How to teach grammar using AAC (what actually works)

Last updated 2026-07-11

The main way to teach grammar through AAC is to model it yourself: build full, correctly formed sentences on the device or board as you talk, work core words into everyday routines, and let the child watch and experiment without pressure to perform. The research keeps landing on the same thing as the real driver of grammar growth: aided language input, where an adult models on the system rather than quizzing the child. Drilling grammar rules in isolation rarely gets traction. And if you're wondering how long to keep modeling before you see anything back, many AAC learners need something like 100 to 200 exposures before they produce a form themselves, so patience matters more here than most parents expect.

Why this isn't like teaching a speaking child

A typically developing child hears "she is running" hundreds of times, from dozens of people, in dozens of settings, long before she ever says it herself. That input is constant and free. For a child who uses AAC, it basically vanishes: nobody around them is casually producing full grammatical sentences on a speech-generating device or board as a byproduct of daily life.

That one difference changes everything. A child using AAC can't absorb grammar by osmosis the way a speaking child does. Every communication partner in that child's life has to build the grammatical input on purpose and keep doing it consistently, which is genuinely hard, and it asks partners to unlearn the habit of simplifying their own language just to get through an interaction faster.

There's also a physical cost to AAC that speech doesn't have. Saying "she is running" out loud costs almost nothing. Producing it on a high-tech device might take three, four, or more separate motor movements across different pages or vocabulary sets. That's part of why David Beukelman and Pat Mirenda, in their foundational AAC textbook, treat symbol-by-symbol sentence building and prestored phrases as two separate, both legitimate, tools for grammar ^[1].

On top of that, the population using AAC is anything but uniform. A child with autism and strong receptive language is learning grammar in a completely different context than a child with apraxia of speech who has strong phonological awareness, or a child with cerebral palsy managing complex access needs. There's no single sequence that fits all of them.

What the research actually shows

The most consistently replicated finding is that aided language input, sometimes called aided language stimulation, increases both vocabulary and grammar in AAC users ^[2]. The communication partner talks as usual but simultaneously points to or activates the matching symbols on the child's system. Nothing is being asked of the child; it's pure modeling.

A systematic review by Snell and colleagues in the journal Augmentative and Alternative Communication looked at naturalistic AAC interventions and found that partner-implemented modeling during daily activities tracked with real growth in multi-symbol utterances, which is essentially the functional stand-in for grammar in AAC ^[3]. The evidence was strongest for children with autism and developmental disabilities, though most of the underlying studies were small and measured things inconsistently.

The Participation Model, first laid out by Beukelman and Mirenda, builds AAC instruction around what a person actually needs to say across real environments rather than a fixed developmental grammar sequence ^[1]. In practice, that means going after the grammatical forms that show up in the child's daily life first, even if they wouldn't be first on a typical developmental chart.

One thing the evidence doesn't support is isolated grammar drilling. Fill-in-the-blank tasks or repeating correct frames outside of real interaction have weak support at best and don't match how AAC grammar actually develops ^[4]. ASHA's own evidence maps consistently favor naturalistic, partner-mediated approaches over structured drill for AAC users ^[4].

There's real uncertainty about exact dosage. The "100 times before expecting production" figure circulates widely among SLPs, but it traces back to practitioner experience and a handful of small case studies rather than a controlled trial. The closest hard data comes from Dada and Alant in Augmentative and Alternative Communication, who found significantly increased symbol use after concentrated aided input sessions compared to control conditions ^[5]. More research here would help.

Which grammar skills to target first

Start with whatever is most functional for the child right now, not wherever a chart tells you to start. That said, most SLPs working in AAC converge on a similar rough order.

Core vocabulary comes first, because grammar has nothing to attach to without words. Work by Gail Van Tatenhove and others shows that core words like "more," "go," "want," "stop," "I," "you," "it," and "that" make up roughly 75 to 80 percent of the words people actually use daily, across ages and disabilities ^[6]. These are also the words doing the grammatical heavy lifting: pronouns, verb markers, prepositions. Without solid core vocabulary on the system, grammar instruction has nothing to stand on.

Once core vocabulary is in place, here's roughly how targets tend to be sequenced, based on both developmental research and everyday communicative payoff:

That table isn't a checklist to work through row by row. In real, naturalistic instruction, an adult can model several targets at once inside a single routine: ten minutes at the table can expose a child to agent-action, negation, and possessives if the adult is being intentional about it.

