Participants

Gavin and Phoebe were recruited from a non-public school serving individuals with ASD. For inclusion, participants had to cooperate with 10 min of instruction, copy text with legible handwriting, tact numbers 0–20, and spell or select 10 names of cities when dictated. Gavin was an 18-year-old Caucasian man diagnosed with ASD. He attended a self-contained classroom with students with similar educational and behavioral presentations. Gavin scored in the extremely low range (full-scale IQ, 67, 1st percentile) on the Wechsler Adult Intelligence Scale Fourth Edition (WAIS-IV; Wechsler, 2008). He could smoothly engage in conversations on a variety of topics. Phoebe was a 17-year-old Caucasian female diagnosed with ASD who attended the same classroom. She scored in the extremely low range (full-scale IQ, 54, below 1st percentile) on the WAIS-IV. Phoebe communicated her wants and needs vocally and had basic conversational exchanges on familiar topics. Phoebe engaged in vocal and motor stereotypy that was easily redirected.

Their caregivers provided informed consent, and participants provided initial assent to participate through a review of a simplified version of the study information, a discussion with the researcher, and an opportunity to review the information with their caregivers. Participants were also offered ongoing opportunities to assent to the procedures at the start of each session (e.g., “Do you want to work with Ms. Sarah today?”).

Settings and Materials

Sessions were conducted in the participants’ school. Participants used an iPad © Mini 4 with the standard Apple © apps of Clock, Weather, and Maps on the home page. We developed customizable probe worksheets with 15 questions for the study (Supplemental Materials A). Each worksheet had spaces for four questions pertaining to time (e.g., “What time is it in ____?”), temperature (e.g., “What temperature is it in ____?”), and distance (e.g., “What’s the distance to_______?” Before each probe session, the researcher used a randomizer to select three cities from a prepopulated list and wrote them in the blank spaces (e.g., “What time is it in Paris?” “What’s the temperature in Boston?” “What’s the distance to Omaha?”). These questions served as EO trials, since they could not be answered by the participants. Each worksheet also included three questions participants could correctly and independently answer without app use to serve as abolishing operation (AO) trials. For Gavin, these consisted of questions comparing relative units of measurement (e.g., “Which distance is farther ____ or ____ miles?). For Phoebe, these consisted of personal information questions. Training worksheets included only nine questions pertaining to one problem type (Supplemental Materials B) and no AO questions.

Response Definitions

The primary dependent variable was the percentage of correct written responses to EO questions about time, temperature, and distance during probe sessions. Units of measurement (e.g., miles, am/pm, degrees) were not required, nor was the use of commas to separate digits (e.g., 5234). The percentage of correct written responses to AO questions was also evaluated.

Response generalization and transfer of stimulus control from textual to vocal stimuli was evaluated in a Vocal Probe condition. In this condition, the percentage of correct, vocal responses to EO and AO questions was measured. Units of measurement and correct number places were not required (e.g., saying “one-zero-one" for 101 degrees).

Reliability and Procedural Fidelity

Sessions were recorded, and a randomizer was used to select sessions within each phase for coding by trained observers. Inter-observer agreement (IOA) was coded using item-by-item agreement for the written responses and trial-by-trial agreement for the vocal responses. For Gavin, IOA was collected on greater than 45% of sessions (46% probe; 57% training) with high levels of agreement (100%; M = 98%, range, 89–100%). For Phoebe, IOA was also collected on greater than 35% of sessions (46% probe; 38% training) with high levels of agreement (M = 96%, range, 83–100%; M = 98%, range, 89–100%). The same sessions were evaluated for procedural fidelity with checklists detailing the steps for each condition. A high percentage of steps were followed with fidelity during probe (Gavin 100%; Phoebe M = 99%, range, 93–100%) and training (Gavin M = 99%, range, 89–100%; Phoebe M = 98%, range, 67–100%) sessions.

Experimental Design and Progression

Each participant progressed through the study following the logic of a multiple baseline design across behaviors (question types). The introduction of training was staggered across participants in a nonconcurrent multiple baseline design. Following the initial baseline, training components hypothesized to serve as part of the precurrent response chain (see Fig. 1) were sequentially introduced in paired steps (e.g., 1a and 1b; then 2a and 2b). Components were paired to sensitively evaluate changes in the dependent variable, while avoiding overly frequent testing. Following mastery of paired training steps, a minimum of two post-training probes (PTPs) were conducted. Mastery criteria for PTPs were 100% correct responding on EO trials across two consecutive sessions or 75–100% across three consecutive sessions.

