Article

Autism Speech Regression at 2: What Actually Causes It

Last March, Priya in Minneapolis counted her son Rohan's words on a sticky note stuck to the fridge. He was 16 months old. The list had 22 entries: mama, dada

When Priya in Minneapolis started tracking her son Rohan's words on a sticky note stuck to the fridge, he was 16 months old and had 22 words: mama, dada, more, ball, dog, milk, shoe, up, go, hi, bye, no, juice, book, bath, car, hat, star, light, yeah, uh-oh, banana. By his second birthday, only three were left. "He just stopped," she told his pediatrician. "It wasn't gradual. It was like someone turned the volume knob down to zero."

Rohan's experience is more common than most parents realize. Speech regression, where a toddler loses words they'd already learned, happens to roughly 25 to 30 percent of autistic kids, usually between 15 and 30 months. The cause isn't fully understood, but current research points to a mix of neurological reorganization, the increased sensory load of a child's expanding world, and a natural shift from rote-learned words to more flexible language. It's not caused by vaccines, parenting style, or trauma in most cases.

If your child just lost words, you deserve more than vague reassurance, so here's what regression actually looks like, what the science does and doesn't explain, what to rule out medically, and what to do in the first month.

What regression looks like

A typical pattern: around 12 to 18 months, a child has a small but real vocabulary, often 5 to 30 words. Between 18 and 30 months, those words start dropping, a few at first, then more, then most or all. The child often becomes harder to engage, eye contact may decrease, and play may narrow to a few favorite objects. This is different from a child who never developed words in the first place. It's a loss of skills that were already there.

Regression is often partial, and the partial versions are what confuse parents most. Some kids lose words but keep their gestures. Some lose social engagement before language goes. Some also lose receptive language, no longer responding to their name or following directions. Each pattern points toward different next steps, which is part of why writing things down early matters so much.

What the research actually supports

The honest answer about causation is that we don't know as much as parents deserve to know, but we're not entirely in the dark either. Regression is a real, well-documented pattern in roughly a quarter to a third of autistic kids, replicated across many studies. Most cases happen between 15 and 30 months, peaking around 18 to 24 months. Something that surprises a lot of parents: regression doesn't predict worse long-term outcomes compared to autism without regression. Over time the trajectories tend to converge. Brain imaging shows some connectivity differences in kids who regress, but the findings aren't clean enough to use diagnostically.

Some explanations are plausible but unsettled. Early words may have been rote-learned, and the child could be shifting toward gestalt language processing, which looks like loss from the outside even though it's really a kind of reorganization. Sensory load also climbs a lot in toddlerhood as a child's world expands beyond the immediate family, and some kids withdraw to manage that load, with language among the first things to go offline. Regression may correlate with sleep disruption or seizure activity in a small subset of kids, which is part of why a medical workup matters.

Some things are worth stating plainly because they've been ruled out. Vaccines do not cause autism or regression; this has been studied extensively and definitively debunked. Parenting style doesn't cause it either, and the discredited "refrigerator mother" theory has been dead for decades. Diet alone typically doesn't cause or cure regression. Individual families sometimes see dietary changes help with specific co-occurring issues, but in the vast majority of cases the language regression itself isn't diet-driven.

The medical workup worth pushing for

When a child loses words, a few rare but treatable conditions need to be ruled out. This part is boring and non-negotiable.

Start with a hearing evaluation. Even kids who passed newborn screening can develop later hearing issues, and a child who responds fine to an iPad may still not be hearing speech clearly: selective hearing and functional hearing aren't the same test. Ask about Landau-Kleffner syndrome too, a rare condition where seizure activity, often silent, causes language regression. An EEG, sometimes with overnight monitoring, can identify it, and it's treatable.

Depending on the broader clinical picture, a developmental pediatrician may also want bloodwork or genetic testing, since metabolic or genetic conditions are a rare but real possibility. It's also worth noting whether a significant illness came before the regression. The link isn't always causal, but the timeline is useful information for every clinician who sees your child afterward. Most kids who regress won't have any of these conditions, but all of them are worth ruling out in that first workup.

The first 30 days

The weeks after you first notice regression are when parents feel most powerless, so here's what actually helps. Write it down: keep a dated list of words that were present, when they were last used, and which ones disappeared. A note on your phone works fine, and this record helps every clinician you talk to. Priya's fridge sticky note ended up being the most useful artifact in Rohan's evaluation.

Call the pediatrician and ask for a hearing test and a developmental referral. Don't let anyone tell you to wait more than 30 days from the first sign of loss to see what happens. If your pediatrician pushes back, get a second opinion. "Wait and see" isn't good advice when a child is actively losing skills.

