Every year, thousands of learners spend a semester and a large amount of money in a country that speaks the language they want, on a premise nobody has ever had to argue for: that being surrounded by a language is the same as learning it.
The premise has a name in the research literature. Sharon Wilkinson called it the language myth — the belief "that study abroad will ultimately and inevitably lead to language acquisition due to the sheer number of hours students spend simply exposed to the language." Her 1998 ethnographic study of four American students on a summer programme in France found that English, not French, was often the primary language they used outside the classroom.
That is not a story about lazy students. It is a story about contact, which is the variable that actually matters, and which turns out to be far less automatic than the postcard version of immersion suggests. This article is about what the study-abroad data really shows, where it disagrees with itself, and how to build the same conditions at home if you cannot go — or how to spend the money well if you can.
First, the Honest Version of the Evidence
Study abroad is not useless. Two meta-analyses have now measured it, and both land in the medium-to-large range.
- A 2016 meta-analysis by Jin-Suk Yang reviewed the methodology of 66 study-abroad studies, then ran the quantitative synthesis on the 11 that compared study-abroad learners with at-home learners. Across 65 effect sizes, the mean weighted effect size was d = 0.75 — a medium effect that comfortably beats most classroom interventions.
- A 2024 multi-level meta-analysis by Tseng, Liu, Hsu and Chu in Language Teaching Research pooled 42 primary studies published between 1995 and 2019 and 283 effect sizes in a three-level random-effects model. The pooled effect was g = 0.87, medium-to-large.
So the average learner who goes abroad does gain — more than the average learner who stays. The interesting question is what is inside that average, because the spread across studies and across individuals is enormous, and the gains are not spread evenly across skills.
What Study Abroad Is Actually Good At
When the research team led by Barbara Freed at Carnegie Mellon split language ability into its components and compared contexts, a consistent profile emerged. A review of that multi-study project in Frontiers described the pattern: study-abroad learners were ahead on oral proficiency and fluency, pronunciation, lexical development and narrative ability — the capacity to produce extended stretches of connected talk.
That profile makes sense mechanically. Fluency is a skill of retrieval speed and automatisation, and it is built by using the language under time pressure with real interlocutors. A programme that floods you with authentic, fast, unpredictable speech and forces you to answer is well designed to push exactly those dials.
It is also worth knowing that gains in the first weeks or months abroad are not uniform in time. Learners commonly report an early period of feeling worse, not better — more tired, less articulate, more aware of what they cannot say. The fluency curves in the aggregate data rise, but the subjective experience of the first month is often a drop.
Where Study Abroad Loses — and the Classroom Wins
This is the finding that should change how you plan a semester overseas. In a study published in Studies in Second Language Acquisition in 2004, Joe Collentine compared 46 learners of Spanish in two contexts for approximately one semester: a study-abroad group in Alicante, Spain, and a formal classroom group at an American university. He tagged transcribed Oral Proficiency Interviews for grammatical and lexical features and ran discriminant analyses on the gains.
The at-home group, not the study-abroad group, showed the greater development on discrete grammatical features. The clearest number: the classroom group increased its accuracy in subordinate conjunction selection by about 22%, while the study-abroad group's accuracy stayed flat.
The Study Abroad experience did not result in overall improved grammatical abilities. As reported in Collentine (2004), the At Home group, if anything, showed superior gains in grammatical performance. — Collentine & Freed, Frontiers: The Interdisciplinary Journal of Study Abroad
This replicates rather than contradicts earlier work. DeKeyser's 1991 study of a semester abroad in Spain, and Freed's 1995 chapter asking bluntly "what makes us think that students who study abroad become fluent?", had already raised doubt about grammatical gains from immersion alone. The honest synthesis of three decades of Spanish studies is that the two contexts produce different learners: the classroom learner accumulates explicit accuracy on the forms the classroom drills, and the study-abroad learner develops the ability to keep talking — more semantically dense, more narratively complex, more automatic.
If you want the accuracy, you have to keep doing accuracy work while you are abroad. Immersion does not take over that job. Which is precisely why learners who arrive abroad with stronger explicit grammar tend to leave with the most dramatic spoken gains: the work at home is what gives the immersion something to automate.
What Students Actually Do Abroad
Here is the part that decides most outcomes. Programme location supplies opportunity; learners supply contact. The two are not the same variable, and the measurement literature is fairly brutal about the gap.