For children with autism who show echolalia, grammar work sometimes needs a different starting point. Echolalic utterances can already contain whole grammatical phrases, and SLPs often use those as anchors for building new combinations. The article on echolalia meaning goes further into working with echolalia instead of trying to eliminate it.

What aided language input looks like at home

Aided language input just means using the device or board while you talk, instead of only speaking. You point to or activate the matching symbols as the words come out of your mouth.

The mechanics are simple. If you're saying "Let's go outside," you point to GO and then OUTSIDE as you say them. You don't need to hit every word. Hitting the content words plus the grammatical linchpins (pronouns, verb markers, negation) is enough. The point is consistent exposure, not a perfect running translation.

Here's what it might look like during a snack. The child reaches for crackers, and you say "you want cracker" while pointing to YOU, WANT, and the cracker symbol. You hand over a cracker and say "here, I give you cracker" while pointing to I, GIVE, YOU. When the child finishes and wants more, you model "more cracker" on the device before prompting anything. And when the child seems done, you say and model "all done eating."

None of this asks anything of the child. You're not telling them to repeat it or pick anything. You're just talking with the device in your hands, or nearby, and making the language visible as you go.

The hardest part for most parents is staying consistent. Pick two or three routines you'll actually model every day. Bath time, meals, and one favorite play activity tend to work well because they happen daily, the vocabulary is predictable, and the child is usually there and reasonably settled. Choose routines you can genuinely keep up: two done consistently will get you further than five you only manage occasionally.

If you're using a high-tech device and don't have your own copy of the same vocabulary layout, ask your SLP about a low-tech or paper backup. Some families keep a printed core board just for the parent to model from while the child uses the device itself. That's a completely reasonable setup.

How do you build two-word and three-word combinations in AAC?

Single words on AAC are communication, full stop. A child who can reliably say "more," "stop," "help," and "go" is communicating, and in the early stages that matters more than grammar. But grammar is what lets a person express nuance, and building toward combinations is a real goal worth working toward.

The approach with the most research behind it is Milieu Teaching, or its modern form, Enhanced Milieu Teaching (EMT). It weaves naturalistic prompting into the child's daily routines and has the strongest evidence base of any AAC grammar intervention in the literature ^[11]. The core technique is a model-prompt-model sequence: model the target combination naturally during the activity ("I want juice"), wait for a moment where the child communicates something related, give a brief prompt ("tell me what you want"), and if the child produces something, expand it grammatically and respond to what they meant. If nothing comes within three to five seconds, model the target again and move on. Expansion is the piece parents often skip. When the child produces "juice," you don't just say "good talking." You say "you want juice" while modeling it on the device, then get them the juice. The expansion supplies the grammar without asking the child to redo what they said. Over hundreds of exposures, those expansions build the internal grammar templates a child eventually uses on their own. The word combination stage can take months or years, and there aren't firm norms for AAC grammar acquisition the way there are for spoken language. According to ASHA's practice portal on AAC, the best predictor of progress is the amount and quality of aided language input from communication partners, not the child's diagnosis or how big their starting vocabulary is ^[4].

Does word order in AAC have to match spoken English grammar?

For most children learning English, keeping standard English word order (subject-verb-object) in your modeling is the standard recommendation. It exposes the child to the grammatical structure of the language they're learning, and their receptive language tends to develop alongside their expressive AAC use. Here's the wrinkle: many AAC systems, especially low-tech core boards, were built to make high-frequency words physically reachable, not to enforce a grammar sequence. A child might hit "want" before "I" simply because "want" sits closer on the board. SLPs debate whether to correct this or accept it. The current thinking, reflected in ASHA's AAC guidance, leans toward accepting the child's word order as a legitimate attempt at communication and modeling the conventional order back in your own expansion, rather than making the child redo their message ^[4]. For children using vocabulary organized by category rather than core word boards, word order slips are common and expected. A child paging through several folders to build a sentence is making tradeoffs with every selection. That's not a grammar error, it's a constraint of the system. Multilingual families carry extra complexity. If a child hears two or more languages at home, the grammar targets need to match the family's actual communication environment. ASHA states plainly that AAC does not cause or worsen language confusion in multilingual children, and that bilingual AAC programming is appropriate and recommended where feasible ^[4]. Nobody should be telling a multilingual family to drop a home language for the sake of AAC grammar instruction.

