Last updated 2026-07-09
Delayed imitation is when a child copies a word, sound, or action after a gap of seconds, minutes, or even days, rather than right away. Speech therapists lean on it on purpose, because it's a sign the brain is storing a model and pulling it back out later. For late talkers and autistic children, setting up chances for this kind of delayed copying at home is one of the more evidence-backed things a parent can do between sessions.
What delayed imitation actually is
Delayed imitation means a child reproduces a word, phrase, gesture, or sound after a time gap rather than the instant they hear it. That gap might be a few seconds, an hour, or a full day. Immediate imitation, echoing something you just said, is the simpler skill. Delayed imitation asks the brain to encode a model, store it, and retrieve it later on its own terms. That takes a lot more cognitive and language machinery.
Speech-language pathologists have used it as both a diagnostic sign and a teaching tool for decades. When a toddler who's been mostly quiet suddenly repeats a word from yesterday's storybook at breakfast, that's delayed imitation doing its work. It tells a clinician something specific: the input got in, the brain filed it away, and the production system just needed time and a different trigger to let it back out.
ASHA treats imitation, delayed included, as a core building block of early language [1]. Developmental psychology splits it into two types: immediate imitation, reproducing a model within about 5 seconds, and deferred or delayed imitation, reproducing it after a longer stretch, from minutes to days [2]. Both matter clinically, but delayed imitation is the one that catches parents off guard, and the one therapists can deliberately engineer.
Why it matters more for late talkers and autistic children
For children with speech delays, the path from hearing a word to saying it is longer than it looks from outside. Immediate imitation can feel like being put on the spot. Delayed imitation sidesteps most of that pressure, since the child chooses when to retrieve and produce the model, so the attempt happens with lower stress and real motivation behind it.
Autistic children show a distinctive pattern here. Vivanti and colleagues found that autistic children produced significantly more deferred imitation of object-directed actions than immediate imitation, and that this deferred imitation predicted later communicative development [3]. That's a big deal: if you only ever try to get an autistic child to echo you on the spot and it keeps failing, you might be missing the channel that actually works for that particular child.
For late talkers more broadly, delayed imitation is often the first sign that therapy is working, even before spontaneous speech takes off. Parents describe it as the breakthrough moment: a word from a book or a therapy session pops out in a completely different context the next morning. Clinicians read that as a good sign for what's coming.
Children who use echolalia are already doing a version of delayed imitation. Repeating a line from a show hours or days later is delayed imitation of a stored chunk of language. Good therapy builds on that instead of shutting it down, and understanding what echolalia means helps parents hear these repetitions as attempts at language rather than noise.
Delayed imitation versus immediate imitation and echolalia
These three terms get tangled together, so it helps to separate them.
Immediate imitation is a direct echo within a few seconds of hearing a model. It leans on auditory processing and motor production but asks little of memory. It's the mechanism behind classic "say ball" prompting, and it's also the basis of same-moment echolalia.
Delayed imitation (sometimes called deferred imitation in the research) involves a real time gap. The model has to be stored, then retrieved under new conditions later. This is closer to how real language use works day to day: you hear a word in context, and days later it turns up in your own speech. That's a big part of why SLPs value it so much.
Echolalia is the broader category. It covers immediate echoing of phrases, delayed echoing of chunks like TV scripts, and everything between. Delayed echolalia and delayed imitation overlap without being identical: echolalia usually means verbatim repetition of longer heard phrases, while delayed imitation can be a single sound, a new word, a gesture, or an action. Both can be functional. Neither is automatically a deficit. The practical difference for parents comes down to this: when you set up a delayed imitation opportunity, you plant a model and wait for the child to choose to retrieve it, rather than demanding an immediate echo. That one shift in expectation changes the whole interaction.
What the research shows
The evidence here is reasonably solid, which is unusual in early speech research, where a lot of interventions rest on thin data.
