A prominent example of this is the Treaty establishing the European Union where every single one of the 20+ language versions is "Canon" and has to be used by the European Court of Justice. There are often judgements with paragraphs devoted to figuring out the hidden differences between certain wordings in different languages. Interestingly this can lead to problems but sometimes it also clears things up if one language is more precise than the others.
From case C-336/03 (easyCar), section 25:
"That interpretation is expressly supported by several language versions of Article 3(2) of the directive, namely the German, Italian and Swedish versions, which mention, respectively, ‘Dienstleistungen in den Bereichen … Beförderung’ (‘services in the transport sector’), ‘servizi relativi … ai trasporti’ (‘services relating to transport’) and ‘tjänster som avser … transport’ (‘services which concern transport’). "
Instances of this are a bit hard to find, since they usually are just one aspect of a larger argument, but they do happen.
"As regards the wording of Article 2(n) of the Dublin III Regulation, a purely textual analysis of the notion of ‘defined by law’ cannot determine whether case-law or a consistent administrative practice are capable of coming within that concept. In the different language versions of that regulation, the term equivalent to the term ‘loi (legislation)’ has a different scope. Thus, the wording used, for example, in the English-, Polish- and Slovak-language versions is similar to the concept of ‘droit (law in the general sense’), which can have a wider scope than ‘loi (legislation)’. Certain other versions, for example, the Bulgarian-, Spanish-, Czech-, German- and French-language versions, have a more restrictive scope.
32 Where the various language versions differ, the scope of the provision in question cannot be determined on the basis of an interpretation which is exclusively textual, but must be interpreted by reference to the purpose and general scheme of the rules of which it forms part (judgment of 26 May 2016, Envirotec Denmark, C‑550/14, EU:C:2016:354, paragraph 28 and the case-law cited)."
As already mentioned this is often just one part of a bigger ruling and the translations are often good enough so that it's not necessary to dissect the whole text like this.
By the way: I think it's an interesting challenge to translate the Treaties (and regulations etc). You don't want to have too much divergence between the different language versions but sometimes it can be very difficult to translate something exactly without using a lot of words.
One tactic that is used in European law in this regard is to start the regulation with a preamble "definition" section (similar to writing a program) where you define all the important words/phrases.
Yes, very correct. That is also the reason why the ECJ has autonomy over interpreting the words of treaties, regulations and so forth. An independent constructor of meaning in the EU.
This seems like a solved problem: agree at the time which translation takes precedence in the event of a dispute. Use a third language (historically French was the language of diplomacy, now English, but for modern treaties you could pick any major world language) if, as is likely, neither side likes the idea of the binding version being in the other side's language.
The same problem has to be solved in countries with more than one official language. In Ireland, the Irish-language version of a law takes precedence. This is a little awkward in theory since parliamentarians, civil servants, judges and lawyers overwhelmingly do not speak Irish to a sufficient level to draft, debate or interpret the law, and laws often exist for years before anyone gets around to translating them and thereby creating the "definitive version". But in practice, the legal system generally works.
Third party language implies that either citizens of country 1 must learn this third party language to the proficiency needed to understand legalese; or to both be expected to know, understand and apply the law but not be able to read it.
Which is my second point: citizens will have to follow and vote on the law of their country, including international treaties, without understanding them since they are not written in their language.
This is not relevant for the political side of the question, e.g. it would not be possible for a citizen on its own to judge whether the merits of a treaty. Say, your senator campaigns on withdrawing from the Iran deal: how are you supposed to assess this position if the deal was in third-party-language Pashto?
In addition, I can present the case of France, where judges must disregard laws that contradict treaties (the hierarchy of norms being Constitution > Treaties > Laws > Executive orders). Therefore treaties, even with no national law to implement them, have a direct effects on the law.
Treaties are accorded to the power of law in the treaty signatories, below their Constitutional documents but above that of legislative laws.
