Forum Moderators: open
Let us take this topic one step further and dissect the art of web translations. Here are some of the issues within this field:
1. The site has advanced markup already. Should the translator provide entirely new markup or just put the translation into existing markup, using e.g. Notepad or some other basic editor? (How then to handle national letters? Hand job?)
2. Most translations will probably be from English to other languages. But many languages take up more space than English - German as much as 20-30%. Suppose the layout is precise, what then? Do a free translation to fill the existing placeholder or make new markup for the page?
3. For which language pairs is it safe to translate the keywords and where are cultural differences so great that this cannot be done at all, so that access to a keyword database is needed.
4. If the source page is poorly optimized, e.g. poor title, no description, no IMG ALT text etc. should the translator fix this problem or deliver translated crap back? What does one have the right to expect?
5. How close should the translation be to the original on a scale that starts with word-for-word and ends with stylish copywriting?
These are just a few of the issues. I am sure there are many, many more.
A good translation to my mind aims at transferring a text with all itīs aspects. That includes the raw informational aspect, the emotional appeal, the style, even cultural references. Try to find a translation from a different culture, like Japan, France or Germany, and more often than not you will immeadiately recognize them being translations, even if they are fairly done. Why? They are translations, they reflect local culture and their communication patterns. Is that bad? No, itīs much better than a literal translation. But even better would be a tranlsation done by a native speaker, who is willing and able to adapt the style and the tone of a given text to his home country. That is what Rencke had as the "stylish copywriting" end, I guess.
So rule No 1: Take by all means a native speaker or at least somebody who lives in that country and is able to express himself in all style registers.
Rule No 2: If possible let him act freely with the text. Think about paying more for adapting the text, not only translating it.
The second best and by far cheaper solution will be to have a good idiomatic translation. This will ensure your users will not be appalled by obvious mistakes, neither in the grammatical area nor in the area of style. ( to illustrate: In Germany you adress people either by the formal "Sie" or by the informal "du" - which one to chose?)
This solution could be chosen by web sites who do not plan to do extensive business in that country and have no office over there.
The third solution is done by many websites right now. They provide a rather literal translation which effectivly turns people off.
Now this accounts for the communicational aspects, what about the technical aspects? I guess from Renckeīs post, it becomes clear that a web site translation involves a lot of html knowledge. The question is: Is it better to let the translator do the job of adapting the code also? or rather take the translated content and code it yourself? Problem is, you might have to ask back the translator ever so often.
To have the translator do it all in one wash should save a lot of extrawork, if the translator really knows how to go about it! Which brings me to another point: How close should translator and website promoters and designers work together?
Rule No 3: Have if you can a translator who is involved in the process of bringing the site to another country from the start to the end.
If that is too expensive: take a translator, who can do the coding along the lines of the sites style by himself.
The next issue is keywords: Rencke, I think itīs a rare case, in which original KWs can be used for both languages or where a direct translation will do. In most cases to get really targeted KWs you will have to look for a local KW database. Itīs the translator who has to put those KWs in the text, itīs him who knows how those KWs are used, what the singulars and plurals and other derivatives are. So he should also be able to put them in the metas.
Rule No 4: Take a translator who has access to a good local KW database. If too expensive, try mining KWs for yourself. There is no real alternative to looking for KWs in the targeted language.
This KW issue directly leads to the SEO aspect: How SEO savvy should a translator be? Again depends on what you need and can pay for.
Basically a translator should be able to deal with the correct stuffing of KW in the text. Should know about metas, about page structuring, about proximity of KWs.
Rule No 5: Do not take a translator coming from print media.
Then not all web site owners read regularly at WebmasterWorld or have the time to take care of each individual country. So for some it might be very helpful to have a translator, who can also provide them with at least basic knowledge of the local SE scenery. And what about submissions to local Yahoo? Having the translator who is already been involved with finding the right KWs helping here might be useful also.
