Late Talker: When to Worry About Speech Delay

A twenty-two-month-old stands on the sidewalk, arm extended, index finger aimed at the dog across the street. The pointing shifts as the dog moves. Stops when the dog stops. Resumes when the dog resumes. The child looks up at the parent, back at the dog, back at the parent again — the whole unspoken question is there, written across the face: are you seeing this? The parent is seeing it. Also visible: the gap where a word should be. No "dog." No "woof." Not even a sound that approximates either. Just the pointing and the looking and the eighteen-month milestone window now firmly in the past, while the question gets louder in the parent's head. Is this toddler speech delay, or is this one of those kids who waits and then talks in full sentences at thirty months, and everyone says they knew all along the child would be fine?

The one-sentence answer: A toddler who says few words but understands everything (follows instructions without gestures, points to named objects, responds to questions) has expressive-only delay, and most of these children reach normal-range language by age four without formal intervention — though comprehension at 24 months predicts long-term outcomes twice as strongly as the number of words spoken, which means understanding is the dividing line between wait-and-see and call-now.

A quick map of what's below:

• Why comprehension at 24 months predicts outcomes twice as strongly as the number of words a toddler says — and what that changes about when to call for help
• The dimensional reframe that flips "will my child catch up" from a yes/no question into a more accurate but harder-to-answer trajectory question
• How conversational turns — not total words heard — build the specific brain tissue that supports language, and why this matters more than most late-talker advice acknowledges
• The three-question screen that separates wait-and-see from call-now, based on the evidence that actually predicts outcomes
• When expressive delay is hiding something motor-based that needs a completely different kind of intervention

If the one-sentence answer above is all you needed, you've got the gist. If you want the mechanism behind each piece, keep reading.

First, the terms. Expressive language is what a child can say. Receptive language is what a child understands. A toddler with expressive-only delay says very few words but understands most of what adults say — can follow two-step instructions like "get the ball and put it in the basket," points to named objects in books, responds to questions without needing visual cues to figure out what you're asking. A toddler with expressive-plus-receptive delay both says few words and shows difficulty understanding language — needs gestures and context to follow instructions, doesn't reliably point to named objects, seems confused by questions. The distinction matters because outcomes differ sharply. Most children with expressive-only delay catch up without intervention. Children with receptive involvement need help, and the earlier that help starts, the better the outcome.

The Distinction That Actually Predicts Outcomes: Expressive-Only vs. Expressive Plus Receptive

Most articles about toddler speech delay frame the question as binary: will my child catch up or won't they? The research over the past fifteen years suggests a more accurate but less comforting answer. Late talkers don't "catch up" in the sense of erasing the gap. They reach the normal range, but they remain on a measurably lower trajectory than socioeconomic-matched peers who were never late.

Leslie Rescorla followed 26 children identified as late talkers at age two all the way to age seventeen. By seventeen, every child in the study scored within the normal range on standardized language tests. That sounds like the reassuring outcome most parents hope for. But Rescorla also compared these former late talkers to peers from similar family backgrounds — matched for income, parental education, and other factors. At every measurement point from preschool through adolescence, the late-talker group scored measurably lower on vocabulary, grammar, and verbal memory. Not below normal, but below what children from similar backgrounds who were never late typically scored.

What this means practically: late talkers don't erase the gap. They reach normal, but they stay on a lower path within normal. A child who starts late often remains a less-strong reader than a sibling or classmate who talked early, even when both are technically fine. Early language at 24 months continued to predict language strength at seventeen. It didn't disappear as a factor. It set a trajectory.

The catch-up rate you'll see cited most often (somewhere between 70% and 90%) comes from studies that define "catch up" as reaching the normal range on a standardized test. That's not the same thing as performing at the level the child would have reached without the delay. The distinction matters for two reasons. First, it changes what "wait and see" actually means in practice. You're not waiting to see whether the delay will resolve. You're waiting to see whether it resolves enough that intervention becomes unnecessary. Second, it reframes the question parents are actually asking. Instead of "will my child be fine," the research-supported version is "will my child's language be strong enough to support reading, writing, and academic work without extra help." For context on what typical language looks like at age two, see How Many Words Should a 2 Year Old Say.

The factor that predicts where a late talker lands on that trajectory is comprehension. A 2017 meta-analysis combined data from 2,134 children across 18 studies and found that how much language a child understands at 24 months predicts later outcomes about twice as strongly as how many words the child says. Family history of language or literacy problems, gender, and whether the child was using two-word phrases all showed weaker effects in the combined data.

