The difficulty of a language is not a matter of opinion and it has been measured. The US Foreign Service Institute has been putting English-speaking diplomats through full-time courses for decades and publishes how long each language takes to reach a working professional level, and the spread is roughly threefold: about 700 classroom hours for Spanish, French, Italian or Dutch, about 900 for German, about 1,100 for Russian, Persian, Turkish or Greek, and about 2,200 for Arabic, Chinese, Japanese and Korean.
Those figures are real and they are also not what decides the outcome, because almost nobody stops for the reason the table describes. They stop in month four, and what determines whether that happens is a different question entirely: whether the language has something to do in your actual week.
The six that decide the choice
- the hardest language you will use beats the easiest you will not. This single rule overrides every other consideration on the list.
- distance from your own language decides the hours. For an English speaker, Dutch and Japanese differ by a factor of roughly four in study time to the same level.
- a reason to use it beats a reason to want it. A person to speak to, a job, a family, a country you will live in.
- access to input matters more than popularity. A language with abundant material and a small speaker count is easier than the reverse.
- script and tones are front loaded costs. They make the first months slow and stop mattering afterwards, so they should not decide the choice.
- related languages transfer. The second Romance or Slavic language costs a fraction of the first.
The table is about somebody who may not be you
Every one of those numbers is measured on a native English speaker, and that qualification is the most ignored sentence in the whole subject. Distance is measured from where you already stand, so the ranking moves once the starting point does. For a Persian speaker, Arabic arrives with the script already known and several thousand words of shared vocabulary already half-learned, which is a large discount the table cannot show. In the other direction, German and French get more expensive rather than less: grammatical gender is not a harder version of something Persian has, it is a category the language does not use at all, and it has to be learned per word for the rest of your life.
The second thing the totals hide is where the cost sits, and it matters more than the size. A new script looks terrifying and is the cheapest kind of difficulty there is: two weeks of unpleasantness, then it is done and never troubles you again. Tone, or a case system, or a verb that has to agree with three things, is spread across the entire first year and shows up in every sentence you try to build. Two languages with the same published total can therefore feel completely different, and that difference is the real content of where the difficulty sits: script, sounds, or grammar.
Month four is where the reason gets tested
This is the stall, and it is the reason difficulty predicts so little. The first three months run on novelty, which is genuine and which expires on schedule. Around month four the material stops being new, the visible progress slows, and the study hour has to compete with an ordinary Tuesday evening on its own merits.
An abstract reason loses that competition every time. Useful for my career is true, and it is a statement about a version of you in five years, and it cannot beat a sofa tonight. What survives is concrete and has a person or a date in it: somebody you want to talk to who does not speak your language, a term abroad with a start date, a family you have never been able to speak to properly, a body of work you want to read in the original. People are often embarrassed by their real reason and give the career one instead, including to themselves, and the substitution is expensive because you cannot plan around a reason you are not admitting to.
The test is worth doing before starting rather than after. Write down what you will actually do with this language in month four, on a specific evening, with a specific person or text. If nothing comes, the language is not wrong; the reason is missing, and it can be arranged before the hours are spent rather than discovered after.
Reasons, tested
Which reasons are still standing in month four
The first three have something specific in them. The next two are true and abstract, and abstract loses to an ordinary Tuesday evening.
Does this reason survive the novelty running out
- YesSomebody you want to talk toA person with a name, not a populationYes
- YesA date you will be thereA ticket or a term, something with a month on itYes
- YesSomething you want to read in itOne named book or one body of workYes
- NoIt is useful for a careerTrue, and a statement about a version of you in five yearsNo
- NoIt is supposed to be the easy onePredicts the first month and nothing after itNo
- Not checkedYou like the way it soundsNobody has measured this, and it may be the best predictor hereNot checked
The three states are this page's judgement rather than a study. The last row is deliberate: liking the sound of a language is the reason people are most embarrassed to give and the one most likely to still be true in a year.
Two things to do before committing a year
The first is to read the Foreign Service Institute's own list rather than the forum argument about it. It is published, it is free, and in 2026 it is still the only estimate built on a large number of adults studying full-time under controlled conditions. It replaced opinion, which is what every other ranking of language difficulty is made of. Read it for the shape and then apply the correction above for your own first language, which no published table does for you.
The second is a week on an exchange app before you commit. Tandem is free at the level this needs and it takes an evening to set up. It replaced choosing from a description and finding out in month four, which is the expensive way to run the same experiment. Spend a week messaging people in each candidate language and notice which one you look forward to opening. That reaction is worth more than the difficulty table, because it is a measurement of the only variable the table leaves out.
The mistake: picking the cheap one and not finishing it
Choosing a language because it is supposed to be easy is the most common decision in this subject and it has a flaw in the arithmetic. An unfinished 700 hours is worth nothing at all: no conversation happens, nothing gets read, and the whole investment stays below the level where the language does anything. A finished 2,200 is worth a great deal, and people do finish those, because nobody sits down to Japanese for a decade without a reason that was there before they started.
The cheapest language, measured in hours per unit of anything you will actually get, is the one you will finish. That is usually the one attached to a person, and only occasionally the one at the top of the easy list. It is also the argument against the tempting third option: two at once looks efficient and mostly halves the rate of both, with the specific exception of pairs distant enough not to interfere, which is settled in when it works, and the pairs that interfere.
Once it is chosen
If the decision is genuinely open and you want the fastest visible return, the shortlist is short and the reasons for it are structural rather than cultural, which is which languages an English speaker gets furthest in fastest. Once a language is picked, the hours in the table become a planning problem rather than a fact about the world, and turning a number that size into weeks is turning a vague intention into hours on a calendar. The method itself does not change with the language, which is the argument of the method, which is the same whichever language you picked, and what does change is what each one asks for first: the sound system, the script, or the grammar.