Looking for very raw, literal translated subtitles

[ssih]

There's an announcement that instructs us not to request subs, but I'm going to do it anyways.

Does anyone have (or know where I can get) subtitles to a jdrama that are very literal, and accurate to the spoken dialog? I'm trying to learn Japanese, and I really thought I'd pick up quite a bit with all the Japanese tv I'm watching (thanks, d-addicts), but I suspect that so much of it is "westernized", and the subs don't accurately match what's being spoken.

For example, "Yoroshiku onegai shimasu" does NOT literally translate to "Nice to meet you", although that's probably what you'll see in the subs.

In many cases, the translation is accurate, but long sentences are rearranged to comply with proper English grammar. For example, you're not going to see subtitles read "I Japanese well understand not."

I know that in fansub groups, often a translator will do a rough first pass, and then a fluent English-speaking editor will redo the translation. Perhaps that rough first pass is what I'm looking for.

Thanks.

[Puppet Princess]

Um no, not even rough translations would be what you want because literal translations would be gibberish in most cases.
Try using actual lessons WITH the drama watching.

[ssih]

literal translations would be gibberish in most cases.

That would be perfect, however I doubt that this would be common from any translator, because they're probably going to do some level of westernizing on their own.

[Puppet Princess]

Well, no. Translators won't put down literal translations not because they are Westernizing it but because their mind says A=B. They know what it means so they don't have to spell it out literally. Sure, they know what they are literally saying but anyone who is actually bi-lingual will completely bypass thinking about the literal part. This is why if you ask a friend a question about languages they have a hard time explaining why something is the way it is. They don't usually think about the words they are using. They just know what to use. Plus expressions and idioms would just confuse you more when taken literally in the middle of a conversation.

The best thing you could do is take the Japanese soft subs and run it through an automated translator. But those are being increasingly programmed to recognize common phrases and translate them correctly as opposed to literally.

[groink]

I know that in fansub groups, often a translator will do a rough first pass, and then a fluent English-speaking editor will redo the translation. Perhaps that rough first pass is what I'm looking for.

This is how I thought things worked. I've been told that nowadays, for the sake of optimizing time, there are no rough first pass translations. Rather, about 90-percent of the the first pass is actually what you see in the final result. The translation will go through a second pass, but not much changes in the second pass. The reason for the quick translations is that the translator is usually a native English speaker, and he'll nail down the westernized lingo and such. Very rarely do we see native Japanese speakers do translations.

There's also a huge increase in using translations in other languages. So lately, we've been seeing a number of Jpn->Korean/Chinese->English translations.

In short, take Puppet Princess' advice and learn Japanese with the lessons available, but use dramas as what I'd call "language labs".

--- groink

[kuro570]

If you want to learn another language its probably best to just take courses in college or something. Learning a language from a tv show would be next to impossible since you probably wouldn't pick up sentence structures, particles, reading/writing, etc, thats not even to mention the many "cant be directly translated" parts of the language. I suggest you take up some courses and use dramas as study material.

[ethidda]

You're probably better off finding the original Japanese translations and using machine translation for that case. Something like Systran would provide less-intelligible, more-accurate gibberish than Google Translate (which uses n-grams).

As for the first-second pass thing: That would only apply to translators who are not fluent in both languages. And the first pass would basically be what machine translation would give you. The reason human translators are generally considered superior to computer translation is because a human understands the nuances and can adjust the translations so that the meanings match and the translation flows, aka non-literal translations.

[ssih]

Yeah, I guess what I'm looking for doesn't exist. It was just a thought. I guess I should just continue with my Pimsleur lessons.

I'm also going to try watching shows that I'm very familiar with WITHOUT the subs turned on.

[Skobb]

Yeah, unless you want to take lessons, the best way to learn is just to watch TV without subs.. Even if you haven't seen it with subs before.. This is how children tend to learn a language, by watching TV for hours on end.. Although a childs brain would be a lot more malleable, I assume that to some degree this would also work for a person of any age..

[ethidda]

Yes, although it should be noted that it takes a child up to five or six years to achieve full fluency, while immersed in the environment. Even if, as an adult, you would only need half the time because you already understand the social protocols and abstract concepts... that's still an awful lot of time.

[ssih]

Even if, as an adult, you would only need half the time because you already understand the social protocols and abstract concepts...

Well, I think that's a double-edged sword. As an English speaking person, I already have a fixed knowledge of how something "should" be said. A child has no pre-existing association of a concept with a verbal term. The child hears "mizu" and pictures clear liquid. The English-speaking adult hears "mizu" and thinks "that means 'water'", and pictures clear liquid.

Another benefit of being a child is that you're generally exposed to simplistic concepts and words. "Dog", "water", "mama", "car", etc. But as an adult, I'm watching mature tv, with a lot of technical and/or complex verbiage - "computer network", "director of human resources", "criminal investigator", etc.

