European word translator: an interactive map | Hacker News Reader

European word translator: an interactive map

ukdataexplorer.com

270 pointsgnabgib109 comments

Wiktionary has dialect maps for common Chinese vocabulary that showcases the differences in terminology across various regions of Chinese, rather than their similarities. Example: sleep -> https://en.wiktionary.org/wiki/Template:zh-dial-map/%E7%9D%A... , hide-and-seek -> https://en.wiktionary.org/wiki/Template:zh-dial-map/%E6%8D%8...

p.s. I'm saying this because most of these terms that has a dial-map are common in daily conversation. The differences in written Chinese vocabulary aren't as significant; how scientific and technical terms are expressed is largely determined by your administrative region.

Note that a very significant amount of this data on Wiktionary comes from the tireless work of a handful of contributors, and especially https://en.wikipedia.org/wiki/User:Justinrleung

Ukrainian and russian words often use the same letters but are pronounced very differently due to distinct phonetics. On the other hand, some Polish and Czech words sound the same or very similar to Ukrainian but look quite different because of their different alphabets. Therefore, phonetic transcription would be a valuable improvement.

I've been using phonetic transcriptions in a parallel text reader application I've been putting together. It seems like they go a long way in allowing a foreign language learner to internalize a word's pronunciation.

The Finnish example translation is wrong: "an example" should be "esimerkki". (There are no articles in Finnish.) The map shows "esimerkiksi" which means "for example". (Prepositions are relatively rare in Finnish.)

Edit: Ah, it says the data is from Google translator. So no suprise here, Google translator produces poor results. It's said that Deepl is much better. I can't really tell because I don't need machine translation Finnish English. Both are roughly equally strong foreign languages for me.

For Slavic languages, it sometimes produces arbitrary case forms, too.

When translating between two languages with gendered nouns, it will also lose the correct gender in translation, since it goes through the intermediate step of English. It’s hilarious how inadequate it is.

The fine-print at the bottom of the page says "The translations were retrieved around 2014". Machine translation has come a long way since then.

Machine translation has come a long way since then, but Google Translate continues to be so far behind it's quite ridiculous.

I've had very mixed luck with all online translation services when translating between danish and english. Google was bad to the point of being unusable, but none were great. ChatGPT however is excellent at this. It's a shame its output is so extremely slow, or it would have been perfect.

DeepL is leagues better than Google Translate, and I can tell because I've worked with a lot of translation to and from Finnish. They are not even comparable. Google Translate will completely garble any translation to or from Finnish. Kagi Translate also does a great job in Finnish translation.

I can mostly speak for German. It seems to mix them all into one general language. But there are a lot of local differences between north and south of Germany, Switzerland and Austria. And it’s not just dialect, but really different words that might not be understood everywhere. If you look at the english part it has at least three different words. Similar in Spanish.

> And it’s not just dialect, but really different words that might not be understood everywhere. If you look at the english part it has at least three different words. Similar in Spanish.

I think you cannot really compare the minuscule differences between "Standard German", "Austrian Standard German", and "Swiss Standard German" to the differences between English, Irish and Welsh, which are not even from the same language family. Also, the tool is based on Google Translate, and AFAIK Google Translate doesn't differentiate between them.

Comparing the tool to this map [0], it seems to do a pretty good job in capturing all major languages in Europe, while ignoring their dialects.

But I agree that I would be great if you could zoom into the map and also show differences in local dialects. ChatGPT seems to be pretty good at translating to different variants of standard German, or German dialects [1]

[0] https://en.wikipedia.org/wiki/Languages_of_Europe#/media/Fil...

[1] https://chatgpt.com/share/67bba4db-9458-800c-b5f8-fd3fa196d4...

