Beyond flashcards: how heritage language learners can build fluency with comprehensible input

Rajanya grew up speaking Sylheti, the language of the Sylhet region in northeast Bangladesh. Standard Bengali is close enough that she follows it, mostly. When she signed up for Reelang+ to watch Bengali, she emailed us about how it was going:

“

I speak Sylheti and understand Bengali okay, but I'm trying to improve my Standard Bengali and eventually learn to read it. I can understand most conversational clips and movie and show dialogue, usually with only a few unfamiliar words. News and more formal or literary content can be considerably harder. Some of the words are uncommon or formal, so I don't hear them in my day-to-day conversation.

She isn't a beginner. Casual conversation and TV dialogue are mostly fine. What she is missing is a register (news, formal speech, literary Bengali) and a script. A course that opens with "hello, my name is" teaches her nothing, and a news broadcast is too far above her level to learn much from. This post is about the ground in between, and about getting from there to reading.

Why heritage learners stall differently than beginners

A "heritage learner" is someone who grew up hearing a language at home, from parents or grandparents, but was mainly schooled in another language (usually English). You understand a lot. You might speak it fluently with family, in one register, about one narrow slice of life: dinner, chores, gossip about relatives. Outside that lane, things fall apart fast. Formal vocabulary, news broadcasts, a coworker from a different region, a novel: any of these can leave you lost, even though you've been "speaking" the language your whole life.

Linguist Silvina Montrul, who has published extensively on heritage bilingualism, describes this as a pattern of uneven or "incomplete" acquisition: strong oral comprehension in one domain, weak or absent literacy, and grammar that's fossilized around whatever input a child got before school-age English took over. It's not that heritage learners know less than a typical beginner. They often know more, just distributed strangely.

That mismatch is exactly what a beginner flashcard deck can't handle. Duolingo-style apps assume you're building vocabulary from nothing, one word at a time, in a fixed order. A heritage learner doing that ends up bored translating "the cat is red" while still unable to follow their own grandmother talking to a neighbor about a doctor's appointment. The gap isn't word count. It's exposure to registers and topics you never got as a kid.

The flashcard problem, specifically

Flashcards and spaced-repetition decks are good at one thing: getting a word into short-term recognition. They're bad at:

• Teaching how a word actually gets used in a sentence, at speed, by a real person
• Exposing you to the register shift between "talking to abuela" and "talking to a doctor"
• Building tolerance for the accent or dialect spoken where your family is actually from, as opposed to a textbook's neutral version

For a true beginner, that's an acceptable tradeoff early on. For a heritage learner, it wastes time relearning what you already have (basic vocabulary) while ignoring what you're actually missing (breadth and register).

What comprehensible input actually means here

Comprehensible input is a term from linguist Stephen Krashen's input hypothesis: people acquire language by understanding messages slightly above their current level, not by memorizing rules or vocabulary lists. Krashen shorthanded it as "i+1": input (i) that's just one notch harder than what you can already handle.

For a heritage learner, this reframes the whole project. You don't need input at a beginner's i+1. You need input at your specific i+1, which is probably somewhere between a native speaker's casual register and their formal one. Watching the news might be i+4. Watching a cooking video where someone explains a recipe out loud might be exactly i+1: real speed, real vocabulary, but a topic and structure you can follow.

This is also why heritage learners often report that the moment things fell into place wasn't a course, it was months of watching one YouTuber, one drama, one family member's voice notes, until the accent and rhythm stopped sounding like noise. That's comprehensible input working the way it's supposed to: repeated, leveled exposure, not memorization.

Building an input routine as a heritage learner

Start from your gap, not from lesson one

Skip beginner content entirely. It will bore you and teach you nothing you don't already know. Instead, find native content on the specific topics where you go blank: workplace conversations, politics, technical explanations, formal writing. That's your actual i+1, even if a placement test would call you "intermediate."

Elizabeth grew up in Dallas with Venezuelan Spanish all around her and parents who wanted her to speak English only. In Spanish-speaking families there is a joking name for kids like her: "no sabo kids", after the wrong verb form a child uses when they mean no sé (I don't know). She did exactly this:

“

I'm what they call a "no sabo kid". Growing up in Dallas, I picked up a lot of everyday Spanish just from hearing my parents talk. Now that I'm an adult I want to learn the parts I'm still missing, so I focus on everything that isn't day-to-day: TV news, documentaries about society and politics. That was the gap I needed to fill, and it feels good to get to the point where my Spanish is at an adult level.

For her, i+1 meant Spanish news and politics, a register that lands around C1 and that her parents' everyday Spanish never covered. Her accent was never the problem.

Match the dialect to your family, deliberately

A huge chunk of heritage learner frustration comes from studying a "standard" version of a language that doesn't match what they grew up hearing. If your family is from northern Vietnam and every course you find teaches southern pronunciation, you'll feel like you're learning a different language. This is one of the more underrated barriers, and it's worth being deliberate about closing it. Reelang organizes its video catalogs by regional accent for exactly this reason: its Vietnamese catalog, for instance, filters clips by northern and southern accent so you can match content to the dialect your family actually speaks, instead of a generic textbook standard. Rajanya's case runs the other way round. Her home variety is Sylheti, and what she wants is the standard Bengali of news and books, so the useful clips for her are the ones a notch more formal than her family's speech, not the ones that match it. Reelang's Bangla catalog is real short-form video from native speakers, graded by difficulty, so she can move up from vlogs and drama dialogue toward explainers and news at her own pace. There is a Sylheti feed too, for the days she wants to hear home.

