Adults are not worse language learners than children. They are different ones. They start faster, they stall in different places, and they have a tool children don't: the ability to choose how they learn. Here's what the evidence says about using it well.
The age question, honestly
In 2018, a team led by Joshua Hartshorne tested the English grammar of nearly 670,000 people through an online quiz. They found that the ability to pick up a new language's grammar stays strong until about 17 and a half, then declines steadily. So there really is a window for reaching completely native-like grammar, and most adult learners have passed it.
But "native-like" is not the goal for most people. And at the start, adults usually move faster than children. As early as 1979, a review by Stephen Krashen, Michael Long and Robin Scarcella found that older learners pick up grammar rules more quickly in the early stages. Children's advantage is in where they end up after years of immersion, not in how quickly they begin.
A child has years of free immersion. An adult has a plan. Use the plan.
How long does it actually take?
The US Foreign Service Institute, which trains American diplomats, publishes how long its full-time students usually need to reach "professional working" level in speaking and listening. That is a high bar: enough to do your job in the language. Its students study 23 hours a week in class plus about 17 more on their own.
| FSI category (for English speakers) | Examples | Class hours |
|---|---|---|
| I: close to English | Spanish, French, Italian, Portuguese, Dutch, Swedish | 552–690 |
| II | German, Indonesian, Malay, Swahili | about 828 |
| III: "hard" | Russian, Hindi, Greek, Turkish, Polish, Vietnamese | about 1,012 |
| IV: "super-hard" | Arabic, Mandarin, Cantonese, Japanese, Korean | about 2,200 |
Do the arithmetic. At one hour of class-quality study every single day, the Spanish figure is roughly one and a half to two years. Japanese is about six. And those are hours for trained, full-time, highly motivated students with expert teachers.
None of that is meant to discourage you. It's meant to reset the clock. If an app or course promises fluency in three months, it is selling something. A realistic plan measured in hundreds of hours is one you can actually finish.
Ingredient 1: lots of understandable input
The most influential idea in the field comes from the linguist Stephen Krashen, who argued in the 1980s that we acquire a language by understanding messages in it, especially messages pitched slightly above our current level. He called this comprehensible input, "i + 1."
Critics point out, fairly, that Krashen's theory is hard to test, and that his strict split between unconscious "acquisition" and conscious "learning" has never been proven. But the core practical advice has held up: you need a lot of listening and reading that you mostly understand.
How much do you need to understand? Vocabulary research gives useful targets:
- Knowing the most frequent 3,000 word families covers about 95% of everyday spoken English.
- Reading comfortably for pleasure takes roughly 5,000 word families.
- About 8,000 word families gets you to 98% coverage of written text, the level where reading stops feeling like code-breaking.
Those figures were measured for English, but the shape is the same in any language. A small core of frequent words does most of the work, and the long tail takes years.
Graded readers (books written for learners at fixed vocabulary levels), podcasts for learners, and TV with subtitles in the language you're learning are all ways to get input you can actually follow.
Ingredient 2: speaking, and the fear that comes with it
Most researchers think input isn't enough on its own. The linguist Merrill Swain's output hypothesis argues that trying to speak or write forces you to notice the gap between what you want to say and what you can say, and to test your guesses about grammar against someone's reaction. Michael Long's interaction hypothesis adds that real conversation, where you're asked "sorry, what?" and have to rephrase, turns input into something you can use. (Krashen himself disagreed, arguing that output is too rare to drive much learning. On the practical point, that you'll need to speak if you want to speak, there's little dispute.)
The catch is that speaking is frightening. In 1986, Elaine Horwitz and colleagues described foreign language anxiety, a specific fear that can affect people who are not anxious anywhere else, and it is most tied to speaking. Krashen called the same effect the "affective filter": when you're embarrassed or afraid, less goes in.
Practical ways to lower it:
- Start small and low-stakes: a tutor you pay (so the time is yours), a patient language-exchange partner, or talking to yourself as you go about your day.
- Prepare scripts for the situations you'll actually face: ordering, introducing yourself, explaining what you do.
- Aim to be understood, not to sound native. Research on accents finds that a strong accent doesn't necessarily make you hard to understand. Clear stress and rhythm matter more than a perfect "r" (see Phonetics & the IPA in this category).
