"halb zwei" ("half two") is the German idiom for "halfway [between one and] two o'clock"
"deux heures et demie" ("two hours and one half") is a much better cognate for the English "half past two o'clock"
We're all used to off-by-one and fencepost errors. To be honest, this is probably a less interesting one of Google translate's mistranslations. But it's one that's very mechanical and we would expect a machine to get something like this right.
And this is very likely the reason for the mistake. See, Google Translate seems to take a shortcut to translating between language A and B, by translating from A to English and English to B.
For instance, for the last five years or so I've been seeing GT translate the Greek word for "swallow", the bird, to the French word for "to swallow", the verb [1]. In Greek and French, the two words have nothing in common (Greek is helidoni/katapino, French is hirondele/avaler). But in English, they are homonyms (swallow/swallow). It seems that GT sees a Greek word, translates it in English, then finds a set of homonyms and simply chooses the translation that's most "likely", meaning the one that's used to most often- which of course means it gets one meaning wrong all the time.
Last year Google made a big todo about its neural nets inventing an "interlingua", an intermediary language to which and from which all other languages it knows can be translated. In practice, I think this "interlingua" ends up being just plain old English.
_________________
[1] Recently I started seeing the translation "machaon", a kind of butterfly, which Google translates in English as "swallowtail". "Avaler" is still available as an alternative.
> And this is very likely the reason for the mistake. See, Google Translate seems to take a shortcut to translating between language A and B, by translating from A to English and English to B.
Although in this case, if you translate "halb zwei" from German→English it correctly translates it to "half past one".
It's plausible that French-English dataset is an order of magnitude larger, mostly due to Canada's bilinguality causing a large trove of parallel data.
Maybe this sounds weird, I have no idea if it has any basis in fact, but I have noticed when you ask Google to translate between two non-English languages, it sometimes introduces errors that an English speaker would make. I've therefore wondered if they somehow, either deliberately or indirectly through some model, translate to English as an intermediate step.
“…a multilingual NMT model trained with Portuguese>English and English>Spanish examples can generate reasonable translation for Portuguese>Spanish although it has not seen any data for that language pair. We show that the quality of zero shot language pairs can easily be improved with little additional data of the language pair in question,” the paper states.
This is a weird thing because you'd be hard pressed to find two languages as similar as Spanish and Portuguese, and whatever this paper has to say on the topic it'd be inevitable that the English translation in the middle would be "lossy".
I sometimes get disappointed when people suggest that machine translation is a substitute for actually learning the languages. It's true that we have limited time in our lives to study languages, and the machine translations can be a quicker way to get the gist, or possibly open some doors. But go much beyond that, like claims I sometimes see that you can wear AR goggles and suddenly communicate like a native, and you set unreasonable expectations. Or perhaps imagine a world where the way to read Shakespeare is to put it through a machine and it tells you, "statistically, Shakespeare meant to say this". Is that really reading it?
Actually there are quite a few, if you include the variants spoken across the peninsula, Galician, Mirandese and Aragon are even closer to Portuguese than Spanish. :)
Yep, I was aware of some of those, not trying to defame the good people of Galicia or anything, just trying to say that Iberian Romance languages are all pretty similar, and notwithstanding many exceptions the probability of easily finding nearly 1-to-1 translations of any given text is much higher than for English where you would inevitably lose more information in transit.
My wife is Russian, and things she says about how Ukrainian sounds to her reminds me a lot of what Spanish speakers say about Portuguese. I don't know enough to say how similar that comparison is, but it passes the "neighboring regions that might have been more mutually intelligible 1000 years ago" test.
As native Portuguese, my biggest beef with Google translate is that it cannot make heads or tails from Portuguese variants when translating.
Same applies to other non-English languages, like Spanish and French, not sure how much this affects English variants as well.
Most of the time one gets a translation that isn't fully correct from the point of view of how the language is actually used in a specific country, sort of mixing American and British English on the same sentence.
Does it? "Half two" would certainly not be common in the US, but certainly folks are familiar with "quarter of 2" which would mean 1:45. If you told me it's "half two" I would assume it means 1:30, not 2:30, as a native (american) English speaker -- although I have taken a couple quarters of German in college so perhaps the German is sneaking in a little bit?