- Peer time beats host time. A Frontiers study by Savicki and colleagues found that study-abroad students spend approximately twice as much time in contact with each other — co-national peers — as with individuals from the host culture. Wilkinson's later work gave the phenomenon a name: the compatriot island.
- The language of the island is your own. In the French fluency study by Freed, Segalowitz and Dewey (2004), the study-abroad group reported using English more than French outside the classroom. In the three-context comparison in the same paper, that group's reported out-of-class language use was skewed towards English in a way the intensive domestic immersion group's was not.
- Community interaction is thin. Mendelson (2004) analysed what learners' out-of-class contact actually consisted of and found "limited spurts to fulfil very specific functions" — bus drivers, shop clerks, travel agents, waiters. Service encounters are real language, but they are short, scripted and forgiving, and they do not require you to hold the floor.
- The homestay is not a guarantee. Magnan and Back (2007) and Rivers (1998) compared learners in homestays and residence halls and found results that contradict the assumption that family contact produces better linguistic outcomes. In Segalowitz and Freed's 40-learner Spanish study, time spent speaking with the host family showed a small negative correlation with gains in the length of the longest speaking turn — the authors' explanation being that host-family exchanges were largely greetings, requests and short formulaic chit-chat, which trains comfort but not holding the floor.
Read those four together and the "immersion" in study abroad often reduces to a few hours a week of low-stakes exchange, wrapped in a country where you buy coffee in the target language. That is not nothing. It is also not what the price tag implies.
The Result That Should Unsettle Everyone
The most instructive comparison in this literature puts study abroad up against an immersion programme at home. Freed, Segalowitz and Dewey compared 28 students of French across three contexts in one semester: a regular classroom at home, an intensive summer immersion programme, and a semester abroad in France.
The intensive domestic group — domestic, meaning on campus in the United States — made significant gains on total words spoken, length of the longest turn, rate of speech, and a composite fluidity measure. The study-abroad group made a significant gain on speech fluidity only. The at-home classroom group made no significant gains on any of them. The immersion students also reported speaking and writing French for significantly more hours per week than either other group.
The lesson is not that a short intensive programme beats a semester overseas. It is that time-under-conditions beats location. A seven-week programme with a language pledge, structured speaking time and no English available delivers more hours of concentrated interaction than a fifteen-week semester in which the learner's social default is a cohort of English speakers.
Who Gains Most — and Why That Is Predictable
The second-most useful study for planning purposes measured which learners improve within one short-term programme. The 2020 Modern Language Journal study of short-term summer study abroad in Spain separated intermediate and advanced learners and tracked grammatical and lexical development, plus the amount of contact each learner had.
The intermediate group improved substantially — on the grammaticality measure the effect size was d = 2.27, very large by any standard. The advanced group's gain was d = 0.44 and not statistically significant. Same programme, same country, same number of weeks: one group moved a great deal, the other essentially held still.
That pattern has two explanations and both are actionable. First, ceiling effects: advanced learners have less headroom on the task being measured. Second, and more interesting, the thing short-term immersion supplies — high-volume, low-stakes, repetitive spoken interaction — is exactly what a mid-proficiency learner needs and exactly what an advanced learner has already banked.
Related work sharpens the individual-difference side. Brecht, Davidson and Ginsberg's multi-year study of American learners of Russian found that pre-programme grammatical ability was a strong predictor of gains in oral performance — the learners who did the most formal preparation before departure were the ones who profited most from being thrown into conversation. And O'Brien, Segalowitz, Freed and Collentine (2007) found that phonological memory, the ability to hold sound sequences in working memory, predicted which adults gained most in oral fluency.
What Contact Actually Has To Look Like
Since contact is the mechanism, the research on contact types matters more than the research on destinations. Freed's 1990 chapter split out-of-class activities into interactive contact (talking with people) and non-interactive contact (watching television, listening to the radio, reading). Interactive contact was associated with oral proficiency gains; non-interactive time largely was not.
There is a complication honest enough to include. In Segalowitz and Freed's Spanish study, the amount of reported out-of-class contact turned out not to explain the difference in gains between the at-home and study-abroad groups, and the median self-reported contact was around 18 hours per week. The authors' own conclusion was that the relationship between contact and gain is weak and indirect, and that the causal story runs partly in reverse — learners with stronger initial oral ability choose more contact, which makes the correlation look better than it is.
