Mobile menu

Pages in topic:   [1 2 3 4] >
Translation - what process do you use?
Thread poster: ViktoriaG

ViktoriaG  Identity Verified
Canada
Local time: 07:30
English to French
+ ...
Aug 13, 2006

Dear fellow translators,

I have been wondering for a while about how exactly other people work when they get translation assignments. I have been wondering because I have my own customized method - which I would be happy to share - that was influenced mainly by information technology.

I would like to know more specifically the steps you execute and in which order. You receive an assignment (...fill in the blank...) and you deliver the translation.

I have seen many bits and pieces of information in the forum. Such things were said as "I take a break between translation and QA", "I read the entire text before translating it", "I start by creating a termbase", and the list goes on.

Would you be so kind as to reveal what steps you perform and in what order in order to make your clients happy? I am not asking about paperwork, prospecting and such, just about the actual work.

I can't wait to see your replies!

Cheers!


Direct link Reply with quote
 

ViktoriaG  Identity Verified
Canada
Local time: 07:30
English to French
+ ...
TOPIC STARTER
My way Aug 13, 2006

OK, here is the way I do it.

1. I use a text analysis tool. This allows me to read the text - without actually reading it. At that point, I get a good idea of how much research this text will involve, as I get statistics on frequently occurring terms/phrases and I can see them in context. I can also export the list of words found in the text so I can build a termbase or glossary with it later. This typically takes half an hour of my time. Of course, if the material is only a few pages long, I skip this part.

2. I do some preliminary research so I don't have to interrupt the translation process too often when I get to it. At this point, if needed, I build a glossary/termbase.

3. I take a look at my TMs and, if needed, I build a new TM for use with the text I am about to translate. If the TM is supplied, then I use my own homemade TM as a read-only TM.

4. I translate. Like I said, I try not to interrupt this process. Depending on the size and level of difficulty of the text, I will put placeholders for tough terms in there, so I can research those later when I'm done with the first round, or I have a Word document open into which I will type comments/questions as I go, again to be researched later.

5. I complete the research that was left to do and incorporate changes. I ask the client all questions I may have that I either couldn't solve or that are clearly for the client to answer.

6. I take a break and wait for the client's reply.

7. I get the client's reply and make some changes to the text.

8. I proofread. I am very anal about proofing and I make many changes, mostly of form and not function.

9. I spellcheck.

10. I turn the document into a PDF and proof a second time. I do this because a PDF looks much more like the final document and therefore I can put all technical aspects of the document aside and see the text as user's will see it. At this point, I find formatting errors if there are any and can correct those formatting errors at the same time.

This is it. Of course, there are variants to this process depending on document formats, subject, level of difficulty, etc., but this is the main method.

What's yours like?


Direct link Reply with quote
 

Henry Hinds  Identity Verified
United States
Local time: 05:30
English to Spanish
+ ...
Wow! Aug 13, 2006

I use a much simpler process, to wit:

I receive an assignment...

1.- I start by doing a search and replace on common, simple words (and continue while translating)

2.- then start translating; when I finish,

3.- I read through and proof it, and

...it's done, I deliver the translation.

I seldom need much research but when I do, it is done immediately when the need arises.

I never read anything before translating it.


Direct link Reply with quote
 

Astrid Elke Witte  Identity Verified
Germany
Local time: 13:30
Member (2002)
German to English
+ ...
My way of doing it Aug 13, 2006

Hi Viktoria,

Your way of translating sounds very thorough! This is my way of doing it:

Firstly, I run a Trados analysis. This is often interesting, because I translate a lot of serial soaps - er, I mean instalments in a court case - so that they have a significant percentage of repetition (citing from previous instalments of the story, for various purposes).

The analysis is useful in at least two respects. Most importantly, it tells me how long the translation is going to take me. For this purpose, I divide the number of words in the new material by 700, and this gives me the maximum number of hours that the document will take to translate (not including proofreading). Sometimes it goes faster, if I am extremely familiar with the story line.

Secondly, I work out from the analysis how much discount I am going to give the lawyer, namely 100% discount for exact matches, 75% discount for 95-99% matches and 50% discount for 85-94% matches. I would not do that with a brand new translation from an agency, but, with these court cases, all the previous material is already in my own TM. I use Trados for all translations.

Prior to translating, I format the document. I first convert it to Word with Abbyy Fine Reader 8.0 if it is a .pdf. Nevertheless, I also prepare Word documents. German legal documents are full of sentences broken up into several paragraphs, along the lines of,

"Please sign the attached undertaking to pay a very large fine for committing the infringement and return it to us no later than

(time and date)."