Pre-Assessment and Pre-training

Consistent with Skinner’s (1957, 1984) conceptualization, we ensured participants could engage in the solution responses in the form of copying a digital stimulus onto the worksheet and tacting the digital stimulus for the Vocal Probe. We also evaluated skills needed as part of the response chain, including independent completion of worksheets with only AO questions, selection of target apps by name (e.g., “Find [app]”), and transcribing city names (e.g., “Spell [city]”). If skills were below the mastery criterion (100% correct responding across two sessions), we taught them using progressive time delay (0 s, 3 s, 6 s) with praise for prompted and independent responses.

Probes: Baseline and Post-training

The researcher presented a probe worksheet (Supplemental Materials A) with three AO questions and 12 EO questions related to time, temperature, and distance for four randomly selected cities from a cultivated list; cities never repeated. The researcher stated, “Time to do your work. You can use this if you want,” while pointing to the tablet. The participant had 15 min to complete the worksheet, but if responses ceased for 1 min, the probe ended. The researcher reviewed the worksheet with the participant and provided praise for each correct response and feedback for incorrect responses (i.e., “Good try, but that’s not right”). The participant was allowed a brief break (2–5 min) with preferred items after completing each worksheet.

Vocal Probe

With the tablet on the table, the researcher said, “It’s time for a quiz,” and read each question on the worksheet (out of the participant’s view). The participant had up to 1 min to respond to each question. The experimenter provided praise and feedback statements following each response, as described above.

Rule Prompts (Phoebe Only)

To occasion stating then following the intraverbal rules for app use during PTPs, we added a prompt (Sautter et al., 2011). When Phoebe reached a target question on the worksheet, the researcher said, “Say your rule about [time, temperature, distance]”.

Training

Sessions consisted of nine trials related to one question type. During the first session of each training step, the researcher provided the most invasive prompt at a 0-s delay, combined with a less invasive prompt. The types of prompts differed according to the skills taught (see below). Following 100% correct, prompted responding in one session, the researcher progressively delayed the less invasive prompt from 0-s to 3-s, to 6-s. The researcher provided the more invasive prompt to correct errors. After two consecutive sessions with 100% correct, independent responding, the training step was mastered. Praise was intermittently provided following responses, and brief breaks with preferred items followed each training session.

Training Step 1a: Using Target App

The researcher presented a city name on a stimulus card and said, “Use the [clock, weather, map] app.” Combined vocal (i.e., “Search it…type it…select it”) and gestural prompts were provided to occasion the responses of opening the app, selecting the search button, typing the city, and selecting the city from the search tab. This produced the “answer” screen showing the time, temperature, or distance.

Training Step 1b: Emitting Intraverbal Rule

The researcher stated, “Say your rule about [time, temperature, distance].” Combined echoic and textual prompts were provided to occasion the target intraverbal rule (e.g., “If the question is about time, use the clock app”).

Training Step 2a: Underlining Keyword

With the worksheet present, the researcher asked, “What is the question about?” and used physical guidance combined with a gestural prompt to occasion underlining the keyword (e.g., time, temperature, distance).

Training Step 2b: Emitting Intraverbal Rule Under Control of Keyword

Once the participant underlined, the researcher provided a vocal prompt (i.e., “Say your rule about [time, temperature, distance]”) combined with a gesture to the underlined keyword to occasion the previously mastered intraverbal rule, now in the context of the worksheet task. As the tablet was present, after the participant stated the rule, the researcher waited up to 5 s for the participant to initiate app use and solve the problem. At the end of the training session, the researcher reviewed the worksheet as in PTPs.

Remedial Training (Phoebe only)

On the basis of Phoebe’s performance during PTPs, additional training steps were developed for links in the precurrent response chain (Training Steps 3a and 3b) and errors in conditional discrimination among keywords (Training Across Apps).

Training Step 3a: Using Target App Following Self-Instruction

After Phoebe underlined the keyword and said the rule, the researcher provided a vocal prompt (i.e., “Use the iPad”) combined with a gesture to the tablet to occasion the use of the tablet. Worksheet responses were reviewed as in the prior step.

Training Step 3b: Copying Digital Text

Once the answer was produced by the app, the researcher provided a vocal prompt (i.e., “Write it”) combined with a gestural prompt (i.e., from the answer produced on the tablet to the worksheet) to occasion copying the answer onto the worksheet. Worksheet responses were reviewed as in the prior step.

Training Across Apps

The researcher conducted training with both time and temperature questions on the same worksheet to promote conditional discrimination. The modified worksheets contained four questions about time and four questions about temperature, in varied order. Training Steps 2a, 2b, 3a, and 3b were repeated with the same prompt fading and mastery criteria described above.

Social Validity

After the study, the participants and their teachers completed anonymous questionnaires about their acceptability of the goals, procedures, and outcomes of the study (Tables 1 and 2).

Table 1 Social validity questionnaire results: participantsTable 2 Social validity questionnaire results: teachers