TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

The Shallowness of Google Translate

217 pointsby ehudlaover 7 years ago

32 comments

YeGoblynQueenneover 7 years ago
For the past 5 years or so, as a kind of benchmark, I&#x27;ve been checking how Google translates a particular word from Greek, to French.<p>The word is &quot;χελιδόνι&quot;, meaning &quot;swallow&quot; (the bird) in Greek. For as long as I&#x27;ve been trying this tiny little experiment, Google has been translating it to the French word &quot;avaler&quot;- the verb &quot;to swallow&quot;.<p><a href="https:&#x2F;&#x2F;translate.google.com&#x2F;#el&#x2F;fr&#x2F;%CF%87%CE%B5%CE%BB%CE%B9%CE%B4%CF%8C%CE%BD%CE%B9" rel="nofollow">https:&#x2F;&#x2F;translate.google.com&#x2F;#el&#x2F;fr&#x2F;%CF%87%CE%B5%CE%BB%CE%B9...</a><p>Once in a while, a different translation appears- &quot;machaon&quot;, which is a kind of butterfly, the Old World swallow-tail. This is slightly closer but still absurd. If GT can make the connection to &quot;swallow-tail&quot;, how can it not see the connection to the bird, &quot;swallow&quot;?<p>The problem seems to be that, in order to translate from Greek to French, Google goes via English. A great big chunk of context is lost in the process, especially since now the translation between two languages that have different forms for male and female nouns goes through a third language that does not.<p>So for example, I&#x27;ve seen the same gender-reversal as Hofstadter reports in his article. The following string in Greek means &quot;I saw my teacher and she said hello to me&quot;.<p><pre><code> Είδα τη δασκάλα μου και μου είπε γειά. </code></pre> The following are the French and English translations by GT:<p><pre><code> J&#x27;ai vu mon professeur et il m&#x27;a dit bonjour. I saw my teacher and he said hello. </code></pre> This seems to happen because Google&#x27;s English language model has learned that &quot;he&quot; is found in the context of &quot;teacher&quot; and &quot;said&quot; more often than &quot;she&quot; is. But of course, such a statistical association is, well, meaningless and for that useless when it comes to translation, where you actually need to know when the least likely case is correct.<p>Generally, it looks like the complete abandonment of any attempt at representing meaning, and relying instead on text statistics to do meaning-intensive work like translation, is producing a lot of nonsense. And that&#x27;s not a criticism of Google Translate only. I think professor Chomsky might &quot;win&quot; that old debate with prof. Norvig, after all.
评论 #16299334 未加载
评论 #16297621 未加载
评论 #16297757 未加载
评论 #16297337 未加载
评论 #16299380 未加载
评论 #16297274 未加载
评论 #16299139 未加载
评论 #16297655 未加载
评论 #16298912 未加载
larrysalibraover 7 years ago
At least Google Translate is a bit better than Baidu Translate which at some point decided that my name is the English translation of 扒饭 (grilled rice).<p>2 years ago, I appeared on the menu of a restaurant on Huawei&#x27;s campus in Shenzhen because someone apparently used Baidu Translate to translate the menu to English: <a href="https:&#x2F;&#x2F;twitter.com&#x2F;larrysalibra&#x2F;status&#x2F;959749866036408320" rel="nofollow">https:&#x2F;&#x2F;twitter.com&#x2F;larrysalibra&#x2F;status&#x2F;959749866036408320</a><p>And 2 years later, I&#x27;m still grilled rice: <a href="http:&#x2F;&#x2F;translate.baidu.com&#x2F;#zh&#x2F;en&#x2F;扒饭" rel="nofollow">http:&#x2F;&#x2F;translate.baidu.com&#x2F;#zh&#x2F;en&#x2F;扒饭</a><p>Human language is hard!