How do you teach verb tenses and word endings through AAC?

This is where AAC grammar instruction gets genuinely hard. Spoken languages are full of small grammatical morphemes: the -ed that marks past tense, the -ing that marks progressive aspect, the -s that marks plural or third-person singular. These are short, fast sounds that even speaking children take years to master. On an AAC device, each one takes a deliberate symbol selection or a prestored word with the ending already attached. SLPs generally take one of two approaches. The first is to provide the ending as its own symbol: some systems include -ing, -ed, and -s as separate cells the user combines with a verb stem, which is how Unity (available on PRC Saltillo devices) and LAMP Words for Life work. This keeps the grammar explicit, but it costs motor effort, since the child has to hit more symbols to say the same thing. The second approach is to store whole inflected words, so "run," "running," and "ran" exist as separate items and the child never has to build the ending themselves. That costs vocabulary size instead: the system grows large, and finding words across pages gets harder. In practice, the right call depends on the child's cognitive profile and motor skills. For children with strong language comprehension who can handle the extra motor step, explicit symbols are worth it because they map directly to the grammar being learned. For children still building fluency with core vocabulary, whole inflected words reduce friction and keep communication moving. Modeling still does most of the teaching. When you model "I am going" on the device with the -ing symbol, you expose the child to that pattern in a meaningful context. The child doesn't need to understand the grammatical rule, just get enough exposures for the pattern to stick. There are essentially no randomized controlled trials isolating this kind of instruction in AAC. What evidence exists is drawn from the broader aided language input literature and a small number of single-subject studies ^[5]. Any SLP or resource promising a clear protocol here is overstating what's actually known.

How do you use routines and scripts to build grammar in AAC?

Scripted routines are one of the most underused grammar tools in AAC. Pick a routine the child does every day, script out the language that happens during it, program that vocabulary into the AAC system, then let the routine generate dozens of grammar exposures a day. A morning routine might run: wake up, want breakfast, I want [food], more please, all done, go get dressed, I need help. Each of those is a grammar target. Across 365 mornings, a child has heard subject-verb, verb-object, and negation modeled thousands of times in a real, motivating context. Scripted routines are also research-supported for children with autism specifically, where predictability lowers anxiety and creates the calm state in which learning actually happens ^[7]. The grammar instruction rides inside that comfort structure. The key is not letting the script become rigid or turning the child's performance into a requirement. Model the script, then follow the child's lead within it. If the child goes off-script and communicates something unexpected, respond to the meaning and model an expansion. The routine gives the predictable frame; the child's real communication lives inside it. For children in early intervention (birth to age 3 under IDEA Part C), routine-based intervention is explicitly recommended in federal guidance because it's the most practical way to embed therapy goals into a family's actual life ^[8]. SLPs working in early intervention often do exactly this: watch a family's real morning or mealtime routine, then help the parent build grammar models directly into what they're already doing.

Should a speech therapist handle this, or can parents do it at home?

Both. The research is clear that parent and caregiver use of AAC strategies matters at least as much as direct therapy time, arguably more, since parents are present for far more communication moments than any therapist can be ^[2][4]. The catch is that home practice works best once a parent has been trained by a qualified SLP. Setting up the vocabulary hierarchy, choosing grammar targets that fit the child's current level, and learning to model without accidentally creating pressure are all skills that benefit from professional guidance. ASHA's scope of practice for SLPs explicitly includes AAC assessment and intervention, and a board-certified AAC specialist (an SLP with extra training) is worth seeking out if a child's needs are complex ^[12]. Access to SLP services varies enormously. Children with IEPs are entitled to speech-language services as a related service under IDEA if the IEP team decides they're necessary ^[8], and children under 3 may qualify for Part C services. Kids who don't qualify for school-based services, or who need more contact than their school schedule allows, can pursue speech therapy privately, sometimes covered partly by insurance, or through online speech therapy platforms that have grown a lot since 2020. If you're doing most of the work at home, get trained first. A few sessions with an SLP focused specifically on aided language input for your child's system beats months of practice without guidance, and most parents who attend even one focused training session report real gains in confidence and consistency. If you want somewhere to practice grammar modeling between therapy sessions, the Little Words app (littlewords.ai/start) is built for exactly this: structured core vocabulary and grammar prompts you can use during daily routines without needing a full speech-generating device on hand. It won't replace an SLP, but it's a useful daily practice tool.

