Is English hard to learn?

Every answer to this question lists the same quirks: the spelling, the phrasal verbs, the articles, the twelve tense forms. None of that tells you whether English will be hard for you. What does is the distance between English and the language you already speak, and that distance has been scored, tested against half a million census records, and can be read against the newest country rankings. This page prints the table.

What is the short answer?

English is hard in proportion to how far your first language sits from it, and for most of the world's learners that is the whole answer. A Dutch or Swedish speaker starts with the same basic grammar, the same alphabet and thousands of shared everyday words. A Japanese, Korean or Arabic speaker starts with none of those. The two learners are not studying the same language in any practical sense, and no list of English's quirks can capture the difference, because the quirks are the same for both and the starting point is not.

That starting point has a number. In 1993 the US State Department's Foreign Service Institute recorded how much speaking proficiency its American students of average ability reached in 43 languages after 24 weeks of full-time training. Economists Barry Chiswick and Paul Miller turned that record into a scale of distance from English, from 1.00 for the languages Americans found hardest to 3.00 for the ones they found easiest, and then asked whether the scale predicted how well immigrants from those languages learned English. It did, strongly. The rest of this page is that table, that test, and what the 2025 country rankings say about it now.

On the Foreign Service Institute language-score scale, Japanese and Korean are the languages farthest from English at 1.00 and Afrikaans, Norwegian, Romanian and Swedish the closest at 3.00, with Hebrew, Turkish, Polish and Thai at 2.00 in the middle (Chiswick and Miller, IZA Discussion Paper 1246, 2004, Table 1).

How far is your language from English?

The full table is below, sorted from farthest to closest. Read it with one thing in mind: the score is the proficiency an English speaker reached in that language after a fixed 24 weeks, so a low number means the language was hard for Americans, and the scale becomes a measure of distance from English only on the assumption, which the authors state openly, that distance runs the same in both directions. The seven languages at the bottom are exactly the ones the State Department still budgets at 88 weeks for its own staff.

Two features of the table are worth pausing on. First, the languages at the bottom share nothing with English at any level: not the writing system, not the word order, not the sound inventory, not the vocabulary. Arabic, Mandarin, Japanese and Korean each use a script that is not the Latin alphabet, and each has a grammar that marks things English does not and ignores things English insists on. Second, the middle of the scale is crowded and diverse. Hebrew, Finnish, Thai, Turkish and Polish all sit at 2.00 despite belonging to five different families, which is a reminder that the score measures how long the learning took, not how the languages are related.

Does that distance actually change who learns English?

This is the part of the study that the SERP never quotes, and it is the part that answers the question. Chiswick and Miller took the 1990 US Census, 237,770 foreign-born men and 243,496 foreign-born women aged 25 to 64 from non-English-speaking countries, and asked what predicted whether a person spoke English well, holding constant their schooling, age, years in the country, marital status, where they lived and how many neighbours spoke their language. Distance from English was, in their words, “a highly statistically significant variable for both men and women”.

Among 480,000 immigrants in the 1990 US Census, moving from a first language closest to English (Swedish) to one farthest from it (Japanese) reduced the probability of being proficient in English by about 17 percentage points, the same effect as about 5.4 fewer years of schooling, other things equal (Chiswick and Miller, 2004).

The Canadian census gives the same result as a timeline, which is the form most learners will recognise. Five years after arriving, one in four immigrant men from a 1.0 language could not yet hold a conversation in English or French; from a 3.0 language it was one in twenty-two. After fifteen years the gap had narrowed but not closed, and the share who had switched to speaking English or French at home was more than twelve times higher at the near end of the scale than at the far end.

One caution belongs here, and the authors give it themselves. When they added a separate variable for each country of birth, the distance effect shrank sharply for women and disappeared for men, because language and country overlap almost perfectly: Korean is spoken in Korea, Italian in Italy. So the census cannot fully separate “my language is far from English” from “I grew up in a country where English was rare”. For a learner the distinction hardly matters, since both travel with you, but it is why this page reports the effect as the authors framed it and not as a law.

Do the 2025 country rankings still follow the scale?

The distance scores are three decades old, so the fair test is whether they still predict anything. The EF English Proficiency Index 2025 scores 123 countries from the tests of 2.2 million adults, with a global average of 488. Below are the 27 countries whose largest first language appears in the 1993 table, sorted by that language's distance score, with the global average marked. The grouping is ours; every score and rank is from the EF country page.

In EF EPI 2025 the five countries whose main language scores 2.75 or higher on the FSI distance scale (Netherlands, Norway, Sweden, Romania, Malaysia) average 606, and the six whose main language scores 1.50 or lower (South Korea, Vietnam, China, Egypt, Japan, Saudi Arabia) average 466, against a global average of 488 (EF EPI 2025, grouping computed by SpeakBase Research).

The gradient runs the right way through all four groups, 606 to 554 to 534 to 466, and the exceptions say something useful too. South Korea sits at the very bottom of the distance scale and above the global average on proficiency, which is what decades of heavy investment in English teaching look like against a 1.00 starting point. Thailand sits in the middle of the scale and second from the bottom of the ranking. Distance sets the price; it does not decide whether a country pays it.

How many hours does English take?

The figure most people are looking for comes from Cambridge English, which publishes an approximate count of guided learning hours per CEFR level. It is cumulative from a standing start and it is an average over every first language on earth, which is exactly the axis this page has been arguing matters. Read it as the middle of a wide band.

Cambridge English estimates approximately 200 guided learning hours per CEFR level, or about 500 to 600 hours from beginner to B2 and 700 to 800 hours to C1 (Cambridge English support article, updated 12 September 2026).

How wide is the band around that average? The best available proxy is the same State Department table read in its usual direction. For an English speaker aiming at a professional level, the FSI budgets 552 to 690 class hours for Dutch, Swedish, Spanish or Italian and 2,200 for Arabic, Mandarin, Japanese or Korean, a factor of four. If the symmetry that the distance scale rests on holds even roughly, a Japanese speaker should expect English to sit well above Cambridge's 700 to 800 hours to C1 and a Dutch speaker well below it. That is the honest version of “how long does English take”: it is a range, and your first language tells you which end of it you are on.

Which part of English is hard for everyone?

The spelling, and there is a measurement for that as well. In 2003 Seymour, Aro and Erskine tested first-year readers across European languages and found that children in most of them, Finnish, Greek, Italian and Spanish among others, were accurate and fluent at basic word reading before the end of the first school year. English was the outlier. The abstract names French, Portuguese and Danish as slower and English as the slowest by a distance: “the rate of development in English is more than twice as slow as in the shallow orthographies”. Those are native-speaking children, not learners. The spelling system is hard before any second-language distance is added.

The reason is that English writes one sound many ways and one spelling many sounds, so a reader cannot rely on letters alone and has to build a second, word-by-word memory alongside the alphabetic one. For a learner this is the one difficulty that does not shrink with a close first language: a Dutch speaker gets the grammar and the vocabulary almost free and still has to learn that though, through, tough and thought share four letters and no sounds. The other famous quirks, phrasal verbs, articles, the tense system, are real but pair-specific: articles are a struggle for a Japanese or Russian speaker and a non-event for a French one.

Frequently asked questions

Related research

The same State Department table read in its own direction, for an English speaker choosing a language, is easiest languages to learn for English speakers. What its hours actually count, alongside the ILR scale and the Cambridge table, is how long does it take to learn a language. The EF EPI skill sub-scores that show reading outscoring speaking in every country measured are in the reading versus speaking gap, and the five primary counts behind the question of how many people speak English are read side by side there.