Aided language modeling

Modeling supplies input on the AAC system

The ASHA AAC Practice Portal describes augmented input, aided language stimulation, natural aided language, and aided language modeling as names for a receptive-language approach. A communication partner speaks while pointing to or selecting AAC symbols during an interaction.

Suppose a child looks toward a closed snack box. A partner might say, “Need help?” while selecting need and help on the child's system, then pause. During play, the partner might select funny, again, or my turn while those messages make sense. Modeling can show requests, refusals, comments, questions, feelings, repairs, greetings, and jokes.

Partners usually select a few meaningful words rather than trying to reproduce every spoken word. The appropriate symbols, pace, amount, and language level depend on the person and system. A fixed rule such as “always model one word more” can miss motor effort, visual load, vocabulary layout, context, and the person's own priorities.

A model and a prompt serve different purposes

A model demonstrates how the system can express a message. A prompt is a cue intended to increase a particular response at that moment. The same partner action can function differently depending on timing and what follows, so teams should define what they are doing.

Modeling should carry no requirement to copy the partner. The person may respond later, communicate through speech, gesture, movement, sign, writing, or AAC, or choose silence. A partner can acknowledge any recognizable message and repair misunderstandings with the person.

The person's system stays available

ASHA says AAC users should always have access to their communication tools or devices. A partner can model on a second copy, a printed version, or the person's system when that use does not block access. Return the system immediately after each selection and preserve the person's positioning, access method, privacy, and vocabulary.

Model words the person can use for more than compliance or requesting. Include refusal, consent, discomfort, pain, breaks, opinions, social connection, information, and correction. Essential needs, communication, relationships, movement, and safety remain available regardless of whether the person watches, imitates, or uses the modeled symbol.

The National Joint Committee AAC resource encourages both aided and unaided communication when useful. Aided modeling should support that multimodal repertoire rather than replace a reliable gesture, sign, vocalization, or spoken message.

Research supports cautious, individualized use

Allen and colleagues reviewed 19 augmented-input studies involving people with developmental disabilities or childhood apraxia of speech. They reported evidence for some single-word vocabulary and multisymbol-expression outcomes. Comprehension beyond the single-word level had not been explored in the included literature, and the authors found diagnosis-specific conclusions difficult.

A 2026 review by Wallace and colleagues found seven school-based studies combining peer instruction with aided AAC modeling. The studies centered mostly on inclusive, nonacademic settings and covered limited participant diversity and communication outcomes. That review supports careful peer training and also shows why a family should ask which people, messages, settings, and outcomes a recommendation actually addresses.

These reviews summarize different bodies of research. Neither defines one protocol for every AAC user. Track the person's experience, communication, participation, access, and partner behavior alongside any target skill.

Families can observe the partner's behavior

Ask the team:

1. Which communication purposes and everyday situations will partners model?
2. Which AAC copy will the partner use, and how will the person's access remain uninterrupted?
3. How were vocabulary, languages, symbol locations, and access methods chosen?
4. What counts as a model, prompt, independent message, partner response, and breakdown repair?
5. How much wait time fits the person, and how will partners respond to dissent or fatigue?
6. Who trains family members, peers, teachers, technicians, and other partners?
7. How will the team measure partner implementation separately from the person's communication?

A speech-language pathologist with AAC expertise can lead communication assessment and language-system decisions within competence and licensure. Other professionals and family members can carry out assigned supports within their roles. A BCBA may help define observable partner actions, contexts, and data while coordinating with the communication lead. The AAC user directs meaningful vocabulary, preferences, and acceptable support wherever possible.

A fictional example keeps the denominators separate

Leo is a fictional eight-year-old who uses gesture, vocalization, and a tablet-based AAC system. The team chooses ten short routines for a one-week partner check. The AAC system or agreed paper copy is available at the start of 9 of 10 routines, or 90%. One missing copy stays recorded as an access failure.

Across those routines, the plan identifies 20 natural moments when a partner can model a relevant comment, request, refusal, question, or repair. Partners provide the defined model in 16 of 20 moments, or 80%. They also record 18 eligible communication opportunities. Leo communicates a recognizable message through any agreed mode in 11 of 18, or 61.1%. Partners acknowledge or act on 10 of 11 messages, or 90.9%, within a fictional ten-second window.

The data show access, partner modeling, communication, and partner response as separate events. A one-week sample without a baseline or comparison cannot show that modeling caused Leo's messages. Leo and the family also rate effort, comfort, usefulness, and whether the selected vocabulary matches daily life.

Measures should detect partner and system barriers

Useful measures include access-ready routines divided by routines checked; defined models delivered divided by planned modeling moments; communication by purpose and mode; partner response divided by emitted messages; prompts; wait time; breakdown repairs; access failures; and the person's rating. Define each event, denominator, clock, setting, inclusion rule, and observer procedure before reviewing percentages.

When fidelity is low, examine training, system availability, vocabulary, partner workload, opportunity design, and plan fit. When the person's communication changes, examine every simultaneous change and gather the person's view before attributing an effect.