Once a treaty enters into effect, a nation is bound by the terms of the treaty even if they do not have implementing laws. However, in many cases treaties don't have any consequences if signatories don't pass implementing laws (see, e.g, the Kyoto Protocol).
I imagine most voters do not read proposed bills in their own language as it stands, instead relying on reporters' summaries as they would for foreign language versions.
Agreed, but that's not a reason to write laws or treaties in a language the general population does not speak. That's too high a barrier to understand the rules of your nation, which should be easily accessible to every citizen. Legalese is complicated enough in your native tongue :).
> Could Latin perhaps fit the bill? It's already popular with lawyers
Lawyer here — very few of us know more Latin than the odd artifact such as ejusdem generis or contra proferentem, which occasionally are relevant to contract interpretation. And those of us of a certain age who grew up Catholic will remember bits and pieces of the old Latin Mass. Otherwise, Latin is probably as dead to lawyers as to anyone else.
English is now the global lingua franca; while I used to be a grammar vigilante, not long ago it occurred to me that in the linguistic "marketplace," if we native speakers want English to outcompete, e.g., Mandarin or Spanish, we need to be willing to let the language evolve so that it's easier for non-native speakers to learn and use confidently.
> In the United Nations Security Council (UNSC) Resolution 242, the French text instructed Israel to withdraw from ‘des territoires’ (the territories) it occupied during the 1967 Six-Day War. The English text, however, merely read ‘territories’, removing the definite article and thus leaving ambiguous how much territory Israel should cede.
This is inaccurate. « Des territoires » would be more accurately translated as 'some territories' or just 'territories', the author is thinking of « les territoires ».
With context it is clear. But the way the author takes « des territoires » out of context, makes it look like it's the indefinite article. They should have translated it to 'of the territories'.
He's not really wrong. As a French speaker myself, seeing "from 'des territoires'" and a translation as a definite article was jarring, since the definite form of the article already contains the "from", so preceding the quote with the "from" strengthens the intuitive parse as an indefinite article.
Isn't it ambiguous? I would translate "withdraw from territories" and "withdraw from the territories" the same way (edit: so does Google Translate). It's a failure by the translator who should have worded it "retrait...de tous les territoires" or "retrait...de certaines territoires" (withdrawal from all, or some of the territories) as appropriate.
It's clear in the context of a good-faith negotiation that Israel wasn't being given the option to withdraw from one square metre and keep the rest. But the language gives some plausible deniability to their supporters.
"withdraw from territories" would be translated as "Retrait des forces armées israéliennes _de_ territoires occupés lors du récent conflit" and "withdraw from the territories" as "Retrait des forces armées israéliennes _des_ territoires occupés lors du récent conflit". It is not ambiguous at all.
This is obvious weasel wording. during the writing of the resolution it was made clear that it meant all the territories.
This is how it was phrased:
>(i) Withdrawal of Israeli armed forces from territories occupied in the recent conflict;
any native enlgish speaker can read the declaration and see from the contest that it clear that all territories taken during the war are meant.
The Israelis came up with this facetious excuse to try to encroach on more land.
indeed, when one says "Dogs must be kept on the lead near ponds in the park."
it does not mean, for any sane person that some dogs shouldn't be kept on the lead near ponds in the park.
Also:
"it is an accepted rule that the various language versions must be considered together, with the ambiguities of one version elucidated by the other"
Plus, the reason that Israel must give back the territory is the prohibition of acquiring land by force.
If you can't acquire land by force. you can't acquire any land by force. it's illogical for the rule to only apply on some but not all land taken by force.
As a software engineer who cares deeply about testing, I sometimes find myself wishing that there was a way to run a binding moot court on a law before it got ratified.
Sure. but "binding moot court" was the most concise way I could think of to say "A legal proceeding where you assume a certain set of facts are known, then argue how the law would apply to them, then a judgement is made which becomes binding precedent on how the law is interpreted unless the law is changed by the legislature to produce more clarity."
