56

There is no one-to-one correspondence between languages and their vocabularies. This means it is impossible for a computer translator to be invertible. The translator's task going from language A to B is fundamentally different from going from language B to A. To understand this, consider the French word "allumette", which in English is "match", that is, ...


10

It is unlikely, since Google translates the Korean original 늙다리(미치광이) (neulgdalimichigwang-i) as "an old man lunatic", where "늙다리" (neulgdali) conveys "old" and "미치광이" (michigwang-i) conveys "maniac". Google maintains that "dotard" is "도살장" (dosaijang) which, however, it translates back to English as "slaughterhouse", so I think Google can be let off the ...


10

A so-called round-trip translation is not a reliable signal that the translation worked well, for a few reasons. Noisy parallel corpora are bi-directional. Machine translation systems train on parallel corpora, which are often noisy. For example, suppose the following noise occurs in the training data: en: "This is a rare sentence." is: "Rænðöm rænðöm ...


9

The source of this mistranslation series was identified by Japanese internet users as parallel corpus data contaminated by a Japanese TV reality show program あいのり Ainori. According to this article, a number of "episodes" other than "78" were observed as well. トナカイさんの贈り物 → Episode 167 星の印 → Episode 58 砂漠の決断 → Episode 60 恋する勇気 → Episode 50 愛の泉 → ...


7

This is a funny question because of 'machine learning' (ML), 'still', and 'better'. Presumably you mean 'machine learning methods in NLP' (Natural Language Processing), because I'm having a hard time thinking of linguistic theory that informs ML uses outside of NLP. 'Still' implies it has a 'better' one now, and 'better' implies a current one is not ...


6

This question is very important and possible to answer empirically, however, words and concepts do not map 1:1 across languages so the mentioned assumption that bilingual dictionaries will have a great impact is speculative. Relative to what we might expect based on economic factors and inherent difficulty, machine translation quality lags for: English to ...


5

Machine translation in general is in its infancy. Even for major languages like Mandarin Chinese and Spanish, computers have trouble with context-dependent concepts such as verb inflection and words with multiple meanings. No machine translation is reliable. All it's good for at this time is to help you get the gist of a text in another language. Google ...


5

You're right to suspect that the accuracy can potentially be very good, but, in practice, unfortunately, as of 2019, most of the major systems - those from Google, Microsoft, Baidu, Yandex, Facebook, Amazon, DeepL and so on - use bridging via English for almost all pairs, even closely related ones. The reason for this is simply pragmatic. Those systems ...


5

I assume your concern is with regard to Norwegians and not compliance with some statutory requirement (if there is any such requirement, which I doubt, I am certain that it wasn't arrived at by opinion polls in Norway). The code "no" refers to any form of Norwegian, and "nb" refers to Bokmål, "nn" referring to Nynorsk which ...


4

MT is hard. Google Translate is based on statistical methods with models trained on large bilingual corpora. There are a few rule-based systems that produce better translations but only in closed specialized domains. As for now nobody has an algorithm or method that performs better. It's impossible to predict the future but I wouldn't hope for ...


4

This is due to one of the mayor disadventages of statistical machine translation systems as the one Google is using: In general nobody knows exactly based on which information a certain mistranslation was created on - so fixing is not a trivial thing. The algorithm just states: Based on hundreds of millions of lines of translated texts in my database A seems ...


4

"The hardest" is difficult to say. I will say that online translators often seem to have especial trouble with Japanese, as its often implicit anaphors (subjectless sentences and so forth) are extremely difficult for a machine algorithm to infer, so you often get lots of confused pronouns, mixed up genders, etc.


4

It depends on generally syntactic structure of two languages and -yes- some pragmatic issues. To answer your question, the easiest languages to be translated into English by any automatic translation software are the ones which are nearly close to syntactic structures of English such as German, Chinese blah blah. the hardest ones are the ones which are far ...


4

Europarl is a classic corpus for research papers, used at the main conference - WMT - and by some of the top people in the field. It would be useful for training a translation system specifically for European parliament domain. But Europarl, like any domain-specific corpus, is not ideal for training a production-strength open-domain machine translation ...


4

If we take into account only P(D|Ai), the probability is similarly wrong/right as P(Ai|D). Since both are taken from the double-language corpora and thus not really good. But because in the right side of the equation we have also the P(Ai) availabel we gain much more realiable result. P(Ai) is much more representative since it is from single-language corpus....


