Hacker News Reader

Comment by phantomathkg | Hacker News Reader

i18n and l10n are something that I have seen a company done right. It is easy developer to assume:

If your IP is coming from country X, you must want the content to be served in language X.

No, there are tourist from country Z, long term resident who prefer language A and people from country X want to learn language B.

- If your browser Accept-Language say X,Y, then you must want all the content to be served in X.

No, I want my search result to be predominantly in X, but when I search for things about Y, show me language Y, and when I search for this band from country Z, please show me in language X.

As a hongkonger (zh_hk + en_gb), living in Singapore (zh_cn + en?), following JPOP. This is the daily fight I have with browser.

I would rather all application, including web app just give me the option to choose and say, interface language, english, content language, follow origin.

View on HN

I also have this problem with apps in my native language. I have my phone set to en,no because I prefer the English version over some bad auto-translated crap where some apps try to translate stuff. But then when I download a Norwegian app, then it prefers to use some badly maintained English translation instead, but here I would of course prefer the native version.

Can't win.

And it would be _so, so, easy_.

Make reasonable assumptions or provide good defaults. Make them overrideable. Make user settings stick.

I have the same problem, by the way — my phone is in English, which means I get to enjoy Apples hilarious English pronunciation of German street names while navigating.

Yeah, I think apps just should have a sorted list (/tiers) of languages they support, and pick whatever first has a match in the user preferences. If the app supports two languages equally well, then choose the one first in the user's preference list.

So for me having no,en (in the future where this works I would dare to have no first):

  app is in english, has auto translated norwegian: choose english
  app is in norwegian, has auto translated english: choose norwegian
  app is in norwegian and english equally: choose norwegian
  app is in french, has english translation: choose english

People figured this out when they specified HTTP 1.1 in 1997. A prioritized list of the languages the user knows. They even allow a q-factor weighing so the user can specify a factor of how comfortable they are with a language. All modern browsers already support this, but 99% of websites fail to use it properly including Google.

It's like modern apps have forgotten all lessons learned about internationalization.

Same in Windows. The "modern" system apps use just one setting (Windows language) for internationalization, ignoring the old date format and time format settings. So if I set my Windows language to English, I get AM/PM and dot as decimal separator in some parts of the UI and not in others.

https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/...

Indeed.

Google is notorious for ignoring browser language preferences not only in Youtube but also in its main product, and inferring from the (often faulty) geolocation.

Same here with Dutch: you could pretty much watch i18n degrade real time by observing the quality of the dutch translations for all major sites and tech products over the years.

Back in the 2000s, dutch wasn't very common but it usually was pretty good (my understanding is that one of the things that helped is that everyone followed the Microsoft style guide for dutch translations?)

Nowadays you get overly literal translations (meaning some form of MTL), translations that don't care for the length of the text (so it gets cut off with ... at the end for interface buttons) and so on and so forth. It all just reeks of automatic translation with little care put into the presentation. This is pretty much a universal experience across every single system I've ever used and why I usually just set all my devices to English. - It's simply not worth it to deal with the botched translations to try and figure out what was actually meant.

I've had the unfortunate mishap of having the perfectly fine English translation of a German site switch to auto translate Dutch crap due to one part using the language preference and the other the IP source, bit weird since as far as I know they usually only offer English. And I can perfectly fine read German so I usually immediately switch to that since some info is not available in English especially with the different Bundesambt websites.

What I would like is a browser and os which allows you to set which languages (multiple) never need translating and the site sticking to that.

On iOS, this is something you can can control on a per-app basis from the (system) Settings.app.

Android has it too.

And for some reason, Apple/iOS doesn't allow you to set this for some system apps. For example, Music (Apple Music) app on iOS doesn't offer this option, which I desperately need because I don't want the music metadata I listened got translated.

Really useful for Google maps. I generally prefer English but I want Japanese for google so I get local people's opinion over tourists who think every ramen is the best ramen they've ever tasted. I'm surprised it's not an option actually. Language aside, tourism and actually living somewhere are totally separate and reviews should reflect that.

Interesting use case! Why won't you just use Tabelog instead?

I don't feel like tabelog is that popular recently. But I could be wrong.

Still miles ahead of Google Maps Review (in Japan) at least.

I live in Japan and am not yet fluent enough, so need to use a VPN to get the Google Search English results, otherwise it's totally unusable. Heck even for programming I pull my hair since I get the translated docs when searching. Just give me the (original) English ones please. I did search in English have Google set to English, but still get the Japanese ones even for those that have English resources.