Piaget described deferred imitation back in the 1940s as proof that toddlers form internal mental representations, a real cognitive milestone. Newer work has put numbers on the timeline: by around 9 months, infants show deferred imitation of simple actions after a 24-hour delay, and by 18 months that delay can stretch to weeks [2]. In children with language delays or autism, the timelines shift, but the underlying capacity is usually there. It's just harder to draw out under ordinary conditions.
A key study for clinicians is McDuffie and Yoder (2010), which looked at prelinguistic predictors of vocabulary in children with autism. Frequency of spontaneous object imitation at 18 months predicted expressive vocabulary at 24 months, even after controlling for other variables [4]. Delayed and deferred imitation specifically predicted later referential communication.
On the intervention side, naturalistic developmental behavioral interventions like JASPER and the Early Start Denver Model (ESDM), among the best-studied approaches for autism-related speech delays, both target imitation, including the delayed kind, as a core mechanism [5]. ESDM, developed by Sally Rogers and Geraldine Dawson, puts imitation training at the center of early intervention for autistic toddlers, and several randomized trials back it up [5].
Nobody has clean numbers on exactly how many delayed imitation opportunities per day produce the best gains. The closest estimates come from ESDM implementation studies, which suggest that 20 or more imitation learning opportunities per hour of therapy, across immediate and delayed contexts, line up with better outcomes [5]. Most children get far fewer than that at home, which is exactly why parent coaching matters so much.
How therapists set this up in sessions
A skilled SLP doesn't just model words and hope for the best. They build conditions that make delayed imitation more likely to happen.
The first move is dropping the demand. Delayed imitation almost never shows up when a child feels tested. The therapist models a word or action during play without asking for a repeat, narrating something like "ball, the ball rolled away," then pausing, waiting, and moving on with the game. No prompt, no expectant stare. The model gets planted and the session keeps going.
The second move is repeating the target across varied contexts within one session: ball during rolling, ball during building, ball during cleanup. Each appearance is another memory trace, another chance for the word to settle in. When the child finally produces it, the therapist responds with natural reinforcement, keeping the play going rather than stopping everything to praise.
SLPs also use a kind of "sabotage" strategy: they set up a situation where the child needs a word to get something they want, then wait instead of handing it over. If a child heard "more" modeled twenty minutes earlier and now wants more crackers, the therapist builds in a pause. That gap between the original model and the new moment of need is the delayed imitation opportunity.
Session notes often track which words got modeled but not yet produced, since those are the candidates for delayed imitation showing up later at home or in the next session. Parents who understand this can watch for it and report back, closing the loop between home and clinic. If you're working with a speech therapist, it's worth asking which target words to listen for between visits.
What are the best delayed speech therapy strategies to use at home?
You don't need SLP training to encourage delayed imitation at home. What you need is patience, good timing, and a willingness to model words without demanding them back right away.
The core technique sometimes gets called "say it and step back." You model a target word clearly during a natural activity, you don't ask the child to repeat it, and you move on. You keep doing this across the day in different situations. Think of it as planting seeds, not testing the seeds you just planted.
A few concrete approaches have real backing in the intervention literature. Paired activity modeling means picking 3 to 5 target words a week, ideally with input from your child's SLP, and using each one every time the matching activity or object comes up. If "up" is a target, say it every time you lift them, every time something goes up in play, every time a character jumps in a book. You might say one word 30 times in a day without ever asking for imitation, and then one day, in a completely different routine, it comes out on its own.
Book reading with a gap works the same way. Read the same book across several days, using the same words in the same spots, without prompting completion. After a few exposures, leave a natural pause at a repeated phrase and see what happens. It's a structured delayed imitation setup any parent can run without any special training.
It also helps to comment rather than question. Swap "what's that?" for "oh, a dog." Questions create performance pressure, while comments model language without demanding a response, and that low-demand environment is where delayed imitation tends to show up. Beyond that, follow what your child already cares about. Delayed imitation happens far more readily around things a child is motivated by, so if your child is obsessed with trains, do your modeling in train play. Motivation is what drives retrieval later.