What survives all of that is a narrower and more useful claim. Contact is necessary. Hours alone are not sufficient. The kind of contact matters — sustained, negotiated, non-formulaic, with someone who will not simplify the moment you look confused.
A Seven-Step Protocol for Engineering Immersion
This is the practical part, and it works whether your budget includes a flight or not.
1. Choose the skill before the destination
Write down which of the two profiles you are buying. If you want fluency and narrative range, exposure and interaction will give it to you. If you want accuracy — gender agreement, case, mood, verb morphology — the data says the classroom is at least as good and possibly better, and you must schedule it deliberately. Learners who book a flight hoping to acquire accuracy passively are the ones who come home fluent and wrong.
2. Measure before you go, with your own voice
Record a two-minute monologue on the same three prompts before and after: one narrative (what happened), one opinion, one explanation (how something works). Transcribe both. Count words per minute, count silent pauses over two seconds, count self-repairs, and count how long your longest uninterrupted run was. The Language Contact Profile developed by Freed, Dewey, Segalowitz and Halter is designed for exactly this kind of self-tracking, and you can keep a simplified version in a note file. Numbers you recorded yourself will keep you honest in week six, when it feels as though nothing is moving.
3. Budget contact hours, not weeks
Multiply the weeks by the daily hours of interactive target-language talk you can realistically engineer, and compare offers on that number. A destination with a cheap flight and a cohort of fellow English speakers scores badly on it. Then track it — the number of real conversation hours per week is the single most useful log a learner can keep, and it is the one that reveals when a housemate arrangement has quietly turned into an English-speaking one.
4. Start every relationship as a target-language relationship
The compatriot island forms in the first fortnight, and once an English-speaking friendship exists, it has a gravitational pull that is very hard to escape in month three. Do the awkward thing early: join the club, the choir, the amateur team, the volunteer rota — anything with recurring meetings, the same faces, and a reason to talk that is not language practice. Recurring is the operative word. Novelty provides interaction; repetition provides automatisation.
5. Keep the awkwardness in the loop, deliberately
Conversation partners who adjust to your level are kind and less useful. The high-value contact is with someone who speaks at normal speed, will not switch to English, and will let a silence sit there while you assemble the sentence. You can also manufacture the pressure domestically: real phone calls instead of chat threads, in-person exchanges with no script, timed speaking on a topic you did not prepare. If it feels mildly uncomfortable, it is probably generating the fluency the data is measuring.
6. Bring the corpus home into your deck
Immersion gives you breadth of input and almost no systematic consolidation. What it does provide, better than any textbook, is a stream of authentic sentences containing words you have just discovered you need. Sentence-mining is the bridge: when a phrase fails you in conversation or when a native speaker says something you could not have produced, capture the whole sentence, not the word. Keep the note file on your phone during the day and convert it into cards in the evening. This is the step that converts exposure into retrieval strength, and it is the step almost everybody skips, because simply being abroad feels like study enough.
7. Build the at-home version as a block, not a lifestyle
The domestic-immersion finding is the best news in this literature, because it is reproducible without a visa. Take a defined block — two to eight weeks — and apply self-imposed conditions: a no-English rule on your own phone and laptop, a fixed daily conversation hour with a tutor or exchange partner, target-language media as your default entertainment, and one public deliverable at the end (a recorded five-minute talk, a letter, a presentation to somebody who will actually listen). The defining feature is not heroism, it is that the conditions stop being optional for a fixed period. Then, when the block ends, expect the fluency to decay unless you keep a maintenance floor of weekly conversation — the attrition literature is unforgiving about unused skill.
The Rule Behind All of It
Study abroad is not a treatment you receive by being somewhere. It is a set of opportunities that must be converted, at a rate that depends on how many hours you spend doing the difficult, socially risky thing — speaking to people who are not making it easy.
What predicts outcomes is therefore not your destination, or even your motivation in the abstract, but the concrete structure you build: how many hours of interactive talk you generate per week, whether the same faces keep appearing, whether anyone refuses to simplify, and whether you write anything down afterwards. Country is a multiplier on that structure. It is not a substitute for it.
And the part of it you can control is larger than you think. The learners who gain most abroad are, in the data, the ones who had accumulated the most explicit knowledge before they left — the ones for whom immersion is not the start of learning but the automation of it. That preparation is a flashcard routine, run consistently, in the weeks before anybody stamps your passport. Build the vocabulary and the patterns first; then find the people who will make you use them at speed.