Last updated 2026-07-11
TL;DR
To teach two-word combinations with AAC, model the combination on the device yourself (a technique called aided language input), then wait expectantly for your child to try it. Start with high-motivation pairs like 'more juice' or 'go fast.' Research suggests most children begin combining words after seeing a combination modeled 50 to 200 times across natural routines, so consistency matters more than any single therapy session.
A two-word combination is two separate symbols or words put together to make a new meaning. 'Want cookie' says something neither word says alone. 'More swing' means something different from 'more bubbles.' That leap matters developmentally, because it shows a child is starting to build language rather than just labeling things around them.
For a child who's a late talker, moving from single words to combinations is often the door to faster vocabulary growth. The American Speech-Language-Hearing Association notes that most typically developing children begin combining words between 18 and 24 months [1]. For kids using AAC, that same milestone is reachable, just on a different timeline and through a different physical route.
AAC (augmentative and alternative communication) covers any system that helps someone communicate beyond natural speech: picture boards, speech-generating apps, high-tech devices. The research is clear that AAC does not slow speech development and often supports it [2]. Teaching two-word combinations on a device isn't a consolation prize. It's real language learning.
Aided language input: modeling on the device itself
Aided language input (sometimes called aided language modeling) means you, the communication partner, point to or activate symbols on the AAC system while you talk, instead of just speaking into the air. It really is the best-supported approach for building symbol combinations. A 2012 study by Drager and colleagues in the American Journal of Speech-Language Pathology found that children with complex communication needs showed significantly more symbol use after aided language input compared to instruction without modeling [3]. A 2021 systematic review in the Journal of Speech, Language, and Hearing Research confirmed that naturalistic AAC intervention, which puts aided modeling at the center, produced positive outcomes across age groups and AAC types [4].
The logic holds up: you wouldn't teach a child piano by talking about piano, you'd play alongside them. Aided language input works the same way. You show the child what their tool can do, in real moments, over and over.
The hard part for parents is actually doing it. Talking and pointing at the same time takes practice, especially on a dynamic display where symbols live on different pages. Start with just two or three target combinations per week so you're not hunting around the device while your child waits.
Choosing your first combinations
Pick combinations your child already wants to say, not the ones a worksheet thinks are important. If your kid loves the swing, 'go swing' beats 'more cracker' even if crackers are the textbook choice.
A useful framework is to combine words from two different categories: an action word plus a noun, or a modifier plus a noun. Here's what the most common first combinations look like in typical development:
| Combination type | Example | Why it works |
|---|---|---|
| Action + object | 'push car' | High motor play, lots of repetition |
| Recurrence + object | 'more juice' | Built-in motivation, clear cause and effect |
| Rejection | 'no bath' | Strong emotion drives communication |
| Location | 'dog outside' | Comments on the environment |
| Attribution | 'big ball' | Describes something visible right now |
Keep the target list short. Most speech therapy approaches for AAC work on three to five core combinations at a time, building repetition across the day rather than breadth across a huge word bank [5].
Check that both words in the combination are already on the device. If your child has to hop to page 3 for 'more' and back to page 1 for 'cookie,' the motor demand will swamp the language goal. Reorganize if needed, or work with your SLP on a core vocabulary layout that puts high-frequency words one tap away.
What this looks like at home, day to day
Forget the word "session." The goal is to weave modeling into what you're already doing, because that's when children learn language best: in context, with real emotional stakes, repeated many times.
Take snack time, with the target 'want more.' Give a small amount of a preferred food. When your child reaches, looks at you, or vocalizes, say 'want more?' while touching 'want' and then 'more' on the device. Then hand over the food. Don't make them repeat it back, and don't withhold food as a bargaining chip, since that creates stress rather than communication. Model, respond, repeat.
Over many days and many repetitions, add an expectant pause before you hand over the food. Look at the device. Wait five to ten seconds. If the child touches either symbol, celebrate it. If they touch both in any order, that's a combination attempt and it counts.
You can run this pattern across half a dozen daily routines: getting dressed ('shoes on'), bath time ('more bubbles'), outdoor play ('go fast'), reading ('turn page'). Research on naturalistic developmental behavioral intervention suggests distributed practice across multiple daily routines works better than massed drill [4].
Aim for roughly 20 to 30 modeling opportunities per target per day, spread across routines. That sounds like a lot until you realize a five-minute snack can hold eight opportunities if you serve tiny portions.