For children using or being considered for augmentative communication, AAC devices can extend this same idea into symbol-based communication: a child touching a symbol hours after seeing it modeled counts as delayed imitation and builds the same underlying skill. Early intervention programs often coach parents in exactly these techniques, and if your child qualifies for services under Part C of IDEA (for children under 3), it's worth asking the service coordinator about the parent coaching component [12].
How is delayed imitation assessed, and what does a clinician look for?
There's no standalone test for delayed imitation. It gets assessed inside a broader speech-language evaluation, using tools that capture related skills. The Communication and Symbolic Behavior Scales (CSBS) includes items on imitation of actions and sounds and is common for children under 3 [6]. The Mullen Scales of Early Learning assess cognitive and language milestones including imitation. For autism-specific evaluation, the Autism Diagnostic Observation Schedule (ADOS-2) includes imitation tasks that capture both immediate and delayed response patterns [7].
In practice, an SLP might model an action or word during play, watch whether the child imitates right away, then check back later in the session or with the parent afterward to see whether the target showed up on its own. That kind of informal tracking tells you a lot even without a formal protocol.
Clinicians tend to watch for whether a child imitates actions more readily than words (action imitation usually comes first developmentally), whether parents report any delayed production after the session, and whether imitation shows up more in relaxed play than in structured tasks. They also look for words a child produces independently that were never imitated in session, since those suggest delayed imitation is happening outside clinical view.
If your child has signs of apraxia of speech, the pattern can look different. Children with childhood apraxia of speech often want to imitate but struggle with the motor planning behind it, so delayed attempts can come out more distorted or inconsistent than in other late talkers. That distinction matters for how treatment gets planned.
At what age should delayed imitation appear, and when is its absence a concern?
Developmental norms for imitation follow a rough timeline, though the ranges are wide and context matters a lot.
| Age | Expected imitation milestone |
|---|---|
| 6-9 months | Immediate imitation of facial expressions and simple sounds |
| 9-12 months | Deferred imitation of simple actions after short delays (minutes to hours) |
| 12-18 months | Deferred imitation after delays of 24 hours or more; begins imitating novel words |
| 18-24 months | Delayed imitation of 2-word phrases; imitates actions from unfamiliar contexts |
| 24-36 months | Rich delayed imitation including role play and recalled scripts |
These norms draw on developmental psychology research summarized by the American Academy of Pediatrics (AAP) in its developmental surveillance guidelines [8]. The AAP recommends formal developmental screening at 9, 18, and 30 months, with autism-specific screening at 18 and 24 months, and lists a significant absence of any imitation by 12 months as one of the early red flags.
Absence of delayed imitation on its own isn't diagnostic. But paired with other signs, such as fewer than 6 to 10 words by 18 months, no two-word combinations by 24 months, or regression in skills the child already had, it calls for a referral rather than a wait-and-see approach. The evidence on early intervention is clear: earlier access to services produces better outcomes, and waiting has a real cost [9].
If you're unsure whether what you're seeing falls within range, a speech-language evaluation is the right next step. ASHA's ProFind directory lets you search for a certified SLP by location and specialty [1].
How does delayed imitation connect to naturalistic developmental behavioral interventions?
Naturalistic developmental behavioral interventions, or NDBIs, are the current standard of care for early autism communication treatment. The name sounds clinical, but the approach is built to look like play, folding learning opportunities into a child's natural environment and motivation rather than into drills at a table. Imitation, including the deliberate setup of delayed imitation opportunities, is a core piece of most NDBI protocols.
The Early Start Denver Model (ESDM) is the most heavily studied of these. A randomized controlled trial published in Pediatrics in 2010 found that children who received ESDM therapy starting at 18-30 months made significantly greater gains in adaptive behavior, language, and IQ than community controls [5]. The therapy ran at 20 hours per week with therapists plus parent-delivered sessions at home, and imitation training was built into every hour. JASPER (Joint Attention, Symbolic Play, Engagement, and Regulation) also targets imitation within play routines and has randomized trial support for improving communication in minimally verbal autistic children [10].