As a native (American) English speaker, "half two" is nonsense to me, but if I had to guess at a time the closest thing would be "half past two" (2:30), so I'd probably guess that first. I'd never heard anyone say "half until two" or "half to two" and they're both incredibly awkward.
"quarter after", "half past", and "quarter til" are the ones we use.
Is that an auto-correct typo or do you say "of"? It's "quarter to 2" in the UK. And "quarter past 2" means 02:15/14:15. It's also perfectly normal to just use "quarter to/past" without an hour, I might say "let's meet again at quarter past", referring to 12:15 if the time were 11:37.
As a Norwegian I was also used to the German variation and it took me a long time to get used to the UK version, but yes, in the UK "half two" is unambiguously always half past two.
"halb zwei" (half two) in German is to be understood as "one half of the second hour has passed".
There are actually regions that use terms like "dreiviertel zwei" (three fourths of two) to indicate 01:45 and "viertel zwei" (one fourth of two) to indicate 01:15. Especially the latter confuses people from other parts of Germany, because regular high German uses the terms "viertel vor zwei" (a quarter to two) and "viertel nach zwei" (a quarter past two).
If you think of the minute hand needing to complete the full circle to reach 2, "Half two" means the minute hand has only completed half that circle, so it's 1:30.
I've even heard of "three quarters (of) two" in German, which means 1:45.
You probably mean the German expression "dreiviertel zwei".
This expression is however not used everywhere in Germany because in other regions it may be said as
- "viertel vor zwei" (approx. "quarter to two") or
- "Dreizehn (Uhr) fünfundvierzig"/"Ein Uhr fünfundvierzig" (The exact time as you would read it off of a (digital) clock/watch)
When I imagine it from the English-speakers perspective, I think "viertel" (quarter), "halb" (half) and "dreiviertel" (three quarters) are probably to be understood like this:
And one problem is that no-one on the European continent learns that at school or hears that on TV. I got quite confused when I recently started to travel to the UK and Ireland a bit more often.
Maybe it's an age thing and that it's relatively new in the UK - I certainly never hear about it when I learned English in Norway from the mid 80's onwards.
Some searches also seems to support that this is something that at least did not become common until the 80's or 90's in the UK, and it makes sense that schools various places would lag behind.
The Oxford English Dictionary [1, university/library subscription needed] gives the definition "In stating the time of day, etc. = half an hour past the hour named." and a quotation from 1791:
> "a1791 F. Grose Olio (1796) 107 C. Pray what's o' clock? W. It will be half ten."
But until looking this up myself, I hadn't realised Americans and Australians always included the "past". Was your teacher British?
--
Assuming you have some familiarity with Danish numbers, you might like another definition: "half, preceded or followed by an ordinal numeral, was formerly used to express a half-unit less than the corresponding cardinal numeral; thus Old English þridda healf, Middle English thridde half or half thrid = two and a half. Obsolete."
Example (c1200 Ormulum (Burchfield transcript) l. 13777) "Þatt sahh. & herrde daȝȝwhammliȝ. Hallf ferþe ȝer þe laferrd" so hallf ferþe means 3½, "halvfjerde".
My teacher was Norwegian. And our English school books all included "to" or "past". We used British dictionaries, predominantly various editions of Oxford's English Dictionary. Probably one of the learner editions.
It's an odd enough thing that I didn't realize that anyone did this until maybe a year or two after moving to London in 2000, as most media as well tend to include the "past", and it's still common enough to include "past" that it didn't really register with me for a while that some people skipped it. It's hard to tell when I noticed, as I would always include "past" because I'd been taught to, and might very well have heard times without it and assumed implicitly the half was there without noticing it was omitted for the same reason....
The Danish definition is the opposite, and consisted with almost all other Germanic languages, and what's giving rise to this discussion. Norwegian has the same as almost every other Germanic language in that respect.
It's the English one that is out of step, and specifically modern English, as the Old English and Middle English examples follows the older Germanic use.
Compare Middle English "half thrid" with modern Norwegian "halv tre" which has the same meaning.