Oh well, I might go rambling on this subject forever and will by any chance :)
>1. The site has advanced markup already. Should the translator provide entirely new markup or just put the translation into existing markup, using e.g. Notepad or some other basic editor? (How then to handle national letters? Hand job?)<
Well, when I provide translation services (I only do to SEO projects) I do the translation in cooperation with translation "experts". I sit down with them and go through the pages explaining what/how I see the content and what it's about. Then the translator takes the site and come up with a translation suggestion. By going back and forth this way I end up with the most appropriate solution. If the client is really serious, I contact some local 3rd parites to do the end proof reading. I then take the translation and do a specific word count on each word with stop words removed. The ones that excist the most times would be the ones that should be the target words. These words can then be the basis of further keywordanalysis. I usually the do some searches on these words to see what comes up. Spiced with realted searches that some engines provide - I end up with a pretty good foundation.
The mark up is always the "interesting" part. Do I deliver in plain text or do I do the HTML mark up?
Often I do the HTML mark up as well. If it's straight forward plain HTML - no problem. But when I enter into dynamic sites or other more complex code, I usually only do basic text translantion with suggestions for titles, descriptions, alt-tags etc.
It is very different from project to project. To do the HTML mark up requires capabillity of handling the excact coding - I rather not do it if I can choose (due the "heavy" price increase for doing advanced/dynamic mark up, I often end up with just plain text translation anyway). It's often a Hand job.
>2. Most translations will probably be from English to other languages. But many languages take up more space than English - German as much as 20-30%. Suppose the layout is precise, what then? Do a free translation to fill the existing placeholder or make new markup for the page?<
IMO not too problematic.
Often the pages do take up more space - the pages turn out to be longer and maybe you have to do some scrolling, but if the text is relevant to the serch term and interessting enough, the visitor would have no worries scolling.
It's very rare that the layout is precise?
>3. For which language pairs is it safe to translate the keywords and where are cultural differences so great that this cannot be done at all, so that access to a keyword database is needed.<
Often the access to a keyword database is needed in an any language.
It is very dangerous to just translate the keyword - especially if you do not glanze at the whole concept of the site. One word phrases could sometimes just be translated as is, but with 2-3 word phrases I wouldn't even consider just translating them. As discussed in the above mentioned threads, consensus was made that you need to get the concept and understanding of the site into a translation job.
If you do it - do it right!
>4. If the source page is poorly optimized, e.g. poor title, no description, no IMG ALT text etc. should the translator fix this problem or deliver translated crap back? What does one have the right to expect?<
I my case I only enter into translations in SEO projects, so the source page would be optimized. If not, I wouldn't "just" optimize it but advice the client to get the original language optimized also.
It depends on the contents of the contract with the client.
>5. How close should the translation be to the original on a scale that starts with word-for-word and ends with stylish copywriting?<
It should end up being stylish copywriting - unfortunately that's not always the case. Particulary when dealing with a very small language as Danish. I run into projects where it is not cost-effective to use a lot of time on perfect translation, copy and SEO.
In these cases we pretty much just do word-by-word translation with a corresponding title and description for the page and directory submissions.
Haven't figured out how to make the nice blue tables at WebmasterWorld yet ;)
I then take the translation and do a specific word count on each word with stop words removed. The ones that excist the most times would be the ones that should be the target words.
I would think you had to go about it the other way round: Start the translation process from the keywords! In this respect a translation should work similar to the original building of a page: you know what the intention of that page is, you set the theme, then decide on the keywords to target and finally write the text around those ( or work them in the text). The only difference in the case of a translation is that the first two steps are done already.
>Suppose the layout is precise, what then?
IMO not too problematic.
If the client is really serious, I contact some local 3rd parites to do the end proof reading.
If you do it - do it right!
did you mean those nice blue tables
You're right - that would also be a good way of getting it together. You leave a lot up to the translator though, basically he would be in charge of putting together the copy.
>A good translation to my mind aims at transferring a text with all itīs aspects.<
Absolutely right! At the end of the day it comes down to creating great copy specifically aimed at the target audience.
>And what about submissions to local Yahoo? Having the translator who is already been involved with finding the right KWs helping here might be useful also.<
In deed. A good way of getting the most out of your translator. When dealing with Asian languages we're completely in the dark and the translator provides not only help with submissions but also with understanding the culture, people and their "habbits" on the internet surfing.