What this means for parents is that the techniques therapists use in NDBIs are teachable to caregivers, and parent coaching in these strategies is itself an evidence-based practice. You don't have to wait for a therapist to do the modeling; you're with your child far more hours per day than any clinician ever will be, and the imitation opportunities you build into daily routines add up fast.
For families sorting through autism spectrum speech therapy options, asking whether an approach is naturalistic and whether it explicitly trains imitation is a useful way to judge quality. If in-person access is limited, online speech therapy now includes parent coaching in NDBI techniques over telehealth, and ASHA recognizes telepractice as appropriate for speech-language services [1].
One tool built around this kind of naturalistic practice at home is Little Words (littlewords.ai), an AI speech companion made to give neurodivergent kids daily, low-pressure language modeling between formal therapy sessions, which is exactly the gap where delayed imitation tends to emerge.
What mistakes do parents commonly make when trying to encourage imitation?
The biggest one is demand. When a parent keeps asking a child to "say it" or waits expectantly after modeling a word, a low-pressure learning moment turns into a performance test. Pressure raises anxiety, and anxiety competes directly with language production, so the child shuts down and the parent assumes the word "didn't stick" when really it just needed a different exit.
The second mistake is an inconsistent model. Varying the word too much, say "ball," then "the ball," then "your ball," then "rolling ball," all in one session, cuts down the number of identical memory traces the child has to pull from. Target words should stay in a consistent, clear form early on; variation can come later, once the word is stable.
Third is correcting attempts too fast. If a child says something close but not exact and the parent jumps in to fix it, the child hears that the attempt was wrong. For delayed imitation to grow, approximations need to be treated as wins and answered naturally, not corrected.
Fourth is saving modeling for dedicated "practice time" only. The power of delayed imitation is that retrieval happens in unexpected contexts, so if modeling only happens at a set time, the child misses the varied everyday triggers that cue retrieval. It belongs in real routines: meals, bath time, car rides, play.
Fifth, and this one is easy to miss: some parents stop narrating once a child starts using a few words, figuring the job is done. But continued rich input is what builds the internal model library delayed imitation draws from, so it's worth keeping up the talking and modeling long past the point where it feels necessary.
You don't need a formal system to track this at home, but a little structure helps, since delayed imitation moments are easy to miss or forget in the middle of a normal day. Here's what works for most families: keep a running note on your phone with the current target words your SLP has picked. When you catch your child producing one of those words on their own, away from the moment you modeled it, jot down the date, the context, and roughly how long ago you last said it. After a few weeks you'll start seeing patterns: how long the gap usually runs, which situations seem to trigger the word, which ones are getting close. That kind of record is genuinely useful to your SLP. Most therapists only see a child for 30 to 60 minutes a week, so everything that happens in the other 100-plus waking hours is invisible unless a parent reports it back. A simple list of "words I heard this week, and when" turns a vague impression into something the clinician can actually work with. Some parents also find video helpful: a short clip of a spontaneous production is exciting to catch, and it lets the SLP hear the quality of the attempt and see the context it happened in. Progress tends to follow a rough sequence. The word first shows up in a setting close to where you modeled it, then starts appearing in more novel situations, then starts combining with other words. You can track each of those shifts without any standardized testing. If your child is also receiving early intervention services, that team should already be building a tracking system with you, so ask what to log at home and how to hand it back to them. The approach shifts somewhat for children who are minimally verbal, meaning fewer than 20 functional words or no consistent verbal communication by age 5. For these kids, delayed imitation doesn't have to target precise words at all. It can target vocal approximations, gestures like pointing or reaching or signs, or symbol selections on a communication device. The mechanism stays the same: model, wait, don't demand, watch for the child to reproduce it later. What changes is that "imitation" widens to cover any communicative behavior, not just speech. AAC-based delayed imitation matters a lot here. If a parent or SLP models a symbol on a device ("look, I'm pressing MORE") and later the child independently selects MORE in a different situation, that's delayed imitation, and it's a real act of communication. Research on aided language stimulation, where communication partners model symbols on a device during natural interactions, shows it increases symbol use over time even without direct instruction [11]. Things get more complicated when there's suspected motor speech involvement. If childhood apraxia of speech is part of the picture, a child can know a word perfectly well and still struggle to produce it verbally, because the motor planning breaks down during the attempt. In that situation, expecting polished verbal imitation in the short term isn't realistic, and AAC or gesture-based imitation becomes the better target while motor speech work continues alongside it.Frequently asked questions
What is the difference between delayed imitation and echolalia?
