Pull Apart, Push Together
You can break a spoken word into its separate sounds and push those sounds back together into the whole word. · 10 min
Here is a trick you can already do with your mouth. Take the word cat and say it as slow as a yawn: /k/... /a/... /t/. You just pulled a word into pieces. Now zip the pieces back up — /k/ /a/ /t/, faster, faster — cat. Same word, taken apart and put back together. This whole lesson lives inside that one trick.
Guess before you learn
Someone stretches a word out for you, slow and sticky: mmm... aaa... p. Push those three sounds together, fast. Which word were they hiding?
It is map: /m/ /a/ /p/, pushed together. If you chose mop, you swapped the middle sound — mop is /m/ /o/ /p/. If you chose nap, you swapped the first — nap starts with /n/. Every sound counts, and every sound stays in its place. Pushing sounds together like this has a name: blending.
K–2
3–5
Some sounds you can hold as long as your breath lasts — /sssss/, /mmmm/, /aaaa/. Stretch those and a word slides apart easily. Other sounds refuse to stretch: /p/, /t/, /k/, /b/, /d/, /g/. Try to hold /p/ and it just stops. For those, bounce them quick and light — /p/ /p/ /p/ — instead of pulling them like taffy.
Raise one finger for each sound you catch. Say cat slowly, /k/ /a/ /t/, and three fingers stand up — so three sounds. Your fingers keep the sounds from blurring together, and they show you exactly how many the word holds.
6–8
Each separate sound is called a phoneme — the smallest piece of speech that can turn one word into another. Change the first phoneme of cat and you get bat, hat, rat: one sound moved, one new word. Blending and segmenting are exact opposites — one builds the word up, the other takes it back apart.
Order is part of the word. The sounds /p/ /a/ /t/ blend into pat; the same three sounds in a new order, /t/ /a/ /p/, blend into tap. A word is not a bag of sounds — it is sounds standing in a line, first to last.
9–12
A phoneme earns its name by making meaning. Pat and bat differ by a single phoneme, /p/ against /b/, yet they name completely different things — that one contrast is what makes /p/ and /b/ separate phonemes of English. A pair alike but for one sound is called a minimal pair.
Segmenting is analysis: resolving the running stream of speech into its separate phonemes. Blending is synthesis: sliding those phonemes back into one gesture of sound. Reading will lean on both — but the ear has to do them first, with no letters anywhere in sight.
K–2
Say sun slow, like warm honey: sssss-uuuu-nnn. Feel it? Three little sounds hiding inside one word: /s/ /u/ /n/. Pulling them apart is called segmenting.
Now push them back together, faster and faster: /s/ /u/ /n/... sun. That is blending. Pull apart, push together — the very same word, held two ways.
Undergrad
The phoneme is an abstraction, not a sound you can point at. The /p/ in pat is aspirated — a small puff of air; the /p/ in spat is not. Acoustically they differ, yet English files them as one phoneme: they are allophones, context-bound variants that never change meaning, so the ear treats them as the same.
Hearing that sun holds three phonemes is a metalinguistic act — language turned back on itself to inspect its own parts. Ordinary speech is coarticulated, each sound smeared into its neighbors with no clean silences marking the seams, so segmenting is genuinely hard. It is also the single best early predictor of who will learn to read with ease.
Postgrad
In generative phonology the phoneme decomposes further, into simultaneous distinctive features — [±voice], [±nasal], and place and manner of articulation. /p/ and /b/ share every feature but [voice], so the minimal pair pat / bat is, underlyingly, a single-feature flip. The phoneme is the bundle; the feature is the atom beneath it.
The acoustic signal hands over no phoneme boundaries ready-made: coarticulation and the absence of invariant cues make the segmentation problem a deep one in speech perception. Phonemic segmentation is therefore a learned, effortful imposition of discrete structure onto a continuous stream — and it is precisely the competence the alphabet was built to exploit, one written mark per imagined phoneme.
segmenting
Breaking a whole spoken word into its separate sounds: sun becomes /s/ /u/ /n/.
blending
Pushing separate sounds back together into one whole word: /s/ /u/ /n/ becomes sun. Segmenting's exact opposite.
Why is this true?
Why do blending and segmenting undo each other?
Because both work on the same thing: the word's sounds standing in a line. Segmenting reads that line apart, one sound at a time; blending runs the very same line back together. Nothing is added or lost, so doing one and then the other lands you right back on the word you started with.
One thing to watch. Keep the sounds in order as you push them together. The same three sounds, /p/ /a/ /t/, make pat one way and tap the other. Blending is not just which sounds — it is which sounds, in which line.
You can now pull any spoken word apart into its sounds, and push those sounds back into the whole word. Your ears do the whole job on their own. Soon each of those sounds will be given a shape you can see — and pulling apart and pushing together is exactly what reading will turn out to be.
Practice — new ink and old, interleaved
1.What is the very first sound in sun?
2.In one sentence: what kind of sound always hums in the middle of these little words?
3.What is the last sound in dog, and how do you find it?
4.A whole spoken word is really made of what?
It is made of separate little sounds, said in order — pull the word apart and you can hear each one.
How close were you? Grade yourself honestly — it sets your review date.
5.How do you find out how many beats a word has?
Say it slowly and clap once for each beat — or feel your chin drop, one drop per beat.
How close were you? Grade yourself honestly — it sets your review date.
6.How many separate sounds are in fan? Say it slowly and count.
7.Blend /b/ /e/ /d/. What word do the sounds make?
8.Say pig slow: /p/ ... /i/ ... /g/. What is the middle sound?
9.How many beats do you clap for the word umbrella (um-brel-la)?