10-Minute Speech Practice That Doesn't Require Sitting Still

Last updated 2026-07-11

What is least to most prompting in speech and communication?

Least-to-most prompting (written L2M or LTM) means you give the mildest cue first, wait, and add a stronger prompt only if the child stalls or errors. The usual ladder runs: expectant pause, gestural cue, verbal model, partial physical assist, full physical assist. You stop the instant the child responds.

The child gets first crack at the skill with zero help, every single time. That first shot at independence is the entire point. A pause of three to five seconds is the usual floor before you climb to the next rung, though some clinicians hold longer for kids who process slowly ^[1].

Say you want a child to request "more bubbles." With L2M you hold up the wand and wait in silence. Then you raise your eyebrows. Then you whisper "more." Then you say "more bubbles." Only last do you guide the child's hands to an AAC device. Once the routine is familiar, most kids never reach that last rung.

L2M sits on the error-correction side of the fence, opposite errorless learning, and it shows up all over the speech therapy literature as the default for teaching new communicative functions like requesting and commenting.

What is most to least prompting and how is it different?

Most-to-least prompting (M2L or MLM) runs the ladder backward. You open with the highest support the child will ever get, often full physical guidance or a complete verbal model delivered before the child can try alone. Then you pull that support back on a schedule, trial by trial or session by session.

The logic is early error prevention. If a child has a history of drilling the wrong word form, the wrong motor pattern for apraxia, or a botched sign, opening with maximum support blocks those errors before they get reinforced. You shape from the top down instead of the bottom up.

For a child learning a consonant-vowel-consonant word with childhood apraxia of speech, M2L might run like this: the clinician moves the child's jaw and models the word at the same time (full support), then models without touching, then drops a partial cue like "starts with /d/," then just waits. Each phase holds until a set accuracy criterion is met, usually 80 percent correct across two or three sessions in a row.

Here independent performance comes last, not first. That is not a cosmetic difference. It changes what error data you collect, when you fade, and how the child's nervous system rehearses the skill.

Which prompting approach does the research actually support?

Honest answer: both have solid evidence, and it's fragmented because the kids studied differ so much.

A 2013 systematic review in the American Journal of Speech-Language Pathology looked at prompting across AAC interventions and reported that "the system of least prompts produced the largest effects for acquisition of requesting behaviors in learners with autism and intellectual disabilities" ^[2]. That's the strongest endorsement of L2M you'll find in a peer-reviewed source.

M2L's strongest base is motor learning and apraxia treatment. Dynamic Temporal and Tactile Cueing (DTTC), one of the two most evidence-supported treatments for childhood apraxia of speech according to Apraxia Kids, is an M2L framework by design. The clinician opens with simultaneous production and fades cues as accuracy climbs ^[3].

For social communication and AAC use in autism, a 2014 study in the Journal of Applied Behavior Analysis found the system of least prompts generalized to novel settings faster than time-delay procedures, though both beat no-prompt control conditions ^[4].

Nobody has clean head-to-head RCT data pitting L2M against M2L on identical outcomes in identical kids. The nearest work sits in AAC prompting comparisons and behavioral skills training, where L2M tends to edge out for novel communicative requests and M2L edges out for motor-production accuracy. Both results make theoretical sense.

When should you choose least to most over most to least?

Choose L2M when the child already has some component skills and the goal is putting them to use. A child who can physically touch a picture symbol but hasn't learned to touch it to request needs chances to initiate, not more adult modeling. Stacking prompts from the start steals those initiations.

L2M is also the safer default when:

• The target is a communicative function (requesting, commenting, greeting, protesting) rather than a specific motor production.
• The child waits for an adult cue before doing anything, the classic prompt-dependency pattern.
• You're teaching AAC devices access and want spontaneous use, more than device touching in response to an adult model.
• Generalization across settings and partners is the main goal.

Prompt dependency is a real clinical problem in this group. Research in augmentative communication has shown that children handed a model before every communication opportunity often can't use their device once the model disappears ^[5]. L2M fights that by keeping the adult quiet first.

One practical warning: L2M lives or dies on wait time. The three-to-five-second pause feels socially awkward to almost every adult, trained clinicians included. If you're running L2M at home, set a phone timer for the first few weeks. You'll almost certainly find you were jumping in faster than you thought.

When does most to least prompting make more sense?

M2L fits when accuracy of the response form matters more than independence of initiation. Motor speech disorders are the clearest case. For a child with apraxia of speech, letting early attempts run unguided often means drilling wrong motor plans, which makes correct production harder down the line. Opening with maximum support and fading tracks with how motor learning research describes "blocked practice with high accuracy" ^[3].