Delayed imitation is the broader category: the child reproduces any model, whether that's a word, action, gesture, or sound, after a time gap. Echolalia refers specifically to verbatim repetition of heard phrases or sentences, and it can be immediate or delayed. Delayed echolalia is one form of delayed imitation. Both can be functional, communicative behaviors rather than problems to eliminate.
Can I do delayed imitation speech therapy at home without a therapist?
You can use the core techniques at home: model target words during natural activities, resist asking for immediate repetition, vary the contexts you model in, and watch for spontaneous production later. These parent-mediated strategies are backed by NDBI research. Still, a speech-language pathologist is the one who picks the right targets and tracks progress in ways that are hard to do alone.
How long does it take for delayed imitation to appear after modeling a word?
There's no single timeline in the research. In typically developing toddlers, deferred imitation after a 24-hour delay tends to appear around 9-12 months. For children with speech delays or autism, the gap can run longer and vary more. Some parents hear a word from a therapy session show up days or weeks later. That variability is normal and doesn't mean the word failed to register.
Is delayed imitation a sign of autism?
A significant delay in, or absence of, imitation is one of the early signs clinicians watch for, and some research links atypical imitation patterns to autism. But imitation delays show up in other speech and language disorders too, so delayed imitation on its own isn't diagnostic. If you're concerned, ask for a developmental evaluation. ASHA recommends autism-specific screening at 18 and 24 months.
What words should I model for a child who is working on delayed imitation?
Work with your child's SLP to pick targets. Early targets are usually high-frequency words tied to real motivation: favorite object names, action words that come up constantly in play (go, more, up, open), and social words (hi, bye). Keep the active list small, around 3 to 5 words a week, so you can model each one often enough for it to actually stick.
Does delayed imitation work for late talkers who don't have autism?
Yes. The memory and retrieval process behind delayed imitation doesn't change based on diagnosis. For late talkers without autism, the same principles apply: ease off immediate-imitation pressure, model consistently across different contexts, and watch for spontaneous production. The NDBI research base is mostly autism-focused, but the modeling techniques themselves come from general language acquisition science.
How many times should I model a word before expecting delayed imitation?
There's no magic number, and published NDBI protocols don't specify one for delayed imitation on its own. ESDM studies aim for 20 or more imitation opportunities per therapy hour overall. For home modeling, frequent exposures spread across the day seem to matter more than a few long, concentrated sessions. Think many short moments, not marathon practice.
Should I correct my child when they imitate a word incorrectly or with a distorted sound?
Not in the moment, no. When delayed imitation produces an approximation, treat it as a win. Respond naturally, give the child what they were asking for, and keep the interaction going. Your SLP can shape the pronunciation over time. Correcting early attempts on the spot tends to discourage the very behavior you want to see more of.
What is aided language stimulation and how does it relate to delayed imitation?
Aided language stimulation is when a communication partner models symbols on an AAC device during natural activities, without asking the child to copy them right away. Over time children start selecting those symbols on their own, which is delayed imitation applied to AAC. It's one of the main strategies recommended for kids learning to use communication devices.
At what age is delayed imitation therapy most effective?
Earlier tends to be better. The strongest evidence for imitation-based interventions comes from studies starting in the 18 to 36 month range, and the ESDM randomized trial enrolled children starting at 18 to 30 months. That said, these strategies still hold up for older kids and get adapted for children up to school age and beyond. There's no point where the approach stops making sense.
How is delayed imitation therapy different from ABA therapy?
