Dictation is a vocabulary test disguised as a listening exercise

30-second version: Dictation — listen to a sentence, type exactly what you heard — can't be gamed by recognizing a word on a page, because there's no page. It tests whether you actually caught the word by ear, which is a stricter and more honest check than reading comprehension. sayit plays a hidden sentence, grades your typed answer by edit distance, shows the diff, and lets you replay it slower.

Most vocabulary tests have an escape hatch. Multiple choice lets you eliminate wrong answers without fully knowing the right one. Reading comprehension lets you infer a word's meaning from context without ever confirming you know how it sounds. Dictation has no escape hatch: you hear a sentence once (or a few times, at your chosen speed), and you type exactly what you heard. If you didn't catch a word — because you don't know it, because it went by too fast, because the sound didn't match what you expected the spelling to be — it shows up wrong, immediately, with nowhere to guess from.

Why this is specifically a vocabulary test, not just a listening one

Dictation errors sort into a small number of causes, and most of them trace back to vocabulary, not raw listening ability:

• You didn't know the word existed, so you heard sounds and typed your best phonetic guess.
• You knew the word but never connected its written form to its sound — you'd recognize it on a page instantly, but hearing it spoken didn't trigger the same recognition.
• A weak or reduced form threw you off — function words and unstressed syllables often don't sound like their dictionary pronunciation in connected speech, which is a vocabulary-adjacent problem covered in connected speech and why vocabulary sounds different in sentences.

Each of these points at a genuine gap that reading practice, by its nature, can never surface — because reading gives you the spelling for free.

How it works in sayit

Listen plays a hidden sentence via TTS; you type what you heard, and the check grades your answer by edit distance against the target, showing exactly where it diverges word by word rather than just a pass/fail score. If a sentence beat you, you can replay it at a slower rate before trying again — useful for isolating whether the miss was a vocabulary gap or a raw-speed listening problem, since slowing it down removes the speed variable entirely.

Reading the diff correctly

The value of dictation practice is almost entirely in what you do after a miss, not the attempt itself:

• A wrong word you'd never seen before → a genuine new-vocabulary gap. Look it up, hear it again, and consider flashcarding it.
• A wrong word you clearly know when written → a sound-to-spelling disconnect. You know the word; you just hadn't connected its spoken form to its written one. Say it out loud a few times, comparing to the reference audio, to close that specific gap.
• A wrong function word or contraction ("would've" heard as "would of", or a dropped "the") → usually a connected-speech issue rather than a vocabulary one — less urgent to drill in isolation.

A practice routine

1. 1.Run a short dictation set — five to ten sentences — without slowing anything down first, to get an honest baseline.
2. 2.Review every miss and sort it into one of the three categories above.
3. 3.For genuine new-vocabulary misses, look the word up and add it to a deck.
4. 4.Re-run the same sentences a day or two later; a dropping error rate on repeated material is a cleaner improvement signal than a single session's raw score.

Where this fits with everything else

Dictation tests recognition-by-ear; reading aloud tests production. Between the two, you get a fuller picture than either alone gives you — a word you can both hear correctly and say correctly is genuinely learned, not just familiar in one direction.

Try it: open Listen, run a short dictation set, and sort your misses into "never knew it" versus "knew it, just didn't hear it."