评论 #16297394 未加载
评论 #16297447 未加载
评论 #16297219 未加载
评论 #16297318 未加载
cs702over 7 years ago
The author is Douglas Hofstadter, author of <i>Gödel, Escher, Bach</i>.[a]<p>In this article, he shows with concrete examples how Google Translate falls short, and then offers two criticisms:<p>* Feeding more data to current models won&#x27;t bring them any closer to understanding, since understanding involves <i>having ideas</i> (including ideas about the state of the world), and this lack of ideas is the root of all the problems for machine translation today. His examples are powerful evidence of this limitation of current state-of-the-art machine translation systems.<p>* Current machine translation systems make no attempt to go beyond the surface level of words and phrases. These systems merely discover statistical regularities that relate words to other words at multiple, hierarchical levels of composition. In Hofstadter&#x27;s words, &quot;there&#x27;s no attempt to create internal structures that could be thought of as ideas, images, memories, or experiences. Such mental etherea are still far too elusive to deal with computationally.&quot; He is right.<p>That said, AI researchers are aware of these limitations, and are exploring possible ways to overcome them. An early, crude example of such research is the multi-modal model of &quot;One Model to Learn Them All&quot; (<a href="https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1706.05137" rel="nofollow">https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1706.05137</a>), which is trained to learn to perform multiple image-recognition, language-translation, image-captioning, speech-recognition, and language-parsing tasks <i>at the same time</i>, using <i>representations that are shared by all tasks</i>.<p>While these early research efforts fall far short of the kind of &quot;understanding of the world&quot; Hofstadter shows is necessary for human-level language translation, it&#x27;s encouraging to see AI researchers actively looking for ways to move beyond the mere discovery of &#x27;hierarchical statistical regularities&#x27; that relate words to other words.<p>This is an exciting time for AI research.<p>[a] <a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Douglas_Hofstadter" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Douglas_Hofstadter</a>
评论 #16298799 未加载
bambaxover 7 years ago
I sell things on Amazon European platforms (UK FR DE ES IT) and they receive reviews.<p>I only speak English and French and therefore for all other languages I try to understand the reviews using Google Translate.<p>The result is never good. Sometimes it&#x27;s barely intelligible; many times it&#x27;s not really. (For some reason, GT is incapable of translating Italian; by which I mean: the resulting translation gives you no idea whatsoever about the original meaning.)<p>It really makes one wonder if the hype&#x2F;fear about AI is maybe misplaced.<p>Google is a self-described &quot;AI company&quot;, with all the money in the world, and staffed with the best people, and access to the most data, and this is all they can come up with?<p>The only explanation is that the problem is simply too hard.
评论 #16303226 未加载
评论 #16298080 未加载
评论 #16297775 未加载
chxover 7 years ago
Ho humm. Of course he is right but is that enough? This reminded me of a quote from Doctorow&#x27;s Microsoft Research DRM talk which I remember well because I translated (hah) it to Hungarian as my last act as a Hungarian journalist -- I sort of resurrected myself as one because by that point I haven&#x27;t written in years.<p><a href="http:&#x2F;&#x2F;craphound.com&#x2F;msftdrm.txt" rel="nofollow">http:&#x2F;&#x2F;craphound.com&#x2F;msftdrm.txt</a><p>&gt; This is the overweening characteristic of every single successful new medium: it is true to itself. The Luther Bible didn&#x27;t succeed on the axes that made a hand-copied monk Bible valuable: they were ugly, they weren&#x27;t in Church Latin, they weren&#x27;t read aloud by someone who could interpret it for his lay audience, they didn&#x27;t represent years of devoted-with-a-capital-D labor by someone who had given his life over to God. The thing that made the Luther Bible a success was its scalability: it was more popular because it was more proliferate: all success factors for a new medium pale beside its profligacy. The most successful organisms on earth are those that reproduce the most: bugs and bacteria, nematodes and virii. Reproduction is the best of all survival strategies.<p>If you want the Hungarian one: <a href="https:&#x2F;&#x2F;www.hwsw.hu&#x2F;hirek&#x2F;40796&#x2F;a-digitalis-jogkezelo-rendszerek-kritikaja.html" rel="nofollow">https:&#x2F;&#x2F;www.hwsw.hu&#x2F;hirek&#x2F;40796&#x2F;a-digitalis-jogkezelo-rendsz...</a>
评论 #16297297 未加载
ddmdover 7 years ago
There is a new online translator, <a href="http:&#x2F;&#x2F;deepl.com" rel="nofollow">http:&#x2F;&#x2F;deepl.com</a>, which relies on deep learning techniques and provides higher (semantic) quality and accuracy of translation. Previously I had quite positive experience with <a href="http:&#x2F;&#x2F;translate.yandex.com" rel="nofollow">http:&#x2F;&#x2F;translate.yandex.com</a> (but I had to manually compare and combine their results with google translate).
