Learn English

Though, through, tough: three words, one ending, three sounds

Read those three words on a page and they look like they should rhyme. Say them out loud and they don't come close: “though” rhymes with “go,” “through” rhymes with “too,” and “tough” rhymes with “cuff.” Same four letters at the end, three completely different vowel sounds. That's not a trick example someone dug up for a listicle — it's a fair sample of how English spelling actually works, which is to say it often doesn't predict pronunciation at all. If you're learning English mainly from a textbook or a grammar app, this is the wall you eventually hit: you can read a sentence perfectly and still not recognise it, or produce it, when someone says it to you at normal speed.

It's not just vowels — stress changes the word itself

English also uses stress to distinguish entire words that are spelled identically. Say “REcord” with the stress on the first syllable and you mean the noun — a vinyl record, a criminal record. Say “reCORD” with the stress on the second syllable and it's the verb — to record something. Nothing on the page changes. English has dozens of these noun/verb pairs (“PREsent” vs. “preSENT,” “PROject” vs. “proJECT”), and no amount of reading practice teaches you to hear or produce the difference — only listening to real speech and having your own attempts corrected does that.

Why this matters more for English than for a lot of languages

Some languages you learn mostly by sight-reading, because their spelling is close to phonetic — learn the rules once and you can predict how almost any new word sounds. English isn't one of those languages. Its spelling froze centuries ago while its pronunciation kept moving, borrowed vocabulary from French, Latin, and Old Norse without normalising the spelling, and ended up with a writing system that's a rough historical record rather than a reliable guide to sound. That gap is exactly why a study method built around reading and grammar drills can leave you able to pass a written exam while still struggling to follow a native speaker or make yourself understood out loud. The fix isn't more vocabulary lists — it's more time actually listening and speaking, with someone (or something) telling you when you got it wrong.

Where Sayelle fits in

Sayelle skips the flashcard-first approach and puts you straight into spoken conversation with an AI conversation partner that listens to what you actually say and answers back, out loud, in natural English. You're not picking from four multiple-choice options — you're ordering food at a restaurant, checking into a hotel, talking your way through a complaint at the front desk, or haggling over a price at a market stall, and the AI responds to whatever you say, not a script. Because the practice is spoken rather than typed, the specific problems English creates — words that don't sound like they're spelled, stress patterns that flip a word's meaning, contractions and connected speech that blur word boundaries when people talk fast — are exactly what you're training against, in real time, instead of studying around them on paper.

Lessons for structure, Scenarios for the unpredictable part

Sayelle's Lessons are story-driven and teach grammar through conversation rather than rule tables, laid out as a roadmap you move through. Scenarios drop the script entirely — free-form situations where you have to understand what's actually said to you and respond appropriately, which is closer to what a real conversation in English feels like than any fixed dialogue can be. After a session, Study turns your own past chats into flashcards, vocabulary review, listening and speaking exercises, and sentence-building practice — so the words and sounds you personally struggled with are the ones that come back around, not a generic deck.

It works without a signal, too

Sayelle's tutor, speech recognition, and text-to-speech models download once over Wi-Fi and then run on your iPhone, so core conversation practice keeps working on a plane, on the subway, or anywhere else without a connection — your voice never has to leave the phone to get a reply. Sign-in, purchases, ads, Cloud Sync, and downloading new content still need internet, but the actual back-and-forth doesn't. Read the full breakdown on the offline AI language learning page.

Getting started

Sayelle is free to start: you get five hearts, spent by beginning a new lesson, scenario, or study session (resuming one already in progress doesn't cost another), and they recharge on their own over time. Sayelle Plus removes that limit, unlocks Plus-tier lessons and scenarios, and removes ads — pricing is shown in the app before you subscribe. You can also learn English as the language you already speak while studying one of Sayelle's other eight languages — see the full language list — or download Sayelle on the App Store and start a conversation.

Sayelle

A voice-first AI language tutor. Speech and AI generation run on your phone; when Cloud Sync is on, saved text history and progress can also sync when connected.

© 2026 Sayelle — made for people who'd rather talk than memorize