In the German version there is also a third "paragraph" to the sentence, namely the verb at the end, so it is very necessary to systematically go through the document and close up all these sentences into one paragraph, ready for translating.

I generally translate in the evening (between 6 p.m. and midnight), the morning being reserved for proofreading and the afternoon for answering e-mails and receiving translations to do. Often the proofreading runs over into the afternoon.

I try to avoid handing in a translation before the next day, because I like to leave a night inbetween and proofread it in a detached manner the following morning. If it is needed for 8 a.m., then I simply get up at 4 or 5 a.m., as necessary.

Firstly, I start up Trados and give a new name to the bilingual back-up document, saving it with a .doc ending. I close it and re-open it to make sure that nothing is lost. I will be making the changes, at intervals during the proofreading (after every few pages) to this document, before cleaning it up a second time and subsequently reformatting the document to match the original.

I proofread the document on paper, with the sun shining in, as I also like good daylight for proofreading. I concentrate very intensely on this part, so as to make sure of retaining the client. Also, with all these legal documents, I cannot afford to make a mistake, even if the client has wrongly dictated, or the secretary typed, "Plaintiff" instead of "Defendant", or vice versa, as happens frequently. I naturally point this out and say that I have translated what was intended. I proofread once thoroughly, time permitting, making the changes, as I said, after every few pages. Before sending it to the client, I then simply read it through once more, at which stage I often still find an odd typing error or make very minor stylistic changes.

I write the invoice immediately after completing the translation. If the client is in Germany, I send it by regular post, and, if the client is anywhere else, I convert it, in seconds, with the "free pdf" program into a .pdf, and send it in the same e-mail as the translation.

Astrid


Direct link Reply with quote
 

ViktoriaG  Identity Verified
Canada
Local time: 07:30
English to French
+ ...
TOPIC STARTER
That also makes sense Aug 13, 2006

Thanks for your replies, Astrid and Henry!

First, I can tell right away that, depending on our respective specializations, our methods vary. Most of my work is technical (engineering, environmental assessments, manuals, etc.) and this is the reason why it is important to me to get a grip on the terminology early on. I can't afford to read through the document, even if many translators would disagree with this. The reason to this is that technical documents tend to be rather long (at the moment, I am working on one that adds up to 200+ pages) and I obviously can't budget the time this would require. That is why I analyse the text with software. In a way, I extract the key parts of the text and take a brief look at that, which is really the same as having read a very condensed version of the original. I guess our methods here differ because you both translate texts of a different nature, hence there is less of a need for you to have such a step in the process.

My method does look very thorough and complicated on first sight, but it is actually much simpler. I think that technical documents need a very strict method that leaves no room for errors. However, if you look closely, you will see that my method favors concentration on one task at once. I basically separate different tasks within the whole translation process. This allows me to concentrate more on the task at hand, and it doesn't take any more time than Henry's, for example. It's just done in a different order.

Also, the second proofing is not meant to pick up errors - hopefully, the errors were taken care of during the first proofing. The second proofing is rather meant to allow me to notice any formatting errors and TOC issues. Therefore, this second reading goes much faster than the first, as I don't actually read all the text - I read the pages diagonally and only pay attention to specific information.

Henry, your method of search and replace is quite surprising. Do you mean you replace source words by their target language equivalents, in the source document, before segmenting? Can you tell me how that can help to speed up the process and ensure that the final product is of good quality? I am very curious to know as this seems to be a rather unusual method - but I am open to it if it can work for me. I think, Henry, that the reason why you can allow yourself to do the research on the spot is because your texts are not as technical - I constantly learn new terms even when I translate texts on subjects I am quite versed in. This is why I'd rather do preliminary research - I might as well because there is always a lot of research involved. This way, I can ensure that I will not constantly interrupt the translation to go browse the web or some dictionaries.

Astrid, I think you are wise to analyse the source text before anything else - I also do that, but only for the purpose of knowing how much time it will take to deliver the final product. However, I do analyse the text with TextSTAT, which also gives me an idea of how hard this document will be to translate, how much research I will need to do and whether the same terms are used over and over again throughout, or if the terms used vary greatly. This also gives me an idea of the "flavor" of the text which is a good indicator of which existing TMs I can mix together to create the read-only TM I will use. In turn, this also gives me a good idea of the time required to finish the job, a bit like a good old Trados analysis will, except that I am looking at other aspects of the information.

Thanks for the input - I would still like more input and do feel free to discuss the subject. I think we have a good chance of learning tricks from each other with this thread.