How long this actually takes
Every parent asks this, and the honest answer is that it varies a lot: nobody has tight population-level data on a clean timeline. Research suggests children typically need to observe a combination modeled in context somewhere between 50 and 200 times before attempting it on their own [3][5]. If you're hitting 20 to 30 modeling opportunities a day across routines, that's roughly 5 to 10 days of work per target before a first spontaneous attempt might show up. Some kids take weeks, some take a month, and a few surprise you on day three.
Things that move the timeline: the child's current language level, how consistently the AAC system is available, how motivated they are by the specific vocabulary targets, and whether they have a motor-planning challenge like childhood apraxia of speech that makes device navigation harder.
If after six to eight weeks of consistent daily practice across multiple routines a child still isn't attempting combinations, bring your data to your SLP and talk about whether the device layout, the vocabulary targets, or the modeling strategy needs to change. Don't just wait longer.
Using wait time without causing a meltdown
Wait time is exactly what it sounds like: you pause after modeling, look at the device or your child expectantly, and give them room to respond before jumping in. Speech-language pathologists often recommend five to ten seconds of silence, which feels agonizingly long in real life.
It works because children with language delays often need more processing time than typical children, and adults have a strong instinct to fill silence by repeating themselves, restating the question, or just handing over the thing the child wanted. All of that short-circuits the attempt before it starts.
To do it without causing frustration, make sure the child is regulated first. If they're already upset, wait time tips into meltdown, so pick calm, motivated moments instead. Get physically close, make eye contact, and look at the device together. If the child reaches for something without touching the device, gently redirect: "can you tell me? I see the button right here." Then wait again. If they touch one symbol, prompt "what do you want more of?" and point to the 'more' symbol, modeling the second word yourself.
The goal isn't to withhold anything from your child. It's to give them a reason to communicate. There's a real difference between the two, and it shows in how the child responds over time.
Should you use errorless learning or prompting with a hierarchy?
Both approaches work, and your SLP should help figure out which one fits your child. Errorless learning means you provide so much support up front that the child almost can't fail: you model the combination, guide their hand to the symbols, and give the reward right away, keeping the error rate near zero. This suits kids who get very distressed by mistakes, or who are still learning that the device does something at all.
A prompting hierarchy works differently. It starts with the most natural prompt, expectant waiting, and only escalates if the child doesn't respond: wait, then an indirect verbal cue ("tell me what you want"), then a direct one ("say more juice"), then pointing, then physical guidance. The idea is to use the least support necessary so the child has room to initiate on their own.
For two-word combination goals specifically, a time-delay procedure has good research support [5]. You start by modeling the combination with zero delay and no expectation of imitation, then gradually add a pause: three seconds after a few weeks, then five, then more. The child learns that the pause is their cue to try.
If your child also shows patterns of echolalia, a prompting hierarchy can sometimes pull more echolalic responses than genuine communication. In those cases, errorless learning paired with natural environment teaching tends to produce more functional communication.
Which AAC systems work best for two-word combinations?
The best system is the one your child will actually use, that carries enough vocabulary to combine, and that your family can learn quickly. For combination goals, a device needs at minimum a full core vocabulary (the 200 to 400 most common words used across contexts), easy access to action words and modifiers, and a layout your child can get through without too many page turns per combination.
High-tech options like Proloquo2Go, TouchChat, and Snap Core First are built around core vocabulary and allow customization. Low-tech options, like a PECS binder or a paper core board, can support combination too, since a child can simply pick up two cards and hand them over together. Our guide to AAC devices breaks down the options in more detail.
Watch out for apps marketed to parents that are really vocabulary-drill tools rather than communication systems. They let kids label pictures but skip the action words and modifiers needed to combine. Before choosing an app, check whether words like "want," "more," "go," "stop," "big," "my," and "help" sit on the home screen or close to it. If it takes three pages to find "go," the combination "go outside" will be nearly impossible to use spontaneously.
How do you coordinate home practice with your child's speech therapist?
The strongest AAC outcomes happen when home practice and therapy pull in the same direction. That sounds obvious, but it's surprisingly rare in practice.
Ask your SLP for three things at each appointment: which two or three combinations are the current targets, exactly how to model them on the device (including which pages the words live on), and what data they'd like you to collect at home. Data collection can be as simple as a tally sheet on the fridge: mark each time you model the target combination, and mark each time the child attempts any part of it. After a week you'll have real information about whether practice is happening and whether the child is moving.
If your child receives early intervention services under IDEA Part C (birth to age three) or school-based services under Part B (ages three to twenty-one), the AAC system and communication goals should show up in the IFSP or IEP. Under IDEA, AAC counts as a related service, and families have the right to be involved in goal setting [6]. If a local SLP isn't available right now, online speech therapy has grown quite a bit and can be a real option for AAC coaching, especially for parents who need guidance on modeling strategy rather than hands-on device work.
