“Or what about elves and unicorns,
or cranberries grown with pairs of horns,
or chocolate cheesecakes laced with thorns,
these things do not exist.”
— ”Things That Don’t Exist”, by Eppink, Hannon, and Taite
“And if the genie were set free
And by the laws of things like that, he’s indebted to me
I’d bury my three wishes
Deep down in the ground
So I will not turn my whole life upside down.”
— “Upside Down”, by Barenaked Laides
Requesting via AAC? Check.
Pointing? Check.
Functional play? Check.
Your child is in a great place. Enjoy the success, but don’t stay here too long. They have vocabulary, which is essential. They can play in the social world, which is a relief. But the real prize — language — is not there yet. It can be though, if you know what to look for, and push for it in a determined way.
What is language that matters?
It is not functional communication. Your child can already tell you what they want using single button presses. “Cookie” means the same as “I want a cookie, please.” “Pee” means the same as “I would like to go to the bathroom, please.” Adding more words to it is social nicety, not language.
It is not core words, words prioritized by how often they occur in natural speech. Your child doesn’t need to press “this” or “here” or “that” or “there” on an AAC. They can convey those ideas perfectly well by pointing.
Other AAC apps have pages for numbers, shapes, colors, seasons, feelings, coping skills, medical terms. All useful, but just vocabulary.
So what is language that matters?
It is word combinations leading to imagination. Here are five nouns: car, cat, dinosaur, banana, mommy. And here are five verbs: throws, eats, runs, attacks, sleeps.
Now pick one of the nouns and one of the verbs. Can you see in your mind what is happening?
Now pick another combination. Suddenly you are in another world, another time, another place. Your mind glides from mommy sleeping to a dinosaur running.
Something amazing happens to typically developing children when they are around 18 months old. By this time, they can express around 50 words. Some are nouns and some are verbs. The remarkable event is that they begin to combine words into pairs: mostly nouns with verbs and verbs with nouns. It may not seem important, but it is the key to everything.
Why? 18 month old children are in the sensorimotor stage — what is within reach, here and now. They live in the present. The world exists as it appears in front of them. Ball. Daddy. Milk.
Over the next months they progress to the preoperational stage — using language to travel through space and time. Ball roll. Daddy go. Milk more. They can plan, they can anticipate, they can imagine.
And this shift happens because of language. Two word combinations are critical.
Many autistic children never make this transition, because it requires such an extreme mental shift. Their mouth moves, but only single words come out. Or scripted phrases. Or bits of songs. Some like the term “gestalt language processing.” But let me state it plainly. It is speech without language. It is equivalent to the expressive language capacity of an 18 month old child. And 40% of autistic children remain at this level when they are 9 years old.
The good news is that two word combinations are a very teachable skill, providing you’ve done all the steps leading up to it — building vocabulary, pointing, and play.
When my son was 26 months old, he was a happy kid. He was good at playing, and good at pointing. His spontaneous speech was limited to “help”, “help please”, and “pick me up”; as well as environmental sounds like “beep”, “whee”, and “honk.” He understood hundreds of verbs and thousands of nouns, and could repeat many of them back. But he was not motivated to use any of those words on his own. Why would he be? A few words got him all he needed: “help” — I would run to him — ”pick me up” — I would lift him — then he would point to what he wanted me to get. In his mind, he was set. But his expressive language was still less than that of an 18 month old child.
That all changed on a Friday. ABA said they wanted to get him saying two word phrases. I had no idea how they could do that, so I thought I would practice with him. I said “eat baby” and nibbled on his head. He found this very funny and started saying “eat baby” over and over again, and looking at me expectantly.
An hour later, I said “eat airplane”, picked up one of his airplanes, and nibbled on its tail. He looked puzzled for a second, then got very excited. He said, “eat helicopter.” I nibbled on his helicopter. He said, “eat fire truck.” I nibbled on his fire truck. He went back and forth through airplane, helicopter, and fire truck for a while. Then he smiled and said, “eat baby!” I nibbled on his head again.
Later, I introduced other verbs like tickle, attack, cut, spin, throw, and honk. He ended up spontaneously saying 2 word phrases and 3 word phrases by the end of the day.
