SpellingIQ

How people learn to spell

What reading science says about how written words get into memory, and stay there.

Spelling is built from sound, not photographed

The intuitive theory of spelling says that good spellers have somehow photographed thousands of words. It is wrong, and the research is unusually clear about it. The dominant account, Linnea Ehri's orthographic mapping, describes a spelling as a set of bonds between the letters in the written form and the sounds in the spoken form. When those bonds form, the written word, the spoken word and its meaning become one unit in memory, which is why a well-learned word can be read instantly and spelled without effort (Ehri, 2014).

The mechanism has a practical consequence: to store a spelling properly, the learner needs the word's pronunciation available and needs to connect letters to sounds. A learner who lacks those tools falls back on rote visual memory, which works for a handful of words and then stops scaling. Most of Ehri's developmental evidence comes from beginning readers, but the mechanism is argued to be general.

Most orthographic learning is self-taught

Nobody is directly taught more than a fraction of the words they can spell. David Share's self-teaching hypothesis holds that word-specific spellings are mostly acquired as a by-product of successfully decoding unfamiliar words while reading: each successful translation of print into sound is a small learning event (Share, 1995). This has been tested directly: boosting the sound-it-out step improves later memory for a word's spelling, and blocking it sharply reduces that memory (Share, 1999). A 2022 systematic review in Educational Psychologist found the effect across writing systems.

One caution: this mechanism is established for reading acquisition. Producing a spelling is the harder inverse task, and the two are not the same skill, a point the instructional literature keeps rediscovering.

We absorb letter patterns nobody teaches

Ask an English speller whether ckat could be a word and they will say no, even though nobody ever told them ck cannot start a word. Rebecca Treiman and colleagues have shown that children and adults pick up these graphotactic regularities (which letter sequences occur, where in a word they are legal, which letters can double) purely from exposure to print, including patterns that have no basis in sound at all (Treiman et al., 2025). Sensitivity appears earlier in development than researchers once assumed, though the most complex, context-dependent patterns are only mastered by advanced spellers.

This is one reason wide reading and wide exposure to words matter for spelling: part of the knowledge is statistical, accumulated across thousands of encounters rather than taught in any lesson.

Is English spelling really that irregular?

Less than its reputation says. The classic computational analysis of 17,310 English words found roughly half spellable from sound–letter correspondence alone and another third wrong in only one letter; once a word's origin and meaningful parts are also taken into account, the share described as genuinely unpredictable falls to a few percent (Hanna et al., 1966). Those exact percentages are quoted far more often than they are re-derived, so treat them as order-of-magnitude. The qualitative conclusion has held up, though: English orthography is morphophonemic. It encodes meaning and history as well as sound, and it is far more consistent at that level than at the level of sound alone.

That is why sign keeps its silent G (it belongs to signature), and why the same past-tense -ed is spelled identically across walked, played and wanted despite three different pronunciations. For the history of how English got this way, and the words where it went spectacularly wrong, see the hardest words to spell.

Meaning is half of spelling

If English spelling encodes meaning, then teaching meaning should improve spelling, and it does. A systematic review of morphological instruction (teaching prefixes, suffixes, bases and roots) found benefits across spelling, word reading and vocabulary, with particular benefit for less able readers and, to the surprise of the time, effects that were not weaker for younger students (Bowers, Kirby & Deacon, 2010). A 2024 meta-analysis agreed that morphological instruction works while finding no clear pattern in who benefits most, so the moderators should be treated as unsettled (Colenbrander et al., 2024).

Competitive spellers arrived at the same place independently: elite spelling-bee preparation is famously organised around Greek and Latin roots and language of origin, not around memorising strings of letters.

Spelling is not one skill

Spelling ability decomposes into layers that can develop separately: phonology (hearing the sounds), orthography (knowing the letter patterns), and morphology (knowing the meaningful parts). Learners genuinely dissociate on them: a speller can be sound-accurate but pattern-inaccurate, or the reverse. The International Dyslexia Association's Structured Literacy framework makes the same carve-up instructionally, and adds a point that matters for anyone designing practice: sound-to-letter must be learned in both directions, because reading a word (letters → sound) and spelling it (sound → letters) are distinct skills that each need their own practice.

How children progress

Children's spelling develops through recognisable, overlapping phases, from treating words as pictures, through partial and then complete sound-letter mapping, to spelling organised around syllables and meaningful parts. The best-known classroom formulation is the Words Their Way stage sequence, whose spelling inventories are scored not by how many words a child gets right but by which orthographic features they control. That is placement by feature, so instruction can target the first feature not yet mastered (What Works Clearinghouse, 2013). The stages are a well-established professional framework rather than hard biological gates; children straddle them in practice.

What about adults?

Adults are not just older children, and the research base for them is thinner. In the populations that have been studied closely (mostly adults with a history of reading difficulty, in clinical and university samples), spelling is the most persistent visible marker of early literacy difficulty, often outlasting the reading problems themselves (Cheema et al., 2023). By adulthood the residual difficulty tends to concentrate in stored word forms and letter order rather than in basic sound skills, and morphological knowledge predicts performance, which is one reason remediation guidance for adults emphasises meaning-based word structure alongside phonics. Encouragingly, the same literature finds adults with long-standing difficulty can still form accurate spelling memories given structured practice. These are findings about selected samples, not the general adult population, and honest writing about adult spelling has to say so.

Keep reading

What makes spelling practice work covers the other half of the story: given how spellings are stored, which kinds of practice actually store them. Or see spelling and digital learning for what software adds, and the rest of the Learn section for the map.

References