Other good M2L candidates:

• A child just starting early intervention with almost no baseline skills, where errors are likely without support and there's no old wrong pattern to interrupt.
• Teaching a brand-new AAC symbol with no context cues attached yet.
• Sign production where the handshape is precise and a wrong shape teaches the wrong thing.
• Any goal where error rehearsal genuinely hurts (certain feeding-adjacent communication goals, for one).

M2L also helps when a child's anxiety spikes after errors. Some kids shut down after a mistake, and that shutdown burns more instructional time than starting supported ever would. You won't find that in most prompting papers, but it's the kind of clinical reality that drives real decisions.

The cost is that fading is hard to do consistently. Systematic M2L fading means tracking performance across sessions and moving to the next prompt level on schedule, not on instinct. Doable in a therapy room with session notes. At home, it takes more structure.

How does prompt fading work and why does it matter so much?

Prompt fading is the gradual removal of support as a child gets more accurate. It's the whole game. A prompting hierarchy that never fades is just dependency wearing a different name.

With L2M, fading happens inside each trial on its own: you quit adding cues the moment the child responds. The structure does the work. What you track over time is whether the child keeps responding at the first level (the pause), which tells you the skill is strengthening.

With M2L, fading is a planned schedule across trials or sessions. A common criterion is 80 percent correct at the current level across two straight sessions before you drop to a lighter prompt. Some clinicians hold out for 90 percent on motor targets. The exact percentage matters less than applying it the same way every time: you move on schedule, not because it feels right today.

A mistake both systems share: fading too slowly because the child succeeds at the current level. Success at a heavily prompted level is not generalization. If a child says "ball" correctly only after a full model, they haven't learned to say "ball," they've learned to imitate "ball." Different skills. Push the fade.

Another mistake: inconsistency across partners. If a parent models before the child has a chance and a clinician waits five seconds, the child gets two conflicting learning environments. Coaching communication partners onto the same prompt level is part of the work, not a bonus.

Can you use both hierarchies at the same time?

Yes, and many good clinicians do. The usual split: M2L for new targets still in acquisition, L2M for targets already partly learned or being generalized to new settings and partners.

A child working on consonant-vowel-consonant words with apraxia might get M2L on a new word set in clinic. That same week, for spontaneous requesting on the playground, the same child gets L2M with familiar vocabulary. Different goals, different mechanisms, and the right hierarchy follows the goal.

There's also a hybrid called graduated guidance, where you start a physical prompt at full intensity but fade it mid-trial as the child picks up momentum. It's common in feeding therapy and motor imitation, and it bleeds into communication work. It's neither pure M2L nor pure L2M but borrows from both.

The honest risk in mixing is that parents and paraprofessionals get confused about which mode to run when, especially without a clear reference. Written instructions that spell out "for these targets, wait 5 seconds first" versus "for these targets, model first" are worth putting on paper in any home program.

What about time delay: is that a third option?

Time delay gets treated sometimes as its own strategy and sometimes as a slice of L2M, depending on who's writing. Worth knowing either way, because it comes up constantly in autism spectrum speech therapy and AAC work.

In constant time delay you always wait the same interval (often four seconds) before giving a controlling prompt. In progressive time delay the interval stretches across sessions, starting at zero (you prompt immediately, before any chance to err) and growing from there. Progressive time delay behaves a lot like M2L in its early phases.

A review of communication interventions for autism (Braddock et al., 2013 ^[6]) found progressive time delay to be one of the most consistently replicated strategies for teaching symbol use and verbal requesting. It's simple to run, which probably helps fidelity.

The practical difference from full M2L is that time delay usually leans on one controlling prompt instead of a ladder of levels. If you're a parent running a home program and want a simpler structure, a four-second wait plus one full verbal model is a defensible, evidence-based starting point for most communication goals.

How do these strategies apply to AAC users specifically?

AAC shifts the picture. The research on AAC devices and prompting is pretty clear that spontaneous, unprompted use is the finish line, and that finish line is harder to reach if heavy prompting ran the whole way through acquisition.

A 2018 study in Augmentative and Alternative Communication followed 22 children using speech-generating devices and found that kids whose partners cut back prompted trials faster showed higher rates of spontaneous communication six months later ^[7]. That's a strong case for L2M as the AAC default, at least past the earliest acquisition phase.

For a child just learning to find their way around a board or device, some early M2L makes sense: hand-over-hand to locate a symbol, then partial physical, then a point, then a wait. Once the child can find the symbol alone, shift to L2M or pure time delay right away.

One pattern worth resisting is adult-directed "show me" prompting with AAC. Asking a child to "find the ball" on their device teaches matching, not communication. The goal is child-initiated requesting in real contexts. Set up communication opportunities (bubbles that run out, a preferred snack held back, a favorite book just out of reach) and let the child start with L2M support instead of quizzing comprehension on demand.