He stopped doing any of this for a while. Then a month later, his language exploded. He progressed at twice the rate of typical development. A few months later, his speech delay was gone. He had caught up.
That experience cemented in my mind the primacy of play. My son didn’t care about the verb “want” and the nouns “snack” or “water.” His world was full of airplanes, fans, wind turbines, space shuttles, cars, trucks, propellers, helicopters, air conditioners, rockets, excavators, shovels, sticks, guns, buttons, keys, motors, lights, switches, and doors. Verbs he cared about were eat, attack, spin, throw, shoot, poke, push, press, jump, turn, climb, go up, turn on, and honk. He was excited when he realized language could make new things happen in his world. Play worked when an emphasis on functional communication did not, because he had learned play, he enjoyed play, and because — to paraphrase Cindi Lauper — Babies Just Want to Have Fun.
The progress my son made in a short amount of time is unusual. You may feel skeptical, and you would be right to — it is one story about one child. There is no guarantee that it will work the same way for your child. The most likely outcome to saying “eat baby” and nibbling your child’s head isn’t that they repeat it back. It’s nothing at all. Silence.
But there is a way you can do what I did, but more deliberately and patiently, which IS likely to work for your child. It’s based on a method called JASPER (Joint Attention Symbolic Play and Emotional Regulation), developed in Prof. Kasari’s lab at UCLA. JASPER has been validated and proven to work over dozens of studies involving over 500 autistic children.
Have you seen the movie Jiro Dreams of Sushi? Chef Jiro, 85 years old, owns a 10-seat restaurant in Tokyo. Every morning at 5 am, his 50 year old son goes to Tsukiji Market by the docks to get the best fish for the day. The food is meticulously prepared and excellent.
JASPER is of the caliber you would expect from a team that has spent the last 35 years perfecting it. I was fortunate enough to get my son into one of their research studies when he was 16 months old. He’s had excellent speech, OT, and ABA therapists since, but none come close to what he received at UCLA. Two one-hour sessions per week — and at the end of each one he was exhausted. He would nap for an hour more than usual.
JASPER is an unusual therapy in that it is intensive on the therapist, not on time. When delivering JASPER, the therapist needs to be ON — attuned, creative, and present. But once they are done, they are done. There is progress every week. Results come quickly. It’s the perfect therapy for a busy parent to deliver — focused, intense, done. You don’t need to carry an iPad around all day, talking and pressing buttons. You don’t need to do it forever with no end in sight. What is asked is that you be there — fully there — for your child 30 minutes per day for 6 weeks. It’s a sprint, not a marathon.
This section can’t do justice to JAPSER — there is a clinician’s manual that describes the method fully, and a training program in Los Angeles for those who want to learn to do the method themselves. But I can provide some strategies based on JASPER. Thankfully, Kasari’s studies have shown that parents can learn enough of the method to boost their child’s language development significantly.
Okay. This is the time when all the pieces come together in one intensive push.
By now, your child can request dozens of single words using AAC. They can point. They can play with you in a functional way.
Imagine what you want your child to be doing — saying “eat bee!” while chewing on a stuffed bee. Saying “teddy cook” while a stuffed bear makes food for his friends. Saying “dinosaur chase” while a T Rex figurine runs after a doll. “Doll hide!” All the while pointing for emphasis.
To make this happen, you will need to turn into a clear language model for your child. Think of a polite butler, making precise exaggerated motions while speaking slowly and clearly. You are not asking questions. You are not telling your child what to say. You are narrating your own side of the play in two-word phrases, while using every modality you can — pointing, AAC, and talking.
For example, let’s say you and your child are making a car out of Magna Tiles, so that Tiger can visit his friends at the zoo.
Then you add the tile to the car. You don’t need to talk through everything you do — make sure to leave room for your child to expand.
Maybe your child grabs a tile and adds it now. And now it’s your turn again. But this time, you narrate a different part of what you’re doing:
Keep at it. Repeat. Repeat. Wait for a miracle.
And then it comes. Your child points at a tile and tries to say “tuh!”