What if my child resists the device or throws it?
This is one of the most common things parents run into, and it's rarely about the device itself. It usually comes down to one of four things: the device is too hard to navigate, the targets aren't motivating, the prompting feels coercive, or the device only shows up at "therapy time" instead of being around all day.
Start by checking availability. AAC works best when it's present constantly, like glasses. A device that only comes out for practice becomes something the child links with work rather than with getting what they want. Next, drop the pressure. Many families accidentally turn AAC into a test ("show me where more is") instead of a tool, and if the child associates the device with being put on the spot, avoiding it makes complete sense. Go back to pure modeling with no expectation of a response for a week: just show them what you'd say on the device and move on. Also make sure the vocabulary is reachable. If the child throws the device after three seconds, they may be frustrated that they can't find the word they want, and a simpler layout with fewer pages often beats a full 60-symbol grid for a child just starting combinations.
If a child is being assessed or already has a diagnosis on the autism spectrum, resistance to the device sometimes connects to sensory sensitivities (sounds too loud, screen too bright) or to dysregulation that needs addressing before communication can happen. Your SLP, and your OT if one is involved, should know about the resistance.
Are there specific strategies for kids with apraxia who also use AAC?
Yes, and this is an area where parents often get conflicting advice. Children with apraxia of speech struggle with the motor planning speech sounds require. AAC can reduce that demand while the motor learning work continues; the two aren't in competition. The American Speech-Language-Hearing Association supports using AAC and speech therapy at the same time for children with apraxia [7].
For two-word combinations, kids with apraxia often find that combining symbols on a device is easier than combining spoken words, since touching two icons is more predictable than coordinating articulation for two words. Combinations can emerge sooner on the device than in speech, which is worth celebrating rather than worrying over.
The strategy shift for apraxia is paying attention to motor teaching on the device too: consistent placement of core words, repeating the same motor sequence over and over, and minimizing navigation that forces re-learning where things are. Many children with childhood apraxia of speech benefit from heavy repetition of the same motor pattern, so drilling a single combination like "want more" across six routines a day fits naturally with apraxia treatment principles. Some families use Little Words (littlewords.ai) as an at-home modeling tool between SLP visits, partly because it adapts vocabulary targets based on what the child is working toward, and that daily repetition matters.
How do you know if the AAC strategy is working?
Progress in two-word combinations looks different than most parents expect. Don't wait for your child to use a combination spontaneously and correctly in a brand new situation, that's the end goal, not the first sign of progress. Earlier signs include touching the device at all when they want something, watching your hand when you model on it, using one symbol from a target combination (touching "more" even though you were targeting "more juice"), or attempting two symbols in any order.
A simple tracking system works well: every day during your target routine, mark whether the child didn't engage with the device, touched one symbol, or touched two or more symbols. After two weeks, a shift from the first to the second, or the second to the third, means the approach is working even if the combination isn't perfect yet. If you see no change across three or four weeks of consistent practice (20-plus modeling opportunities a day), something needs to change, whether that's the vocabulary targets, the device layout, the motivation in the context, or the language level the goals assume. Bring your data to that conversation with your SLP.
The Vineland Adaptive Behavior Scales and the MacArthur-Bates Communicative Development Inventories are two standardized tools your SLP might use to track communication growth over time [8][9]. Both are research-validated and give you a baseline to measure against.
