Day 2 of 5

The extra “uh” that stops words coming together

The extra “uh” is the whole problem. A child taught “buh” for b says five sounds in “bat” when the word has three, and they refuse to fit together. Teach the clean, clipped sound instead, and the sounds slide into a word on their own, with no gap for her to fall into.

Why does "buh" make blending harder?

Hi there, yesterday was about the ear. Today is about your mouth, because one small habit in the way grown-ups say letter sounds costs children more time than almost anything else at this stage.

When an adult says the letter b as "buh", they have actually said two things: the sound /b/, and a little vowel grunt tacked on the end. Linguists call that grunt a schwa. On its own it is harmless. It stops being harmless the second a child tries to build a word out of it.

Ask a child who learned it that way to read bat and you get "buh, a, tuh". That is five sounds where the word has three, and the two extra ones sit exactly where the joins are meant to be. She is following your instructions perfectly and the word still refuses to arrive. Most children draw the obvious conclusion: reading is a thing they cannot do.

The child is not getting it wrong. The sounds she was handed do not fit together.

The evidence on this is unusually tidy. Gonzalez-Frey and Ehri (2021), in Scientific Studies of Reading, taught kindergarten children to decode in two different ways: one group broke the speech stream between sounds ("sss, aaa, nnn") before blending, the other ran the sounds together with no break at all ("sssaaannn"). The connected group learned to decode faster and read unfamiliar words more accurately afterwards.

Their error analysis is the part worth carrying home: breaking between the sounds made children forget the first one by the time they reached the end of the word. The gap itself is the leak. And the words saved for the hardest part of that study were exactly the ones built on stop consonants like b, d and p, which are hard to blend precisely because saying one on its own drags a schwa in with it.

A much smaller 2025 study points the same way. Hassler, Cook and Meng, in Reading & Writing Quarterly, followed three first-graders through six weeks of decoding practice. Both ways of sounding out helped, they report; connected phonation may suit short words with a stretchy first sound better. Three children is not a rule. It is the same direction of travel.

Which letters carry an extra sound, and which are safe?

Every letter can pick up the extra vowel, but they are not equally hard to fix. Some sounds can be held like a note: hum "mmmm" and you cannot add an "uh" without stopping first. Adults still say "muh" for m out of habit, and children still copy it, but the repair is instant, because the held version is audibly right the moment you try it.

The six stop sounds are where the habit actually costs a child something. A stop sound is over the instant it begins, which is precisely why adults reach for a vowel to make it audible. Those six are worth your attention; the stretchy ones mostly fix themselves.

The sounds that slip

LetterThe clean soundThe common slipA cue that fixes it
bA quick pop of the lips, almost nothing after it"buh"Whisper it. A whisper has no voice, so the vowel cannot survive
dA light tap of the tongue behind the top teeth"duh"Say "dog", then cut yourself off after the first flick
gA short catch at the back of the mouth"guh"Make it shorter than feels reasonable. It should sound unfinished
pA puff of air with no voice in it at all"puh"Hold a tissue in front of your lips. The puff moves it; the "uh" does not
tA light tap, no voice, gone immediately"tuh"Say "top" and stop after the very first sound
kA short catch at the back, no voice"kuh"Same tissue test as p: one puff, then silence
mA hum you can hold for two seconds"muh", or the name "em"If you can hum it, you already have it right
nA hum through the nose, equally holdable"nuh", or the name "en"Stretch it out. Stretchy sounds are the forgiving ones

The same rule covers the pairs of letters that make a single sound. When your child meets sh, ch or th, that is two letters making one sound, so it gets one clean pronunciation too: a long quiet "shhh", never "shuh".

How do I say each sound cleanly myself?

You only have to retrain six sounds, and each one has a physical check that takes a second.