Hm.. I don't seem to have that particular problem. I use VPN sometimes, but for other reasons, so I don't normally do my searches there. With the native IP I get both English and Japanese results, depending on what it finds, but for multi-language sites in Japan I actually get English sometimes. If I only want results from Japan I prefix my Google search with site:.jp, or whatever for other places. I never get auto-translated documents. I get what I would expect, actually. With a bias from the country (in this case Japan) I am in, but that's about it.

Google Search seem to be... becoming hyper-localized in the last 5-10 years. They seem to simply refuse to give back a lot of pages, by 15-ish-mile IP locality. It's sort of super alarming.

> i18n and l10n are something that I have seen a company done right. It is easy developer to assume:

> If your IP is coming from country X, you must want the content to be served in language X.

I would assume there are multilingual speakers in mostly every single team at YouTube. Or at the very least enough nerds who just like some random content from another country.

People who would both want their UI to be in a language A but also to consume content from languages B, C...

I do not understand how that assumption holds in any product decision except in one where the YT product teams are entirely and totally separated from the engineering teams.

I have all of my language settings configured to en_US. I've explicitly configured YouTube's country setting as well. I still get autotranslated titles for the local language in the search results. There are so many irrelevant results that YT search has become unusable. I'm increasingly noticing similar behavior in Google search, especially around news and current events.

But sometimes you actually have to search for something in your native language. Plants and mushrooms for instance: I know them by their Norwegian names, I wouldn't often not know how to search for them in English.

But then Google serves up wikipedia articles auto-translated from English, often with made up (but plausible looking) domain terminology: "Russula cyanoxantha, ofte kjent som kullbrenneren eller spraglete russula" - No, that's not true, Google, Russula cyanoxantha may be known as the coal burner in English, but I have NEVER heard anyone call it kullbrenneren in Norwegian, and "spraglete russula" is also not something it's ever called.

And of course it weights its own AI translated garbage above the search results.

Another specialty seems to be a half page of "related" questions. I'm not a stickler for grammar, but I frequently find questions here with a syntax suitable for a cartoon caveman.

Windows took a nosedive with recent Windows releases, due to the "apps".

Oh, you're using Norwegian keyboard layout? Well surely you want Norwegian display text to go with that.

Oh, you're using US keyboard layout? Well surely you'll want US date and time formatting to go with that, along with the display text.

All the while ignoring the setting that's been there for ages and worked reasonably well.

Danskjævel here, and I feel your pain. MS Windows has always had a pretty terrible danish translation. Partly because a lot of tech language translates poorly to danish, and partly because MS are lazy and incompetent. Back in Windows 2000 days, I found instances where "Windows" the product name was translated to "vinduer", the danish word for an actual window. I actually found a ~25 year old screenshot I took --> https://0x0.st/80r5.png

Worse than that is, as you say, software that makes dumb assumptions about my language preferences. It gets especially interesting when graphic drivers translate strings like ambient occlusion, screen space reflections, temporal anti-aliasing, subsurface scattering. The result is incomprehensible gibberish - I mean levels far beyond the baseline gibberish danish already is :D

Win10 regional settings are so terrible, and it's pretty much impossible to set keyboard, date/number formatting and basically everything but display language to danish, while keeping the display language english (The King's English, thankyouverymuch). Thankfully Linux has the locale "en_DK.UTF-8" which instantly makes everything show like I want it.

Hah, that "skjermbilde" is hilarious!

One thing I've always found a bit peculiar between Norwegian and Danish computer words, is that in Danish they're often not translated but the English word is just used. My Danish family would say words like computer, download, password, cloud, software, while I would say datamaskin, nedlasting, passord, sky, programvare. And then they would mock our silly Norwegian words, heh.

Yea, german is the same way, they translate a lot of tech words as well. Danish is taking on english loan words at an alarming rate. Many products have english labels on them, even though they're manufactured for the danish (or scandinavian) market. Seing a store having "udsalg" is getting rarer, now it's a "sale". Many younger people use english words in regular conversation, and I hate how it affects me, and I start doing it as well. I try very hard not to. Last place I worked had english as the official corporate language, and all meetings were in english. Even if I'm very proficient in english, my brain works in danish and it takes much less effort for me to converse in danish, and switching between languages all day makes the mental load worse.

A language is much more than a protocol for communication - it's culture. I'm not a purist (only a little), but I think this is a concerning development. I can only imagine how much worse it is in Copenhagen, because that's a very silly place.

I used to think this way, i.e. assuming a misunderstanding on the product development side.