My best guess:

- Swiss German and Austrian German didn't make the cut because Switzerland and Austria are on good terms with Germany and don't mind if we call their languages a dialect of German. Not only is that justification to exclude them, they are also not in Google translate for this reason (which this map uses)

- Luxembourg did mind and went to great lengths to get their German dialect recognized as a separate language, is in Google translate, but Wikipedia lists them as only 300k speakers

- Frisian is seen as a distinct language because of how different it is, is in Google translate, but has about 200k speakers

- Similarly, Scottish Garlic is in Google translate has only 70k-200k speakers

The map is consistent if you set the goal of only considering languages that are in Google translate and have at least 500k speakers.

I do think these rules detract from the map. Frisian and Luxembourgish are interesting as "in-between" languages (Luxemburgish has a lot of French influence, Frisian is closer related to English). And Swiss German has many distinct words that are very different from their German counterparts, so for the purposes of this map it really should be a language.

IIUC, the Swiss German can't make a cut as there's no standard written form (and with it, not much resources), and the variations between the cities are pretty significant.

There really isn't a single "Swiss German" dialect. It is rather a family of dialects, and this family is again part of the larger family of "Alemannic German" dialects, which are spoken in most of southwestern Germany, Switzerland and western parts of Austria [0]. It is really very hard to clearly demarcate "Swiss German" from dialects spoken for example in the Black Forest, around the city of Freiburg im Breisgau, in Vorarlberg or even (historically) in Alsace. My own dialect is Swabian (also Alemannic), and I never had trouble understanding the local dialects around Basel, Berne or Zurich. It is easier for me to understand these Swiss German dialects than, for example, Bavarian dialects.

[0] https://en.wikipedia.org/wiki/Alemannic_German#/media/File:A...

As per some folks I've met (I live in french Romandie, almost 0 variation here they just speak slower than french), for ie folks from Zurich its almost impossible to understand folks from Bern. And thise are 2 big cities pretty close to each other, not some remote mountain valley.

But they can easily switch to more modest verion or even high german if needed.

> for ie folks from Zurich its almost impossible to understand folks from Bern. And thise are 2 big cities pretty close to each other, not some remote mountain valley.

This is absolutely not true. Bern is the capital and many people travel there for work or other reasons. It's also a dialect very heavily featured on TV (e.g. I remember there was a weather reporter from Bern, don't know if she still does this), a lot of famous politicians are/were from Bern (e.g. former Federal Council member Adolf Ogi) and many famous musicians also sang/sing in this dialect (Mani Matter, Züri West, Gölä, etc.)

Almost all Swiss dialects are mutually intelligible simply due to the high level of exposure to the diversity (and also their relative similarity). There are some people who don't understand Walliserdeutsch well, because it's less represented and also linguistically more removed from the rest - but even that's something you get used to quickly.

Alemannic is still spoken in Alsace. Albeit it has some of the same issue your listed: no standard written form (Hochdeutsch was used for that) and wide difference even between close villages. In particular, Northern and Southern varieties have a different vocalic systems.

Similiar to Slovenian - we have 400 dialects, grouped in 7 larger groups based on similarity. Given that there is only like 2 million speakers that may feel like a large number, but it's a consequence of rather hilly geography.

Differences between some of them are rather extreme, especially Prekmurje dialects feel like their own language - so we need to fallback to "book" Slovenian when talking with people from different regions.

I think for ‘Scottish Garlic’ you meant ‘Scottish Gaelic’…

It’s hilarious that an English language website has so many enthusiasts for the regional differences of their favourite foreign languages while all pretending that English is monolithic and consistent everywhere. Try driving around the north of England for an hour or two and see how many different words for bread roll you encounter. Baps, barm cakes, oven bottom muffins…

In my experience, people mostly tend to hyperbolize the differences between their local dialect and "everything else" for patriotic reasons. Usually, they give some singular words that are vastly different as examples (I suspect you can find such examples in most languages and most regions), and ignore that 99% of the vocabulary, plus the grammar and most daily sentence constructs, are equivalent (modulo the accent). A standard example in German is how the outermost bread slice is called, which differs completely from region to region, town to town, and sometimes even family to family [0].