评论 #16297156 未加载
评论 #16297079 未加载
评论 #16297108 未加载
评论 #16297531 未加载
评论 #16297169 未加载
评论 #16297122 未加载
评论 #16297125 未加载
btrettelover 7 years ago
Having used Google Translate to translate at least 6 Russian academic papers in full and about 40% of a Russian dissertation (along with a far larger number of partial translations of papers), I can say that Google Translate works decently for this task. The result won&#x27;t win any awards for elegance, but it&#x27;s by large intelligible. I submitted many corrections over the past years and that seems to have noticeably improved the quality, at the very least reducing the number of gibberish sentences. My contributions are limited to fluid dynamics, so perhaps they won&#x27;t generalize, but I am very happy with the results.<p>Google in particular had problems with technical phrases which it translated literally, where the literal translation does not correspond to the equivalent English phrase. In one case there was no equivalent English phrase and Google Translate returned a phrase that seemed like it had a Latin etymology. After reverse engineering the word based on the Latin, I recognized the concept and found it interesting that there was a Russian word for this. If I recall I added a footnote explaining the word. Anyway, I doubt your average human translator without subject knowledge could do this. So I don&#x27;t blame Google Translate for this too much.<p>I have a list of problem sentences that I&#x27;ll take to a human translator some time in the future. Machine translation has been convenient but does not yet replace human translation even when one sets a fairly low bar as I do.
评论 #16298617 未加载
nabla9over 7 years ago
Hofstadter&#x27;s strength is his ability to stay focused on the highest level cognition problem. His theory is that analogies are the core of human cognition. His main work revolves around discovering how extensive human ability to think using analogies is.<p>I especially like his attempts to understand and capture the high level cognition in simple toy problems. His `copycat` and `Letter Spirit` programs illuminate the problem and his thinking in very clever way.<p>The current AI and ML research is building things bottom up and there is still significant gap before we reach the high level cognition that Hofstadter is interested in. What is the representation that binds low level and high level cognition and allows high level fluid concepts to be used as analogies from one domain to next with just one or two examples? Style transfer, variational autoencoders and transfer learning are very limited in this regard.<p>Challenges for deep learning:<p>* Deep Letter Spirit. Show 1-3 examples of lowercase letters of the roman alphabet in some font and have an algorithm that understands the style and completes the rest of the alphabet in the same style.<p>* Bongard problems solver.
jhanschooover 7 years ago
In defense of Google Translate, I will have to point out two constraints of machine translation that human translation usually do not face.<p>1. Machine translation lacks a direct conception of the physical world: it only understands the &quot;grammar&quot; imposed by physical constraints indirectly through digitized verbal corpora and hand-constructed parameters.<p>2. Machine translation does not have the luxury of understanding their target audience&#x27;s domain knowledge. Much of the jargon Google Translate does not understand comes from very specific situations that few people generally experience, e.g. Pan-Germanic. Normally, if such words were used in a news article, the journalist should have to spend a sentence or two describing what is meant by using those words. If Google Translate was tuned to favor translating jargon into jargon, it is likely that its translations would contain much jargon from various domains of experience and be very difficult to read in general.