If you want an app that mirrors this at home, Little Words builds practice around natural communication routines instead of drill.

What do speech-language pathologists say about prompting in real practice?

ASHA's Practice Portal on Augmentative and Alternative Communication lists the system of least prompts and time delay as evidence-based practices for AAC intervention ^[1]. ASHA doesn't name M2L as an AAC strategy, though it's baked into structured motor learning protocols like DTTC.

For motor speech, the Apraxia Kids treatment guidance notes that DTTC "begins with maximum support and systematically fades" ^[3], which is M2L in plain language.

The practical SLP reality, from published survey data, is that most clinicians run L2M by default and reach for heavier prompting on new or hard targets. A 2016 survey of 312 AAC-using clinicians in Augmentative and Alternative Communication found over 70 percent used L2M or time delay as their primary approach for generalization goals, while about 45 percent used higher-support prompting for initial acquisition ^[8].

The research-to-practice gap is real but narrower in prompting than in some other corners of speech therapy. Prompting hierarchies get taught in most graduate programs, and the logic is intuitive enough that even less formally trained home providers can learn to apply it.

If your child's therapist uses one approach only, it's fair to ask which targets they're applying it to and whether the other approach got considered. That's not second-guessing. It's informed collaboration.

A side-by-side comparison of least to most and most to least prompting

This table lays out the differences across the dimensions that matter most for communication goals.

Neither column is a verdict. Read across the row for your child's specific target, and let that row point you to the column you start in.

How can parents apply prompting hierarchies at home without formal training?

You don't need a graduate degree to use these well. You need two things: a consistent routine and a willingness to wait.

The simplest L2M at home: set up the opportunity, count silently to five (use your fingers if it helps), give a gesture if needed, then your word or phrase, then physical help only if it's truly needed. Stop the moment the child communicates anything close to the target. Don't correct the approximation unless the therapist has asked you to.

The simplest M2L at home: during new-skill practice, model first, every time. Don't wait. Say the word or phrase before the child tries. Once the child is accurate five times running, wait one second before you model. Then two. Then three. The fading is the whole job.

Consistency across caregivers is genuinely hard. If a speech therapist has set up a home program, the highest-value move a parent can make is pulling the other caregivers, teachers, or aides into one 20-minute conversation about which approach is running and for which goals. Research on treatment fidelity in early communication intervention shows inconsistent prompting across partners slows skill acquisition ^[9].

Some of the best home practice happens inside routines you already have: mealtime, bath time, car rides. You don't need special therapy time. You need the time you already spend, used with more intentional structure. Little Words is built around exactly this, fitting prompting-aware practice into daily routines.

For children who also show echolalia, the same prompting principles hold, but what counts as a correct response gets broader. An echolalic approximation moving toward functional communication counts.

Frequently asked questions

What is the system of least prompts in speech therapy?

The system of least prompts is a teaching method where a therapist or parent starts with the smallest possible cue, such as a silent pause or expectant look, and adds more support only if the child doesn't respond correctly within a set wait time. It's one of the most widely used strategies in AAC intervention and is listed as an evidence-based practice by ASHA.

Is most to least prompting the same as errorless learning?

They overlap but aren't identical. Errorless learning structures tasks so errors are nearly impossible, often by giving the correct answer before any attempt. Most-to-least prompting starts with maximum support and fades it over time, which limits early errors but does allow them once fading begins. Errorless learning is a stricter version of the same protective logic.

Which prompting strategy reduces prompt dependency?

Least-to-most prompting reduces prompt dependency more reliably because the child always gets an unprompted opportunity first. If the child only communicates after an adult cues them, they never practice independent initiation. Most-to-least can also reduce dependency, but only if fading runs consistently and on schedule, which is harder to hold across caregivers.

How long should a wait time be in least to most prompting?

Most protocols suggest three to five seconds. Some children who process language more slowly benefit from longer waits, up to ten seconds. Set the wait time in advance and apply it the same way each time, rather than adjusting based on how the adult feels in the moment. Many parents underestimate how long five seconds actually feels during a communication opportunity.

Can most to least prompting cause prompt dependency?

Yes, if fading is delayed or inconsistent. When a child gets full models on every trial without criteria-based fading, they learn to wait for the model instead of initiating. Research on AAC users shows that slower fading of adult-provided models correlates with lower rates of spontaneous communication months later. The strategy itself isn't the problem. The fading failure is.

Which prompting hierarchy is best for a child learning to use an AAC device?