Nowadays I think its more of a conscious decision many times. Like "We know someone could travel to france as a tourist, but its a small fraction of french IP addresses so screw these people". etc.

It baffles me that a ton of sites that have been translated into multiple languages still set the language based on IP rather than trying to determine it based on the client settings or defaulting to a set language with an easy way to switch it.

Countless times I landed on websites I use relatively frequently in foreign countries to see them in a language I don't understand, having to rely on my browser's translation functionality to find the language switcher. My operating system + browser are set to English, yet I still get served the one in the language I don't understand.

The worst offenders in my opinion are the ones assuming language based on IP for multi-lingual countries like Switzerland. People living in the French or Italian parts almost always get served the German content. It's bad UX.

>client settings

But even setting the client right is not really possible. I'm danish. I understand english and german. Norwegian and swedish are similar enough to danish that I can read it without too much trouble. Websites, if they're translated at all, usually offer their native language plus an english translation. So if I visit a website in any language I understand, I'd prefer the original. But for any other language I want the english version.

If I set my accept-language to "da,en", I get a lot of horrible machine translated danish on a lot of websites. If I set it to "en,da", all danish government websites are now english. I can't win.

How very strange. I don't get any kind of translations of native sites (government or otherwise).

I recall my British colleague’s disbelief when I told her there are many people in the US whose first language is not English…

Similarly, plenty of folk born in the UK whose first language isn't English. I have many friends who grew up speaking Scots, and a few whose first language is Gaelic.

> whose first language is Gaelic

Is that really true though? I think it's a bit optimistic to say "first language". I know that many people do speak Irish, but the number of people who can speak Irish better than English is abysmal.

No Béarla[1] is one of the saddest documentaries I've ever seen. It follows a native speaker who tries to do mundane tasks in Ireland using only Irish, but can barely get anything done.

[1]: https://www.youtube.com/watch?v=eyll-bBZzyk

Likely not for Gaelic, even in the Gaeltacht. I have a couple of friends who deliberately speak Gaelic to their kids, but neither of them would consider it their first language. The kids themselves are properly bilingual, but tend to default to English when visitors are present.

Scots is a bit different, because for many speakers it's just seen as an informal register of English. My grandparents, for example, were primarily Ulster Scots speakers - they could understand English but would have struggled to speak it naturally. My own internal monologue is peppered with plenty of Ulster Scots but fowk wid luck at me quare an funny if I were to use it much in my daily life in London!

Sorry, I should have been more specific: Scots Gaelic.

And yes, it's not exactly common but there are still people who were taught Gaelic as their first language. Some of them I consider to be friends.

We've got lists of falsehoods programmers believe about all kinds of things, we could probably do the same for all groups, including falsehoods the British believe about the USA, or about languages.

I think the UK is something like 9% English as a second language, and the USA is 22% ESL?

(Wnes i fy ngradd prifysgol yn Aberystwyth).

I get this a lot in restaurant menus nowadays. My phone is set to English because I am more familiar with techy terms in English than my native language with its weird translations. I go to a restaurant, open the QR menu and tell them what I want. But they don't understand what I want because the menu is set to English because of my phone settings. A lot of places have English names in their regular menu (because cool) so I am not talking about speaking English to a nonspeaker. Like sometimes a wrap is called a wrap and sometimes it's called whatever the native thing is. That kind of thing.

No need to do these assumptions, youtube accounts are google accounts and google already supports setting multiple languages that you speak/read.

It doesn't work, you can set all the languages you speak on your Google account settings, and Youtube will still change every video title and audio track to either your default language on settings, or your system language. At this point I don't even know what stupid decision between the two above the Youtube backend is choosing.

Yes, this. Google still doesn't get even the Accept-Language header right, which is shocking, because they're Google.

that doesn't mean a single thing - google would auto-default to IP country resolution every step. It's funny in an awful way to have a road trip in Europe - every day it's a different language. While I speak 3 languages that's far from sufficient, morealso google translations leave a lot to be desired outside English (as everything is designed in US,incl. date formats [mm/dd], units - inches and miles, etc.)

Accept-language doesn't do anything at all for google, either.

Would be great if they actually use it in their products instead of showing me some insane mixup of stuff I don't want :)

It's gotten to the point where searching Google for word definitions is actively hampering me by defaulting to showing the definition in my "regional language" (which isn't correct in the first place since it's spoken by a majority in my country but not in the region where I live) as well as English when I only want the English meaning. Stuff like this has had me switch to DDG where even if the results are strictly worse, at least it's not going out of its way to annoy me

The very fact that Google allows multiple language setting yet still fearlessly screw you at every turn is the biggest sign IMHO that they just hate multi-linguals.