评论 #16297414 未加载
paganelover 7 years ago
I heartily recommend Umberto Eco&#x27;s text &quot;Experiences in Translation&quot; (full text here, a little long: <a href="https:&#x2F;&#x2F;archive.org&#x2F;stream&#x2F;UmbertoEcoEXPERIENCESINTRANSLATION&#x2F;Umberto%20Eco%20EXPERIENCES%20IN%20TRANSLATION_djvu.txt" rel="nofollow">https:&#x2F;&#x2F;archive.org&#x2F;stream&#x2F;UmbertoEcoEXPERIENCESINTRANSLATIO...</a>), which perfectly describes the difficulty (and I&#x27;d say beauty) of the act of translation.<p>&gt; In this book Umberto Eco argues that translation is not about comparing two languages, but about the interpretation of a text in two different languages, thus involving a shift between cultures.<p>The above short presentation of the book says it all, basically, but surprisingly enough is really hard to understand for lost of technical people who only think that translation is just &quot;de-coding&quot;, as Hofstadter explains. We won&#x27;t have proper translations until AGI is here.
olasaustraliaover 7 years ago
Jeffrey Shallit has a response - <a href="http:&#x2F;&#x2F;recursed.blogspot.com&#x2F;2018&#x2F;02&#x2F;doug-hofstadter-flight-and-ai.html" rel="nofollow">http:&#x2F;&#x2F;recursed.blogspot.com&#x2F;2018&#x2F;02&#x2F;doug-hofstadter-flight-...</a>
评论 #16297072 未加载
评论 #16297087 未加载
评论 #16297064 未加载
评论 #16297159 未加载
评论 #16297042 未加载
评论 #16297068 未加载
评论 #16297196 未加载
lukas099over 7 years ago
&gt; The bailingual engine isn’t reading anything—not in the normal human sense of the verb “to read.” It’s processing text. The symbols it’s processing are disconnected from experiences in the world. It has no memories on which to draw, no imagery, no understanding, no meaning residing behind the words it so rapidly flings around.<p>I had forgotten how edifying Hofstadter&#x27;s writing was about topics such as this.
ehudlaover 7 years ago
DRH&#x27;s &#x2F;Le Ton Beau De Marot: In Praise Of The Music Of Language&#x2F; is a remarkable book about translation of literature (all of his books are must reads, btw). Still, it seems to me that shallowness is a feature, not a bug, of automatic translation.
mehrdadnover 7 years ago
Related heads-up: if you ever want to use or test Google Translate, don&#x27;t make the mistake of assuming Gmail&#x27;s translator would return its results. I have no idea what it uses, but it sure isn&#x27;t the same Google Translate. I&#x27;ve seen it produce output far inferior to that of Google Translate for the exact same text.
Yetanfouover 7 years ago
Google Translate - and all the other translation engines bar some experimental ones - are a step in the right direction. No, they do not translate like a human translator does as they are better compared to transpilers than translators. Be that as it may the emergence of the likes of Google Translate has made it possible for just about anyone who can read and write to get the gist of what is written in another language without needing to get outside help. While the translation might be rickety it generally is possible to get what was written. The next step will be taken when the technology is ready for it, no sooner. Giving the rate at which machine learning or &#x27;AI&#x27; is being pushed it won&#x27;t be that long before human translators are taken out of the loop for most tasks. They&#x27;ll still have a job translating literature and some legal texts [1] but most business communications will be translated by machine.<p>[1] even though legal texts should be a prime candidate for machine translation as they are written in something resembling human byte code to start with.