[Kakijun]

In many cases, the translation is accurate, but long sentences are rearranged to comply with proper English grammar. For example, you're not going to see subtitles read "I Japanese well understand not."
I know that in fansub groups, often a translator will do a rough first pass, and then a fluent English-speaking editor will redo the translation. Perhaps that rough first pass is what I'm looking for.

I don't think this really happens at this level. If you put "I Japanese well understand not" no one is going to understand exactly what in the world you wrote. Maybe it works there but there are going to be a lot of errors in the subs.

If you're looking for literal subs try my subs of Ryusei no Kizuna; I thought I went a little too literal with the translation. They might be what you're looking for.

[ssih]

Thanks, Kakijun! I'll take a look at them.

fwiw, I couldn't find them in the list of soft subs. Doing a search on d-addicts brought them up, though.

[peacht]

Moving to General Discussion Forum.

[qop123]

Learning another language through watching videos with subs is an intresting topic for me.

I learned at least some of my English by watching TV dramas and movies with English subs. Especially, as it was really hard for me a Japanese to recognize some of the pronunciations of the English language, watching English videos with English subs helped me a lot.


Now I'm preparing to provide Japanese subs for some of the dramas for the next summer quarter.

And, since there seems to be at least some demand for the subs as a langage study tool, I think I will be able to publish the romanized Japanese subs too. Though it may be hard for foreign people to read the full Japanese text with Kanas and Kanjis, the romanized Japanese text is easier to read and can work as a good starter.

And I might be able to provide hybrid subs too, though I haven't found a good way to program it yet, and so I can't promise it. (I'm not an educated programmer, but I'm learning Perl to accomlish it.)

I have made some of Chinese subs hybrid myself with a tool, to learn some Chinese.

Like this.

136
00:10:28,156 --> 00:10:31,188
妈, 听 老 村长 说 你 想 把 爸 抬 回来
mā, tīng lǎo cūnzhǎng shuō nǐ xiǎng bǎ bà tái huílái

137
00:10:31,260 --> 00:10:33,714
我 就 想 把 他 给 抬 回来, 是
wǒ jiù xiǎng bǎ tā gěi tái huílái, shì


Currently, I'm considering to provide Japanese subs at least for Buzzer Beat, Otomen and Maid Deka, and if possible Dandy Daddy? and other dramas too.

But the problem is the CM cutting.

Though now I know a basic way to extract Japanese subs from TV broadcast, as the dramas uploaded are the CM-cutted dramas, I have to also cut the subs in accordance with the CM-cutted dramas. I may be able to deduct the cutting points, assuming the Raw makers would cut the video by frame by frame, or maybe I can download the raws myself and resync the subs.

It depends on how much I can automate the process, to provide the subs for as many dramas as possible. And so now I'm working on to find a good way to do it.


And can anyone make a small calculation assistant program to help me?

In my current workflow, I have to do some calculations with frame numbers. But this might be able to be resolved in others ways. I'm still searching for a better way.

[bugmenot3]

That would be awesome qop123!! Japanese subs are like gold dust for us at http://forum.koohii.com ever since subs2srs.

As for cutting the CM's, it shouldn't be much more than 10 minutes work in a program like aegisub. You can easily shift the timings of a block of subtitles by a set amount.
First pull back all the lines to match the beginning, then select all the subs from the first CM to the end and then pull them back. Repeat as necessary.
This is what I did to retime subs extracted from the occasional .ts file I could find a torrent of.

I can write a simple python script to help calculate the timings if you'd like =) To calculate how far to pull back the timings based on the amount of frames?

*Ps*
Sorry about using a bugmenot account, but for some reason register page is broken for me =(

[qop123]

Hello,

Thank you for letting me know the koohii forum and subs2srs. As I took a glance at them, they looked very interesting and wonderful.

When I cut the times for CMs manually, I use the shift time command of Aegisub. Fisrt, seek a hiatus of subs, for if there is a CM, usually subs don't exist for at least over a minute, and second, shift the timings with the amount of the CMs from there onward, which is almost just as you wrote.

As I have been cutting the times for CMs for some time, I found out that most of the CMs in Japan are either 90 seconds or 120 seconds long, and so there is practically no need to calculate the amount of CMs from the number of frames unless we try to be very precise.

In addition, Japanese people made a nice little program called CutSrt that auto-cuts the times for CMs utilizing the delete list produced by a made-in-Japan AVI editing program called Aviutl. And there is also another similar utility called SrtSync that utilizes the list produced by TMPGEnc 4.0 XPress. (SrtSync doesn't work correctly with the delete list from Aviutl though it says it does. That is why another person made CutSrt.)