Article

Modeling AAC: How to Talk Through the Device

Last spring in Portland, a mom named Rachel sat across from me at a coffee shop with her four-year-old daughter's iPad propped between our mugs. The device ha

Last spring in Portland, a mom named Rachel sat across from me at a coffee shop with her four-year-old daughter's iPad propped between our mugs. The device had a full AAC system loaded, 84 buttons on the main screen. It had been set up by their SLP three months earlier. "She hasn't touched it once," Rachel said. "Not once. I put it in front of her at meals and just... wait." I asked Rachel how many times she'd tapped on the device herself in the past week. She thought about it. "Maybe twice?"

That was the whole problem.

Modeling AAC means tapping words on your child's device while you talk, so they see how the device works in real conversation. The technical term is aided language stimulation. And it is, by a wide margin, the single most important variable in whether AAC works. Without modeling, a device sits unused. With consistent modeling, the device becomes the child's voice.

LittleWords is a speech-practice companion, not an AAC system. This article is for families using a dedicated AAC device.

The logic is obvious once you see it

A child does not learn to use AAC by being told to use AAC. They learn by watching adults use it.

Think about how spoken language develops. Adults talk around babies constantly, in every context, about everything and nothing. A child hears thousands of hours of speech before producing a single word. Then they imitate. Then they generate their own sentences. Nobody sits an infant down and says, "OK, now talk."

AAC follows the same pattern. The child needs to see the device used in every context, by every person, for every kind of message. Imitation comes after. Generation comes after that.

Here's the thing: if the only person who ever touches the device is the child, the child has no model. They see the device as a homework assignment, not as a way the world communicates. But if adults model on the device alongside their verbal speech, the child begins to see it as talking. They map the visible words on screen to the spoken words in the air. They learn where words live on the board. They learn what words do.

Modeling is input. No input, no output. Simple as that.

What it actually looks like at breakfast

You're eating. Your child has their AAC device next to them.

You say, "I want cereal." As you say it, you tap "I" then "want" then "cereal" on the device.

You finish your bowl. You say, "All done." You tap "all done."

Your child reaches for the juice. You say, "You want juice?" You tap "want" and "juice."

That's it. You're speaking out loud. You're tapping on the device at the same time, matching the key words you just said. You're not telling the child to use the device. You're not quizzing them. You're not pausing with eyebrows raised, waiting for them to perform. You're just showing them, over and over, that this is how words work.

How much is enough

The aided language stimulation literature doesn't give you a single magic number, but the practical translation is this: tap at least one key word per spoken sentence in your child's hearing. More when you can. Throughout the day.

You will not get every word. You won't even get most words at first. But over time, looking at the device when you speak becomes habit, and the tapping becomes almost automatic, like gesturing when you talk.

A reasonable starting target for parents new to this: ten modeled phrases in the morning, ten in the evening. Build from there. Within a few weeks, you'll find yourself tapping without thinking about it.

One word ahead, not ten

When you model, match your tapping to your child's current production level, then add one word. This is sometimes called the one-up principle.

If your child has no words on the device yet: model single words. "More." "Eat." "Done."

If your child taps one word: model two. "More cereal." "All done." "Want juice."

If your child taps two words: model three. "I want more." "All done eating." "Mommy go work."

You're showing them the next step. Not the summit, just the next step. It's like teaching someone to cook by standing at the stove with them, not by handing them a Julia Child recipe and leaving the room.

Five traps that quietly kill progress

Only modeling when you want them to communicate. This is the most common mistake. Modeling should happen during your communication, not just during theirs. When you talk, you tap. Not only when you want them to tap.

Only modeling requests. So many parents model "want" and "more" because they've internalized the idea that AAC is for asking for things. AAC is for all communication. Model comments ("look, big truck"), feelings ("I'm sad"), social words ("hi," "bye"), and protests ("stop," "no"). A child who can only request is a child who's been taught that their voice only matters when they need something. That's a bleak lesson.

Modeling and then quizzing. "I tapped 'want.' Can you find 'want'?" This turns modeling into a test. The child feels the pressure shift. Skip the quiz. Just model.