cannamover 7 years ago
In 1968 the Sunday Times ran a competition to translate a poem by Baudelaire (&quot;Je suis comme le roi d&#x27;un pays pluvieux&quot;) into English.<p>The poet Nicholas Moore read about the competition and, apparently angered by the fruitlessness of the task, entered it 31 separate times with 31 different poems, many submitted pseudonymously and in the styles of other poets.<p>One of them even begins &quot;I&#x27;m like the Winner of the Competition &#x2F; the one who wrote the strong, rewarding phrase...&quot;<p>None of his entries won, but his anger (&quot;All I have against translation is that it can&#x27;t be done!&quot;) carried him a surprisingly long way.<p>You can read his entries (and the original) here <a href="http:&#x2F;&#x2F;www.ubu.com&#x2F;ubu&#x2F;pdf&#x2F;moore_spleen.pdf" rel="nofollow">http:&#x2F;&#x2F;www.ubu.com&#x2F;ubu&#x2F;pdf&#x2F;moore_spleen.pdf</a>
jvvwover 7 years ago
My sister is a professional translator and when companies pay her they generally do it because they want really good translations not because they just want people to understand the content. This can be for semi-legalistic reasons (though she&#x27;s not a legal translator per se, but she has done a fair bit of work related to the EU) or for reasons of professional reputation. Literary translation, although not her mainstay, is also obviously really hard to do well. She&#x27;s not worried about Google Translate at all yet, although she does have the advantage that one of her languages is rare for native English speakers and she is an extremely good writer in English.
评论 #16297273 未加载
kccqzyover 7 years ago
The author makes one point that is wrong. The current deep learning version of Google Translate specifically <i>didn’t</i> all of the available corpus and data that the old translation uses, simply because the new engine takes much longer to train with the all the data. Google internally looked at statistics and decided that the new engine is aleeady better so there’s no need to feed it more data. I believe they decided that a reasonable time for training (around two weeks IIRC) is more important.<p>So any want of quality is definitely a pragmatic choice made by Google, but not for want of data.
anotherevanover 7 years ago
He never did explain why Frank and his Danish friend used Google Translate despite both being fluent in each other&#x27;s native languages. I read right through mainly to try and find that out.
pooya72over 7 years ago
On the other hand, it&#x27;s interesting how well Google Translate translates philosophical text. Just compare the translations below.<p>Gadamer&#x27;s text from Truth and Method:<p>Die folgenden Untersuchungen haben es mit dem hermeneutischen Problem zu tun. Das Phänomen des Verstehens und der rechten Auslegung des Verstandenen ist nicht nur ein Spezialproblem der geisteswissenschaftlichen Methodenlehre. Es hat von alters her auch eine theologische und eine juristische Hermeneutik gegeben, die nicht so sehr wissenschaftstheoretischen Charakters waren, als&quot;vielmehr dem praktischen Verhalten des durch die Wissenschaft ausgebildeten Richters oder Pfarrers entsprachen und ihm dienten.<p><i>Weinsheimer and Marshall translation:</i><p>These studies are concerned with the problem of hermeneutics. The phenomenon of understanding and of the correct interpretation of what has been understood is not a problem specific to the methodology of the human sciences alone. There has long been a theological and a legal hermeneutics, which were not so much theoretical as corrolary and ancillary to the practical activity of the judge or clergyman who had completed his theoretical trainin<p><i>Google translate:</i><p>The following investigations have to do with the hermeneutic problem. The phenomenon of understanding and the right interpretation of the understanding is not only a special problem of the humanistic methodology. There has also been a theological and juridical hermeneutics from ancient times, which were not so much scientific-theoretical in character as they corresponded to and served the practical behavior of the scientist or pastor trained by science.
评论 #16297200 未加载
SZJXover 7 years ago
It already works very well, especially after its improvement in the recent years after using neural networks. I&#x27;ve frequently been surprised by the quality of the translation it provides to Japanese sentences, not to mention between German and English. The test cases I put in in some other languages (e.g. Chinese) are really promising too.<p>Machine translation is a really hard problem. The messiness of language as a system, the importance of context in daily conversation etc. all play a part. Another layer of complexity is the gap between everyday usage and official, written form, which is also being tackled by researchers. You have to put this thing into perspective. Many of the old rule-based&#x2F;Chomskyan software have been simply unusable for decades. New statistical approaches have been in use for barely 10 years, and industrial deep learning less than half a decade. There are still much more to come. The hype IMO is well justified.
