CODESWITCH · A bilingual CS education program
Every programming language is written in English. if while return class — the keywords are English words and every error message is an English sentence. So a student who's still learning English is asked to learn two languages at once, and then graded as though they were only learning one.
They aren't less capable. They're being tested on vocabulary while we tell them we're testing them on logic. CODESWITCH fixes the room, not the student.
01 · The method
Translating an English curriculum gives you a translated curriculum with English pedagogy still underneath it. Four rules change the teaching itself.
A loop is explained in Spanish or Vietnamese, in plain words, before any English keyword appears on a screen.
for, if, return aren't translated. They're treated as proper nouns, the way you don't translate a person's name. Students are told outright: this isn't English you need to understand, it's a symbol you need to recognise. That single reframe removes an enormous amount of anxiety.
Every session decodes one real error message word by word, in both languages. Reading errors is the most transferable skill in the entire course, and it's the one monolingual teaching skips.
Including mid-sentence. The tutor follows them. Nobody's English gets corrected during a CS explanation — do that and you teach a student that speaking is risky, and they stop asking.
02 · Three programs
Eight weeks, 90 minutes a week, after school. Eight to twelve students, two bilingual tutors. Everyone finishes having built one working thing they chose themselves — and demos it to their family in week eight, in whichever language they want.
First cohort · fall 2026A free, open library of bilingual CS materials: terminology mapped across languages with real explanations instead of dictionary glosses, error-message decoders, and side-by-side lesson materials. Spanish and Vietnamese first, Arabic next.
Mirror Docs 01 is live — the bilingual glossary →Bilingual undergraduates trained as tutors. Near-peer is the pedagogy, not a budget compromise — someone two years ahead still remembers what was confusing, and the social distance is small enough that a student will admit being lost.
Recruiting now03 · The Bridge, week by week
Python because its keywords are close to plain words, its error messages are short, and it does something on the very first line — no boilerplate standing between a student and their first result.
Students write instructions in their home language; a tutor follows them literally and wrongly. Then: their own name printed in a terminal, in their own script, diacritics intact.
A variable is a labelled box, and the label is theirs to name. Students discover they can name variables edad, tuổi, عمر — legal Python that almost none of them knew about. Ownership of the code changes right there.
Decisions drawn on paper before any code. Then if/elif/else, and the = versus == trap named early.
Physical demo first: clap until I say stop. Then for, while, and one deliberate infinite loop so they learn that Ctrl-C exists and nothing is unrecoverable.
They build a real two-language dictionary inside a Python dict. The data structure and the metaphor turn out to be the same object.
def, parameters, return — and a full segment on printing versus returning, because it's the most common conceptual break at this level. Students choose their final project here.
No new concepts. Tutors circulate with hands off the keyboard, because a student whose tutor fixed their bug hasn't learned to fix bugs. Debugging taught as a procedure: read the last line first.
Every student presents to the room and to invited family, in their language of choice. Certificates, next steps into the school's AP CS pathway, and an open invitation to come back as a tutor in two years.
04 · The economics
Cheap because the expensive part — the tutors — are trained volunteers who are two years ahead of the students instead of twenty. Everything below is a plan, not a result. No outcome numbers appear on this page until a cohort has finished and been measured.
What we measure — and publish. Not attendance and not satisfaction; both are easy to collect and nearly meaningless. We measure persistence (does the student enrol in a CS course afterwards), self-efficacy (a six-item pre/post survey given in the student's home language), and conceptual understanding (a bilingual concept check, plus a project the student can explain). Every cohort gets a two-page report — the school sees it first, then it's published here. Including the cohorts where it doesn't work.
05 · Working with minors
These are hard gates. No tutor runs a session until every one of them is complete, and they're listed here rather than buried because they're the first thing any school needs answered.
District volunteer application, background check, TB screening where the district requires it, and mandated-reporter training — all completed and verified before a tutor is in a room.
Never one tutor alone with a student. Sessions happen in the school's classroom or library, never anywhere else, and never in a vehicle.
No personal numbers, no DMs, no private social media between tutors and students. All communication routes through the school's named staff contact.
Held by the school. No student data with identifiers ever leaves the campus — evaluation results are aggregate only.
06 · Two things we need
You need conversational Spanish, Vietnamese, or Arabic, and to be comfortable with loops and functions. You don't need to be an expert — we specifically want people who learned this recently and still remember what was confusing. It's unpaid, it requires a district background check, and it asks for all eight weeks.
We'll send the handbook and the clearance steps. Legal Privacy
What we need from you: a room, one afternoon a week, and access to laptops you already own. What we don't need: any curriculum work, any budget, or any of your teaching time. What you get back is 8–12 more students who've built something real, plus the pre/post data on whether it worked.
Or email directly: yusuf.gadelrab06@gmail.com
Status, stated honestly. CODESWITCH is a program in launch, not an established organisation. The curriculum, volunteer handbook, safeguarding requirements, partnership kit and evaluation design are written and public. What doesn't exist yet: a signed school partner, a cleared tutor corps, and a single completed cohort. Those are the next three things, in that order. Nothing on this page claims otherwise.
Legal status. CODESWITCH is not an incorporated organization and is not a registered charity. It has not applied for recognition of exemption under section 501(c)(3), no application is pending, and contributions are not tax-deductible. There is no board of directors and no insurance policy. It is the program now being developed as CodeBridge under the planned Miftah Foundation, which is also not yet incorporated.
07 · Why this, from me
I'm Yusuf Gadelrab — CS at San José State, and an undergraduate researcher in Dr. Ethel Tshukudu's CS education lab. This year I co-authored a mixed-methods study on bilingual programming environments and inclusive CS learning, accepted at SIGCSE TS 2026. I already tutor CS at SJSU and help run CS programming at a San Jose high school with a large English-learner population. I'm bilingual myself — Arabic and English.
The finding was clear and the paper is published. A published finding that changes nothing in a classroom isn't worth much. This is the attempt to change something.
FAQ
A free eight-week introductory computer science program for high school students who are still learning English. Sessions are taught bilingually by trained near-peer tutors who speak the students' home languages — concepts are introduced in the home language first, and English keywords are attached to ideas the student already holds.
Translating an English curriculum produces a translated curriculum with English-language pedagogy underneath it. Bilingual teaching changes the instruction itself: the concept is explained in the language the student thinks in, English keywords are treated as symbols to recognise rather than vocabulary to master, and students may ask and answer in whichever language they choose.
It is free for students and for partner schools. Delivering one cohort costs roughly two hundred dollars, or about ninety-two dollars per student across a full year, because the tutors are trained volunteers rather than paid staff.
Every tutor completes the partner district's volunteer clearance, background check, TB screening where required, and mandated-reporter training before their first session. Two adults are present at all times, sessions happen only in school spaces, and there is no private contact between tutors and students.
The first cohorts are planned in Spanish and Vietnamese, the largest English-learner populations in the target San Jose schools, with Arabic following. Language coverage is set by which bilingual tutors are trained and cleared.
Three measures: persistence, meaning whether students enrol in a computer science course afterwards; self-efficacy, measured with a short pre and post survey administered in the student's home language; and conceptual understanding, assessed through a bilingual concept check and a final project the student can explain. Results are published whether or not they support the hypothesis.