  • Whisper it. Whispering removes your voice entirely, and the schwa needs voice to exist. If the whisper still ends in a vowel, it is not a whisper yet.
  • Borrow a word, then chop. Say "dog" at normal speed, then say it again and stop after the first flick of your tongue. The word puts your mouth in the right shape for you.
  • Time it. Stretchy sounds (m, n, s, f, l, r, v, z) should last two seconds. Stop sounds (b, d, g, p, t, k) cannot be stretched at all. If yours lasts, there is a vowel hiding in it.
  • Model, do not correct. When she says "buh", say "/b/" back warmly and carry on. Children move toward the version they hear most, and being told off is not what moves them.

There is a second move worth stealing from that 2021 study. When you sound out a short word for her, do not chop it into separate beats. Run the sounds into each other in one breath, "sssaaat", then say "sat" at normal speed. The gap between beats is where children lose the first sound, so removing the gap removes the problem.

Should we still teach letter names at all?

Yes. This is the part that gets overstated online. The evidence does not say to hide letter names; it says the sound is what does the reading work. Plenty of names actually help, because they contain their own sound: "bee" starts with /b/, "tee" starts with /t/. Names like "aitch" and "double-you" carry no clue at all, which is why sounds have to be taught outright rather than inferred. There is more on how the two fit together in letter names versus letter sounds.

Say both, in one breath, every time you pass a letter: "that is b, and it says /b/". Two seconds, and the child gets the label and the tool at once.

Do it with the letter visible, too. Erbeli and colleagues (2024), in Scientific Studies of Reading, pooled sixteen studies of early phonemic awareness teaching and found the benefit rose up to about ten hours, then gave less back. Where the same instruction put letters in front of children alongside the sounds, that useful stretch ran considerably longer. Sound and shape learned as a pair go further than either alone.

What do I do when school or an app says it the other way?

Take it calmly, because this one genuinely varies. The UK Department for Education's phonics programme criteria (2023) require a validated programme to work in single phonemes rather than in clusters. How those single sounds get said out loud in any particular classroom is a separate question, though, and plenty of well-loved alphabet toys still sing "buh" and "cuh".

You do not need to referee it. A child can hold both versions at once for a while; she settles on the one she hears most from the person she reads with. So do not correct the teacher, and do not correct your child in front of anyone. Just be consistent in your own five minutes, and let the clean version quietly become the normal one.

There is some comfort in how this plays out in classrooms. Lane and colleagues (2025), in Reading Research Quarterly, followed kindergarten and first-grade classes through a full year of one structured foundational-skills programme and measured early literacy gains of more than a standard deviation over matched classes, with the largest gains where teachers stuck closest to the programme as written. Sticking with one clear way of doing it is what that study rewards, and one clear way is exactly what you can offer in five minutes a night.


A child who can say six sounds cleanly can build more real words tonight than a child who can name all twenty-six.

If you would like the sounds on paper, the letter-sound cards in the free printable packs carry a short "say it like this" line under every letter, so nobody has to remember which ones stretch and which ones stop.

Tomorrow: why five minutes most nights beats thirty minutes on a Sunday, and what goes inside them.

Sources

  • Gonzalez-Frey and Ehri (2021), Scientific Studies of Reading 25(3), connected versus segmented phonation: https://eric.ed.gov/?id=EJ1295469
  • Hassler, Cook and Meng (2025), Reading & Writing Quarterly 41(5) 398-411, connected phonation as a strategy for word reading fluency: https://eric.ed.gov/?id=EJ1489760
  • Erbeli, Rice, Xu, Bishop and Goodrich (2024), Scientific Studies of Reading 28(4), optimal dosage of early phonemic awareness instruction: https://eric.ed.gov/?id=EJ1427043
  • Lane, Contesse, Gage and Burns (2025), Reading Research Quarterly 60(1), foundational reading skills in kindergarten and first grade: https://eric.ed.gov/?id=EJ1458523
  • UK Department for Education (2023), Validation of systematic synthetic phonics programmes, supporting documentation: https://www.gov.uk/government/publications/phonics-teaching-materials-core-criteria-and-self-assessment/validation-of-systematic-synthetic-phonics-programmes-supporting-documentation