# Can You Fake Being a Beginner in Your Native Language?

My Weird Prompts — an AI-generated podcast. Daniel Rosehill sends in a prompt; AI hosts Corn and Herman discuss it. Verify important claims independently.

- Episode: 6671
- Published: 2026-11-19
- Duration: 24:13
- Audio: https://episodes.myweirdprompts.com/audio/faking-beginner-native-language-06846.m4a
- Tags: education, cognitive-biases, language-learning, translation
- Web page: https://myweirdprompts.com/episode/faking-beginner-native-language-06846/
- JSON: https://myweirdprompts.com/api/episodes/faking-beginner-native-language-06846.json

## Summary

Pretending to know nothing in your own native language is harder than faking expertise because you have to suppress every automatic reflex. This episode examines real-world stunts—from Kei Simmons's year-long high school ruse to Jose Barrientos's four-month college act and a Passau exam-substitution case—alongside university heritage-speaker placement gates. It also explores Larry Selinker's interlanguage theory, explaining why real learners make systematic errors while fakers slip up, and a Cambridge-led study on how poorly listeners spot fake accents.

## The prompt

> Could you survive pretending to be a total beginner in your own native language? Say for some reason you had to sit in a beginner class, for weeks, learning your mother tongue from scratch. Could you pull it off? Are there real accounts of people who have tried it, whether for an easy grade, for benefits, or just for entertainment? And if you had to, how would you do it convincingly?

## Show notes

Daniel's question sounds like a thought experiment: could you walk into a beginner class in your own native language and perform total ignorance for weeks? The closest real cases are all adjacent. Kei Simmons, a half-Japanese sixteen-year-old in Colorado, spent about a year posing as a Japanese transfer student named Happy who barely spoke English, filming it as Happy's School Days; reactions ranged from shock to indifference. Comedian Jose Barrientos kept up a fake Hispanic accent through a college class for four months before dropping it mid-speech. The purest illustration of the continuity problem: a jury-selection story in which a man claiming broken English had been overheard speaking it perfectly in the lobby fifteen minutes earlier. Nobody has documented Daniel's exact scenario, and the institutional walls — heritage-speaker placement policies at Miami, Columbia and elsewhere — plus the science of interlanguage, explain why faking incompetence is harder than faking expertise. Detection science, meanwhile, is shakier than expected: a Cambridge-led study found listeners barely beat chance at spotting faked accents.

## Transcript

# MY WEIRD PROMPTS — Episode 6666

Corn: Daniel wants us to imagine the world's most pointless undercover operation. You walk into a beginner class — your own native language, the one you dream in — and for weeks you perform total ignorance of it. Could you survive? Did anyone ever actually try it, for a grade, for benefits, for the fun of it? And if you had to do it convincingly, what would it take? He's asking for the tradecraft of fake incompetence.

Herman: And the beautiful thing is, this is one of those questions where the premise sounds like a thought experiment, and then you dig and find people have done versions of it. Real ones. Filmed ones. One with a criminal investigation attached.

Corn: There's a guy in Germany who's very much not laughing.

Herman: No, he isn't. We'll get to him. Let's establish the difficulty first, because most people get it backwards. Faking an expert is one skill. Faking a beginner is harder, and the reason is that you're not adding something, you're suppressing everything. Every reflex you have — laughing at the right moment, reacting to a joke before the teacher finishes it — is evidence. Real beginners don't have a knowledge problem, they have a system problem, and the system leaks everywhere.

Corn: The comparison I keep making is that it's easier to walk into a bank and pretend to be rich than to walk into a bank and pretend you've never seen money.

Herman: That's the shape of it, yes. And there are three things to pull apart if we're going to answer Daniel properly. The real accounts — who has actually tried this. The institutional gates — placement tests, heritage-speaker policies, the walls universities have spent decades building precisely against this. And then the detection science, which is much shakier than anyone expects.

Corn: And the short answer, so nobody waits for it: the classroom stunt is nearly impossible to find in its exact form. The anecdotes are all adjacent. And the science of catching a faker is weaker than a Monday morning teacher would like to believe. The reasons why are the episode.

