This is a complete 11-day pronunciation course for Level 1 (beginner) adult ESL students: about 50 slide decks, a video version of nearly every lesson on YouTube, and 70-plus phone practice games with real recorded voices, so students can keep working at home. The decks run on Zoom. Everything is free for you to download, use, change however you like, and make your own. My teaching notes are tucked into the speaker notes of every deck — what to say, the questions to ask, the little tricks.
I built a version of this course last summer — by hand, generating images one at a time in ChatGPT, with a little help here and there. It took most of a summer. This year's version was rebuilt far faster, and then refined more carefully than I could ever have managed alone.
I'm not a linguist, and I have very little formal teacher training — just a lot of years in the room, noticing things. One thing I'd noticed: I have a growing number of Afghan students, many of them Farsi/Dari speakers, who are wonderful at repeating and understanding once a word is said and, in a sense, decoded for them — but who don't yet know the sounds of English letters and words very well. The Roman alphabet is new to them.
So this year I wanted to go beyond memorizing letters and their sounds, and actually help students with the code — the decoding that turns letters on a page into spoken words. I knew the symptom from years of watching it. I didn't know how to name it or build a method around it.
On our optional teacher-prep day, I sat down with Claude and asked it to rethink my whole course from last summer with that concern in mind. We talked it through. It helped me see the pattern clearly, shape a pedagogy around it — front-loading the alphabet and letter-sounds, stretching decoding across the early days, pushing the harder schwa and reduction work to the end — and then it built the media to teach it. Not a plan for me to execute. The actual, ready-to-use decks.
That first build was fast. But the part I'm proudest of came next.
A fast first draft is one thing; a course that actually flows is another. So I went back through the whole thing with Claude, deck by deck, and we cut hard. We pruned decks that repeated each other — four near-identical clapping drills became one; a 26-slide lesson on a low-value sound became eight tight slides. We caught a teaching habit I'd baked in without noticing: asking beginners to label a vowel as "long" or "short," when what actually helps them is simpler — just hear the word and point to the picture. We built new lessons for the genuinely hard vowel pairs my Spanish-speaking students struggle with — ship and sheep, pen and pan. And we stripped out the live-class chatter ("type in the chat," "go to your breakout room") that doesn't belong in a deck someone else will teach.
I taught the course four times this summer — two sections a day, twice through. That's where this stops being a story about building materials and becomes a story about teaching. Every class showed me something, and because edits were now nearly free, every "something" got fixed by the next morning. A picture of a rope that students kept reading as a garden hose — regenerated overnight. Speed games that outran the whole room, including me — repaced. A set of real-life dialogues I'd been proud of — cut entirely, because the room showed me my students weren't ready for them, and tongue twisters did the integrating job better.
The biggest lesson was structural. One day of the course kept failing — I taught it twice and it dragged both times, and one section skipped it with no harm done. Talking it through with Claude, we found the reason: every good day in the course teaches at one level of sound — letters, blending, vowels, syllables, stress. That day was organized around something else entirely (the ends of words), cutting across all the levels at once. No wonder it limped. We dissolved it, moved its best material where it belonged, and at summer's end renumbered the entire course — every deck, the website, the videos, the games — in one afternoon. Under the old economics, a restructure like that is the kind of thing a teacher knows is right and never does, because it would cost a week. It cost a conversation.
I also asked it to make practice games students could use at home on their phones — listen to a word, tap the right picture. What started as a small experiment is now 70-plus games covering all eleven days, and my students genuinely use them.
The strange part is how little of the actual work was mine. Claude built the decks, ready to use. I recorded the lessons; it ran the whole video pipeline — leveling the audio, repairing slides, publishing to YouTube. It built this website and pointed my domain name — murraycohen.com — at it. When I started, updating the site meant dragging a folder into my hosting service; these days I just say what changed, and it's live before I've finished my coffee.
The piece that still surprises me: Claude Code and Cowork work right on my own computer — reading, writing, arranging, and organizing my files where they actually live. I'm not copying and pasting things back and forth. It does the work in place.
If you're a teacher and you're not using Claude, Claude Code, and Cowork yet — you should be. I mean that plainly. It helped me find things in my own teaching I'd half-noticed for years, turn them into a method, build the materials to teach that method, and then go back and make all of it better — fast, and at a quality I couldn't have produced alone.
Take these decks. Use them. Change them. And if they help even one student crack the code, that's exactly what they're for.