Cats, dogs, boxes: why doesn't -s sound the same on any?

It follows the same logic as the -ed ending, just with a different set of three sounds — /s/, /z/, or a full extra syllable /ɪz/ — decided by the same voicing rule: what the base word's final sound already is.

30-second version: The plural or third-person-singular -s is pronounced /s/ after a voiceless consonant (cats → /kæts/), /z/ after a voiced consonant or vowel (dogs → /dɒgz/), and a full extra syllable /ɪz/ after a sibilant sound — /s, z, ʃ, ʒ, tʃ, dʒ/ (boxes → /ˈbɒksɪz/). If you've already worked through the -ed ending, this is the same rule wearing a different sound — worth confirming it generalized rather than assuming it did.

Why the same rule shows up twice

English keeps consecutive consonants matched in voicing wherever it comfortably can, which is exactly the mechanism behind the -ed rule too — the language isn't inventing a new principle for plurals, it's applying the same voicing-agreement instinct to a different ending. After a voiceless sound, the -s devoices to match: /s/. After a voiced sound or a vowel, it stays voiced: /z/. And when the base word already ends on a sibilant — a hissing or buzzing sound like /s, z, ʃ, tʃ/ — adding another sibilant directly on top would be nearly impossible to hear as a separate ending, so English inserts a full syllable, /ɪz/, to keep it audible. This is precisely why "boxes" isn't "box" plus a quiet /s/ — it's a genuinely different, longer word by one syllable.

The consequence for learners who don't know the rule: applying a single sound to every -s ending (usually /s/, because that's the letter) produces "dogz" as "dogss" and "boxes" as one syllable short — both understandable, but both consistently marked as errors in fast or careful listening.

How sayit checks it

Just as with -ed, the target IPA for a word like "dogs" ends specifically on /z/, and your recording's actual final sound is compared against that target directly. Producing /s/ instead of /z/ registers as a substitution at that slot; skipping the extra syllable on a word like "boxes" registers as a deletion of the whole missing syllable, which is a more serious, more audible gap than a single misplaced consonant.

The rule, applied

Drill it

Record: "The cats chased the dogs past the boxes." All three endings — /s/, /z/, and the extra /ɪz/ syllable — in one sentence, so a single take checks the whole rule.

If "boxes" is the one that keeps coming out one syllable short, build it the same way you'd build a consonant cluster: say "box," then just the added "-iz," then join them, rather than trying to fix the whole word at once.

How do I know it's improving?

Test the rule on genuinely new words, the way you would for -ed — five plural or third-person-singular words you haven't specifically drilled, predicting the ending sound before you say each one. A rule that only works on the memorized drill sentence hasn't actually transferred yet.

Try it

Open sayit free — no install, no card — and record the drill sentence above. Then check third-person verbs too — "she watches," "he pushes" — since the same rule governs both plurals and verb agreement, and learners often fix one without noticing the other needs the identical fix.