Language Matching.

Via John Cowan, who writes:

I thought the chart at Language Matching might interest you and the Hattics. It gives a list of language pairs, the one you want and the one you can get, and gives you a distance metric between them from 1 (basically no issue) to 80 (maximally distant pair). This is not necessarily about genetic relatedness or mutual intelligibility: for example, the distance from Breton to French is 20, the same as from Catalan to Spanish, because if you want the first you can probably cope with the second, even though the genetic distance is much higher in the first pair than in the second.

I’m not quite sure how it works, but it’s certainly impressive, so take a look.


  1. David Eddyshaw says

    Took me a while to see what this is about, but I think I’ve got it: basically, if I ask for a webpage or something like that in Breton, and you haven’t got one, it’s better to fob me off with French than (say) Russian (or even Welsh.)

    What language that we’ve actually got is the user most likely to understand, given what they’ve actually requested? It’s an interesting question …

  2. John Cowan says

    Just so.

  3. David Eddyshaw says

    One interesting thing is that this can’t work like a distance metric at all, because it’s clearly asymmetric: on the whole, most users requesting French would not be content to be presented with Breton instead (rather, the obvious second choice would of course be Mooré.)

  4. As is well known, I learned French (well, just reading) initially from French Breton textbooks. I don’t regret of course, French is wonderful. But if the textbooks were in Breton, I would also have learned Breton.

  5. Czech-Slovak is 20 but Yiddish-English (cross-scriptually, even) is 10? I don’t get this at all.

  6. Yiddish to English is 30, but the chart is still odd. I gather this is measuring “cultural proximity”. I.e the odds that a Mongolian speaker can use a Russian language application are higher than for other languages. Most Yiddish speakers today have a better command of English than German (or Modern Hebrew?), but the odds a Yiddish speaker can default to English are somewhat lower than the odds a Czech speaker can default to Slovak. Still unclear how these distances are measured.

  7. “What language that we’ve actually got is the user most likely to understand,”

    Reminds me about the concept of Dachsprache.

    Consider Moscow 1995. Every computer user is familiar with English (knows words “File”, “Copy”, “MkDir”…) but even those who studied English in school don’t know it well (that is: when they see a line, they struggle with it and often defeat it. When they see a text, that would mean struggling with every line).
    Belorusian is a “funny” language.

    If you see an unfamiliar English name (a button, or a class in a video game) you can look it up in a dictionary but no one ever does, or you can ask around, or you can just remember that “rename” is the name of this operation (“move the fail to another directory”:-/), whatever the word means (and that’s what I did).

    When you see a name in Belarusian it is much more likely to be familiar than a name in English. But if it is unfamiliar, you giggle (and then ask around or just remember that it is the name of the operation).

  8. January First-of-May says

    move the fail to another directory

    The file, of course, but I’m pretty sure I’ve seen this exact transliteration of Russian файл in example file names in some programming book.

  9. I think in 90s I would not make this typo or mistake or what it was.
    I did not know “fail”, I wrote in English slowly and much more accurately and when you see “file” first you read it as if it were in Latin. In 1995 “файл” already entered Russian, but I still remembered some English spellings in Latin reading – in case I need to explain the spelling to someone.

Speak Your Mind