What early dyslexia looks like — and what helps

At a playgroup last spring, I sat next to a five-year-old who couldn't rhyme "cat" and "bat." Her mom played a few rounds of the game with her — cat, bat, hat, mat — and the girl named the animals on the cards instead. Cat. Baseball bat. Cowboy hat. Doormat. The mom laughed and shrugged. "She's just younger than the rest. She'll catch up."

She might. About four in five kids will. But roughly one in five won't, and the rhyming gap at five is one of the earliest, cleanest signals we have. The International Dyslexia Association (IDA) and the Yale Center for Dyslexia & Creativity both estimate that around 15–20% of the population has some degree of dyslexia. Most of those kids are not identified until second or third grade. By then, the gap between their reading and their classmates' is wide enough that they've already decided something about themselves: I'm bad at reading.

This piece is for parents who are sitting where that mom sat and want to know what to actually look for, what to do, and what to skip.

What dyslexia is, and isn't

Dyslexia is a difference in how the brain processes the sounds of language and links them to letters. It is not a vision problem. It is not low intelligence — IDA's working definition is explicit that dyslexia is "unexpected" given a child's other cognitive abilities. It is not lazy. It is not something a kid grows out of by reading more.

It runs in families. If one parent has dyslexia, the chance a child does is roughly 30–50%, depending on the study. If a sibling has been diagnosed, the odds for the younger ones go up too. Family history alone is one of the strongest predictors we have, and it's a question most pediatricians don't ask.

What the early signs actually look like

You can see signs of dyslexia before a child reads a single word. Roughly, in the 4–6 range:

• Difficulty rhyming. Not "occasionally gets one wrong" — unable to hear that cat and bat sound alike, even after several rounds of the game. This is the playgroup moment.
• Trouble breaking words into sounds. Ask "what's the first sound in moon?" A child building typical phonemic awareness will say "mmm." A child who's struggling will say "moon" or freeze.
• Letter-sound confusion that lingers. Most kids reverse b/d, p/q, was/saw around kindergarten and sort it out by mid-first grade. Reversals that persist past age 6–7, paired with other signs, are worth noting.
• Slow vocabulary recall. Knowing the word but pulling up "the thing you sit on" instead of "chair." Sometimes called word-finding difficulty.
• A parent or sibling with reading difficulty. Family history.
• Mispronouncing common words longer than peers. "Pasghetti" past age 5, "aminal" past age 6.

No single sign is diagnostic. Three or four of them clustered together, especially with family history, is the pattern. The National Institute of Child Health and Human Development (NICHD) has been funding research on these early markers for thirty years, and the consensus is durable: the earlier you screen, the better the outcomes.

The wait-to-fail problem

Here is the part that frustrates the most parents I talk to. In many schools — public and private — the formal process for evaluating a child for a learning disability does not start until the child is visibly behind. The phrase educators use is "Response to Intervention." In practice it means: give the child general classroom instruction, watch them fall behind, give them small-group help, watch them fall behind again, then refer for evaluation. By the time the paperwork moves, the child is in second or third grade and a year or two below grade level.

There are good reasons the system was built this way — schools want to avoid over-identifying kids who just need more time. But the cost of waiting is high. Reading interventions are most effective when they start in kindergarten or first grade. Research summarized by IDA and Yale shows that children identified and supported before second grade often close the gap entirely. Children identified after third grade rarely do, even with excellent instruction.

If you suspect something at five, you don't have to wait for the school to suspect it at eight.

What evidence-based help actually looks like

The phrase to know is structured literacy. It's an umbrella term for instruction that is:

• Explicit. The teacher names the sound, names the letter, shows the mouth shape, and connects them. Nothing is left to be inferred from context.
• Systematic and cumulative. Sounds and patterns are introduced in a fixed order, each building on the last. Short vowels before long vowels. Single consonants before blends.
• Multisensory. Say it, hear it, see it, write it, sometimes trace it in sand. The kid uses several channels at once.
• Diagnostic. The teacher knows exactly which sounds and patterns the child has and hasn't mastered, and the next session targets the gap.

The classic structured literacy method is Orton-Gillingham (OG), developed in the 1930s and refined ever since. Programs in the OG family include Wilson, Barton, Lindamood-Bell's LiPS, and Logic of English. They share a backbone: small steps, immediate feedback, repeated practice on the specific sound-letter mappings the child needs.

If your child is in a structured literacy program, you'll hear them saying sounds out loud, tapping fingers for each sound in a word, and reading words they've explicitly been taught the patterns for. You will not hear them being asked to guess from pictures or "use context clues" as a primary strategy. (For more on the underlying research — and why guessing got into early reading instruction in the first place — see The Science of Reading, explained without jargon.)

