You're chatting with a friend in your second language. You know the word. You used it last week. Then your brain opens the drawer and finds a vast blank space, so it grabs a clunky backup word and hopes nobody notices. Your mouth fills with the wrong consonants. You laugh, you move on. But something just happened.
That something has a name: language attrition. It is the surprisingly predictable way language skills fade when we stop using them. It happens to immigrants who lose their childhood tongue, to study-abroad students six months after they fly home, and to the rest of us in a quieter way, when a verb conjugation you nailed in college suddenly refuses to come.
The good news for anyone working to keep their languages: attrition is not a mystery, and it is not your language getting deleted from the hard drive. Decades of research have mapped what disappears first, what stays remarkably intact, and what kinds of practice keep the system humming. This post pulls that research together, with the practical question in mind the whole way: what does a learner actually need to do to hold on to what they've gained?
What Is Language Attrition?
Language attrition is the gradual, structured decline of language skills when a language falls out of regular use. It happens to immigrants who lose their first language, to learners six months after an immersion program ends, and to anyone who builds a language and then stops maintaining it. Crucially, the loss of language through attrition is rarely total: research consistently shows the language becomes harder to access rather than absent, and that relearning is far faster than learning from zero. Vocabulary goes first; grammar tends to survive.
What attrition actually is (and what it isn't)
First language attrition vs second language attrition
First language attrition (L1 attrition): changes in your native language when you move into a place where another language takes over daily life. Research on first-language attrition shows this can happen even to adults, though usually in small and structured ways.
Second-language attrition (L2 attrition): the fade-out of a language you learned later, usually after classes, tutoring, or immersion end. A sweeping look at what we actually know about second language attrition finds that classroom-learned language loss has been oddly under-studied, and pulls together the evidence on why some languages stay reachable while others get harder to grab.
What do you forget first in a second language?
What fades first: individual words. Behavioral studies consistently show that attrition is easiest to spot in word meaning and word recall, while grammar, sentence patterns, and pronunciation hold up better, at least in speakers who acquired the language past puberty. In plain terms: the word bank gets messy first. The grammar usually stays standing.
Why words wobble: vocabulary mostly lives in fact memory, the same system that stores names, dates, and what you ate last Tuesday. Grammar leans more on habit memory, the system behind skills like biking, typing, and pronouncing sounds without thinking. Researchers have technical names for these two systems, but the plain-English version is facts versus habits, and the two systems are distinct. Habit memory tends to last longer. Riding a bike after twenty years is wobbly but real. Recalling the Mandarin word for "screwdriver" after twenty years is standing in front of a locked filing cabinet with no key.
What "forgotten" usually means: "I've forgotten my Spanish" is almost always an overstatement. What people usually mean is that word-finding is slow and effortful, that words fail to surface on demand, and that small grammatical bits feel shaky. It rarely means the system is gone.
Why tip-of-the-tongue moments are a warning sign
- The everyday sign: tip-of-the-tongue moments. Across studies, bilinguals report more of them than monolinguals, and people losing easy access to a language report them most of all. One study followed a multilingual speaker for ten years and found that these moments rose and fell with how much the speaker used each language. When you stop using a language, the words don't vanish. They show up late to the party.
The biology of language attrition
Under the hood, attrition starts with memory. Your brain does not store every part of a language in the same way. Some parts act like sticky notes: useful, but easier to misplace. Other parts act more like biking or typing: wobbly after a break, but still there.
Why vocabulary disappears before grammar
| Memory system | What it stores | Attrition pattern |
|---|---|---|
| Fact memory | Vocabulary, idioms, names, and irregular forms you had to memorize | More fragile: words get slower and harder to grab |
| Habit memory | Pronunciation, rhythm, and sentence patterns you run without thinking | Sturdier: the grammar engine usually keeps rumbling along |
That is why "speaking a language" can keep running while "knowing the exact word" gets hard.
Why your stronger language blocks the weaker one
There's a second brain story worth knowing. Brain imaging of adults losing easy access to a language shows changes in how a first language is processed when a second language has become dominant. The first language still lights up the same general brain areas, but with different timing and effort. The system isn't damaged. It's been deprioritized, like an app still on your phone, just shoved onto the last screen and slow to open.
The brain also has a language-control system that works like a traffic cop: it lets one language speak while telling the other to wait. When one language takes over daily life, the traffic cop gets too good at shushing the other one. The bar gets higher. Words stall.
How fast does it happen?
Attrition is a free fall at first, then a long glide: faster than most people guess, then much slower than they fear.