Houshalterover 7 years ago
These NNs have fewer neurons than an insect brain. They have to use so little computing power they can be provided for free to everyone in the world. Of course they have limitations. But there&#x27;s been exponential progress in the last few years, and it&#x27;s amazing they can do as much as they do.
half-kh-hackerover 7 years ago
I&#x27;ve found, for Korean, at least, Naver Papago is miles ahead of Google Translate - Especially while decoding some unnatural phrases.<p>&#x27;백조가 연못에 있지만 나는 안봐&#x27; has its [implicit] object pronoun dropped for its second verb when translated through Google, but not when translated through Papago.
buovjagaover 7 years ago
Regarding the &quot;gender blunder&quot;: <a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Anaphora_(linguistics)" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Anaphora_(linguistics)</a><p>I noticed Apertium is looking to solve the problem in their system: <a href="http:&#x2F;&#x2F;wiki.apertium.org&#x2F;wiki&#x2F;Anaphora_resolution" rel="nofollow">http:&#x2F;&#x2F;wiki.apertium.org&#x2F;wiki&#x2F;Anaphora_resolution</a><p>It&#x27;s a potential GSoC project for them: <a href="http:&#x2F;&#x2F;wiki.apertium.org&#x2F;wiki&#x2F;Ideas_for_Google_Summer_of_Code&#x2F;Anaphora_resolution" rel="nofollow">http:&#x2F;&#x2F;wiki.apertium.org&#x2F;wiki&#x2F;Ideas_for_Google_Summer_of_Cod...</a>
jmadsenover 7 years ago
I use Google Translate extensively to help write things in Japanese - considered (one of) the most difficult languages in the world.<p>I show the results to Japanese people who nearly always tell me it looks just fine to them.<p>I have also passed off translations as my own work on tutoring sites like Lang8 and had natives correct &quot;my work&quot;. They often will give a slightly different wording, but I have never had an &quot;WTF does this mean?&quot; type response.<p>----<p>I think it is important to distinguish the difference between a perfect translation with nuance, and a simple &quot;I need to make this point&quot; translation that is what we need most of the time.
评论 #16297152 未加载
评论 #16297198 未加载
评论 #16297331 未加载
Stranger43over 7 years ago
It might be worth looking into the pidgin languages that tend to emerge when two cultures meet and start to trade without the intervention of a officially appointed&#x2F;educated class of arbitrators i.e. machine translated text might be seen as a kind of pidgin language thats not really any of the two languages in question.
vanderZwanover 7 years ago
&gt;&gt; <i>“South study walking” is not an official position, before the Qing era this is just a “messenger,” generally by the then imperial intellectuals Hanlin to serve as. South study in the Hanlin officials in the “select chencai only goods and excellent” into the value, called “South study walking.” Because of the close to the emperor, the emperor’s decision to have a certain influence. Yongzheng later set up “military aircraft,” the Minister of the military machine, full-time, although the study is still Hanlin into the value, but has no participation in government affairs. Scholars in the Qing Dynasty into the value of the South study proud. Many scholars and scholars in the early Qing Dynasty into the south through the study.</i><p>&gt; <i>Is this actually in English? Of course we all agree that it’s made of English words (for the most part, anyway), but does that imply that it’s a passage in English? To my mind, since the above paragraph contains no meaning, it’s not in English; it’s just a jumble made of English ingredients—a random word salad, an incoherent hodgepodge.</i><p>&gt; <i>In case you’re curious, here’s my version of the same passage (it took me hours)</i><p>I stopped reading here for now, to avoid having his translation affect what I am about to do.<p>What Hofstadter doesn&#x27;t really go into is that I can still manage to extract <i>some</i> information from the machine translation, compared to <i>none</i> for the original Chinese. Not only that, interpreting machine translations itself is a skill. In a sense, instead of learning a second language, one learns to translate <i>poorly machine translated English</i>. Of course, one can still ask whether that&#x27;s a good thing or not. Here&#x27;s my attempt:<p>&gt; <i>“South study walking” is not an official position, before the Qing era this is just a “messenger,” generally by the then imperial intellectuals Hanlin to serve as.