Hand-over-hand prompting. Physically moving the child's hand onto buttons. This often produces aversion, sometimes lasting aversion. Model with your own hand. Let the child come to it when they're ready.

Modeling only familiar words. Model rich language. Use words the child doesn't have yet. Their brain is mapping the whole system, not just the five words you think they need this week. Expose them to the full vocabulary.

The whole household has to be in on it

If only one adult models, the child learns the device is for talking to that one adult. AAC needs to be the language of the whole environment, not a tool for a specific relationship.

This is the hard part. Getting grandparents, babysitters, siblings, and teachers to model takes patience. Some will resist. Some will feel awkward. Some will say the device is "not their thing."

What works: show, don't lecture. Model in front of them so they see how casual it is. Give them one specific instruction ("When you hand her the cup, tap 'cup' on her device"). Ask them to try just one word per visit. And be patient, because adults learning new communication habits takes weeks, sometimes longer.

Schools and therapists should be modeling too. If they're not, raise it directly. The IEP can and should require AAC modeling by staff. That's a reasonable accommodation, and a necessary one.

When your child also uses some verbal speech

Many AAC users have partial verbal speech. They say some words out loud and use the device for others, depending on the situation, fatigue level, or complexity of the message.

Modeling continues regardless. You model on the device even when the child can sometimes say the word verbally. The model reinforces the device as a real communication tool and gives them a reliable alternative for the moments when speech is hard.

Don't hide the device when verbal speech shows up. Both modes coexist. That's the whole point.

You never really stop

Modeling continues as long as AAC is part of the child's life.

What changes is the ratio. As the child becomes more proficient, the adult does less modeling per interaction. But it never hits zero. Adults keep using the device with the child as a normal part of conversation, the same way you don't stop talking to your child once they can talk. You keep modeling more complex language at higher and higher levels.

Where LittleWords fits (and doesn't)

LittleWords is not AAC. For an AAC user, LittleWords is a separate tool for emerging verbal speech practice. The AAC device handles daily communication and gets modeled by you and everyone else. LittleWords is for ten minutes of structured verbal practice alongside all of that.

Getting help with your modeling

The SLP who set up your child's AAC should be coaching you on modeling. If they're not, ask for it. Modeling coaching is a specific skill, and honestly, some SLPs are much better at it than others. A good one will observe your natural interactions, give you targeted feedback, and help you troubleshoot the moments where modeling breaks down.

If your child has been on AAC for months and you're not seeing engagement, the most likely culprit is insufficient modeling. Not the device, not the layout, not the child. The modeling. An SLP can observe and adjust.

Rachel, the mom in Portland? Her SLP started joining breakfast twice a week over video, watching Rachel model and coaching in real time. Six weeks later, her daughter tapped "more" independently for the first time. Then "want more." Then, unprompted during a car ride, "go park." Rachel cried. She told me about it in a voice memo that was mostly laughing and crying at once.

The device was never the problem. The input was.

FAQs

How long until my child starts tapping the device? It varies widely. Some kids tap within the first few weeks. Others take months. Consistent modeling is the strongest predictor of engagement. Don't stop modeling because you haven't seen results yet.

My child doesn't look at the device when I model. Should I stop? No. They're absorbing more than they appear to be. Continue modeling. Engagement often increases after extended exposure, sometimes quite suddenly.

Is it ok to model on the device with my own finger? Yes, that's exactly what you should do. Your finger on their device is fine. The device works no matter who taps it. Modeling means you tapping while you speak.

What if I tap the wrong word by accident? Just tap the correct one next. The child sees you make a mistake and fix it. That's actually useful. It shows them errors are normal and recoverable.

Do I need to use the device every single time I speak? No. Aim for partial coverage, gradually increasing over time. Start small. Build the habit. Perfection isn't the goal; consistency is.

---

Related reading: AAC for autism hub · Speech therapy at home for autistic kids (pillar guide) · AAC for toddlers · AAC myths autism

Related Little Words guides

Important: Little Words is educational support for home practice. It is not a medical device, not an AAC replacement, and not a substitute for a licensed speech-language pathologist, pediatrician, or developmental evaluation.