Searching Google Books, I came across "A Dictionary of the Scottish Language", John Jamieson (1846) that contains this:
"HALF, s. This term frequently occurs in a Scottish idiom, which affords mirth to our Southern neighbours. If you ask "what's o'clock," when it is half-past-three, a Scotsman replies "Half four", i.e. half an hour to four. "Ha!" says the Englishman, "then I must wait dinner a long while, for it is only two o'clock!" But this is a good Gothic idiom, yet common in Sweden; half fyra "half past three; half an hour after three;" Wideg.; literally "half four."
So apparently until at least the 1840's, Scotland held on to the Germanic way, while it seems likely it had fallen out of use in England given that the author expected it to be confused for "half of four" rather than "half past four".
So now I'm curious if this has changed meaning in Scotland since then too, and if so when. I'm assuming it must have, or I'd assume there'd be a greater awareness that this would cause confusion.
There seems to be very few uses of the modern English version in books, and the oldest unambiguous mention I've come across so far other than your example and the above dictionary (which uses it in the wider Germanic way) and a Scottish book from 1841, seems to be from 1997.
I've also found one from 1968 where I can't tell whether or not it uses the modern English or not, and whether or not it reflects genuine English usage or a mistake, as it's published in an American student publication (Generation, volumes 20-21, published by students at University of Michigan) and written in a broad English dialect that I can't tell for sure if is genuine or "adopted" by an American for the purposes of the story.
I'm sure there must be more, but it does seem like its more wider use is a a very recent thing - most mentions I can find are 2010 and later.
I've just sampled, but e.g. specifically searching for "half four" in in the 19th and 20th century gave me just a handful examples, while "half past four" gave me thousands. But of course that could mean it was just predominantly used orally. I've tried "half five", "half three", "half six" too and skimmed, with similar results, but not looked exhaustively. And of course this is Google Books - it's certainly not exhaustive.
You are thoroughly demonstrating that "[common] sense" is a product of environment and teaching; what makes sense to one person does not necessarily make sense to another.
Correct for everything but the first, not sure why it added "du soir". For the a complete sentence "Um halb drei gehen wir." it's also correct ("On part à 14h30.").
As far as I can tell it's _in_correct in every instance except the first:
À trois heures et demie = À trois heures et demie
Aller à deux heures et demie =\= 14 h 30 du soir [1]
A quatre heures et demie nous partons =\= À 14 h 30 [2]
A trois heures et demie nous allons au cinéma. =\= On va au cinéma à 14h30 [3]
[1] misses the "to go" part and also gets the time of day wrong- which is important because it uses 24-hour clock notation (and so it's obvious that it has no idea what 14h30 means).
[2] completely misses the action ("nous partons") and gets the time wrong ("quatre heures" is "four o'clock", not 14).
[3] Still gets the hour wrong (14h30 to "trois heures", i.e. three).
You'd certainly wouldn't want those translations to "help" you with your appointments in France, or Germany.
I was only referring to the time. "halb drei" is two thirty, which usually means 14:30 unless you specifically say "in the morning". The action in [2] would probably be added if you add a full stop.
and the 4th one is 3:30. I'm also not sure why you introduced "afternoon" into this; it's not said anywhere and it's easier to understand the translation if you just pretend like this is all early in the morning.
which in one part of austria (vienna) means 1:15pm (or 1:15am), in the other half (90% of austria) it is understood as 1:45pm (or 1:45am), sometimes as 2:15pm (or 2:15am) - but never as 1:15pm/am
"viertel [vor/nach] zwei" - quarter [to/after] two. the [] are optional.
When I first moved to Stuttgart I was confused by the Swabian "viertel" (quarter) and "dreiviertel" (three quarters). The first means quarter after and the second quarter before the hour.
Where in Austria does "viertel zwei" mean 2:15? My impression was that the places that don't use the Eastern ("Viennese", if you will) system always explicitly use "viertel vor" or "viertel nach" (or "viertel über"), leaving nothing optional/implicit.
Viertel zwei = a quarter of the second hour has passed, i.e., it's 1:15. It's a logical extension of the standard-language "halb zwei". But note that this is only used and understood in some German-speaking regions, not all!