What this means practically: if a 24-month-old says fewer than ten words but understands everything adults say — follows two-step instructions like "get your shoes and bring them here," points to named objects in a book, responds to questions without needing to see the adult gesture — the delay is expressive-only. If the child is not understanding language at an age-appropriate level, or if comprehension is hard to assess because the child relies heavily on context and routine to figure out what people want, that's the signal to call for an evaluation now rather than wait.

Why the Back-and-Forth Matters More Than Total Words Heard

For years, the dominant narrative around language development was the "30 million word gap" — the finding from Betty Hart and Todd Risley's 1995 study that children from higher-income families heard dramatically more words by age three than children from lower-income families, and that this gap predicted vocabulary size and school readiness. The takeaway most parents absorbed was that quantity mattered: talk more, and your child will learn more words.

Research from the past decade has added an important correction. It's not the total number of words a child hears that drives language development. It's the number of conversational turns — the back-and-forth exchanges where the child says something, the adult responds, and the child says something again. Rachel Romeo and her colleagues at MIT recorded naturalistic language samples from 36 families matched for socioeconomic status and measured both the total number of adult words and the number of conversational exchanges. Then they scanned the children's brains using MRI.

Conversational turn count — not total adult words — predicted activation in Broca's area and structural growth in the left perisylvian language regions, the cortical areas that support speech production and comprehension. The effect held even after controlling for family income, parental education, and the child's own vocabulary size. A follow-up study in 2021 showed that the association between conversational turns and brain structure was specific to language regions. Turn-taking didn't predict growth in visual cortex or motor areas. The mechanism is selective.

What this changes for parents of late talkers is the intervention target. If the problem were purely input volume, the fix would be to narrate more (describe what you're doing, label objects, provide a running commentary on the day). That helps, but it's not sufficient. The brain is specifically tuned to interactive exchange. When a toddler points at the dog and the parent says "yes, that's a dog, the dog is walking," and the toddler points again or makes a sound, that loop is doing something qualitatively different from hearing the parent describe the dog while the child is looking at something else.

The practical version: instead of aiming for more words, aim for more turns. When a late-talking toddler gestures, vocalizes, or makes eye contact, treat it as the child's side of the conversation and respond as if the child just said something specific. Then pause and give the child a chance to respond back. That pause (the space where the next turn could go) is where the learning happens. A 2020 randomized trial of parent coaching in conversational turn-taking showed that 18-month-olds in the intervention group gained 100 words on average by 24 months, compared to 60 words in the control group. The intervention didn't teach parents to talk more. It taught them to pause more and respond more contingently to what the child was already doing. For more strategies on building these conversational exchanges, see How to Help Your Baby Talk: Evidence-Based Tips.

Simple tools can help create these turn-taking moments. First Words Flashcards were designed specifically for this: each card shows a clear image that gives parent and child something concrete to point at and name, creating a natural pause where the toddler can respond with a sound, gesture, or attempt at the word before the conversation continues.

When "Late Talker" Is Hiding Childhood Apraxia of Speech

Some children who look like late talkers at 18 or 24 months turn out to have childhood apraxia of speech, a motor-planning disorder where the brain has difficulty coordinating the precise movements needed for speech. The child knows what to say. The problem is getting the mouth to cooperate. Apraxia requires motor-based speech therapy, not the same language-stimulation strategies that work for expressive delay alone.

The red flags that separate typical late talking from something motor-based: fewer than five different consonant sounds by 24 months, vowel errors (vowels are easiest to produce and typically stabilize early), and inconsistency where the child says "ball" clearly once and then produces something completely different for the same word five minutes later.

Children with apraxia often show oral motor difficulty beyond speech. Trouble imitating simple mouth movements like sticking out the tongue or blowing a kiss, avoidance of crunchy or chewy foods, or drooling more than expected for age. None of these signs is diagnostic on its own, but the cluster is worth flagging.

The wait-and-see approach makes sense for expressive language delay but not for apraxia. A 2010 neuroimaging study found that former late talkers showed different brain activation patterns during language tasks, even after reaching the normal range on tests. The specific brain regions involved support automatic, habitual skills — the system that makes learned tasks feel effortless with practice. For former late talkers, language appears to remain more deliberate work even when it looks normal on the surface. This helps explain why difficulties with reading and writing often show up in school. The spoken language normalized, but the underlying processing stayed effortful.