>designers will hate you for that<
Most of them already do ;)
Jumping up and down - yes! yes! Exactly those blue tables.
the translator DOESN'T touch the mark up...it's a specialised job that I don't trust anyone else with
we send a word document to the translator with colour coded sections to represent links, menus, titles etc...if they stay with the colour coding then I can mark the document up even without understanding the language
2:
you can't do precise layout...all it takes is somebody overruling the font size and any precision is gone...since Macs and PCs interpret pixels and points differently any assumption of precise layout was fantasy in the first place
the differences in sizes of languages is simply another reason to build fluid pages...don't hire dead tree designers :)
3:
the translation has to be done by concept not by a set of precise specifications...the most important thing is that it works in itself
we start the document that goes to the translator with a note of the general concepts that we want stressed by repeating a single word or phrase...we tie that in with teh concept of the words likely to be searched for...after that we have to trust the translator or we'll end up with a document that reads as if it was written by committee
we finish the document with a set of key words/phrases and a few site descriptions...I can then use these to compare with the translated text to get some idea of where the keywords are
4:
the translator translates the text, no more no less...mark up should be an entirely separate job
both the writing of text and the marking up of documents involve a certain amount of personal style and flair if they are to be done at the highest level...I'm not going to ask the translator to produce poetry, but I'm also not going to get in the way of them producing readable prose by insisting on too many specifications...likewise the site is marked up MY WAY...it's maybe not great art, but it must all work the way I want it to
5:
not at all...for some languages it can't be close...English to or from Chinese for example
-----
I like Heini's rules for a good translation...but I think it is very dangerous to have too many different people working on mark up or on SEO...these are things that also have an element of style and personal choice...but most of all, unless you will have permanent access to the translator you will need to be able to do some fine tuning anyway...may as well take the whole job on :)
----
proofreading...we get the draft translation read by an independent native speaker...we then send their crit to the translator who makes any appropriate changes
submissions I do with the aid of babelfish and systrans etc...(anyone know a good English-Finnish-English one?)...I'm now at the stage where the only Chinese SEs we are not listed with are using an encoding we haven't set up yet...I do not speak a word of Chinese
it doesn't take long to learn enough of most European languages (except for Finnish, Hungarian, Greek and the ones that use a Cyrillic alphabet) to submit without help...as long as the original translation was done well I suspect most people could handle it if they could overcome the fear of not understanding the words
just my thoughts
On your approach on seperating markup and translation completely:
I see markup in a translation also as a way of communication between translator and webmaster: A translator could by doing basic markup, <h><p> etc tags, provide a much better base for discussion with the client. Another way could be to comment in heavily in the source, respectively the text document.
Generally seperating translation and markup is a recommendable way to go about it only when the webmaster has a basic understanding of the language.
we start the document that goes to the translator with a note of the general concepts that we want stressed by repeating a single word or phrase...we tie that in with teh concept of the words likely to be searched for.
Also dangerous it might be to do important directory submissions by yourself. How often did yahoo submissions fail because an editor quite simply didnīt like the style of the inquiry?
I liked the part about precise layout :)
I agree with you on the most parts, but Iīd recommend "Erics Way" only to the really talented multilingual folks out there.
Also dangerous it might be to do important directory submissions by yourself. How often did yahoo submissions fail because an editor quite simply didnīt like the style of the inquiry?"
since we have an existing site we give the URL to the translator...but the way we mark up the original document seems to be enough for most of the translators we have used...in our case we have a limited number of appropriate key words and phrases anyway...but the translator chooses them, not us...but are asked to use the word/phrase most likely to be searched for...if one ever comes back to us and says there are equally valid options I'll check it up for them...the only query we have had so far that implies there might have been another option is with German, where a Swiss German has been quite adamant that a secondary search word should have been our prime key word (he's wrong, over the German speaking world as a whole the one we have used is searched for at least ten times more often)
the only problem I've had with Yahoo so far is yahoo.hk who need a type of Chinese encoding we haven't uploaded yet...the translator writes a set of descriptions to be used in different circumstances...I just cut and paste
so we get somebody independent to check ALL translated pages before they are published
