Using the phonics decoder to say any new word, cold

30-second version: Most vocabulary lookups assume you'll hear the word spoken and imitate it. The phonics decoder is built for the case that method can't handle — a word you've only ever seen written, with no audio to imitate. It segments the spelling against the word's real IPA and explains, piece by piece, how to physically produce each sound, so you can say an unfamiliar word correctly the first time without ever having heard it.

Every method of learning pronunciation up to this point in this series assumes one thing: you hear the word, then you try to reproduce it. That covers most vocabulary. It doesn't cover the specific, common situation where you meet a word only in writing — an unusual name, a technical term in a paper, a word from a book with no audio edition — and need to say it before you've ever heard it correctly. That's the exact gap the phonics decoder is built to close.

What decoding actually does

Feed the phonics decoder any written word and it segments the spelling into pieces, aligns each piece to the word's real phonetic sounds (drawn from the same ground-truth pronunciation data used across sayit, not a guess), and explains how to physically shape each sound — where your tongue goes, whether your lips round, whether the sound is voiced. This is fundamentally different from a dictionary pronunciation guide, which tells you what the word sounds like assuming you can already produce the symbols it uses. Decoding tells you how, mechanically, to produce each sound, which is what makes it usable on a word you've genuinely never heard.

Why this matters even if you usually rely on audio

Audio-first learning — hear it, imitate it — is efficient and usually the right default (see how to learn a new word and pronounce it the first time). But it has a hard dependency: a native model to imitate. Decoding removes that dependency entirely. It's the fallback that works when there's no audio to lean on, and it's also a faster path for a word whose spelling is genuinely ambiguous — where hearing it once doesn't tell you why it's said that way, only that it is.

How to use it well

1. 1.Try to say the word yourself first, based on your own instinct from the spelling. This is your diagnostic — it tells you where your existing spelling-to-sound assumptions are already solid versus where they'll mislead you.
2. 2.Decode it and compare. Look at where the segmented sounds diverge from your guess — that divergence is the specific thing worth understanding, not the whole word.
3. 3.Read the articulation description for any genuinely new sound, not just the ones you got wrong. Understanding how a sound is produced (place, manner, mouth shape) generalizes to other words containing that same sound.
4. 4.Say it back and check. Decoding tells you what to do; recording and scoring confirms whether you actually did it.

Where this compounds with sound families

Decoding a handful of words that share a spelling pattern back to back is how a sound family actually becomes visible as a pattern rather than a one-off fact — three -tion words decoded in a row make the shared stress rule obvious in a way that decoding them individually, on different days, doesn't.

What decoding doesn't replace

Decoding tells you the target — it doesn't replace saying it out loud and getting corrected. A word decoded but never spoken is still just information, same as a definition read but never used; see why definitions alone don't make a word usable for the fuller version of that gap.

Try it: pick a word you've only ever read, never said out loud, and run it through Phonics before attempting it yourself.