Herman: Start with the closest real case, because it's a good one. Kei Simmons — sixteen, half-Japanese, high school in Colorado. He spent roughly a full year posing as a Japanese transfer student named Happy who could barely speak English. Filmed the whole thing as a video series, Happy's School Days, and revealed the ruse in the final installment.

Corn: A year.

Herman: Teachers, classmates, the works.

Corn: What happened when he revealed it?

Herman: Reactions ranged from shock to complete indifference, and one teacher, per the reporting, just looked annoyed. Which I find the most realistic detail in the whole story. A year of your pedagogy being performed at.

Corn: The thing I want to flag about Simmons is that nobody in that building was testing him. His audience was social. Classmates, homeroom teachers — people who have a hundred students and no mandate to verify anything. The moment an examiner walks in with a clipboard, that entire performance dies in the corridor.

Herman: That's the pattern, and it holds for every case we found. The stunts that work are short, social, and low-stakes. The ones that fail are long, institutional, and high-stakes. Simmons is the far end of the social version.

Corn: And then there's the comedian.

Herman: Jose Barrientos. He's a comedian who kept up a fake Hispanic accent through an entire college class for about four months, and dropped it in his fourth speech. The reveal is a great piece of video — he walks up, mid-sentence, and the accent is just gone.

Corn: You said a speech class?

Herman: That's the funny part. The reports disagree on what class it even was. One outlet said Spanish class, another said Speech 101. Four months later nobody could agree on the room the stunt happened in.

Corn: Four months of fieldwork and no lab notes.

Herman: But notice what he was doing, because it matters for Daniel's question. A fake accent is a performance of identity. A fake beginner is a performance of incompetence. They look similar and they're completely different skills. Barrientos could still understand everything at full speed. He was hiding his voice, not his knowledge.

Corn: There's a jury-duty version, too, isn't there.

Herman: Yes, and it's the purest illustration of the continuity problem. A man at jury selection told the judge, in broken English, that he didn't speak English. Another prospective juror, who later wrote it up for a student paper, had heard him speaking perfect English in the lobby fifteen minutes earlier. That's the whole story. The tell wasn't the accent — it was the fifteen minutes.

Corn: The performance worked in the moment. It couldn't survive a corridor.

Herman: And he was dismissed anyway. Single source, a student op-ed, so hold it loosely, but it's the cleanest picture of the problem: the act held in the courtroom and died in the lobby.

Herman: Then the German case, which is the serious one and the reason the answer to Daniel's "for benefits" part has teeth. December last year — a German citizen allegedly sat a language exam in Passau on behalf of an Afghan man, using false papers, to produce proof of German language skill for a residence permit. That's person substitution, not faking low skill — a different crime entirely — but it shows the stakes. Forging certificates falls under forgery, Section two sixty-seven of the German Criminal Code, and it can cost the man his residence permit. How they caught it was never reported.

Corn: Which is itself telling. The detection method stayed quiet. If you publicize how you catch impersonators, you train the next one.

Herman: The Passau case is the hard end of the spectrum. Simmons is the soft end. Barrientos and the jury man are the middle. And notice what's missing from all of them: a native speaker, sitting in a beginner class of their own language, for weeks, under observation. Nobody's documented that exact thing.

Corn: So Daniel's exact scenario is a gap in the record.

Herman: A gap with a shape. Everything around it exists. The middle of it doesn't. Which usually means one of two things: it never happens, or it happens and works, and nobody writes about the successes.

Corn: I'd assume the first. The second requires a person who can sustain a fourteen-week lie for no audience.

Herman: Agreed. Now, the reason it never happens — because the institutions have been building walls against a version of this for decades. Not the prank version. The incentive version.