What should AAC grammar instruction look like for a child with autism specifically?

There's no single autism-specific grammar protocol, and that's because autistic children who use AAC don't fit one language profile. Some have strong receptive grammar but limited expressive output. Some produce echolalia that includes complete grammatical forms. Some process language in ways that shape how they build grammar internally.

Still, a few things hold up across the research. Visual supports help: a grammar pattern shown as a simple strip, PERSON + ACTION + OBJECT, can scaffold instruction, especially for kids who process visually well. That doesn't mean drilling the strip over and over. It means keeping it available as a reference during actual communication.

Motivation may be the strongest grammar teacher there is. When a child badly wants to talk about something they love, they'll put in the motor effort to build longer, more complete messages. Grammar instruction tied to a topic the child actually cares about beats instruction in a neutral, made-up context every time.

It also matters that you don't treat atypical grammar as a problem to fix. Many autistic communicators produce grammatically unconventional sentences for years before their grammar shifts toward standard forms, and that's not a failure. A child who says "want it Thomas train go on track blue now" is communicating something complex with non-standard grammar. The right move is to expand on it and respond, not correct it.

If you want the wider picture of what services and approaches tend to work well for AAC and autism together, the piece on autism spectrum speech therapy covers that ground.

One more thing: grammar instruction has to account for regulation. A dysregulated child cannot learn grammar, full stop. Any partner who has pushed grammar targets during a meltdown or a hard transition already knows how that session ends. Regulation comes first, communication second, grammar third.

How do you measure progress without formal testing?

Standardized tests are often a poor fit for AAC users, mostly because most normed language tests were normed on speaking children ^[9]. That doesn't mean grammar progress is unmeasurable. It just means you need different, more realistic measures.

Mean Length of Utterance in morphemes (MLU-m) is one SLPs lean on often: the average number of morphemes per communication act, sampled during natural interaction. Collect a sample of 50 to 100 utterances during play or a meal and calculate the average. ASHA's practice portal has reference ranges ^[4]. A child moving from an MLU of 1.2 to 2.5 over six months has made real, measurable grammar progress, even if no standardized test would ever catch it.

Another option is tracking the percentage of utterances with a grammatical marker. Pick one target, say verb -ing forms, and count how many utterances in a sample include it (in speech or on the device) versus how many use a verb with no tense marker at all. Track that percentage over time.

A third measure is communicative function diversity: is the child only requesting, or are they now commenting, protesting, asking questions, narrating? Grammar growth tends to track with a child's ability to serve more of these functions ^[4].

A practical way to do this: video ten minutes of a daily routine once a month and count these measures afterward. It takes maybe twenty minutes of analysis, and it gives a real picture of where the child is heading, which beats a standardized score that can't account for the child's AAC access limits anyway. Parents can do informal MLU tracking at home too, once their SLP teaches them the method, which usually takes a single session. It's not complicated math. Mostly it's careful counting.

Common mistakes adults make teaching grammar through AAC

A few patterns keep showing up, both in the research and in what SLPs describe seeing in practice.

Asking before modeling is a common one. Parents and teachers sometimes jump straight to prompting a child for a grammatically complete message without having modeled it many times first. A child can't produce language they haven't been exposed to. Model first, and ask much later, if at all.

Another mistake is simplifying your own language to match the child's level. When a child is producing single words, partners often slip into telegraphic speech, dropping articles, verb endings, and pronouns to make things "easier." The research says this actually weakens the grammatical input the child gets ^[2]. Speak in full sentences, model full phrases on the device, and don't assume that because the child's output is simple, your input should be too.

Some approaches also treat grammar as something a child has to earn before they're allowed to communicate, quietly demanding subject-verb combinations before a child can request something. That's backwards. Communication function should come first, with grammar emerging as a natural byproduct of rich modeling, not a gate in front of it.