[0] https://www.atlas-alltagssprache.de/wp-content/uploads/2014/...

For an example take a look at this map of the different words used in German for "meatballs":

https://www.atlas-alltagssprache.de/runde-7/f01f/

Same for Belgium, Google/Apple translate has never been able to correctly translate French and Dutch for us while our vocabulary choices are drastically different from neighbouring France and Netherlands.

For people with their iDevice set to the Swiss/Austrian German or Belgian Dutch locales, Apple Translate initially didn't even offer their languages in Apple Translate (i.e. not even German or Dutch). Only after internal complaining did they allow Swiss/Austrian German users to use German in Apple Translate.

Italy is the same. Each region has it's own variations of words, which can be very different or mean different things in other regions.

For example in Rome a grocery store bag is "busta", but in Milan it is "sachetto" with "busta" being the word they use for an envelope.

The Spain map is not showing Spanish only, those are separate languages spoken in Spain

One-to-one word translation doesn’t make too much sense, because words tend to have more than one meaning, and they don’t map 1:1 between languages, both in meaning and usage. For example, “nice” is translated here into words that would (depending on language and context) more commonly translate back to “beautiful”, “pleasant”, “good”, or “fun”. They aren’t necessarily wrong as translations, but the website’s premise of “word A in language X is equivalent to word B in language Y” is.

You can put phrase, e.g. https://ukdataexplorer.com/european-translator/?word=glass+o...

You immediately see the difference (or similarly) of languages when using words that are very old, such as "iron", or "stone", which are words that have existed from the origins of that language.

Also "cow". And "sun", "mama" and "papa" seem to transcended most European languages.

salt, tea, ..

one can follow migrations.. and criss-crosses..

btw, "orange" as color in Bulgarian is still "orange" (оранжев/а/о/и), but "orange" as fruit is портокал ("portokal") - so that's tricky..

"oranges" seems more correct, vs "orange color" maybe

Salt and tea are good examples for the 2 reasons that can be the cause for finding the same word in many languages.

Salt is an ancient Indo-European word that was already in use several millennia ago, so it has been inherited in most Indo-European languages.

Tea is a relatively recent borrowing in the European languages, which has spread from one language to another, with a few pronunciation variants, across all Europe, regardless of the genetic relationships between languages.

Poland took issue with the story that worldwide there are only two words for tee, and which one you use depends on whether you got introduced to tea via sea or via land

For some words, their identity is not obvious when you do not know the rules for the changes of sounds between the Indo-European language subfamilies.

For instance "cow" and "Kuh" come from the same word as "boeuf" and "buey" (also despite the gender difference).

Mama and papa is a whole other phenomenon.

Arabic: mama babi. Mandarin: mama baba. Swahili: mama baba. Inuktitut: anaana ataata. English: mama papa. Tamil: amma appa.

These languages are not known to be related.

The first vowel sound a child makes is approximately "a" and the first consonant they form tends to be a nasal plosive "mba mba mba" and the second distinct sound tends to be a dental or labial plosive "pa ta pa ta". And the first thing a baby says is "mommy" of course and the second thing a baby says is "daddy" of course. So mama is mommy and papa or tata is daddy. That's the usual explanation, anyway.

> Inuktitut: anaana ataata

That's interesting. Ana/ata means mother/father in Turkic languages

Check this out:

https://polatkaya.net/Native_Americans.htm

Here is an interactive card of how Swiss people (Only the German speaking part) call the left over piece of an Apple: https://www.kleinersprachatlas.ch/interaktive_karten/F15_Apf...

Here some other words as well: https://www.kleinersprachatlas.ch/karte-1-butter

In lower Austria we would call this a "Burzen" -> Apfelburzen. But this is colloquial and somewhat outdated.