4

I see now… what was the reason that they blocked me if I did not have any rechasado would seem to come from (Latin American) Spanish (rechasado not being translated because it should be written with z instead of s; Latin Americans get this wrong because z and s both sound like /s/, unlike in Peninsular Spanish, where z sounds like /θ/). The original ...


3

One potential source of this kind of seemingly incomprehensible corpus-poisoning could be the use of translation memories from Computer-Aided Translation tools. Human translators not infrequently have to deal with badly formatted and segmented source texts, and in particular with texts where hard line breaks (etc.) have been used for layout purposes. This ...


3

Well, a word like 'bank' that is ambiguous between 'financial institution' and 'side of a river' will often be disambiguated by context. So, the first sense is talked about near words like 'money', while the second sense is talked about near words like 'water'. An SMT model will pick up on this regularity, and when it sees something like "money....bank", it'...


3

The BLEU score is between 0 and 1, but is sometime expressed as a percentage, i.e. ranging from 0 to 100%. E.g. http://www.statmt.org/moses/?n=Moses.SupportTools#ntoc5 returns a score between 0 and 100 (code). Misc: Original BLEU paper: http://www.aclweb.org/anthology/P02-1040.pdf Some technical issues in BLEU: https://github.com/nltk/nltk/issues/1268


3

I see this often (in Europe), it is just a direct translation. It is a peculiarity of Anglophone language or culture that it sounds so strange - and does indeed sound strange in English, being said from one anonymous man to another - as such an expression is common in many languages of Europe, Central Asia and South Asia, even relatively close cousins of ...


3

Your question makes some broadly misleading assumptions about both translation and machine translation. The problem here is not "grammatical" correctness. It's not a problem for Google to generate superficially grammatical sentences. But sentences only appear as part of text/utterance which is the really carrier of communicative meaning. Any one word or ...


3

After reading Mitch's answer and some of his references, I think a prior question needs to be addressed: Does machine learning need a theory? In the essay from Norvig, Chomsky, I don't see any theory going on, at all, much less a linguistic theory. Does Norvig know what a theory is? He talks a lot about statistical modeling, and I think here he uses the ...


3

Translating isn't a one-to-one process. For a simple example, consider the French words si and oui, both of which can correspond to English "yes" in different contexts. Once you've translated them into English "yes", you need to choose which one to use when going back into French. A human translator could use context to figure this out, but even cutting-edge ...


2

Without your receiving a response from the contact form, and there being no answer when I call the phone number listed in business directories online, let's see what clues we can find on the website... There is a "submit a translation" feature. This is ambiguous since individual translations could contribute in a statistical way (i.e. enlarging ...


2

I know it's late, BUT I did find something else useful in case someone else is searching for this information. It's a bit of a process and only works for single words, though. Go to coptic-dictionary.org and type in the word you need in the Quick Search bar. If the word is common enough, the site will give you different ways it is translated into Coptic. ...


2

As Locoluis explained in his excellent answer, machine translation in general is unreliable. For dead languages, the best attempts are laughable: Google Translate's English-to-Latin, for instance, tends to produce incomprehensible gibberish. If you really want your students to pronounce something in Ancient Egyptian, the Book of the Dead is the obvious ...


2

The big name in computer-based dictionaries is WordNet which groups English lexical items by concept (called synsets). I can be downloaded and used offline. Obviously, this tool is quite powerful and is used extensively a lot in various Computational Linguistic and Natural Language Processing applications. There are even projects to create WordNet ...


2

Look if the surrounding context has anything to do with Japanese pop-culture. The "花*花" duet had a song in 2012 titled "さよなら 大好きな人". "Take on Me" and "When the Going get Tough, the Tough Get Going" are also song titles.


2

There is no one-to-one correspondence between languages and their vocabularies. It is fundamentally different from translating A to B and translating B to A in any machine translation software. For example, when I translate "What is the nearby station?" into Chinese with Google Translate, Google Translate translates this sentence to "什么是附近的车站?" and when I ...


2

Already, Google Translate has reduced the need for knowledge of foreign languages and for translators. There is the possibility that human translators will become much less useful, especial for major languages, and combined with the prospects of radical reduction in the number of languages spoken, needs for human translation may shrivel significantly. ...


Only top voted, non community-wiki answers of a minimum length are eligible