I see it as the other side of Hanlon's razor: we know there's competent people there, we should fully attribute this to malice.

> As a hongkonger (zh_hk + en_gb), living in Singapore (zh_cn + en?), following JPOP. This is the daily fight I have with browser.

Ah yes as a Korean living in Japan with locale set to English, this truly is a daily fight.

> I would rather all application, including web app just give me the option to choose

I've left websites for other competitors because they wouldn't have a button to change language.

I think you overestimate amount of multilingual users and tourists.

If you have 40 million country and you have 10 mil tourists over whole year given week you might 200k users that happen to be in that country.

Even if you have another 1mil expats living in that country. it still is 35mil of people vs 1mil of people for whom "your IP is from country X you get language X" is pretty good heuristic.

That said I am also pissed off by that approach but I do understand there is much more people who happen to use only that language in that country.

Optimizing for Expats or Tourists would be stupid as those are exceptions not the norm.

I'm a tourist 4 weeks per year. Excluding any business travel. That's 7.6% of the time. Now, I travel internationally more than most, but it's not out of the ordinary and it's 100x your estimate.

Of course, I'm multilingual even if I stay at home. Do we know how many people are multilingual? About half of Europeans speak more than one language. That's hundreds of millions of Youtube users. 22% of Americans for 76 million in the US. That's just numbers from Google's own Gemini (which is doing a bit of a half-assed job). Note that all EU countries except Ireland mandate second language teaching in schools (23 out of 27 mandate at least two foreign languages); you don't need to be an ex-pat, tourist, longterm resident or second generation immigrant to be multilingual.

Not taking input from multi-lingual users is not just bad practice from Google, it's actually impressive how they manage to ignore people they probably know and are working with.

> all EU countries except Ireland mandate second language teaching in schools

Ireland does mandate second language teaching in schools, it's just not a foreign language, it's Irish (separate language, not a dialect of English). And it's taught atrociously badly. The vast majority of graduates, after about 13 years of mandatory Irish classes, cannot functionally speak it.

> separate language, not a dialect of English

Of all the languages spoken in the UK and Ireland that argument attracts, I've never heard anyone claim that about Irish - an actual Gaelic language.

> The vast majority of graduates, after about 13 years of mandatory Irish classes, cannot functionally speak it.

They just need to get out their word book - that's right, their focal leabhair

♫ oh get out your focal leabhair, get out your focal leabhair

and we'll have you speaking Irish in a half-an-hour

Sea means yes and Níl means no

get out your focal leabhair and we'll all have a go ♫

https://www.youtube.com/watch?v=gYNZbmrMg7M

> I've never heard anyone claim that about Irish - an actual Gaelic language.

Many people from other parts of the world don't know it exists, so when you say "Irish language", they just think you mean hiberno-english. I've never heard anyone claim it either, just assume, then get confused in conversation until I clarify.

> And it's taught atrociously badly. The vast majority of graduates, after about 13 years of mandatory Irish classes, cannot functionally speak it.

I grew up in a country where we had mandatory lessons in a little-used local language imposed on us by politicians.

That I learned nothing isn’t the teachers’ fault - they were simply set an impossible task.

Monolinguals consistently underestimate how hard it is to reach proficiency in another language. The exception to this being being raised in a multilingual home.

I don't know what their reasoning is. They can plainly see around them foreigners struggling to speak, but they somehow think that it won't happen to them?

> Note that all EU countries except Ireland mandate second language teaching in schools (23 out of 27 mandate at least two foreign languages); you don't need to be an ex-pat, tourist, longterm resident or second generation immigrant to be multilingual.

Yes, but learned at school doesn't necessarily count for much. I grew up in the UK, secondary school was post-Maastricht so I was in the EU at the time; French was mandatory, but my grasp of the language is still so bad that when I tried to say "I don't speak French" in French in front of a French person, she couldn't tell if I was trying to say "I can" or "I can't".

I have pushed myself quite a bit in other languages since school, but I'm also a nerd who likes learning for its own sake. I suspect a lot of people are only just about barely able to function in tourist settings in the languages they learned at school.

That said, auto-translation of videos is kinda an existential threat to hosting a language course on YouTube.

Huh, basically every European aged <~40 is going to know their native language + English at the very least.

Knowing English is not equal to enjoying content in English.

Besides that they also get German, Dutch, Finnish or Indian stuff translated that they don't understand.

Reply on news.ycombinator.com