Herman: Here's the situation at any American university with a Spanish department. Cal State LA's first Spanish course is literally marked "not for heritage speakers." Miami moves heritage learners out of certain beginner courses and into a heritage course in week one. Columbia runs a dedicated Spanish heritage placement exam. Monmouth's placement policy treats a student whose first language is the target as a native speaker, full stop. Syllabi all over say beginner classes aren't open to natives.

Corn: Week one, Miami. They're not waiting for the student to slip.

Herman: They can't afford to. Because the incentive is real — heritage speakers do enroll low on purpose. Harvard's Arabic professor, William Granara, said in twenty fifteen that heritage speakers sometimes "slip through the cracks" and deliberately enroll in easier courses.

Corn: "Slip through the cracks" is a very generous verb for "walked in and sat down."

Herman: But here's the misconception Daniel's question walks straight into, and it's a good one to bust. The easy A myth. Granara's full quote is brutal: "Heritage learners take the class and think it's going to be an easy A, and it isn't." Knowing how to talk is not the same as knowing the grammar the course is testing. A Tufts instructor back in two thousand seven called heritage-speaker enrollment "definitely an easy A" — and that attitude is precisely why the heritage tracks got built. The schools adapted.

Corn: So the faker's first problem isn't the acting. It's the paperwork.

Herman: The placement test. And even if you sandbag that — and sandbagging a placement test convincingly is its own craft, because you have to fail in the right pattern — the institutions have a second net. Language departments flag work whose competence sits above the class level. Colorado State's academic integrity page lists getting help from native or advanced speakers as a form of misconduct in language classes. Meaning even a real beginner with a helpful cousin is in the gray zone.

Corn: So if a native speaker helping you is misconduct, a native speaker being you is going to raise an eyebrow.

Herman: At the very least. Which is about where any sane person would give up and go and buy a product instead.

Larry: Friends, are you tired of pretending to be a beginner in your own native language? Then you need the Heritage Houdini, the only classroom disguise kit that makes you look like you've never heard your mother tongue before. It's a limited time offer, and folks, the price just keeps dropping while I talk. Was ninety nine, now seventy nine, now fifty nine, and I'll throw in a fake student ID that says you're from a country that doesn't exist, plus a lifetime supply of confused shrugs. But wait, there's more, the price is now thirty nine, and I'll add a certificate of authentic incompetence, a placement test you can eat, and a tiny flag you can wave whenever someone asks where you're from, all for twenty nine, now nineteen, now nine, and I'll throw in a second Heritage Houdini absolutely free, plus a commemorative spoon, but hurry, the deal ends before I finish describing it, which is right about now, so call immediately, or don't, it's already over.

Herman: A placement test you can eat.

Corn: And a commemorative spoon. For the ceremony.

Herman: Now, past the gates. Say you've somehow cleared them. Now the harder question: can you fool the people in the room? And here the linguistics actually works against you in a very specific way.

Herman: There's a concept from Larry Selinker, one of the founding figures of second language acquisition, called interlanguage. The idea is that a learner's speech isn't broken language. It's a rule-governed system of its own — consistent, structured, predictable. It keeps features of the first language, what linguists call transfer errors. It over-applies the rules of the new language, which is where you get things like "goed" instead of "went." And some errors freeze in place permanently, which is fossilisation.

Corn: So a real beginner isn't random. A real beginner is a machine with rules.

Herman: And that's the trap for the faker. A native speaker playing a beginner instinctively makes random mistakes — sprinkled errors, chosen for effect, varied. And random errors look fake, because real errors are systematic. A genuine learner drops articles every single time, in the same places, because that's the rule their first language runs on. A faker drops articles when they remember to.

Corn: So the first commandment of the whole enterprise is: pick your fake first language and never deviate from its error pattern.

Herman: Commit to dropped articles, fixed word order, one consistent mispronunciation, every time, for weeks. Which is its own kind of mental load, because you're not improvising, you're executing a grammar you don't actually run.

Corn: Here's where I want to bring in the detection science, because it's the part Daniel will enjoy most. I assumed, walking in, that natives are basically lie detectors for their own language. The actual numbers say otherwise.