Now you repeat back what they meant to say, but better:
If they say the word clearly, or use the AAC to say the word, you can expand what they’re saying:
All of this can be hard to juggle while you’re trying to play with your child. Don’t worry about perfection. Keep your eye on the prize — getting your child to say two-word combinations on their own. And you do this by saying two-word phrases during play in a way that makes you super-easy to copy. Point. Talk. Use AAC.
When they say their first two-word phrase to you, respond enthusiastically! And say it right back to them. The enthusiastic part will come naturally, I promise.
Once the floodgates open, keep going. You can allow yourself some relief at this point. Once your child gets how 2 word combinations work, they have that forever. Something unlocks in their brain. It is like riding a bike — only better. They are flying into the preoperational stage on the wings of language.
Once they see how nouns and verbs can swap in and out, it’s a small step to give other objects agency too. I eat. Teddy eats. I drive. Teddy drives. Symbolic play, imagination, it all comes together now. Their language grows and grows, and now their mouth runs nonstop because it has plenty to say.
Have you ever made rock candy? You bring water to a boil, then stir in more and more sugar until the liquid can hold no more (you’ll know once sugar starts settling to the bottom rather than dissolving). Pour just the liquid part into a separate bowl, then let it cool down. Cold water can’t hold as much sugar as hot water, so eventually the sugar needs somewhere to go. Bit by bit it comes out as a beautiful crystal. It can happen in a minute; it can happen in an hour; it can happen in a day. Once the crystal starts to grow, it is unstoppable.
You may want to know when your language teaching efforts will pay off. Will two-word combinations happen in week 1? Week 6? Or not at all?
I don’t know. It’s not possible to know when it will happen, when chaos shifts to order, when the pattern coalesces without your child’s mind. All you can do is create the right conditions for growth to happen. And that means saturating your child’s world with as many two-word combinations as you can.
That is the idea behind aided language modeling. Speak normally to your child, narrating your actions throughout the day. Have your tablet or phone with you at all times. As you talk, press buttons for key words, e.g.
In time, your child may begin to tap their own word combinations into the AAC.
Seems reasonable right? Tap while you talk. Tap while you talk. Carry a device with you, and capture every opportunity to tap out two-word combinations. The more language models you can provide, the better.
Here’s the challenge: it’s really hard to do. Not the tapping while talking part. You’ll get the hang of it quickly. But the psychological aspect is brutal. You may end up modeling without expectation, spending the day tapping while you talk — all the while getting nothing back from your child. It feels like the definition of insanity, doing the same thing over and over, with the same result each time.
That is why I recommend aided language modeling as a secondary method, not as a primary method.
Teaching your child how to request? That’s just a procedure, with plenty of motivation on both sides. You hand-prompt, they get a cookie, everyone's happy.
Using AAC during play? Two-word combinations fly by fast and furious, a testament to the dynamic nature of play. It’s a target-rich environment. You tap "throw" + "ball," ball flies through the air, child laughs.
Doing aided language modeling through your daily routines and life? You tap "put" + "shoe" while putting on their shoe. Nothing happens. You tap "open" + "door" while opening the door. Nothing happens. You tap "eat" + "banana" while they eat a banana. Nothing happens. If you can manage it, please do it. It will help your child by giving them more language models, like pouring more sugar into that boiling water. But if it’s too much to handle right now? I understand. Don’t worry about it. Focus on modeling language during play instead.
Now for something a bit lighter — music! Play is one way to get children excited about language. Music is another. Words inform, but music moves, so much so that some children who cannot talk will nonetheless sing. Who knows why? Perhaps it has to do with different areas of the brain being activated. Or perhaps the emotional resonance of music pushes through to convey a deeper message: this means something. Even if I don’t know the words, this means something, and it is worth paying attention to. Music can be a bridge to language.
Visual Rhythm is a supplemental game that combines Aided Language Modeling with children’s songs. As each song plays, picture tiles of key words appear on screen. Your child taps each tile in rhythm to continue the song. Miss a beat, and the song pauses until they tap. They get immediate feedback — tap the word, and the song continues (similar to the Sensory Social Routines you did in the first section).
Here’s an example of pulling out key words to help build language:
If you’re HAPPY and you know it CLAP your HANDS.
This distills the song into a phrase your child could say on their own later:
HAPPY. CLAP HANDS.