The six-month danger zone after immersion ends
Hermann Ebbinghaus's forgetting curve showed steep early loss followed by a long, flat tail. Language attrition follows the same shape. A six-month study that followed German university students returning from study abroad in Spain found measurable Spanish forgetting within the first half-year home, tied directly to how much each student kept using Spanish. A recent Language Learning journal investigation found the same pattern in advanced classroom-taught learners: the first few years after lessons stop are the danger zone.
Why well-learned languages can survive for decades
Then the curve flattens. Harry Bahrick tested 733 Americans on their school Spanish over a 50-year window. People lost a lot in the first few years, then hit a plateau that lasted roughly thirty years, followed by a gentle final drop. Whatever survived the early drop often stayed there for decades. The parachute opened. The question was how high. The biggest predictor was not vacation Spanish or the occasional song lyric. It was how deeply people learned Spanish in the first place.
The implication is freeing: the language you knew well is mostly still there. It is not gone. It is just being rude about coming when called.
The dormant-versus-lost debate
This is where the field splits, and the split matters for practice.
Is forgotten vocabulary still in your brain?
One view says words are usually still stored, just harder to wake up. In the technical theory behind this view, every word has a kind of mental doorbell. Use it often and one tap is enough. Stop using it and weeds grow over the trail, so the word stalls, arrives late, or gets stuck on the tip of your tongue. On this view, attrition is mostly a reach problem, not an erase problem. A related practice-gap idea says bilinguals get more tip-of-the-tongue moments because each language gets fewer reps than a monolingual's one language.
The best evidence for the "dormant, not lost" view is the relearning bonus researchers call savings. Words you once knew come back faster than brand-new words, even when people swear they remember nothing. The old path is there; it just needs to be walked again.
When childhood language loss really can be total
The counterexample is striking. Christophe Pallier's brain-imaging study of Korean-born adults adopted into French families between ages 3 and 10 found no sign that their adult brains still recognized Korean. The adoptees could not tell Korean sounds apart any better than French adults in the comparison group who had never heard Korean. When they heard Korean sentences, their brain scans looked the same as when they heard a random unfamiliar language. If anything was dormant, it was sleeping very deeply.
The likely resolution: adults often have a paved path under the weeds. Very young children may not. Their concrete was still wet when the language disappeared from daily life. What survives the first few years of disuse may depend on whether the language was ever locked in deeply enough to make it into deep memory in the first place.
Is language attrition loss or interference?
Another research idea predicts that languages fade in reverse order: the newest, trickiest pieces go first. Evidence is mixed. Some rarer verb forms do seem to fade early, but not everything fits. Some researchers argue the cleaner explanation is bleed-through: your other language leaks in and changes how easily you reach the first one.
That sets up the bigger question: is attrition mostly loss, or competition? The competition view, increasingly common in bilingualism research, says both languages are awake at the same time in a bilingual brain. What we call "attrition" is often the stronger language blocking the weaker one from getting through. Practicing your weaker language is partly training the brain's bouncer to let it back into the room. A long stretch abroad can feel like it scrambles your native language less because the first language faded, and more because the second language muscled in.
This same idea explains a strange medical pattern. Older migrants often return toward their first language even after decades of mostly using their second language, as if the brain's control system weakens with age and the original language rises back up. The first language wasn't gone. It was being held down.
Can language attrition be reversed?
For most adults, yes. The relearning evidence is consistent: words and grammar that feel gone are usually dormant rather than erased. The loss of language skills from a period of disuse is almost always recoverable, not permanent. Reactivation moves much faster than original learning, which is why people who "forgot" a language can pick it back up in weeks rather than years. The exception is early childhood: a language lost before it fully settles, as Pallier's adoption study showed, may leave little retrievable trace. But for adults who reached genuine intermediate ability or higher, reversal is the rule.
What attrition looks like in practice
Translating the literature into recognizable scenes:
Losing Japanese after college
A college senior finishes four years of intensive Japanese, scores well on a solid intermediate Japanese exam (JLPT N2), and starts a job with no Japanese in it. Two years later they can still read a menu and follow a TV drama, but speaking has gone choppy and English grammar creeps in where Japanese marker words belong. Classic intermediate attrition: the word bank is mostly intact, but the speaking machine sputters, so small grammar words and endings get shaky.
Understanding a heritage language but not speaking it
A heritage speaker grew up hearing Mandarin at home, switched to English at school around age six, and at twenty-two understands their parents perfectly but cannot hold a fluent conversation. This is not quite ordinary attrition. It is closer to a language built for listening, but never fully built for speaking. Recovery is more like learning to speak with a head start than simple maintenance.
Getting rusty after reaching C1 fluency
An advanced learner (C1) reached fluent Spanish in their thirties through years of work and travel, stopped traveling at forty, and notices at forty-five that friends say their grammar sounds slightly off. This is mild adult attrition. Bahrick's long-term retention work predicts that almost everything they had at that advanced level (C1) is still in there, recoverable with weeks rather than years of focused contact.