Access constraints get ignored too often. If a grammatical word sits three pages deep in a device, the child won't produce it. Vocabulary needs to be reachable within a second or two for grammar instruction to have a chance. When grammar work isn't landing, check the device layout before deciding the child isn't ready.

And it's easy to give up on morphology. Small grammatical words and endings are genuinely hard to teach through AAC, so they get deprioritized. But morphology is what separates "I want" from "I wanted," and that distinction matters for real communication. Keep modeling those forms even when it feels tedious.

Frequently asked questions

At what age should you start teaching grammar through AAC?

As soon as a child starts using AAC meaningfully, grammar modeling can begin. Under IDEA, children can qualify for early intervention services from birth to age 3, and the AAC research base includes children as young as 18 months. There is no developmental floor for aided language input. The earlier communication partners begin modeling grammatically complete messages on the device, the earlier the child has rich grammar exposure to absorb.

Does using AAC slow down speech development or grammar in spoken language?

No. This concern is well studied and consistently unsupported by evidence. ASHA's position is that AAC does not impede speech development and may support it. A 2006 research review by Millar, Light, and Schlosser found no evidence that AAC use suppressed speech output, and many children showed increases in spoken language after AAC introduction. The fear that AAC will replace speech has no empirical basis and should not delay access to AAC for any child.

How do you teach pronouns through AAC?

Pronouns are among the most important and most confusing grammar targets in AAC. Model them in first and second person first (I, you, my, your) because these are most frequent and most functional. During aided language input, consistently point to yourself when you say and model "I" or "my," and point to the child when you say and model "you" or "your." Repetition across hundreds of naturally occurring moments is what builds pronoun mapping. Third-person pronouns (he, she, they) can come later.

What if the child just uses single words on their AAC and won't combine?

Continue modeling multi-word combinations and expand the child's single-word productions without requiring them to revise. Single words can serve full communicative functions, and pushing combination too early backfires. Check that the vocabulary needed for combination is actually accessible on the device. Then increase the density of aided language input during high-motivation routines. Sustained rich modeling over weeks to months typically precedes the emergence of combinations. There is no shortcut.

How do you teach question formation through AAC?

Place wh-words (what, where, who, when, why, how) in a consistent, accessible location on the AAC system. Model questions yourself on the device during daily interactions: "what is that?" when encountering something new, "where did it go?" during a disappearance game. Comment on and reinforce the child's use of wh-words in any communicative context. Question-asking is strongly linked to social motivation, so pair it with highly preferred activities where the child genuinely wants information.

Can children with complex physical access needs still learn grammar through AAC?

Yes, though access constraints shape the instruction significantly. For children using switch scanning or eye-gaze access, every symbol selection has a higher motor and cognitive cost. This increases the value of prestored phrases and decreases the feasibility of building sentences symbol-by-symbol. Grammar instruction for these children focuses on modeling, on expanding the vocabulary of prestored grammatically complete phrases, and on making high-frequency grammatical words the fastest-access items in the layout. An AAC specialist can help optimize the system layout for grammar access.

How long does it take for AAC grammar instruction to show results?

Honest answer: it varies widely and nobody has solid norms. The small studies available suggest that consistent aided language input during daily routines over several months produces measurable increases in multi-symbol utterances. Some children show movement in 8 to 12 weeks of daily partner modeling. Others take much longer. Severity of underlying condition, quality and consistency of input, and device accessibility all affect the timeline. Monthly informal language samples are the best way to track individual trajectory.

Should I correct my child's grammar errors on their AAC device?

Generally, no. Correction as a direct instructional move (asking the child to redo a message correctly) has weak evidence and can reduce communication motivation. The recommended alternative is expansive modeling: respond to the meaning of what the child communicated, then model the target grammatical form on the device yourself. This provides the correct model without making the child feel their communication was wrong. Consistent expansion over time shapes grammar more effectively than correction.

What AAC systems are best designed for teaching grammar?

Systems with a strong core vocabulary, morpheme symbols, and a consistent layout across vocabulary sets are best for grammar instruction. LAMP Words for Life, PRC Unity, and Snap Core First are frequently cited by SLPs for their grammar-accessible vocabulary organization. What matters most is that the system has been customized for the specific child, that the child's communication team has been trained on it, and that it is actually available and accessible throughout the child's day. The best system is the one the child uses.

How do I coordinate grammar targets between school and home?