It works the same in Catalan: un quart de dues is 1:15, as in "one quarter of the hour 2". If we go full into it, we have as valid and used constructs "un quart i mig de dues" for anything around 1:21-1:23 (one quarter and a half of 2, which would be 1:22.5)
Germans like to reference time relative to the half hour, e.g. "fuenf vor halb drei" for 2:25 or even "zehn vor halb drei" for 2:20, etc.
Do other cultures do that?
Also, google gets "fuenf vor halb drei" wrong but "fuenf vor halb zwei" correct, it seems pretty random. "Fuenf nach halb eins" is 1:35 instead of 12:35 and "fuenf nach halb drei" is 5:30 instead of 2:35!
Swedish does it but only the five before/after half case, not 10. And generally no other number than five actually. Examples: "Fem över halv två (13:35)" or "fem i halv fyra (15:25)".
Norwegian too. Though other numbers than 5 but rarely more than 5 tends to be acceptable. E.g. "tre på halv fire" (15:27) works, but it's rare to use it to be more precise than with five. If you want to be that precise people are more likely to resort to saying the specific minute of the hour ("tre tjuesju" or "tre syv og tyve"; the latter is "old fashioned" and from Danish but still frequently used), alternatively with 15 instead of 3 if even remotely ambiguous.
Interestingly it does do the Dutch "half twee" correctly, to French and back. Moreover, halb zwei is translated to half twee and back normally. So there is no inter-dependency or translation "via English" or something.
I don't think anyone would claim that it does, except maybe Google Marketing and a few blue-eyed computational linguists - who would probably do themselves a favour by focusing a little more on the linguistics of the endeavour.
It is a text-to-text machine translation system. There is, as far as I know, no attempt at an internal representation of anything like semantics or knowledge about the world. They may have added some grammatical facts, but there is no model of the world and no notion of time like a person would use to make or understand statements in natural language.
Case in point: The artifacts of the training processes behind machine translation systems like Google Translate cannot be used to generate sentences from pictures or notions or anything like that. What is happening is nothing like a human learning a language, or even a pair of languages.
That's not to say it's not impressive, useful and important tech, mind, but it really does not do all of what it appears to do at first glance.
I have wondered if it's not an increasingly difficult problem for Google Translate that, as it becomes more popular, it is likely to see more and more training data which it itself helped generate.
My go-to bad machine-learning translation is from Swedish "Män som hatar kvinnor" to "The Girl with the Dragon Tattoo". The correct translation is "Men who hate women" the book's original title. We can deduce why it picked up the wrong translation in that case, harder to say here.
Machine translation is so frequently and so badly wrong that it can't be trusted at all. I guess it keeps translators in a job when it can't even get basic things right, let alone complex translation.
There are two different conventions, but I never saw any ambiguity:
One convention is "viertel vor zwei" (13.45) / "viertel nach zwei" (14.15), the other convention popular in the east and southwest is "dreiviertel zwei" (13.45) and "viertel drei" (14.15).
The ambiguity comes from the problem, when talking to someone from “the other system“. I remember that my “viertel drei“ was misinterpreted as 15.15 by someone, who autocompleted this to “viertel nach drei“ following his system.
Interesting. I never had that problem. Some people can't tell the time in the 'other' system, but they always recognised that it's a system they don't know.
(in France, the good students are sent to learn German as a third language, and the bad ones are sent to learn Spanish as a third language, knowing German is a marker of class)
I can confirm that in my generation, at least, learning German was seen as a more elitist choice than Spanish. Latin/Greek were optional and very few chose to study them. Ironically at the time I felt that Latin classes were easier, and even more interesting than modern languages, e.g.: we were reading already Cicero in our first year.
I have studied in France and this is not true from my experience. Greek and Latin were and will be for the good students (Hollande removed that but it'll be back).
The choice between German and Spanish is made by the parents except if there is a lack of German teachers.
Depending on where in germany you are, this is actually correct.
The term "halb zwei" can mean 13h30 or 14h30 depending on which region in germany the speaker originates from.
"halb zwei" ("half two") is the German idiom for "halfway [between one and] two o'clock"
"deux heures et demie" ("two hours and one half") is a much better cognate for the English "half past two o'clock"
We're all used to off-by-one and fencepost errors. To be honest, this is probably a less interesting one of Google translate's mistranslations. But it's one that's very mechanical and we would expect a machine to get something like this right.