As a software engineer, I laugh bitterly every time a lawyer tells me that they are being rigorous in their interpretation of a law. Even when all the lawyers agree on interpretation, their notion of "rigor" bears no resemblance to the way a computer applies the term. A test engineer would have zero problem fuzz-testing a law into nonexistence.
To be fair, that's how most people feel when programmers tell people that they rigorously test their software.
What programmers consider rigorous testing is literally just automated, predetermined situations that often have little resemblance to the ways end users actually use (and break) the software.
The way I see it, that's what a developer is: a tool for translating fuzzy requirements into rigid definitions. Sometimes they're just plain lazy about that task, but even when it's done well, they discover vagueness and incompleteness in the specifications. They use their judgment to fill that in, but they're often wrong, especially when they don't really know the domain.
Programmers want rigid requirements, because that makes their job easier. Indeed, it does their real job for them. If we had a really rigid way of writing specs, you'd just write a compiler for it. The real job of a programmer is precisely about playing intermediary between the true rigor of the computer and the pseudo-rigor of requirements.
I was hinting toward frustrating programmers, testers and manager, who don't bother reading requirements (or think they know better) and proceed to code and test against own imagination. Despite requirements being unambiguous about this or that point.
Oh, definitely agreed. We programmers do rather suck at our profession, especially as I define our profession. Not all of us, and not all the time, but we get rather feted for our technical sk1llz while failing to deliver the thing we're actually supposed to be doing -- even just outright ignoring it.
The problems discussed in the article are not relevant anymore today:
We can now have many professional translators work on the same documents. It's not like we have to copy treaties manually, go on a months-long journey by boat, horse or foot just to get a second opinion on a translation.
I don't think availability of documents was at all implied to be the issue. It's more about two actors arguing over semantics, but each actor has a different set of rules to make their case.
Which one would not sign in a time where translations can be checked instantaneously by a legion of translators. You don't have to trust a single translation anymore, you have enough translators to discuss every possible way of misinterpretations before accepting a translation.
The treaties mentioned on the article were signed in 1889 at the earliest. By that time, telegraph lines were already connection most nations on the world.
In Japan, the constitution was originally written by the Americans before being translated, and there are slight wording differences between the Japanese and English.
I wonder if the cost of having people learn to use a logical engineered language is greater or less than than the cost of hashing out ambiguities in other languages later.
There's a huge chicken-and-egg problem: Few people want to be the first mover in an area as dark and mysterious as "The Law" (I'm being sarcastic). Add to that, the capture of contract drafting by lawyers, many of whom prefer to write hard-to-understand legalese to try to keep the mystery going and to bolster their own employment prospects.
45 comments
[ 4.4 ms ] story [ 97.6 ms ] threadFrom case C-336/03 (easyCar), section 25: "That interpretation is expressly supported by several language versions of Article 3(2) of the directive, namely the German, Italian and Swedish versions, which mention, respectively, ‘Dienstleistungen in den Bereichen … Beförderung’ (‘services in the transport sector’), ‘servizi relativi … ai trasporti’ (‘services relating to transport’) and ‘tjänster som avser … transport’ (‘services which concern transport’). "
Instances of this are a bit hard to find, since they usually are just one aspect of a larger argument, but they do happen.
"As regards the wording of Article 2(n) of the Dublin III Regulation, a purely textual analysis of the notion of ‘defined by law’ cannot determine whether case-law or a consistent administrative practice are capable of coming within that concept. In the different language versions of that regulation, the term equivalent to the term ‘loi (legislation)’ has a different scope. Thus, the wording used, for example, in the English-, Polish- and Slovak-language versions is similar to the concept of ‘droit (law in the general sense’), which can have a wider scope than ‘loi (legislation)’. Certain other versions, for example, the Bulgarian-, Spanish-, Czech-, German- and French-language versions, have a more restrictive scope.
32 Where the various language versions differ, the scope of the provision in question cannot be determined on the basis of an interpretation which is exclusively textual, but must be interpreted by reference to the purpose and general scheme of the rules of which it forms part (judgment of 26 May 2016, Envirotec Denmark, C‑550/14, EU:C:2016:354, paragraph 28 and the case-law cited)."