</i><p>“South study walking” is GT&#x27;s best attempt at labelling an unofficial position taken by intellectuals, comparable to being a messenger for the emperor.<p>&gt; <i>South study in the Hanlin officials in the “select chencai only goods and excellent” into the value, called “South study walking.”</i><p>It was a position only available to highly-qualified &lt;Hanlin officials&gt;. Quick google search for &quot;Hanlin&quot;: <i>The Hanlin Academy (Chinese: 翰林院; pinyin: Hànlín Yuàn; literally: &quot;Brush Wood Court&quot;; Manchu: bithei yamun) was an academic and administrative institution founded in the eighth-century Tang China by Emperor Xuanzong in Chang&#x27;an.</i> <a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Hanlin_Academy" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Hanlin_Academy</a><p>.. so people from Hanlin academy, suggesting the position was administrative in nature.<p>&gt; <i>Because of the close to the emperor, the emperor’s decision to have a certain influence.</i><p>The position was close to the emperor, giving those who held it some influence over him.<p>&gt; <i>Yongzheng later set up “military aircraft,” the Minister of the military machine, full-time, although the study is still Hanlin into the value, but has no participation in government affairs.</i><p>&quot;Study is still Hanlin&quot; is likely referring to the “South study walking” position, since we established the connection to Hanlin earlier. With that, this reads as: Yongzheng set up a ministry of defence, which meant the position was excluded from direct government affairs, although there was still value in having the position.<p>&gt; <i>Scholars in the Qing Dynasty into the value of the South study proud. Many scholars and scholars in the early Qing Dynasty into the south through the study.</i><p>Many scholars in the Qing Dynasty have taken the position of &quot;south study walking&quot;, and it was a prestigious position.<p>I&#x27;m sure this is terrible, full of errors, and even the information I correctly inferred undoubtedly misses a lot of nuance, but again: it gives <i>some</i> sense of the information the original passage contains.<p>So here is Hofstadter&#x27;s translation.<p>&gt;&gt; <i>The nan-shufang-xingzou (“South Study special aide”) was not an official position, but in the early Qing Dynasty it was a special role generally filled by whoever was the emperor’s current intellectual academician. The group of academicians who worked in the imperial palace’s south study would choose, among themselves, someone of great talent and good character to serve as ghostwriter for the emperor, and always to be at the emperor’s beck and call; that is why this role was called “South Study special aide.” The South Study aide, being so close to the emperor, was clearly in a position to influence the latter’s policy decisions. However, after Emperor Yongzheng established an official military ministry with a minister and various lower positions, the South Study aide, despite still being in the service of the emperor, no longer played a major role in governmental decision-making. Nonetheless, Qing Dynasty scholars were eager for the glory of working in the emperor’s south study, and during the early part of that dynasty, quite a few famous scholars served the emperor as South Study special aides.</i><p>Well, that definitely reads a lot better, but I wasn&#x27;t that far off in terms of meaning of the text. And it didn&#x27;t take me hours (writing this comment took a long time though).<p>I absolutely agree that human translation by experts is an art, that it produces much, much better results, and that we should not let it be devalued. But the value in getting a quick impression, even if flawed, through machine translation should not be undervalued either. It has a very different application. On social platforms, for example, it is the difference between being <i>completely</i> out of the loop of a conversation or still somewhat following it and being able to ask a question for clarification in a shared language.
babuskovover 7 years ago
I don&#x27;t know what translation engine Bing is using, but I find that combining Google Translate and Bing results gets me pretty close to the real thing.<p>I wish teams making those two could work together :(
EvgeniyZhover 7 years ago
And what is &quot;real understanding&quot;?