Most kids combining words with AAC start doing so around the same window as typical spoken language, roughly 18 to 24 months [1], though the timeline for AAC users varies more. If your child is past 24 months and still stuck on single symbols (or isn't using symbols at all), that's worth raising with an SLP now rather than later. AAC intervention tends to work better the earlier it starts, so there's no real benefit to waiting for a child to seem "ready." You don't need a high-tech device to get started. A low-tech core vocabulary board works fine for two-word combinations, since the child can just point to or pick up two symbols in sequence. Some families actually prefer starting low-tech to build the concept before moving to a high-tech system. What matters more than the technology is that both target words are easy to find, not buried behind extra pages or clutter. If your child uses single words but won't try two, you're probably not doing anything wrong. Going from one word to two is a real cognitive leap and takes time. The most common issue is expecting combinations before modeling them enough. Make sure you're modeling the target combination 20 or more times a day across different routines before expecting your child to attempt it themselves, and double check that both words are easy to reach on the device layout. Most SLPs begin two-word combination work once a child has a functional vocabulary of around 50 words, whether spoken or on AAC. That threshold lines up with when combinations tend to emerge naturally in typical development [1]. Many therapists start modeling combinations earlier than that, though, so the child has heard plenty before being expected to produce them. You can absolutely do the daily modeling and practice yourself, and it makes a real difference. But an SLP should be the one setting goals, choosing the device or board layout, and tracking progress. School-age children, and infants and toddlers under IDEA Part C, are guaranteed access to speech-language services if they qualify [6]. If in-person services aren't available to you, telehealth SLP coaching for AAC is a legitimate option. Don't stop working on speech to focus only on AAC. They're not in competition: research consistently shows AAC supports speech development rather than replacing it [2]. A good SLP builds both at once, working on speech motor skills alongside AAC combination use. Most children end up using a mix of speech and AAC as their skills grow, and that's the goal, not an either-or choice. As for PECS versus core vocabulary AAC: PECS was originally built around exchanging single pictures for items, and while its Phase IV introduces sentence strips for combinations, the vocabulary stays mostly fringe (specific items) rather than core. Core vocabulary AAC gives kids words like "want," "go," "more," and "stop" that combine across all kinds of contexts, which is why most SLPs now see it as the stronger foundation for teaching combinations. If your child's IEP doesn't mention AAC, request a meeting and ask for an AAC evaluation to be part of it. Under IDEA, the team is required to consider AAC for any child with a communication need [6]. Bring what you have: a list of your child's current communication attempts, any SLP reports, and a written request. You don't need to be the expert in the room, just the one asking whether AAC supports would help your child reach their communication goals. If you're not confident using the device yourself, ask your SLP for a 15-minute tutorial focused only on your target words. You don't need to learn the whole system. Find where "more," "want," "go," "stop," "my," and five or six high-frequency nouns live, and practice reaching them quickly. Most families get comfortable with just three to five two-word targets without ever mastering the full layout, so start small and build your own fluency as your child builds theirs. There's research on AAC for late talkers beyond autism, too, even though much of the field's research has historically centered on children with complex communication needs including autism. A 2021 systematic review in the Journal of Speech, Language, and Hearing Research found positive outcomes for naturalistic AAC intervention across diverse populations [4], and ASHA's evidence map on AAC includes children with developmental language disorder, Down syndrome, and idiopathic late talking as well. "Full AAC" refers to a system with a large core vocabulary, typically 200-plus words, organized so a child can say nearly anything they want, not just request items. For two-word combination goals, you need at least a solid core of action words, modifiers, and pronouns alongside nouns. A system with only 20 to 30 symbols meant for requesting isn't really built for combination work. To track progress at home between appointments, keep a simple daily log during your target routine: note whether your child didn't engage with the device, used one symbol, or used two or more. Add a 30-second video once or twice a week during that routine. Bringing the log and a clip to each SLP appointment gives your therapist enough real data to spot trends and adjust the approach as needed. This article is for general information and isn't a substitute for an individual evaluation or advice from your child's speech-language pathologist.Sources
- ASHA, Late Language Emergence: Most typically developing children begin combining words between 18 and 24 months
- ASHA, Augmentative and Alternative Communication: AAC does not slow speech development and often supports it
- Drager et al. (2012), American Journal of Speech-Language Pathology, 'Aided language modeling with children with autism': Children with complex communication needs showed significantly more symbol use after aided language input compared to instruction without modeling
- Alzrayer et al. (2021), Journal of Speech, Language, and Hearing Research, systematic review of naturalistic AAC intervention: Naturalistic AAC intervention produced positive outcomes across age groups and AAC types; distributed practice across routines more effective than massed drill
- Schlosser & Wendt (2008), American Journal of Speech-Language Pathology, effects of augmentative and alternative communication on speech production: Children typically need to observe a combination modeled in context between 50 and 200 times before attempting it independently; time-delay procedure is well-supported
- U.S. Department of Education, Individuals with Disabilities Education Act (IDEA): AAC is considered a related service under IDEA; families have the right to be involved in goal setting; IDEA Part C covers birth to age three, Part B covers ages three to twenty-one
- ASHA, Childhood Apraxia of Speech: ASHA supports concurrent use of AAC and speech therapy for children with apraxia of speech
- Vineland Adaptive Behavior Scales, Third Edition, Pearson Assessments: The Vineland Adaptive Behavior Scales is a standardized tool used to track communication growth over time
- MacArthur-Bates Communicative Development Inventories (MB-CDI), Stanford: The MacArthur-Bates CDI is a research-validated tool used to establish communication baselines and measure growth
- American Academy of Pediatrics: The AAP recommends referral for speech-language evaluation if a child is not using two-word combinations by 24 months