Elision: the sounds English speakers drop, and why "chocolate" has two syllables
Chocolate is spelled with three syllables and pronounced with two, and if you say all three you will sound like someone reading the word rather than using it.
Elision is the deletion of a sound that the spelling still shows. It is not the same as assimilation, where a sound changes to match its neighbour; here the sound is simply gone. English does this constantly, and it does it according to describable patterns rather than at random. Learning where sounds drop is what turns a spelling-shaped pronunciation into a speech-shaped one, and it is also the difference between hearing a fast sentence as words and hearing it as a smear.
/t/ and /d/ vanish inside consonant clusters
The single most productive rule in English elision: when /t/ or /d/ sits between two other consonants, it drops. English tolerates long consonant clusters better than most languages, but not that well, so the alveolar stop in the middle gives way.
- next day → /neks deɪ/
- last night → /lɑːs naɪt/
- most people → /məʊs piːpl̩/
- friendship → /ˈfrenʃɪp/
- hand bag → /hæn bæɡ/
- old man → /əʊl mæn/
- exactly → /ɪɡˈzækli/
- sandwich → /ˈsænwɪdʒ/
Notice that this works both inside a word and across a boundary. Friendship and sandwich have lost their /d/ permanently for most speakers, while next day loses its /t/ only in that phrase, and next April keeps it — because /t/ before a vowel is not in a cluster and has nothing to give way to.
There is one constraint worth knowing: the /t/ or /d/ that marks past tense resists deletion more strongly than a lexical one, because it carries grammatical information. Passed by holds its /t/ better than last bus does. But in fast speech even that gives way, and listeners recover the tense from context.
Schwa drops out of unstressed syllables
The second pattern removes whole syllables. When an unstressed schwa sits between a stressed syllable and a following consonant, English often deletes it, and the word contracts.
- chocolate → /ˈtʃɒklət/
- camera → /ˈkæmrə/
- every → /ˈevri/
- comfortable → /ˈkʌmftəbl̩/
- interesting → /ˈɪntrəstɪŋ/
- family → /ˈfæmli/
- different → /ˈdɪfrənt/
- vegetable → /ˈvedʒtəbl̩/
- separate as an adjective → /ˈseprət/
Say chocolate with three full syllables and then with two, and listen to which one sounds like English. The three-syllable version is not wrong in the sense of being unintelligible, but it is not what anyone says, and a spelling-faithful pronunciation of this particular word is one of the most recognisable markers of a learner. The same goes for comfortable, which has four spelled syllables and three spoken ones for most speakers, with the -fort- collapsing entirely.
The environment matters. The schwa drops when it follows a stressed syllable and precedes /r/, /l/, /n/ or another sonorant — which is why camera contracts but cinema does not, and why history becomes /ˈhɪstri/ but mystery often holds on to more of its middle.
/h/ disappears from unstressed pronouns
The third pattern is the one that makes pronouns almost invisible in running speech. The words he, him, her, his, has, have and had lose their /h/ whenever they are unstressed and not at the start of a phrase.
- tell her → /telə/
- ask him → /ɑːskɪm/
- give her a call → /ɡɪv ər ə kɔːl/
- what has he done → /wɒt əz i dʌn/
- I saw him → /aɪ sɔː ɪm/
Combine this with the weak forms and tell her becomes a single two-syllable unit that sounds nothing like its two spelled words. This is why learners often report that they hear a verb and then nothing — the pronoun is still there, but it has been reduced to a vowel and glued onto the end of the word before it.
The /h/ comes back the moment the word is stressed or begins a phrase. He told her keeps the first /h/ and drops the second. I gave it to HIM, not her keeps both, because contrast restores the full form.
The point of learning all three patterns is not to start deleting sounds on purpose. Careful speech is clear speech, and nobody is harmed by a fully pronounced next day. The point is that elision is regular enough to predict, so when it arrives in someone else's speech you can reconstruct what was said. Pick one pattern this week — the schwa deletions are the easiest to hear — and check your own pronunciation of a dozen common words against it.
(CTA) Want to know which syllables a word actually keeps? PHONO shows per-word IPA for any English text from an offline dictionary, and reads a sentence or a single tapped word aloud so you can count the syllables by ear. Download: https://apps.apple.com/us/app/phono-english-phonetics-ipa/id6783243417 · phono.douhouse.com