Least-to-most is generally preferred for AAC, especially for generalization and spontaneous requesting. ASHA's Practice Portal lists it as an evidence-based AAC strategy. Full models and hand-over-hand may fit the very first sessions when a child has never touched a device, but fading should start as soon as the child can locate symbols.

Does prompting hierarchy matter for children with autism?

Yes, and the research is reasonably strong. A 2013 systematic review in the American Journal of Speech-Language Pathology found the system of least prompts produced the largest effects for acquisition of requesting in learners with autism and intellectual disabilities. That doesn't make M2L wrong for every goal, but L2M has the better evidence base for communicative requesting in autistic children specifically.

How do I fade prompts without losing accuracy?

Set a criterion before you start, such as 80 percent correct over two consecutive sessions, and move to the next lower prompt level when the child hits it, no matter how uncertain you feel. The data overrule the instinct. Fading too slowly is the more common mistake. If accuracy drops when you fade, step back up one level, stabilize, then fade again more gradually.

Is most to least prompting used for apraxia of speech?

Yes. Dynamic Temporal and Tactile Cueing, one of the most evidence-supported treatments for childhood apraxia of speech, is an explicit M2L framework. It starts with simultaneous production, where clinician and child say the word together with tactile cues, and systematically fades to independent production. Motor accuracy early in treatment is the reason for starting with maximum support.

What happens if I use the wrong prompting hierarchy for my child?

In most cases a mismatch slows progress rather than causing harm. Using M2L for a child who already has the skill mostly means they don't get practice initiating. Using L2M for a child who needs motor accuracy support may mean they rehearse errors more than needed. Neither is catastrophic, but correcting course when you notice it matters. A speech-language pathologist can help identify which way the mismatch runs.

Should teachers and aides use the same prompting approach as the speech therapist?

Ideally yes. Inconsistent prompting across settings is one of the documented reasons communication skills fail to generalize. If a child gets L2M in therapy and full models from a classroom aide, they learn two different responses to two contexts rather than one transferable skill. Ask the SLP for a one-page prompt guide that names the level and wait time for each communication target.

What age do children usually start learning with prompting hierarchies?

There's no minimum age. Least-to-most and most-to-least strategies are used in early intervention from 12 to 18 months onward, including in naturalistic developmental behavioral interventions. Prompt intensity and wait times get adjusted for developmental level, but the underlying logic of starting at the right support level and fading toward independence applies at any age.

How do I know which prompt level to start at with my child?

Run a brief probe before starting a new goal: set up the opportunity and give no prompt at all. If the child responds correctly at least 20 percent of the time without help, L2M fits. If the child never responds without support, a higher starting level, either M2L or time delay with a controlling prompt, makes more sense. Your SLP can formalize this as a baseline session.

Sources

1. ASHA Practice Portal, Augmentative and Alternative Communication: ASHA lists the system of least prompts and time delay as evidence-based practices for AAC intervention
2. American Journal of Speech-Language Pathology, Ganz et al. 2013, systematic review of AAC interventions: The system of least prompts produced the largest effects for acquisition of requesting behaviors in learners with autism and intellectual disabilities
3. Apraxia Kids, treatment guidance on Dynamic Temporal and Tactile Cueing (DTTC): DTTC begins with maximum support and systematically fades, an M2L framework for childhood apraxia of speech
4. Journal of Applied Behavior Analysis, Lorah et al. 2014, AAC and prompting in autism: The system of least prompts produced faster generalization to novel settings compared to time-delay procedures for AAC use
5. Augmentative and Alternative Communication, Romski & Sevcik, prompt dependency in AAC users: Children who receive models before every communication opportunity often fail to use their device without the model present
6. Augmentative and Alternative Communication, Braddock et al. 2013, progressive time delay review: Progressive time delay is one of the most consistently replicated strategies for teaching symbol use and verbal requesting in autism
7. Augmentative and Alternative Communication, 2018, prompting and spontaneous communication in SGD users: Children whose communication partners reduced prompted trials faster showed higher rates of spontaneous communication six months later
8. Augmentative and Alternative Communication, 2016 survey of 312 AAC clinicians on prompting practices: Over 70 percent of AAC clinicians reported using L2M or time delay as their primary prompting approach for generalization goals
9. National Institute on Deafness and Other Communication Disorders (NIDCD), Autism Spectrum Disorder: Communication Problems in Children: Inconsistent prompting across partners slows skill acquisition in early communication intervention
10. ASHA, Evidence-Based Practice in Communication Disorders: ASHA framework for applying evidence-based prompting strategies in speech-language pathology
11. CDC, Learn the Signs. Act Early. Early Intervention: Early structured communication support improves long-term communication outcomes for children with developmental differences