[Edited at 2006-08-13 19:55]


Direct link Reply with quote
 

Williamson  Identity Verified
United Kingdom
Local time: 12:30
Flemish to English
+ ...
A method. Aug 13, 2006

1. Read through the text
2. Use a CAT. Not to calculate rate reductions, but to see if parts of the text have not been translated before
3. Highlight all the terms you don't know.
4. Send it the text to a specialist, who writes specific terminology between brackets in the source text or just above the source-text.
4. Use dragon dictate to translate the text.
5. Format the text
6. Review the text (you and another person linguistically proficient person).


[Edited at 2006-08-13 20:08]


Direct link Reply with quote
 

Astrid Elke Witte  Identity Verified
Germany
Local time: 13:30
Member (2002)
German to English
+ ...
I'm learning something new from this thread Aug 13, 2006

I have never heard of your TextSTAT tool, Viktoria. Already when you first mentioned a text analysis tool, I was curious. It is something I would find interesting to investigate - perhaps not in regard to my usual legal translations, but for use when I get a "strange" translation, on a new topic, from an agency, or for when I get one of those innovations (patents).

I have also not heard of Williamson's "dragon dictate". I take it that this must be some sort of speech recognition software?

This is the last evening of my fortnight's holiday, before I start work again tomorrow morning, and I partly like to use holidays to learn new things that may help with the work in future.

Have a nice Sunday evening, all of you!

Astrid


Direct link Reply with quote
 

ViktoriaG  Identity Verified
Canada
Local time: 07:30
English to French
+ ...
TOPIC STARTER
TextSTAT and Dragon Naturally Speaking Aug 13, 2006

Hi Astrid,

I am glad you find this thread useful, it was to learn of such tools and maybe of some methods as well that I opened it.

TextSTAT is a free text analysis tool I've been using for some time now to analyze texts with. It is useful mainly to rapidly capture the terminology of a text and be able to make up a termbase or a glossary using this information. I also use it so I don'T have to read through all the text, not that I am lazy, but rather because my clients simply can'T afford to give me the time to read through it all. You start by creating something called corpora. You add to it the text(s) you would like to analyze, typically your source text. Then, the software will make a list of all words used (yes, ALL words) in the text, by order of frequency. Their frequency in the text is listed next to each of them. Typically, the first words that come up are "the", "a", "and", etc., but these are followed later by words that are frequently used in the text that are specific to the text - in your case, the most frequent words would be "plaintiff", "court", "defendant", etc. These are then followed by the words that don't occur as frequently and are generally more technical. Usually, these are the words I want to pay close attention to, as they are more technical and may not be part of my old TMs - in fact, I may not know them at all. This is usually what I base my preliminary research on. Finally, at the end of the list, you get words that occur very rarely that may or may not be highly technical. These are often proper nouns, among other things. Then, when you right-click one of these words, you can choose to have them displayed in context. In the text I am working on, I can right-click the word "program", and I discover that there are many occurrences of "field program", "sampling program" and "herbicide application program" in my text. I can define how big the context should be, and specify the number of cheracters before and after the word I would like to see. I can then sort the context alphabetically to the left of the word or to its right in order to get a quick glance of terms and phrases that use the word in question. The results of all this can be exported into different file formats. I find this very useful and it saves me a lot of time - the 200-page text I am working on took me days less to read through, obviously, so the difference is easy to appreciate.

Incidentally, it was made at the University of Berlin, and you can read the documentation in German if you prefer. You can get TestSTAT here: http://www.niederlandistik.fu-berlin.de/textstat/software-en.html

Dragon Naturally Speaking is a speech recognition program. If you search the forum on ProZ, you will find lots of information on DNS in a translation environment. You basically stop typing and dictate away. You can use it with CAT tools. It comes down to dictating your translation rather than typing it - major good vibes for your wrists! Also, you can tell your PC to do things, for example, you can say "check mail", and Outlook opens automatically. Consider that we type in average about 40 words per minute - but we speak at around 160 words per minute. We can all definitely use some speed! The preferred version - most popular with translators - costs around 200 Euros. You can get information on it here: http://www.nuance.com/naturallyspeaking/preferred/

Don't forget to check back for more findings on our respective methods - this could be good!


Direct link Reply with quote
 

Marijke Singer  Identity Verified
United Kingdom
Local time: 12:30
Dutch to English
+ ...
My method is similar to Henry's Aug 13, 2006

Henry Hinds wrote:

I use a much simpler process, to wit:

I receive an assignment...