Or for another song:
The DOOR on the BUS goes OPEN and SHUT.
which becomes:
DOOR BUS. OPEN. SHUT.
The game may help with speech as well as language. Babies between 6 to 12 months old go through a "babbling and banging" stage, simultaneously babbling while clapping their hands or banging an object in a coordinated manner. This integration of vocal and motor systems is thought to help speech develop at a later age. There is a therapy called Auditory Motor Mapping Training involving singing and drumming that has been shown to help minimally verbal autistic children produce speech. Singing and tapping on words may help in a similar way.
Finally, if your child likes listening to the same song over and over again on repeat, this game may satisfy that itch. But instead of passively listening, your child will be actively engaging with words and language.
Will all this actually help your child? I don’t know. The game is a supplement both in the sense of “extra” and also in the sense of “not proven to work.” But it does combine elements of methods known to help with language and speech. I encourage you to give it a try. It can’t hurt, it may help, and it is likely to be fun for them.
I was talking with Prof. Howard Shane from Harvard Medical School, who has worked with hundreds of minimally verbal autistic children.
I was showing him Visual Choices, and he was mildly dismissive. Any app could get kids to request single items, he explained. The real challenge was that for any AAC system, most children ended up stuck at that level.
The reason? There was a lack of emphasis on 2 word combinations in traditional speech therapy. Core words, yes. Functional communication, yes. But learning how to mix nouns, verbs, and adjectives together in flexible ways? No.
He advocated for what he called language arts instruction — explicitly teaching how words worked together. To this end, he made activity boards — collections of nouns, verbs, adjectives, and prepositions curated to a specific activity. And then he would act out an activity and model how to make the correct word combinations. Or he would pull up a photo or video of a scene, and have the child describe it in the same way. (I later talked to Prof. Binger at the University of New Mexico, who pioneered using activity boards to model two-word combinations; she expressed the same frustrations as Prof. Shane at the current state of AAC language instruction, though she focused more on aided language modeling during play).
Sound familiar? It followed the same idea as JASPER. Find out how to model lots of word combinations, and encourage the child to make some back. But the difference was that it didn’t follow naturally from play. He would show a scene, model it, show another scene, model it, then show another one, and wait for a response. It was explicit language instruction — and it worked. Many of his patients were able to progress rapidly in language this way.
I created Visual Language as a way for your child to practice language in a way aligned with Prof. Shane’s approach. The game has two parts. First is a set of structured lessons, each focused on a particular type of word combination. It follows closely the constructs published by Brown in his original work on toddler language development. For example, we show a picture of a big cat, and we ask
[big or small?] cat
The child needs to select the correct adjective. Then we show a picture of a small bear:
small [cat or bear?]
The child needs to select the correct noun. Finally, we show the child a picture of a small cat, and have them select the correct adjective and the correct verb:
[big or small?] [cat or bear?]
Do they get it right? Do they understand how words can mix and match? With repeated practice, they do.
The game then moves on to verb-objects, subject-verb, subject-verb-object, prepositional phrases, and other constructs.
The second part to Visual Language is a “magic AAC”, which is a grid of 5 nouns, 5 verbs, 4 adjectives, 2 prepositions, and the extra verb “is”. Your child taps in a sentence of up to 5 words, and an image instantly appears. For example:
Instead of learning language through play, your child is playing with language. They experience first-hand how a handful of words can combine to make thousands of possibilities (around 3,000 images in this case).
You may object — my child is barely making sounds with his mouth, and you’re advocating for explicit grammar instruction. Are you out of your mind?
That is fair. And I empathize that it seems hard and unnatural.
But here’s the thing — we are already asking your child to do something hard. Toddlers work hard. Making that mental leap at 18 months old is hard work. There may be many ways to help a child make this leap, but at the end of the day, the leap still has to be made. If they don’t learn it today, it doesn’t get any easier tomorrow. It may well get harder.
As for natural versus unnatural, there is something seductive about the idea that to teach language, all you need to do is expose your child to enough samples of it. Eventually they will pay attention to enough of it, figure out the pattern, and start talking in sentences, like every typically developing child does. But this just doesn’t happen for a large number of autistic children. A little extra support, deployed in a modality your child enjoys, may be beneficial.