Get on the speech-language pathologist waitlist even before you have a diagnosis. Waitlists run long, and you can always cancel later if you don't need it. Start using AAC at home too, even something as simple as a picture board with the lost words on it. Your child still knows what those words mean; it's the output channel that's offline, and AAC gives them a parallel route to communicate (our guide to speech therapy at home for autistic kids goes into more detail on building this).

Keep saying the lost words yourself. Don't stop talking about "milk" just because your child stopped saying it: receptive understanding usually stays intact even when speech doesn't. And ease off the pressure to talk. Stop quizzing, stop asking for repetition, lower the cost of trying to communicate. Many regressed kids slowly come back to spoken language once that pressure drops. Think of it like a muscle in spasm: you don't fix it by pushing harder, you release the tension first.

The gestalt connection nobody mentions

This is, in my opinion, one of the most overlooked pieces of the regression puzzle. A subset of kids who "regress" are actually shifting from analytic to gestalt language acquisition. They had a few rote-learned single words, and those drop off; what replaces them is chunked, intonation-based, often quoted language instead.

If your child stopped saying "ball" and "milk" but started repeating phrases from shows, or using intonation that sounds like sentences without clear words, look into gestalt language processing. The "regression" may be a shift in how language is being acquired rather than a loss of it overall, a bit like watching someone switch from typing to dictation: the output looks completely different, but they're still communicating. A speech-language pathologist trained in gestalt language acquisition (sometimes called Natural Language Acquisition, or NLA) can help you tell the difference. Not all SLPs have this training, so ask specifically.

What recovery actually looks like

Recovery patterns vary a lot, and I won't pretend otherwise. Some kids start regaining words within 3 to 6 months of consistent intervention. Others don't get the original words back but build language differently, often through gestalts. Some continue with limited spoken language and become full communicators through AAC instead. Some develop fluent spoken language by age 5 or 6, with the regression turning out to have been a temporary reorganization.

There's no reliable way to predict which path a specific child will take in the first weeks. What matters more than any single week is the pattern over months and years. Progress is rarely a straight line: two steps forward, one step sideways, a plateau, then a sudden burst. That's normal, not a warning sign.

When to push for more

Get an evaluation this month, not next, if your child has lost 3 or more words, has stopped responding to their name when they used to, has lost gestures like pointing or waving they used to use, or if the regression shows up alongside new sensory or sleep issues. The same goes if you suspect autism and haven't had an evaluation yet.

Push for a full workup rather than just an SLP visit. A developmental pediatrician, an audiologist, and, if seizure-like activity is suspected, a neurologist should all be part of the first round. Asking for all three isn't overreacting. It's being thorough.

Frequently Asked Questions

Did I cause my child's regression? No. There is no parenting behavior that causes autism regression. The current research points to neurological factors that are not parenting-driven. You did not cause this.

Will my child get the words back? Many kids regain at least some spoken language over time. Some regain all of it. Some develop full communication through a mix of speech and AAC. No one can predict the specific path for a specific child, but the tools that help (AAC, modeling, reduced pressure, real evaluation) are well established. Show up with those tools and trust the process over months and years.

Should I be worried if regression keeps happening? A single regression is common. Multiple regressions, or a regression that keeps progressing past 6 months without any forward movement, warrants more medical workup. Ask about EEG and a developmental neurology consult.

Is regression worse than never having had words? The long-term outcomes are roughly similar between regressed and non-regressed autistic kids. Regression feels worse to parents because of the visible loss, but the trajectories tend to converge over time.

Should we cut screens during regression? Probably not as a blanket strategy. Most current research does not support screens as a cause of regression. Some kids regulate better with familiar shows. Some kids dysregulate with too much input. Adjust to your specific child rather than following a universal rule.

Should I keep talking even though my child does not respond? Yes. Receptive language usually stays intact during regression. Your input matters. Keep narrating, keep reading, keep including your child in conversation. The output channel coming back online needs years of input to draw from.

How soon should we start AAC? Now. You do not need a diagnosis to start with a simple picture board or a free AAC app. There is no evidence that AAC delays spoken language. It supports it.

Related Reading

Related Little Words guides

Important: Little Words is educational support for home practice. It is not a medical device, not an AAC replacement, and not a substitute for a licensed speech-language pathologist, pediatrician, or developmental evaluation.
Buddy is a speech companion built for neurodivergent kids.

Little Words is a voice-first app where your child talks and plays with Buddy, at their own pace and in their own way. It is free to download.

See your child's planor download on the App Store