It's from https://de.wiktionary.org/wiki/Apfelbutzen

my favorite example of trade routes influenced the spread of a word is "tea". the word for “tea” comes from either a variation of “cha” or a variation of “te,” reflecting distinct dialect pronunciations in China.

countries receiving tea overland (e.g., via the Silk Road) adopted forms of “cha,” while those trading by sea through Fujian ports adopted forms of “te.”

The project visualise perfectly this distinction.

Not doubting about this but then, how come that Portuguese uses the "chai" version, being on the extreme west of Europe, and with all the other countries in between Portugal and the end of the Silk Road using "te"? Not to mention the fact that Portuguese were a naval power for many years, with colonies in Asia as well.

Good question, Portuguese traded not through Fujian but Macao, where chá is used.

The term cha (茶) is “Sinitic,” meaning it is common to many varieties of Chinese dialects. Meanwhile, the word tea comes from the Min Nan variety of Chinese, spoken in the coastal Fujian province, where the character 茶 is pronounced te.

Russia got it from Mongolia, so it uses the “chai” variation. Move just a tiny bit West though and Poland got it from France, which got it via sea routes, therefore Poland uses the “te” variation, even though a lot of the culture of drinking tea in Poland has been influenced by Russia, with boiled tea being a thing in more eastern parts, and variations on that with citruses and honey being very popular (even becoming more popular today). Curiously, contrary to the word for tea, Polish word for kettle borrows from Russian influence – “czajnik”, where the first syllable comes obviously from “chai”.

Love that the numbers in Catalan are represented as numerals, not as words.

EDIT: playing with it, it's a bit sad that large numbers do not work at all (in any language); and that not all common forms of a word are shown. For example, I tried to see how "ninety six" is said in french in France, Belgium and Switzerland, but it does not work.

As a French, I always found that the way Wallons or Swiss word out numbers >69 makes way more sense than ours

As a no French, I love French numbers you dislike. 90 being 4 20 10 is something sort of awesome and funny.

I often wondered if the fact your number system forces you to multiply somehow affects your mathematical competence. France has won a lot of Fields medals.

The English number system kind of also forces you to multiply. Ninety-six is nine tens plus six, or 9*10+6. French is just special because they randomly sneak in base 20. But I doubt they really think more about saying 4 score plus sixteen then you do about saying nine tens plus six.

What is more influential (in a detrimental way) is German randomly switching reading direction. They read 2196 as 2000+100+6+90 instead of the more reasonable 2000+100+90+6

Dutch does that too, and I've tried out what happens if you say it correctly ("negentig 'n zes", ninety and six, instead of "zes 'n negentig")

It takes a second to process and then they'll ask "do you mean [reverse order variant]?" so they do kinda get it and I think transitioning to the sane version could be possible without much trouble, but people would have to want to

It's because if it's out of order, your "partial interpretation so far" in your head gets messed up. A relative of mine pulls a version of this in English (that I find hilarious): in American English we usually read phone numbers one digit at a time, but sometimes group pairs of digits, and it doesn't really matter because "forty five" and "four five" present in the same order. Their phone number has a 1 in it, so to prank people that are writing down or typing the number they read it as (numbers made up, but the effect is):

   five
   six
   seven
   four
   -teen

...

and then cracks up. I cracked up when it was done to me, although apparently not everyone finds it so funny. ;)

In France we read our numbers in pairs: 01 23 45 67 89. The initial zero is complicated, it is a butchered-up solution when we were transitioning the numbers and in an international context one never knows whether to say +33 0 1 23 45... or +33 1 23 45... (the latter one is correct, though both will likely work). When you remove the zero in the international version, lots of systems will format the number by triplets, which seems to be more common in Europe (+33 123 456 789)

Sing a song of sixpence, a pocket full of rye, four-and-twenty blackbirds, baked in a pie…

I had schoolteachers who still spoke like this in 1970s Yorkshire. I don’t know if it was a regional dialect thing or a generational thing across all England, but among the over-40s back then it was still pretty common to hear German-style backwards numbers in English.