1.- I start by doing a search and replace on common, simple words (and continue while translating)

2.- then start translating; when I finish,

3.- I read through and proof it, and

...it's done, I deliver the translation.

I seldom need much research but when I do, it is done immediately when the need arises.

I never read anything before translating it.


I also use this system. If the subject is something completely new (i.e. new technology as in leading edge), I will also do some preliminary research into it (a couple of months ago I had one about a special application of concrete and another translation involving wind energy in a very novel fashion). If time allows, I will also give the text an additional read (aloud) after a few days/hours have passed (rarely the case though). I rarely print anything and I usually use a CAT tool.


Direct link Reply with quote
 

ViktoriaG  Identity Verified
Canada
Local time: 07:30
English to French
+ ...
TOPIC STARTER
Reading text out loud... Aug 13, 2006

It is great to read text out loud, it really does make you appreciate any errors that could be left over, or problems with style.

However, with texts as large as the one I am working on, I could seriously use my voice. This is where Acrobat comes in - you can have it read the text to you. I know, it sounds like a mean robot, but it is clear enough to understand, and after a while, you can really get used to it.

Having Acrobat read the text out loud can help a lot with eye strain!


Direct link Reply with quote
 
xxxmediamatrix
Local time: 07:30
Spanish to English
+ ...
Two situations - Two methods Aug 13, 2006

I have two methods, one applicable to around 60% of my translation workload, and one for the remaining 40%.

The 60% is for clients in the ordinary sense, i.e. people/agencies/clients paying me to translate stuff for them.

1 Receive document.

2 Print it on paper - I never, ever, translate from a text displayed on screen.

While it's coming off the printer, check it's all there (if clients send jobs as several files there's sometimes a file missing), get a quick idea of the likely difficulty and decide about time-scales and priorities.

3 Dive straight into the job. No prior research, no pre-read of the document, no nothing.

Translate, using my PC as if it were a stupid typewriter. I always translate into an empty Word file using my standard template (variants of Normal.dot, set up for the appropriate language). I do not try to match the client's layout in any way unless they have specifically asked for that - AND provided a proper fully-functional template (most don't and many don't even know what a template is anyway), AND are paying me to do both translation and 'type-setting' (most won't, so I don't). I don't use any CAT tools - and won't take any work that requires me to do so.

If terminology needs to be researched (very rare in the stuff I do), I do that as and when it arises, giving preference to paper resources over on-line (quicker and more reliable).

4 Scan the entire translation rapidly on-screen, checking mainly that the structure matches the source document (no missing paragraphs, footnotes, etc.).

5 Run Word's spell and grammar checker on the entire document.

6 Do something entirely different for at least 3 hours, if deadlines permit. Ideally, for short jobs, I translate in the morning, break for lunch and a siesta, and deal with steps 7 et seq. in the afternoon. Most jobs I get can be done in a single day (although it's occasionally a 12 or 15-hour day ...).

7 Print the translation and proof-read *as if the text had been authored in that language*. I do not refer to the source text while proof-reading; I've already translated it, I'm sure the facts are right, now I only need to check it 'reads like native'. Any changes are keyboarded immediately, without bothering to mark up the hard copy.

8 Send file to client.

9 Now the nice bit: Prepare the bill!

.......

The single client for the other 40% of the workload (for translation, leaving aside other professional activities) is - myself. I produce written materials - including technical documents and a multi-thematic tri-lingual website - all in three languages. The translation process is rather different:

A Draft the 'source' text - and that will be done in any (or several) of the three languages, depending on the subject-matter and the languages of any reference materials I'm using. When I'm writing about broadcasting, I write quite indiscriminately in English, French or Spanish, changing language from one paragraph to the next; if it's something legal (contracts, web privacy policy, etc.) I write mostly in Spanish - and if it's about international organizations, then in French or occasionally English. Why? - because when I gained experience in those fields I was living and working in countries using those languages, so they come most naturally as authoring languages.

B Proceed as per steps 2 to 7, above, for the first translation. There's obviously (?) nothing in the source text I don't understand, so translation ought to be a breeze...

Actually, it's not always that easy, even if the source was, say, Spanish and I'm translating into my native English. If I come across a sentence in my own Spanish that I can't put nicely in English, then . . . I re-write the Spanish to make things easier. And, sometimes - I can be quite ruthless - I'll just cut out the offending passage altogether.

C Repeat for the third language - translating from the source text, not from the first translation. Again, if there's a problem, I'll re-phrase/chop the source text and adapt the translations to suit.

D Publish and be d....d!

E Hmmmm. No-one to send the bill to!