Early intervention can rebuild these automatic systems while the brain is most plastic. Motor-based therapy for apraxia targets this directly, training motor sequences through intensive repetition. It's not the same thing as giving a late talker more language input. If a late-talking toddler has a very limited set of consonants, inconsistent productions, vowel errors, or oral motor difficulty, those are the patterns worth mentioning explicitly during an evaluation.

The Three-Question Screen That Separates Wait-and-See From Call-Now

If you're trying to decide whether a 24-month-old's toddler speech delay is the kind that resolves on its own or the kind that needs intervention, the research points to three questions that carry more weight than the raw word count.

First: Does the child understand language at an age-appropriate level? Can the child follow two-step instructions without gestures (not just "get your shoes" when the adult is pointing at the shoes, but "go to your room and bring me the blue blanket" when the child is in the kitchen and the adult is in the living room)? Can the child point to named objects in a book when asked "where's the truck?" without the adult looking at the truck first? Does the child respond to simple questions like "are you hungry?" or "do you want more?" without relying entirely on context to figure out what's being asked? If the answer to these questions is no, or if it's hard to assess because the child is so skilled at reading context that it's unclear how much of the language is actually being processed, that's the clearest reason to call for an evaluation. Receptive delay predicts poorer outcomes more strongly than expressive delay alone, and it's the dividing line between wait-and-see and intervene-now in most research protocols.

Second: Does the child point, gesture, and share attention the way other toddlers do? Children with pure expressive language delay typically have strong social communication. They point to things they want adults to see, not just things they want. They bring adults objects to show them, check the adult's face to see if they're watching, and shift their gaze back and forth between an interesting object and the adult's face to make sure everyone is looking at the same thing. These are the joint attention skills that differentiate late talkers from children on the autism spectrum. If a toddler rarely points, doesn't bring things to show adults, and seems to play in their own world without checking in, that's a different kind of concern and it needs a developmental evaluation, not just a speech evaluation.

Third: Is there a family history of language or literacy problems? If a parent or sibling had late language, speech therapy as a child, dyslexia, or significant reading or writing difficulty in school, that shifts the calculation. A 2009 study from the Western Australian Pregnancy Cohort followed 1,392 children from age two through age seventeen and found that children with both late talking at age two and a positive family history were more likely to have persistent language difficulties than late talkers without the family loading. The genetic component of language delay is substantial: research suggests about half of language ability differences are inherited, and the effect is even stronger at the lowest end of the ability range. Family history doesn't mean a child will definitely have the same difficulties, but it does mean the wait-and-see period should be shorter and more closely monitored than it would be otherwise.

If all three answers are reassuring (strong comprehension, strong social engagement, no family history), the evidence supports a short period of active monitoring with strategies to boost conversational turns at home. If any of the three raises a flag, particularly the first one, that's when the research-backed recommendation shifts to evaluation rather than waiting. Most states offer free developmental screenings through early intervention programs, and a speech-language pathologist can assess whether the delay is isolated to expressive language or whether there's a receptive or motor component that needs direct treatment.

What the Research Says About Bilingualism and Toddler Speech Delay

One of the most persistent myths about late talking is that exposure to two languages causes delay. The research does not support this.

Children learning two languages simultaneously from infancy follow the same developmental timeline as monolingual children when you count words across both languages. A bilingual 24-month-old might have 30 words in English and 40 words in Spanish for a total conceptual vocabulary of 70 words, which puts the child in the typical range even though the English-only vocabulary looks small. When researchers measure total conceptual vocabulary (the number of unique meanings a child can express across both languages), bilingual toddlers perform at the same level as monolinguals.

A child who is significantly delayed in both languages is not delayed because of bilingualism. The child is delayed for the same reasons monolingual children are delayed (genetic factors, limited conversational input, hearing problems, or an underlying developmental issue), and the bilingual exposure is incidental. Dropping one language won't fix the delay, and it will cut the child off from communication with family members.

The current position statements from the American Speech-Language-Hearing Association, the American Academy of Pediatrics, and the CATALISE consortium all state explicitly that bilingualism does not cause language delay and that families should not be advised to reduce language exposure. If a bilingual toddler is late talking, the evaluation and intervention proceed the same way they would for a monolingual child.