Losing study-abroad gains after coming home
A returning study-abroad student watches their hard-won Italian fade within months of going home. Research that follows students after study abroad is clear: this initial dip is real, steep, and highly reversible. Small daily practice during this window prevents most of the loss.
Who loses what, and why
A few factors do most of the predictive work.
Why children lose languages faster than adults
Children are far more vulnerable than adults. Heritage-language research consistently finds that kids who switch dominant languages before adolescence can lose a lot of first-language ability, sometimes dramatically. Adults who emigrate after puberty rarely lose the core of their first language.
Why C1 languages last longer than B1 languages
How well you learned the language in the first place may be the most useful finding here. Bahrick's long-term retention work says it plainly: the deeper you learned it, the longer it lasts. Intermediate plateaus matter, because a language pushed to advanced (C1) fades much more slowly than a language stopped at intermediate (B1). The work you do now to make the language solid can pay off for decades.
How often do you need to use a language to keep it?
Frequency of use is the factor this research keeps returning to. In studies of German speakers living abroad, many outside factors mattered less than expected: age when they moved, attitudes toward German, and years spent away. What consistently predicted outcomes was how much the speaker actually used German. There is no shortcut around contact. The brain keeps what gets used and discounts what sits in storage.
Why similar languages interfere with each other
Similarity between your languages shapes how attrition shows up. Similar languages can be trickier because they behave like wet paint in close colors. Spanish and Italian, for example, do not always vanish cleanly. They can bleed into each other. Spanish-English bilinguals may lose precision on similar-looking words and subtle grammar differences. Distant pairs like Mandarin and English fade differently than close pairs like Spanish and Italian, where the danger isn't loss so much as bleed-through.
Which language skills fade first?
Speaking, listening, and reading fade at different speeds, too. Speaking goes first because it is the high-wire act of language: find the word, build the sentence, move your mouth, check yourself, and do it all before the conversation dies. Listening holds up better because recognizing a word is easier than producing it from scratch. Reading is most stable because the page waits for you, with context clues, rereads, and no awkward silence. So when learners say "my Italian is rusty," they usually mean their speaking has slowed. Their listening and reading are often nearly intact.
| Factor | Effect on Attrition | Practical Implication |
|---|---|---|
| Age when fading starts | Children lose far more, far faster; adults keep the core structure even after decades of disuse | Language gets sturdier with age: adult-learned languages are more durable |
| How well you learned it | The single strongest predictor of long-term retention; advanced (C1) leaks much slower than intermediate (B1) | Pushing to a higher level before reducing use is the highest-leverage move |
| Frequency of use | Most actionable factor; regular contact keeps words easy to reach | 15-20 minutes several times a week significantly slows decline |
| How similar your languages are | Close pairs (Spanish/Italian) show more bleed-through; distant pairs show cleaner but larger gaps | Target maintenance toward the features that differ between your pairs |
| Skill type | Vocabulary fades first; grammar and pronunciation are more stable | Vocabulary review and speaking or writing practice matter more than grammar drilling for maintenance |
| Output vs. input | Speaking fades faster than listening or reading | Keep some speaking practice even when other contact is minimal |
A common worry the research does not support: adult native speakers losing their first language entirely. They essentially cannot. Even after decades of living mostly through a second language, adult-onset first-language attrition produces measurable but small effects. Words may come more slowly. A few grammar choices may get nudged by the other language. But the accent stays. The grammar stays. Dramatic first-language loss in international adoptees is different, because the language had not fully settled in childhood.
A side benefit worth noting
While we're discussing what bilingualism costs you in occasional word-finding lag, it's worth saying what it pays back. Multiple large studies have now found that lifelong bilingual experience appears to delay the onset of dementia symptoms by roughly four to five years on average. The likely reason is mental backup capacity: a bilingual brain gets years of practice choosing one language and quieting the other. The brain that hunts for the right word in two languages is also doing tiny attention-and-control pushups. The tradeoff, slower word-finding in exchange for sturdier scaffolding, seems to favor the multilingual.
Some myths worth retiring
A few stories about attrition circulate that the research doesn't quite back up.
"Once you stop, you lose it forever." False for anyone who reached genuine fluency as an adult. Both the relearning-bonus research and Bahrick's long-term retention work point the other way. The loss is real, but it is mostly slower access, and the relearning curve is steep in your favor.
"You have to immerse to maintain." Misleading. Immersion helps you build the language. Routine helps you keep it. Maintenance can happen at a small fraction of immersion intensity. The polyglots who hold four or more languages over decades are not living in four countries. They've built ordinary daily contact.