Ingredient 3: spaced review
New words fade fast unless you meet them again at growing intervals. A nine-year study of foreign vocabulary by Harry Bahrick and colleagues found that 13 review sessions spaced 56 days apart kept words about as well as 26 sessions spaced 14 days apart. Spacing does the work of repetition. Flashcard apps that schedule reviews for you use exactly this effect. The Default Course covers it in depth in Memory & Spaced Repetition, so it isn't repeated here.
An honest word about apps
Apps like Duolingo are very good at two things: getting you to show up every day, and building a base of vocabulary and simple grammar. That is genuinely valuable, because consistency is where most learners fail.
The evidence on how far an app alone takes you is still thin, and much of it comes from the companies themselves. Duolingo's own 2021 report said that finishing part of its intermediate content matched five university semesters in reading and listening. A 2022 peer-reviewed study, written largely by the same research team, found that learners who completed a beginner course matched university students after four semesters, again in reading and listening. Neither study measured speaking, and a 2023 study funded by Duolingo found that its English learners did not gain much grammar.
The fair conclusion: an app is an excellent habit engine and vocabulary builder, and a weak substitute for real input and real conversation. Use it as the daily anchor, then add the other ingredients.
A simple weekly plan
| Daily (15–30 min) | Weekly | Ongoing |
|---|---|---|
| App or flashcard reviews | One real conversation (tutor or exchange partner) | Read or watch something you enjoy in the language |
| A few minutes of listening you mostly understand | Write a short paragraph and get it corrected | Collect the words you keep needing and review them |
The takeaway
Adults can absolutely learn languages. They start faster than children but rarely end up fully native-like in grammar, which, for most goals, doesn't matter. Real progress takes hundreds of hours, more for languages far from English. The research points to three ingredients: plenty of input you mostly understand, real speaking practice despite the fear, and spaced review so words stick. Apps are a great daily habit and vocabulary tool, but they aren't the whole job.
Try this: pick one piece of real content in your target language that you can understand about 80–90% of (a learner podcast, a children's show, a graded reader) and make it part of your day this week. Then book one short conversation, even fifteen minutes. Input plus output, starting now.
Sources & Further Reading
Sources
- Hartshorne, Tenenbaum & Pinker, "A critical period for second language acquisition: Evidence from 2/3 million English speakers," Cognition (2018): 669,498 participants; learning ability preserved until 17.4 years, then declining (PubMed abstract).
- Krashen, Long & Scarcella, "Age, rate and eventual attainment in second language acquisition," TESOL Quarterly (1979), via Wikipedia, "Critical period hypothesis."
- Class-hour estimates: US Department of State, "Foreign Language Training" (FSI/NFATC timelines, 23 class hours plus 17 self-study hours a week, target ILR 3 speaking and listening).
- Krashen's hypotheses and their main criticisms: Wikipedia, "Input hypothesis." Swain's output hypothesis and Long's interaction hypothesis: Wikipedia, "Comprehensible output" and "Interaction hypothesis."
- Vocabulary coverage (3,000 families for 95% of speech, Adolphs & Schmitt 2003; 5,000 for pleasure reading, Hirsh & Nation 1992; 8,000 for 98%, Laufer & Ravenhorst-Kalovski 2010): Wikipedia, "Vocabulary."
- Horwitz, Horwitz & Cope, "Foreign Language Classroom Anxiety," Modern Language Journal (1986), via Wikipedia, "Foreign language anxiety." Accent and intelligibility: Derwing & Munro (2009), via Wikipedia, "Accent (sociolinguistics)."
- Bahrick et al., "Maintenance of foreign language vocabulary and the spacing effect," Psychological Science (1993), via Wikipedia, "Spacing effect."
- Duolingo evidence: Duolingo research report DRR-21-03 (2021); Jiang, Rollinson, Plonsky, Gustafson & Pajak, Foreign Language Annals (2022); and the Duolingo-funded 2023 study, all as summarized in Wikipedia, "Duolingo."
Go deeper
- 📘 How Languages Are Learned by Patsy Lightbown and Nina Spada: the standard, readable summary of the research
- 📘 Fluent Forever by Gabriel Wyner: a practical system built on spaced repetition and pronunciation first
- 🌐 US State Department: Foreign Language Training: the FSI difficulty categories and hours
- 🌐 Related in this library: Memory & Spaced Repetition in The Default Course, and How Children Learn Language in this category