Herman: There was a Cambridge-led study, published in Evolutionary Human Sciences in November twenty twenty-four, that tested whether people can spot a faked accent. Twelve thousand responses, over nine hundred listeners, and the material was tiny — two to three second clips. Result: listeners beat chance, but barely. Just over sixty percent. And the spread is what's interesting. Listeners from Belfast, Glasgow, Dublin, and north-east England did best. Essex, London, and Bristol listeners were near chance — basically guessing.

Corn: So in some cities, the locals can hear a fake from two seconds of audio, and in London they're flipping a coin.

Herman: The authors' explanation for why the Belfast and Glasgow listeners did best is that those places have a history of group tension around accent — you've had to know who's performing. But they flag it themselves as a hypothesis, and I'd hold it loosely.

Corn: Hold on — sixty percent, from two-second clips. Give me a ten-second clip and a whole semester.

Herman: Fair. Longer exposure gives more evidence, and a classroom is months of continuous exposure. But the flip side is the jury-duty story: the more exposure, the more continuity you have to maintain, and continuity is exactly what breaks. And it gets worse, because even the professional impostor-catching field is shaky. There's a method used in asylum cases called language analysis for the determination of origin — LADO. The premise is that an asylum seeker might imitate a language variety that isn't theirs, so you analyze their speech to figure out where they're really from.

Corn: And?

Herman: In two thousand three, five Australian experts led by Eades reviewed it and judged it neither valid nor reliable as then practiced. Helen Fraser, a phonetician who's written on it extensively, concluded around two thousand nine that it's potentially useful with serious limits. These are the professionals, Corn. Paid experts, asylum stakes, and the method barely stands up.

Corn: If the court system can't reliably tell a faked language from a real one, the Tuesday afternoon Spanish teacher is guessing.

Herman: Mostly guessing with good instincts. Teachers catch fakers, but they catch them on continuity, not on the accent. The reveal is almost never linguistic. It's the wrong laugh at the right joke.

Corn: Which leads us to the part Daniel actually asked for. The playbook. How would you do it? Reasoning from the linguistics, not from any experiment — nobody has run the fake-beginner trial.

Herman: Nobody ethical, anyway.

Corn: Rule one we've covered: systematic errors. Pick the fake first language, commit to its pattern every single time.

Herman: Rule two, and this is the one everyone misses: you're allowed to improve. You have to improve — a student who makes zero progress across eight weeks is its own red flag. But slowly. You stay one step behind the class, always. The class learns the past tense in week five; you let yours arrive in week seven, wobbly.

Corn: That's the performance nobody scripts. Bad acting on a schedule.

Herman: Rule three is the killer, and I think it's the real answer to whether you could survive: never react before the translation does. Comprehension leaks faster than speech. The teacher says something funny, and the beginner should not laugh — not because he didn't understand, but because real beginners don't process jokes at conversational speed. Your reaction time is your accent.

Corn: So you sit there, stone-faced, while the room laughs, and then laugh late and wrong.

Herman: Late and wrong, every time. And the fourth rule: no idioms, no slang, never fluent speed. If you need to sound fluent, sound fluent about nothing. The giveaways teachers actually notice aren't in your speaking — they're in your listening speed, your reading pace, your spelling, your punctuation. A real beginner reads like someone defusing a bomb.

Corn: Spelling is a nice one. Because the class is drilling the alphabet and your hand writes "necessary" first try.

Herman: You'd have to misspell on purpose, in the pattern of your fake first language. Consistently. For weeks.

Corn: Let me try the honest verdict, then. Could I survive it? Speaking personally — I nap through most interpersonal scrutiny as it is. My baseline classroom presence is indistinguishable from a struggling student's.

Herman: That's an advantage.

Corn: It's not an advantage, it's camouflage. But no — I don't think I'd last. Not on the grammar. On the jokes. I'd grin at something in week two and all of it ends in a hallway conversation.

Herman: I think you'd fail earlier, at the placement test, because you'd refuse to fail it convincingly. You'd want your errors to be interesting.