As already mentioned this is often just one part of a bigger ruling and the translations are often good enough so that it's not necessary to dissect the whole text like this.
By the way: I think it's an interesting challenge to translate the Treaties (and regulations etc). You don't want to have too much divergence between the different language versions but sometimes it can be very difficult to translate something exactly without using a lot of words.
One tactic that is used in European law in this regard is to start the regulation with a preamble "definition" section (similar to writing a program) where you define all the important words/phrases.
The same problem has to be solved in countries with more than one official language. In Ireland, the Irish-language version of a law takes precedence. This is a little awkward in theory since parliamentarians, civil servants, judges and lawyers overwhelmingly do not speak Irish to a sufficient level to draft, debate or interpret the law, and laws often exist for years before anyone gets around to translating them and thereby creating the "definitive version". But in practice, the legal system generally works.
In addition, I can present the case of France, where judges must disregard laws that contradict treaties (the hierarchy of norms being Constitution > Treaties > Laws > Executive orders). Therefore treaties, even with no national law to implement them, have a direct effects on the law.
Once a treaty enters into effect, a nation is bound by the terms of the treaty even if they do not have implementing laws. However, in many cases treaties don't have any consequences if signatories don't pass implementing laws (see, e.g, the Kyoto Protocol).
Lawyer here — very few of us know more Latin than the odd artifact such as ejusdem generis or contra proferentem, which occasionally are relevant to contract interpretation. And those of us of a certain age who grew up Catholic will remember bits and pieces of the old Latin Mass. Otherwise, Latin is probably as dead to lawyers as to anyone else.
English is now the global lingua franca; while I used to be a grammar vigilante, not long ago it occurred to me that in the linguistic "marketplace," if we native speakers want English to outcompete, e.g., Mandarin or Spanish, we need to be willing to let the language evolve so that it's easier for non-native speakers to learn and use confidently.
This is inaccurate. « Des territoires » would be more accurately translated as 'some territories' or just 'territories', the author is thinking of « les territoires ».
> Retrait des forces armées israéliennes des territoires occupés lors du récent conflit.
In this context, "des" means "de les". It is indeed a definite article.
It's clear in the context of a good-faith negotiation that Israel wasn't being given the option to withdraw from one square metre and keep the rest. But the language gives some plausible deniability to their supporters.
This is how it was phrased:
>(i) Withdrawal of Israeli armed forces from territories occupied in the recent conflict;
any native enlgish speaker can read the declaration and see from the contest that it clear that all territories taken during the war are meant.
The Israelis came up with this facetious excuse to try to encroach on more land.
indeed, when one says "Dogs must be kept on the lead near ponds in the park."
it does not mean, for any sane person that some dogs shouldn't be kept on the lead near ponds in the park.
Also:
"it is an accepted rule that the various language versions must be considered together, with the ambiguities of one version elucidated by the other"
Plus, the reason that Israel must give back the territory is the prohibition of acquiring land by force.
If you can't acquire land by force. you can't acquire any land by force. it's illogical for the rule to only apply on some but not all land taken by force.
What programmers consider rigorous testing is literally just automated, predetermined situations that often have little resemblance to the ways end users actually use (and break) the software.
Programmers want rigid requirements, because that makes their job easier. Indeed, it does their real job for them. If we had a really rigid way of writing specs, you'd just write a compiler for it. The real job of a programmer is precisely about playing intermediary between the true rigor of the computer and the pseudo-rigor of requirements.
We can now have many professional translators work on the same documents. It's not like we have to copy treaties manually, go on a months-long journey by boat, horse or foot just to get a second opinion on a translation.
That's just 25 years before WW1.
https://en.wikipedia.org/wiki/Lojban
I wonder if the cost of having people learn to use a logical engineered language is greater or less than than the cost of hashing out ambiguities in other languages later.