What not to do

A short list, because every dyslexia parent I know has been talked into at least one of these:

• Drill without feedback. Twenty minutes of flashcards a night with no one correcting the errors hardens the wrong mappings. Practice without feedback is worse than no practice.
• Switching programs every month. Structured literacy is cumulative. Bouncing between three approaches in a year means the child never finishes the early scope and sequence in any of them.
• "She'll grow out of it." Some kids do. The ones who don't — and you usually can't tell which group your kid is in at five — lose ground every month.
• Vision therapy, colored overlays, eye exercises. IDA, the American Academy of Pediatrics, and the American Academy of Ophthalmology have issued joint statements that these do not treat dyslexia. The reading brain isn't an eye problem.
• Shame. Kids pick up on parental frustration faster than we think. The kid who decides she's stupid at five carries that into middle school.

Where adaptive AI fits — and where it doesn't

An AI tutor cannot diagnose dyslexia. It shouldn't try. Diagnosis is a clinical process involving a psychologist or educational diagnostician, a battery of tests, and a written report.

What a well-designed adaptive tutor can do is the daily practice piece. Structured literacy works because it's explicit, repetitive, and responsive. A patient adult sitting next to a child for fifteen minutes, naming sounds, listening carefully, correcting gently — that's the gold standard, and most families can't afford an hour a day of it. Software can fill some of that gap if the feedback is fast enough to matter, adjusts to where the child actually is rather than where the curriculum says she should be, and doesn't punish her for being slow.

This is the reason we built Lumi the way we did — voice-first, sub-second response, no streaks or pressure, the difficulty re-derived from what the child just said rather than what level the app thinks she's on. It's not a substitute for evaluation or for a trained tutor. It's the part in between, the daily practice a parent can actually sustain. For more on why we treat adaptation as the product rather than a setting, see Adaptive learning isn't a setting — it's the whole product.

When to seek a formal evaluation

IDA's guidance, distilled: if you see a cluster of early signs at age 5–6, especially with family history, ask for screening. You have two paths.

• Through the school. In the US, you can request an evaluation in writing. The school has a legal timeline to respond. Free, but slower and sometimes narrower in scope.
• Privately. A psychologist or neuropsychologist who specializes in learning differences. Typically $1,500–$4,000, sometimes covered partially by insurance, and faster. You get a written report you can bring to the school.

The IDA's Find a Provider directory and the Yale Center site list specialists by region.

The takeaway

The signs are visible early. Most kids who have dyslexia are missed early. The cost of being missed is not academic — it's the story the kid tells herself about who she is. If something feels off at four or five, trust it. Ask the question. The worst case is you're wrong and your kid gets a year of extra phonics practice she didn't need. The best case is the difference between a child who decides reading is hard and a child who decides reading is hers.

If you want a daily voice-first practice tool that adapts in real time and is built for kids who learn differently, join the Lumi beta waitlist.

Image brief

• Hero image: A warm picture-book illustration of a small child holding a card with the letters "cat" while a parent kneels nearby listening, in the Lumi cocoa and sun palette.
• Inline image 1: A simple checklist-style graphic of the six early signs (rhyming, sound-blending, letter-sound, vocabulary recall, family history, mispronunciation), placed after the "What the early signs actually look like" section.
• Inline image 2: A side-by-side comparison illustration of "wait-to-fail" timeline vs. early-screening timeline, with two child silhouettes and a widening gap, placed after the "The wait-to-fail problem" section.

Internal link suggestions

• "The Science of Reading, explained without jargon" — anchor: the underlying research — and why guessing got into early reading instruction in the first place
• "Adaptive learning isn't a setting — it's the whole product" — anchor: why we treat adaptation as the product rather than a setting
• "Reluctant readers: when your kid says 'I hate reading'" — anchor (optional, for sidebar): what to do when a struggling reader has already decided to hate it

Editor's note

A few items for Tim's review. (1) The playgroup anecdote is composited — please confirm whether to keep it generic or swap in a real Remi-era moment. (2) The 15–20% prevalence figure is the Yale/IDA range; if you want a single number, IDA most commonly cites "up to 20%." (3) The 30–50% heritability range is the band most often cited but varies by study — flag if you'd rather drop the range and say "significantly elevated." (4) The private evaluation cost range ($1,500–$4,000) is US-specific and city-dependent — confirm whether to localize or remove. (5) "Join the Lumi beta waitlist" CTA link is set to "/" — replace with the real waitlist URL when available.