022) state that maturation effects on internal validity reduce if each subject is “exposed to different amounts of time” in baseline with sufficient lag prior to the introduction of the intervention and that the baseline exposures do not need to synchronize strictly. They also state that if the immediate within-subject demonstration of effect from baseline is maintained during the intervention, concurrence is “not necessary to detect and control for maturation.” Kennedy (2022) also states that maturation confounds typically evolve over time and, provided the study is short, the impact on internal validity may be minimal. Therefore, the late addition of Andy should not have materially impacted on the internal validity of the multiple baseline single case design. Further, the between-subjects effects on internal validity would be minimal because the conduct of the study was not visible to others, and between-subject discussions did not take place because of the limited communication and social skills of the participants. The study became a non-concurrent multiple baseline across participants design.
All participants increased their initiations when using I-Connect for self-monitoring, and visual analysis of the data showed an immediate effect when moving from baseline to I-Connect training. All participants maintained high levels of initiations during the self-monitoring intervention phase. The similarity of outcomes between the present study and the reference study using alternate means for self-monitoring is consistent with the findings in Bouck et al. (2014) and Crutchfield et al. (2015) that self-monitoring is effective whether using pencil and paper or technology. Each participant learned to watch for the I-Connect prompt “Did you ask?” and listen for the chime to remind them to self-record, demonstrating strengths in responding to auditory cues. Although there was variability in the accuracy of I-Connect “Yes” responses among the participants, the rate of I-Connect responding was consistently high across all phases of the study for all participants. The variability of accurate I-Connect responding is consistent with Newman and Ten Eyck (2005) who found that accuracy did not correlate with the number of initiations made. However, the accuracy of recording is indicative of the degree of participant awareness during I-Connect interactions. The steady rate of I-Connect responding may indicate that I-Connect was a good choice for the participants, aligning with the observation by Valencia et al. (2019) that technology is appealing to individuals on the autism spectrum. An explanation for the high rate of responding could be a behavior chain (Cooper et al., 2019), as observed by Romans et al. (2020), or by the self-awareness produced by self-monitoring (Bellini, 2016; Ganz et al., 2013; Kim et al., 2019; Schulze, 2016).
All participants generalized their use of I-Connect for self-monitoring of initiations in multiple school settings, such as in the library, art room, and at recess. When the time came for the participants to begin using I-Connect, they quickly went to the designated area with an apparent motivation and willingness to use the device. Throughout the study, Ronnie and Vicky focused on the I-Connect device on the table before them rather than looking around the room. Andy remained seated rather than trying to elope to other parts of the classroom. The participants’ positive reactions to I-Connect are consistent with previous research (Crutchfield et al., 2015; Romans et al., 2020; Rosenbloom et al., 2016).
Initiations decreased in the generalization phase for all participants, as did responses to the I-Connect prompt. Ronnie recovered from an initial drop and increased his initiations to 100% of intervals and developed consistency in I-Connect responding. Vicky and Andy’s performances were more variable and decreased compared to the self-monitoring intervention phase, but their levels of initiations remained higher than in baseline.
When submitting the social validity survey the parents and teacher’s aide provided some insights into how they scored the survey. For instance, at home, Ronnie’s parents noticed he used more expansive spontaneous language, vocalized requests, and acknowledged family members more often. The teacher’s aide observed Vicky using her AAC device more frequently and spontaneously at school. Andy’s mother reported that he began independently washing dishes, which he had never done without prompting, although he did not consistently perform this task. Ronnie demonstrated improvements in fine motor skills in classroom writing assignments during the self-monitoring intervention, as reported by the teacher’s aide and his parents. The occupational therapist also observed improvements in Ronnie’s fine motor skills. These observations align with the findings of Koegel et al. (2001) that initiations and self-management are pivotal areas, and improvements in these core areas can positively affect various aspects of functioning.
Limitations
The present study has limited scope due to the small sample size and the variability of the communicative levels of the participants. The population of non-speaking and minimally speaking children is known to be difficult to assess (Kasari et al., 2013). Another limitation of the present study was the contrived nature of the 15-minute sessions and 90-s intervals. Although the participants independently self-recorded on I-Connect and self-reinforced, the presence of an adult may have influenced their performance. Further, the interval of 90 s was too long for Vicky, who quickly lost interest in the reinforcers. Intervals could have been varied to prevent anticipation of a prompt.
The researcher, who was also the classroom teacher, selected the participants, which may reflect researcher bias on whether the students met the present study criteria. The generalizability of the results did not include settings outside of the school.
Implications for Practitioners
Research has shown that self-monitoring of initiations using I-Connect during 15 min of instructional time in a classroom setting can increase communication initiations for students on the autism spectrum with limited language skills, including those who use an AAC device.
The I-Connect application is available to students, parents, and teachers to reinforce communication and other skills in multiple settings. Practitioners can integrate I-Connect into the daily routines of their students.
Contributions To the Literature
All participants increased their overall initiations from baseline to the self-monitoring intervention phase and maintained a higher level of initiations over baseline during the generalization phase. Further, even though the accuracy of I-Connect responding varied across participants, there was encouraging acceptance of I-Connect by all participants.
This present study adds to the literature on technology-based interventions by showing that I-Connect may be effective for the population of elementary school students on the autism spectrum with limited language skills, including those who use an AAC device. Further, this present study was conducted primarily in the familiar classroom setting of the participants and in other general education locations on the school campus.
Future Research
Future studies could use individualized forms of self-monitoring of initiations with I-Connect based on the results of an assessment such as VB-MAPP. Future research could examine the effect of using I-Connect for multiple targets, expanded language functions, or the year-to-year effect on the participant’s performance on the alternate assessment.
Research in natural environments for participants may include new and evolving technologies including advanced AAC devices and artificial intelligence.
Conclusions
There is a lack of current research in technology-based self-monitoring of initiations for elementary school students on the autism spectrum with limited language skills. This present study showed four areas of improvements for the participants: (a) initiations increased for all participants using self-monitoring with I-Connect, (b) the accuracy of self-recording communication initiations on I-Connect improved for all participants as the present study progressed from I-Connect training to the self-monitoring intervention, (c) all participants generalized self-monitoring of their initiations using I-Connect to other school settings, and (d) the participants, parents of the participants, and the teacher’s aide all reported satisfaction with the social validity of communication initiations for participants, and were generally supportive of I-Connect for increasing initiations. Observations of the participants at school and home indicated increases in daily functioning skills, as evidenced by expanded spontaneous language, independent completion of tasks, increases in the efficiency of fine motor skills, increases in AAC use, and improvements in areas not related to the self-monitoring intervention.
I-Connect proved to be an effective means for the participants to self-monitor initiations. Two participants who used an AAC device were able to use I-Connect concurrently. The present study concluded that I-Connect was a valid alternative to the token system used in the reference study, with comparable results.
Declarations
Ethics Approval
The International Review Board granted ethics approval for this study, which was performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards.
Consent and Assent
Consent was obtained from the parents of the participants in this study. Assent was obtained from the participants.
Competing Interests
The authors have no competing interests.
Inclusive Language
This manuscript generally complies with the JADD Inclusive Language Guide except when citing research by other authors who may have used different language.
Disclosure
The authors do not have anything else to disclose.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.