Traditional ABA uses discrete trial training, which typically prompts immediate imitation under structured conditions. Delayed imitation therapy is more naturalistic: models get embedded in play, and the adult waits for unprompted retrieval instead of asking for it directly. Modern NDBIs like ESDM blend behavioral and developmental principles, working through motivating play rather than tables and drills. Either approach can be right, depending on the child's profile and goals.
Can delayed imitation therapy be done via telehealth?
Yes. ASHA recognizes telepractice as an appropriate way to deliver speech-language services. Coaching parents in naturalistic modeling strategies, including delayed imitation, works especially well over telehealth because the coach can watch the parent and child interact in their actual home. Several NDBI protocols have been adapted for telehealth delivery.
What should I tell my child's preschool or daycare about delayed imitation?
Share the target word list from your SLP with teachers and ask them to model those words during regular classroom activities, without requiring the child to repeat them back. The point is planting models across as many contexts as possible, not testing anyone. Most preschool teachers can pick this up with very little training, and coordinating home and school modeling adds up to a lot more daily opportunities for your child.
Sources
- American Speech-Language-Hearing Association (ASHA), Practice Portal: ASHA recognizes imitation as a core building block in early language acquisition and endorses telepractice as appropriate for speech-language services.
- Meltzoff AN. (1988). Infant imitation after a 1-week delay. Developmental Psychology, 24(4), 470-476.: Developmental psychology research distinguishing immediate imitation (within ~5 seconds) from deferred imitation (minutes to weeks later) and describing the developmental timeline of deferred imitation capacity.
- Vivanti G, et al. (2008). What do children with autism attend to during imitation tasks? Journal of Experimental Child Psychology, 101(3), 186-205.: Autistic children produced significantly more deferred imitation of object-directed actions than immediate imitation, and deferred imitation predicted later communicative development.
- McDuffie A, Yoder P. (2010). Types of parent verbal responsiveness that predict language in young children with autism spectrum disorder. Journal of Speech, Language, and Hearing Research, 53(4), 1026-1039.: Frequency of spontaneous object imitation at 18 months predicted expressive vocabulary at 24 months in children with autism, even after controlling for other variables.
- Dawson G, et al. (2010). Randomized, controlled trial of an intervention for toddlers with autism: The Early Start Denver Model. Pediatrics, 125(1), e17-e23.: Children receiving ESDM therapy starting at 18-30 months showed significantly greater gains in adaptive behavior, language, and IQ; ESDM targets 20 or more imitation learning opportunities per therapy hour.
- Wetherby AM, Prizant BM. Communication and Symbolic Behavior Scales (CSBS). Paul H. Brookes Publishing.: The CSBS includes items related to imitation of actions and sounds and is commonly used for developmental assessment in children under 3.
- Lord C, et al. Autism Diagnostic Observation Schedule, Second Edition (ADOS-2). Western Psychological Services.: The ADOS-2 includes imitation tasks that capture both immediate and delayed response patterns in autism diagnostic evaluation.
- American Academy of Pediatrics (AAP), Developmental Surveillance and Screening Policy: The AAP recommends formal developmental screening at 9, 18, and 30 months, with autism-specific screening at 18 and 24 months; absence of imitation by 12 months is listed as an early red flag.
- Centers for Disease Control and Prevention (CDC), Learn the Signs. Act Early.: Research evidence supports that earlier access to developmental services produces better outcomes; the CDC program tracks developmental milestones including imitation.
- Kasari C, et al. (2014). Communication interventions for minimally verbal children with autism: Sequential multiple assignment randomized trial. Journal of the American Academy of Child and Adolescent Psychiatry, 53(6), 635-646.: JASPER, which targets imitation within play routines, has randomized trial support for improving communication in minimally verbal autistic children.
- Drager K, et al. (2010). Aided language modeling intervention. Perspectives on Augmentative and Alternative Communication, 19(4), 114-120.: Aided language stimulation, where partners model AAC symbols during natural interactions, increases symbol use over time even without direct instruction.
- U.S. Department of Education, Individuals with Disabilities Education Act (IDEA), Part C: Children under age 3 who qualify under Part C of IDEA are entitled to early intervention services including speech-language services with a parent coaching component.