We can test your theory by checking how the Danes fared.

https://blogs.transparent.com/language-news/2016/08/29/danis...

I once read an article claiming ghat Chinese are do good at math for two reasons:

1. Their words for Numbers are based on Base-10 system (so no nonsense such as „eleven” and „twelve”)

2. Their words for Numbers are short, one syllabe, so they can keep morę of them in their short term memory at once

Not saying there’s any truth to that, but sounds interesting

No, because nobody thinks 4*20+16 when saying 96, this is just a string in your head, without any links to 4, 20, 16. This is just a word like "ghdtehbdf"

Growing in Belgium, we learned that our Walloon brethren use septante (70), quatre-vingts (4 * 20 or 80) and nonante (90).

We never learned huitante (80), but here are apparently parts of Belgium that use is. We did learn soixante-dix and quatre-vingts-dix, and were allowed to use both. [0]

The Swiss also use huitante, and Nova Scotia uses octante.

[0]: Funnily enough, writing American English was a no-go. We had to write centre, colour, metre, lift (elevator), ticket (receipt).

Ha, I only knew of octante, I dd not know there was a "huitante". I worked at CERN and it was octante.

This is very cool. Also, it seems like Romanian is the only language where the word for turtle translates literally to "shelled frog".

German’s Schildkröte, “shield-toad”, is quite similar.

Sköldpadda - Swedish too

I love when languages have funny words like that.

Like how in Japanese, "mushroom" can roughly be translates as "tree child".

Hedgehog is literally "needle-mouse" (hari-nezumi).

In Dutch, a porcupine is called a "spike-pig" (stekelvarken)

Edit: and a turtle is also a "shield-toad" (schildpad)

Porcupine is actually the same meaning, porcus (pork) spina (thorn/spike). It's just a bit obfuscated in modern English compared to e.g. Italian (porcospino)

Similarly, in many Slavic languages the bat could be translated as “flying mouse”.

There are examples from five language families shown here: Indo-European, Basque, Uralic, Turkic, and Afro-Asiatic.

The words for bridge split neatly into language subfamilies. The only exception appears to be Welsh.

Ugro-finnic too.

https://en.wikipedia.org/wiki/Finno-Ugric_languages

The Hungarian translation of "en example" is incorrect, the translation there would be "as an example" or "for example".

"Egy példa" would be more literal.

Why translate 1-2 words? That makes no sense. Words don't map 1-1 or 1-2 or 2-1. It would make sense to enter a sentence, select 1-3 words as important, then highlight the same 1-3 words in the translated sentence. But translating the english words "current", "right", "bark", "date", etc. are good examples of why 1-1 word translations don't make sense.

The random words had "lead", I tried it and some translations are e.g. "plumb" which is about the chemical element, while others are e.g. "conduire" which is about the verb "to lead"

So it seems to not interpret the English word as the same one for all languages in case of words with different meanings!

You are coloring it by 4 colors like map but you should color countries phonetically (speex, levenshtein or something similar)

It seems like it would be more useful if the Russian and Greek words were transliterated into the Latin alphabet.

Huh, the example "she runs" is not correct in Polish. Currently translates to "ona działa" = "she functions."

She runs, as in the form of locomotion, is "ona biega/biegnie."

The site says that in a blue bubble below the input field (I agree it's not very noticeable at all):

> This example demonstrates that the map should be interpreted with care; some translations have the meaning "she lasts" or "it works".

Another mistake for this example, although subtler, is the Dutch version, which is translated to the meaning of "she walks"

Thanks, didn't notice that. Not very thorough of me.

It's interesting that the site says it uses Google Translate, because using it via the web UI, it does give the correct answer.

https://translate.google.com/?sl=iw&tl=pl&text=she%20runs&op...