Manishearthover 7 years ago
The 锺书 example matches my experiences with Chinese through Google Translate.<p>Chinese is a language where you&#x27;ll often redundantly say things because each character may have (a) multiple meanings and (b) a lot of homophones, even accounting for tone.<p>So the word for vegetable(s) is &quot;蔬菜&quot;, which really is two different words for &quot;vegetable&quot; put together.<p>I <i>suspect</i> that the engine has &quot;learned&quot; that words can be thrown away sometimes.<p>Which leads to fun cases where it tells you the <i>literal opposite</i> of what you said. An example is in <a href="https:&#x2F;&#x2F;twitter.com&#x2F;ManishEarth&#x2F;status&#x2F;919434569446776832" rel="nofollow">https:&#x2F;&#x2F;twitter.com&#x2F;ManishEarth&#x2F;status&#x2F;919434569446776832</a> -- it has since been fixed.<p>I bet what happened with 锺书 was that Translate decided to throw away 锺 in one of the cases because it didn&#x27;t know what to do with it, leaving it with 书 (&quot;book&quot;).<p>-------------<p>Regarding the anecdote about the Danish-speaking friend, this isn&#x27;t too uncommon. For example, I natively speak Marathi but spoken&#x2F;written Marathi differ a lot and I don&#x27;t have much practice reading&#x2F;writing, so I make spelling&#x2F;grammar mistakes. So using Google Translate as a crutch (and then modifying&#x2F;verifying the output) is great. Though I&#x27;ve found out that it&#x27;s not really good at non-EU languages (The EU translates all of its documents into the languages of its members so this forms an excellent corpus of professional translations for Google Translate) and that my written Marathi is often better.<p>-------------<p>My favorite example of Google Translate limitations is what happens if you ask it to translate &quot;Yes.&quot; and &quot;No.&quot; to Chinese. You get the answer 是 and 没有, which literally translate to &quot;am&quot; and &quot;don&#x27;t have&quot;. (there are no conjugations and kind of no tenses, so by &quot;am&quot; I mean &quot;whatever conjugation of the English word &#x27;to be&#x27; makes sense in this context&quot;)<p>The thing is, Chinese doesn&#x27;t really have words for Yes and No. 是的&#x2F;不是 <i>sometimes</i> works (I think this is more because there are implicit 是s in verbless sentences). But the basic idea is that if you want to answer a question, you repeat the verb. &quot;Do you eat meat?&quot; &quot;Don&#x27;t eat.&quot; or &quot;Do you read books?&quot; &quot;Read.&quot;. 是 and 有 are pretty common (as are &quot;to be&quot; and &quot;to have&quot; in English), so it seems like it picked up on the most common yes&#x2F;no answers. For some reason, it picked a different one for no than it did for yes.<p>Now, this isn&#x27;t exactly an example of the machine translation sucking, but more of the translation not having an outlet to express ambiguity at all. The article gives examples involving this as well, but the basic idea is that some languages do not allow for the same kind of ambiguity that others do. You should be able to display something about that when translating from an ambiguous term to a specific one.<p>Yes&#x2F;No is an example of a set of ambiguous terms in English where Chinese has no similar ambiguous terms. In the other direction, as I already explained, 是 (or any verb!) in Chinese is an example of an ambiguous term that doesn&#x27;t translate to English because the tense need not be fully specified (Chinese does tense from context, and even then it may not be as fully specified as it is in english), and English <i>requires</i> you to handle tense.
ColinWrightover 7 years ago
For those who might be interested to see the discussion by the HN on this article, it&#x27;s been submitted and discussed before:<p><a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=16287171" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=16287171</a> (23 comments)<p><a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=16267363" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=16267363</a> (12 comments)<p>There are other submissions without comments, showing that the article is of interest:<p><a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=16285196" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=16285196</a><p><a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=16279656" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=16279656</a><p><a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=16265302" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=16265302</a><p><a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=16294491" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=16294491</a><p><a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=16296792" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=16296792</a>
评论 #16297050 未加载
评论 #16297197 未加载
87ert3746over 7 years ago
What&#x27;s going to happen is that we humans are going to become more machine-like, as we loose our appreciation of our own sophistication &amp; subtlety, so that most people will be satisfied with machine-translation, writing, art and music never having known that there are higher levels