Use real short-form video, not content made for learners

Heritage learners have an advantage most beginners don't: real native speech, at real speed, is usually within reach faster for you, because your ear already has the phonology. TikToks and Reels made by native speakers, about ordinary topics (cooking, family drama, complaining about the weather), tend to hit that i+1 sweet spot better than textbook audio, which is often slowed down and artificially clean in a way real speech never is.

This is the specific gap Reelang was built to close: a scrollable feed of real TikToks, Reels, and YouTube clips from native speakers, graded by difficulty and filterable by accent and topic, in 40 languages. For a heritage learner, that means you can skip beginner material entirely and start with clips at your actual level, on subjects you care about, in the dialect your family speaks. Focus mode lets you slow down and study a sentence at a time when something's genuinely new; Flow mode lets you just watch, the way you'd watch anything else, once a video's easy enough to follow without stopping.

Read after you listen, not before

If you grew up with a language only spoken, not read, resist the urge to start with a grammar book. Krashen's model and most subsequent heritage-language research agree on this: listening comprehension built through exposure comes first, and literacy and formal grammar layer on top of it far more easily once the ear is trained. Trying to learn to read a language you can't yet parse by ear is backwards, and it's a common reason heritage learners give up on formal classes that start with the alphabet and grammar tables.

The script: the part your family couldn't give you

For many heritage learners the biggest gap isn't vocabulary at all, it's reading. Rajanya put it last on her list, after vocabulary and formal Bengali, and asked for transliteration to help her get there. That ordering is sound, because a heritage speaker comes to the script with something a beginner doesn't have: they already know most of the words they are trying to decode.

Persian shows this clearly. Children of Iranian families in Los Angeles, Toronto or Hamburg often speak Farsi fluently at home and can't read a menu in Tehran. Persian script usually leaves short vowels unwritten: کتاب (ketâb, book) is spelled k-t-â-b, and the reader supplies the missing e from knowing the word. For a beginner that is a wall: you can't pronounce a word you have never heard. For a heritage speaker it is closer to a puzzle with the answer already in their head. Sound out the consonants, and the word they have heard a thousand times usually surfaces.

The same holds for Bengali's script, Arabic, Hindi's Devanagari, Korean Hangul and most other scripts heritage learners run into. A few things make the step easier:

• Learn the letters in a week or two, then stop drilling them. The alphabet itself is small. What takes time is recognizing letters fast enough to read words, and that comes from reading, not from charts.
• Read things you have just heard. Subtitles in the original script are ideal: you hear a sentence, then see how it is written, with the sound still in your ear. On Reelang the captions are in the language's own script, you can tap any word to see what it means, and Focus mode goes through a clip one sentence at a time, so a video you already half understand doubles as a reading lesson.
• Use romanization as a bridge, not a home. Plenty of diaspora families text in romanized Farsi ("Finglish") or romanized Bangla, so the Latin letters are already familiar. For languages written in another script, Reelang shows a second subtitle line under the original: a romanization, or for Russian the same Cyrillic with stress marks added. Read the script first and glance down only when you are stuck. If you only ever read the second line, you stop learning the script.
• Expect the formal written register to feel like a second language. Written Bengali and written Persian both carry vocabulary that rarely comes up at home, which is the register gap from the start of this post showing up in print.

Let speaking lag behind on purpose

You don't need to force output before you're ready. Comprehension expands first; production catches up once you've absorbed enough patterns to reuse them instead of translating word by word in your head. If you're already comfortable speaking in one narrow register (home, family), that's a head start most learners don't get. Widening your listening range will widen what you can say, in roughly that order.

FAQ

Is comprehensible input enough on its own for a heritage learner? It builds the listening and intuitive grammar base that formal study can't. Most heritage learners still benefit from targeted grammar or literacy work once comprehension is solid, but input should come first, not last.

What if I can't find content in my family's exact dialect? Get as close as you can. A closely related regional accent will still train your ear better than a "neutral" studio voice designed for absolute beginners. Catalogs organized by region, rather than one generic feed, make this easier to search for.

Why do I understand my grandmother but not the news? Different registers use different vocabulary, sentence structure, and pacing. Childhood exposure usually covers one register deeply and others not at all. Deliberately seeking out content in the missing registers (workplace talk, formal media, technical topics) closes that gap faster than general study.

I speak my family's language but can't read it. Where do I start? Learn the alphabet quickly, then read things you have already heard: subtitles, song lyrics, captions on videos from creators you follow. Your spoken vocabulary does most of the decoding work, which is why heritage learners often learn to read faster than beginners who studied the language for years.

Should I use a flashcard app alongside video input? Sure, for looking up specific words you keep missing. Just don't expect a flashcard deck to teach you the thing you're actually missing, which is fluency across registers and dialects, not raw vocabulary count.

The specific takeaway

If you're a heritage learner, the fastest progress usually comes from admitting you're not a beginner and stopping the beginner exercises. Go find the register you're missing, in the accent your family actually speaks, and spend time with it regularly, even at 60% comprehension. That discomfort is the work. Reelang's video feed, filtered by accent and topic across 40 languages, is one place to go looking for exactly that kind of input once you know what you're missing.