The site also says (at the bottom) that it grabbed the translations from Google Translate back in 2014 and hasn't updated them since.

So I tried "folk", and it comes up with the translation for "people" in Dutch and German, "folkoric" in French, "popular" in Italian, "folk music" in Danish, and fails for a whole bunch of languages. And it's these kind of words for which its use is most interesting. Words like "linguistics" have a much greater similarity across languages than common, old words (although surprisingly called glossologia in Greek).

i wonder how long until google translate just swaps out for gemini

LLMs are so much better at this

(chatgpt 4o gave these answers:

English: folk

French: peuple

German: Volk

Dutch: volk

Italian: popolo

Danish: folk

)

> LLMs are so much better at this

Yeah, IME as well LLMs really shine at translation of sentences and getting the right meaning depending on the context for words. Way, way better than Google Translate via web UI or app.

I guess that right now there might be some high level IC/manager trying to get a promotion by switching Google Translate to use Gemini in a cheap and effective way :)

Neither of them is right or wrong, really. "Volk" can have very specific connotations in German (e.g. "völkisch" nationalism is similar to ethno-nationalism but the concepts of "Volk" and ethnic group are not entirely identical which is why a separate term exists for the concept and why the English term uses it as a loanword). "Folk" can mean many different things in English. There is no exact overlap between words across languages most of the time so it depends on the context. For example, while "Leute" in German also translates to "people", in "power from the people" that word would be correctly translated as "Volk" not "Leute" as it refers to the people as an abstract political force from which a state's power is derived.

This is not surprising, because many English words are ambiguous, being used not only as nouns, but also as adjectives or even as verbs.

All such uses must be translated into different words in other languages. When the word to be translated has no context, a random translation choice is possible.

"Folk" as in "folk music" or "folk dances" may be translated correctly as "folkloric" or "popular".

I know, but it's a pity. There isn't a natural language without ambiguous words. It would be more valuable if e.g. it would list all translations.

Perhaps of interest, the translation guesses things like age relations and genders. This is accurate when the word has that same meaning in English, like nun and monk have a gender, but e.g. the word hairdresser in English is translated to a specific gender in German and Dutch even though the original didn't have one. Similarly for diminutives: "brother" (broer) usually means "big brother" in Dutch because for a younger brother you'd add a diminutive suffix. It's hard to define, but maybe: words that are not synonyms, yet all translate back to your input. (The reverse also exists, of course: insulation and isolation isn't differentiated in Dutch)

The map would be more complete with this information because it may be very similar or completely different and can be interesting to compare, for example:

- EN: receptionist for both, NL: receptionist and receptioniste, DE: rezeptionist and empfangsdame. The map currently just shows the female version for German, without indication that they also use a transliteration of the English.

- EN: little brother, NL: broertje (the submission shows a doubled up version of kleine broertje), DE: kleiner Bruder. Although German has the diminutive suffix to make Brüderchen, they don't use it the way that we do, which I find interesting to see.

Google Translate's API can output multiple options, <https://cloud.google.com/translate/docs/reference/rest/v3bet...>, and Google's own website seems to indeed provide these different variants, but there is no label to say what the different array entries mean the way that Google's own website shows

I got curious which gender it guesses that you might mean. It seems to assume a male unless it's also very heavily female-connotated in English. In Dutch and German, it outputs male for hairdresser and doctor, female for nurse and receptionist (German translations mean "sick-sister" and "reception lady", respectively), and mixed for secretary (female in Dutch, male in German) because Dutch doesn't have a male word for it anymore (only workarounds)

Excellent, nice work.

Here are a few feedback for improvement,

- looking up for "user" gives some results, but starting from a term in other languages doesn’t work; also it doesn’t display multiple nouns that can apply as a translation (ex: ulisatrice/utilisateur in French)

- if the target user is English speaking user centered (at least for now), probably providing transliterations (along there Cyrillic/Greek correspondence) would probably make more sense

It mixes different word meanings on the same screen:

Input: cross

Russian: пересекать (as in verb "to cross")

Polish: Krzyż (noun, as a christian symbol).