Ask the child's SLP to write the current grammar targets explicitly into the IEP or therapy plan with enough specificity that a non-SLP can implement them. A one-page home practice sheet showing the three current grammar targets, the vocabulary on the device to target them, and the two or three routines to practice them in is more useful than a generic IEP goal. Request a brief training session where the SLP models the aided language input strategy on the child's actual device. Video that session if possible.

Do multilingual children using AAC learn grammar differently?

Multilingual AAC users learn grammar across two or more systems at once, and that process is not well studied. ASHA's position is that bilingualism does not impede AAC use or language development and that AAC systems should reflect the child's full linguistic environment where possible. If the family communicates in Spanish at home and English at school, the ideal is a device with both language vocabularies. Grammar targets may differ between languages. An SLP familiar with the family's specific languages is the best resource.

How does grammar instruction through AAC differ for a child with apraxia of speech?

Children with childhood apraxia of speech often have strong language comprehension and grammar knowledge that exceeds what they can produce motorically in speech. For these children, AAC can serve as a bridge or supplement while speech motor skills develop. Grammar targets through AAC should track the child's receptive language level, which may be well above their spoken output. The article on childhood apraxia of speech covers how to align AAC and speech motor treatment goals for these children.

Sources

1. Beukelman, D. & Mirenda, P., Augmentative and Alternative Communication: Supporting Children and Adults with Complex Communication Needs, 4th ed., Brookes Publishing, 2013: Distinction between symbol-by-symbol sentence construction and prestored phrases as legitimate grammar tools in AAC; the Participation Model framework
2. Snell, M. et al., Twenty Years of Communication Intervention Research with Individuals Who Have Severe Intellectual and Developmental Disabilities, AAC, 2010: Partner-implemented modeling during daily activities associated with meaningful growth in multi-symbol utterances in AAC users
3. American Speech-Language-Hearing Association, Augmentative and Alternative Communication (AAC) Practice Portal: Naturalistic, partner-mediated approaches outperform structured drill for AAC users; volume and quality of aided language input is primary predictor of grammar growth; AAC does not impede speech development; multilingual AAC is appropriate
4. Dada, S. & Alant, E., The Effect of Aided Language Stimulation on Vocabulary Acquisition in Children with Little or No Functional Speech, AAC, 2009: Significantly increased symbol use following concentrated aided input sessions compared to control conditions
5. Van Tatenhove, G., Building Language Competence with Students Using AAC Devices, ASHA Perspectives on Augmentative and Alternative Communication, 2009: Core vocabulary words account for roughly 75 to 80 percent of words used in daily communication across speakers of all ages and disabilities
6. Nunes, D. & Hanline, M. F., Enhancing the Alternative and Augmentative Communication Use of a Child with Autism through a Parent-Implemented Naturalistic Intervention, International Journal of Disability, Development and Education, 2007: Scripted routines reduce anxiety in children with autism and create the regulated state in which AAC learning occurs
7. U.S. Department of Education, IDEA Statute, 20 U.S.C. 1400 et seq., Individuals with Disabilities Education Act: Children with IEPs are entitled to speech-language services as a related service; Part C covers birth to age 3; routine-based intervention recommended in early intervention
8. Prath, S., Evaluating the Language of Students Who Use AAC, Perspectives of the ASHA Special Interest Groups, 2021: Most normed language tests were normed on speaking children and are poorly suited to AAC users; ecologically valid sampling measures recommended
9. Millar, D., Light, J. & Schlosser, R., The Impact of AAC on the Speech Production of Individuals with Developmental Disabilities: A Research Review, Journal of Speech, Language, and Hearing Research, 2006: No evidence that AAC use suppressed speech output; many children showed increases in spoken language after AAC introduction
10. Kaiser, A. P. & Roberts, M. Y., Parent-Implemented Enhanced Milieu Teaching with Preschool Children who have Intellectual Disabilities, Journal of Speech, Language, and Hearing Research, 2013: Enhanced Milieu Teaching has the strongest evidence base of AAC grammar interventions; model-prompt-model sequence increases multi-word utterances
11. ASHA, Scope of Practice in Speech-Language Pathology, 2016: AAC assessment and intervention is within SLP scope of practice; board-certified specialists in AAC available for complex cases