"Children pick up languages effortlessly, so they keep them effortlessly too." Wrong, and inverted from the truth. Children learn fast but also lose fast, because their language system has not fully settled yet. Adult-learned languages are more durable in this specific sense: adults build more conscious support around them.
"If you're rusty, you must not have really known it." Probably wrong. Rust can look similar whether you knew the language at intermediate (B1) or near-native advanced (C2); the difference is how deep the recovery has to go. A near-native advanced speaker (C2) with three years of disuse and an intermediate student (B1) fresh out of class can sound similarly halting in one conversation. They are not in the same place at all.
"Using two languages confuses kids and adults." The bilingualism research has spent decades refuting this. If anything, the long-term brain-health findings suggest the opposite.
What actually works for maintenance
Now the practical part. The research converges on a small number of strategies for any language learner, almost all of which exploit the same underlying truth: words and patterns stay easier to reach when you use them.
Can 15 minutes a day prevent language attrition?
Use it often. The memory research is clear: small sessions spread across the week beat one big cram session. Fifteen to twenty minutes of real engagement, several times a week, beats the classic guilt-powered three-hour Saturday cram. That is the spacing effect: flashcards work better when they return just as you are about to forget.
How spaced repetition protects vocabulary
Review vocabulary on a schedule. After the long-term retention work, Bahrick tested what happens when vocabulary reviews are spaced farther apart over time and found that 56-day intervals beat 28-day intervals once learners already knew the words well enough to recall them reliably. Once a word has stuck, reviews can be far apart and still preserve it. Spaced repetition isn't cramming. It's the smallest amount of review that still works.
Why podcasts and TV can maintain a language
Consume input you'd enjoy anyway. The idea behind mostly understandable input is simple: content you mostly understand gives your brain useful practice without making you feel like you're grinding. That's why podcasts and TV in the language you're trying to keep can preserve so much. If the content is interesting and roughly at your level, you're walking the trail often enough to keep the weeds down. Polyglots who maintain four or more languages consistently report that they've folded each language into something they already enjoy doing.
Why speaking practice still matters
Keep speaking. Reading and listening keep recognition strong, but speaking fades faster than understanding. If you want to keep speaking fluency, you have to keep speaking. Even brief conversations with a language partner, tutor, or yourself while cooking preserve the speaking system in a way pure input cannot.
How polyglots maintain multiple languages
Rotate your languages. The polyglot community's practical wisdom matches the research: rotate. One language gets active study; the others get maintenance contact. When a maintained language starts to feel stiff, swap it into the focus slot for a few weeks. Attrition happens by degrees, so small reactivations keep words from getting too hard to reach.
Which weak spots should you review first?
Target shaky zones. Generic vocabulary review is fine, but research on how one language nudges another suggests predictable weak points for each combination. For Spanish-English speakers, that means gender on nouns, the ser/estar split, and little pronouns like lo and le. For Japanese-English speakers, it means particles like wa, ga, and o, plus counters for people, flat objects, and small animals. Practicing those zones intentionally pays more than drilling random words.
The best long-term strategy: push proficiency once
This is the strategic implication of Bahrick's long-term retention finding. A language that was solidly advanced (C1) before you stopped using it daily will sit in your head for decades with minimal touch-ups. A language that was intermediate (B1) will leak. If you're at intermediate now and worried about attrition, the highest-leverage move is a focused push to make the language sturdy. Those gains keep paying you back.
Why relearning a language is faster than starting over
The relearning-bonus finding deserves emphasis because it removes the catastrophizing. If your French has gone fuzzy after three years away, bringing it back takes a small fraction of the work it took the first time. The mental scaffolding is still in place. You are not starting over.
Our experience with language attrition
Our team has an average of 4 languages in maintenance mode each.
What works for us, day to day, is simple: reading and listening in maintained languages on a rotating schedule, at least one conversation partner per language, and quick spaced-repetition reviews for specific vocabulary and grammar patterns we've flagged as fading. The thirty-minutes-a-day rule, distributed across languages, is real. It takes effort, but maintenance stays fun with a system that makes maintenance more automatic.
How Atlas Runa fits
Maintenance in Atlas Runa is built around exactly these findings. The app remembers the vocabulary you've actually learned, works with media you already consume, brings words back before they get fuzzy, and gives you short lessons aimed at the grammar and word zones most likely to get shaky. For advanced learners, where maintenance is the real game, this turns the fifteen-to-thirty-minute commitment from an open-ended chore into a clear routine you can actually keep.
You can't easily prevent the brain from doing what brains do. But you can keep words easy to reach, your hard-won fluency banked deeply enough to last, and your practice frequent enough that the word you want shows up when you want it. The science is on your side. The daily reps are the hard part, and that part we can help with.