The notes clearly say that it only shows one meaning, and I it would be more difficult (five years ago before LLMs I would have just said impossible) to discern between different meanings of a word and pick the proper one.

Edit: "translation" -> "meaning".

Its great they acknowledge this in their notes, but it does not make the service more useful/reliable or accurate.

At least they could have tried picking translations for matching parts of speech (nouns, verbs etc), and it would have been a great improvement, even if they ignored homonyms. Doing so does not require LLM.

I was excited to try this, but it seems to have a very tiny dictionary built in. It came up with nothing for any combination of "small {noun}", or some other common words. I sometimes struggled to find two words that would work.

There is something wrong or that I don't understand. I put "the example", the Italian translatione is "nell'esempio" which is wrong, but in google translate the translation is correct "l'esempio"

Nice idea, quite a novel to represent a dictionary.

Somehow it breaks on words "Monday", "January" and "Italy", for example - doesn't show any of the translations.

Such a neat little tool! I shall use it and share it. Clever idea using the map to simultaneously display words. Small feature but hugely piques the interest of the user.

> The translations were retrieved around 2014

How does that work? Are they cached somewhere? Devtools doesn't indicate a call to google translate.

I tried with month names and it doesn't work, it only show the English word. Too bad because month names in other languages are sometimes interesting.

It appears to be using a downloaded/pre-cached sub-set of words from Google Translate rather than a live Google Translate query.

I assume it doesn't include any proper nouns. I tried putting in country names because I always find it interesting what different countries are called in different places, but it didn't return those either.

She runs is translated as she walks in several languages....should be ona trči in Croatian and ze rent in Dutch (German is also translated as she walks)

Missed opportunity to colour countries by the similarity (phonetical) of the translations. Now as i understand all country colours are hardcoded.

What is the most "diverse" word? Fewest seemingly etymologically related translations?

"girl" is my best so far

Off topic, but I recently learned that "girl" in Middle or Old English originally referred to a child of either gender, and I believe only acquired its gender specific meaning in the late middle English or early modern period.

I typed in "nazi". Only UK has the word appear over it.

I highly recommend people go and translate "pineapple".

For Portuguese, it gives the Brazilian variant, although the translation is placed over Portugal.

French is the only language where Company and Society are the same word: société. It's fun to watch

No, in French you can use both "une compagnie" or "une société", as well as "une entreprise", "une firm", "une corporation" and "un business" or even "le biz" are all pretty common.

But of course, none of them are fully interchangeable in all contexts. You will typically not expect to hear "salut la compagnie" in a formal meeting with "les gens de la bonne société."

If you like synonyms, CRISPO gives 77 for société and 33 for compagnie.

https://crisco4.unicaen.fr/des/synonymes/soci%C3%A9t%C3%A9

https://crisco4.unicaen.fr/des/synonymes/compagnie

Étant français, je n’ignore pas les différents synonymes disponibles. Mais le mot société a les deux sens, et est TRÈS utilisé dans les deux sens. Au point que cette app montrait société comme étant la traduction de society et de company. Je voulais voir si d’autres langues européennes étaient similaires.

Didn't know Welsh is bit challenging. e.g. awareness in Welsh is "ymwybyddiaeth", I'm sorry what?

Other than that, great implementation.

Search for Welsh road signs, on most you have Welsh and English, and the differences are night and day, I find it quite amusing => "Drive safely / Gyrrwch yn ddiogel".

Or this village name: "Llanfairpwllgwyngyllgogerychwyrndrobwllllantysiliogogogoch", in it's short form just a smooth "Llanfairpwllgwyngyll" (https://en.wikipedia.org/wiki/Llanfairpwllgwyngyll).

Reply on news.ycombinator.com