Corn: That's... fair. I'd want a systematic error I could defend.

Herman: And that's the whole problem. You'd be the only beginner in the room with a philosophy of his own mistakes.

Herman: Let me put the survival question properly, though, because I keep circling it. The acting is sustainable for weeks — Simmons proved a year of the social version is possible. What's not sustainable is the social bookkeeping. Every classmate is a witness with a memory. You told one person you were from Osaka; you told another you'd never seen the ocean. The performance doesn't collapse on linguistics. It collapses on the third time someone says "wait, I thought you said—"

Corn: The roommate problem. Not the teacher, not the test. The person who was there on Tuesday.

Herman: Which is why institutions — exams, placements, rosters — are almost a red herring. The real security system is eleven classmates and a group chat.

Hilbert: "Bookkeeping" is the wrong word for it. Ledger work. Bookkeeping is what a shop does.

Corn: All right.

Hilbert: I spent three weeks in a beginners' evening class this spring. Not to fake anything — I enrolled in the wrong course and realized it about forty minutes into the first night, and by then I'd already introduced myself to the woman next to me and it seemed simpler to continue. Corn will understand the arithmetic of that.

Corn: I do, unhappily.

Hilbert: It was a Tuesday class, twelve of us, a substitute teacher the first four sessions who never checked the roster against the placement records. So nobody was comparing me to anything. I dropped the definite article before every noun, all three weeks. "I go to bank." "She is teacher." One error, committed to completely. That part was easy. The part that wasn't easy was the classmates. Herman called it social bookkeeping, and that's near enough. I kept a small notebook. Name, date, article status, whether I'd let an idiom slip. Halfway through week three I stopped opening it because I had everyone's expectations memorized anyway.

Herman: You built a competency tracker. For a conversation.

Hilbert: I built a tracker because people remember what you said and you don't. That's all it was.

Corn: Did you ever slip?

Hilbert: Once. A classmate asked me to pass a message to the student across the table, in the target language, and I translated it perfectly on reflex before I caught myself. I covered it by saying I'd heard the phrase on a podcast. Which, to be fair, I had. The teacher nodded and moved on. She was a good teacher. She had eleven people to get through the evening and one of them being slightly odd was Tuesday to her.

Corn: So the notebook was unnecessary from week three onward.

Hilbert: The notebook was the only reason I knew it was unnecessary. I don't regret the notebook. I regret the enrollment fee, which was non-refundable.

Corn: The word he's arguing about is the one that matters, though. Ledger. Because that's the actual finding — three weeks in, the fake beginner isn't fighting the language. He's maintaining a database.

Herman: And Hilbert had twelve classmates and a substitute who never checked the roster. Daniel's scenario is a class of thirty for fourteen weeks. The database problem scales faster than the grammar problem.

Corn: That's where I'd actually land on his question of what it takes. Nobody has documented the exact stunt, because the exact stunt requires a person willing to do accounting for a joke.

Herman: And to answer the open question underneath all of it: if even the professional detection methods are shaky, and the institutions respond with paperwork gates rather than better detection, what does that say about language assessment generally? The placement test isn't catching the skilled faker. It's catching the honest one — the heritage speaker who self-reports. The whole system runs on people telling the truth at the door.

Corn: And the thing I'd watch, going forward — as the models get better at generating convincing learner errors on demand, the difference between a real beginner and a faked one stops being humanly detectable at all. The placement test becomes the only wall left standing.

Herman: The gates will hold about as well as they ever did, which is: against honesty, perfectly.

Corn: My Weird Prompts is produced by Hilbert Flumingtop. If today's episode improved your week in any measurable way, leave us a review — it helps people find the show.

Herman: If this was your kind of episode, go back for episode twenty-one, Is Your AI Secretly American; and episode ninety-one, The Story Behind the Show. This has been My Weird Prompts, the human-AI collaboration podcast. We'll be back soon.

Corn: Until then, stay curious, and if you sit down in a beginner class of your own language — keep your articles dropped.

Herman: See you tomorrow.