MediaMatrix


Direct link Reply with quote
 

Henry Hinds  Identity Verified
United States
Local time: 05:30
English to Spanish
+ ...
Reply to Viktoria Aug 14, 2006

How I use search and replace:

I do replace source words by their target language equivalents, in the source document. I don't even know what "segmenting" is.

Can you tell me how that can help to speed up the process and ensure that the final product is of good quality?

It speeds up the process because I sometimes don't have to type them again. As I go along I also pick up on words that are common to the specific document and do them.

I say "sometimes don't have to type them again", because often they can be out of place so the process doesn't always help. It helps a little. For most work the value is rather limited.

Of course it helps quality because those are words I can no longer overlook, they're changed and they're uniform.

No, I do the research on the spot because very little is necessary, the work I do is often very technical but mostly in fields in which I am experienced and I have a good memory. I am also a conference intepreter so I've trained myself that way. No time for research there, it's do or die.


Direct link Reply with quote
 

Riccardo Schiaffino  Identity Verified
United States
Local time: 05:30
Member (2003)
English to Italian
+ ...
A few important steps that seems to be missing here Aug 14, 2006

I think we should also remember our ousekeeping/administrative steps:

1) Receive availability request from customer

a) If existing customer, GOTO 2)

b) ELSE

1a) Check prospect on Blue Board, TCR, PP and any other payment practice lists

a) If reputation = satisfactory, GOTO 2)

b) ELSE

1b) Refuse assignment

2) Confirm availability to customer

3) Receive material to translate

4) Save material to translate in a folder (under the "Current work" folder) together with any ancillary material (translation memory, glossary, special instructions, etc.), this folder is created automatically by a script, so that all projects always have the same structure (I adopted this system a few years ago, after reading of a similar system suggested by Gianfranco Manca)

5) Record project in translation management system (indicating customer, rate, date received, date due, number of words, etc.)

6) Start actual translation work

... (several steps here, which I may describe more in detail later)

7) Send translated assignment back to customer (bcc myself to make sure the file was actually sent)

8) Move project folder to "Completed projects" folder

9) (At the end of the month) Invoice project

[Edited at 2006-08-14 01:15]


Direct link Reply with quote
 

ViktoriaG  Identity Verified
Canada
Local time: 07:30
English to French
+ ...
TOPIC STARTER
About sending the final file to the client... Aug 14, 2006

Nice trick, Riccardo! I haven't thought of that.

My trick is I send the document in the same e-mail as the invoice. If they have received the document, that necessarily means they also received the invoice. The invoice is in PDF format and is locked for editing My e-mail is set to automatically request a read receipt, and clients usually comply with this. Also, I backup all my work-related e-mail with special software for e-mail and attachment backup, so it is a lot like recommended mail - I always have proof that the stuff was sent AND received. For clients that are not yet established, I ask for some advance payment.

How about your translation process? Do you also read before actually translating? Do you do any preliminary research? Do you have any tricks to share with us that you haven't yet seen in this thread?

Thanks for participating!

Henry, I think then the real reason why we have completely different ways of doing this is that you don't use a CAT tool. At least, so I understood as you mentioned you don't know what segmentation is. Segmentation is when a CAT tool breaks up a text into segments (sentences) which will later be turned into translation units as you translate (TUs are what make up a translation memory or TM). I guess your unique search and replace trick is what makes up for the time you probably would have saved by using a CAT tool. In fact, if I had to work without Trados, I would probably eventually come up with something similar myself.

Still, I think our respective specializations also dictate, at least to some extent, the steps we take in translating.

Cheers!

Keep'em coming!

P.S. to Henry: I just thought of something! If you batch translate stuff by using search and replace, then you could probably use TextSTAT, as it will tell you which words are frequent in your text, by order of frequency. You would probably be able to replace much more text this way - which means you will have much less text to type. Why not give it a try?

[Edited at 2006-08-14 02:28]


Direct link Reply with quote
 

Henry Hinds  Identity Verified
United States
Local time: 05:30
English to Spanish
+ ...
No CATS Aug 14, 2006

Although I do favor them as animals, no CATS for me. I just dig in and go until I finish. Memory serves me quite well.

Like I say, search and replace is only of limited value, such as for "and, of, or, but, during, in", etc., plus others that may be common in certain texts.

Mostly it's just dig in and go. Translation is a human process, and so is mine, 100%.

Bit I do realize that many need CATS.


Direct link Reply with quote
 
Pages in topic:   [1 2 3 4] >


To report site rules violations or get help, contact a site moderator:

Moderator(s) of this forum
Maria Castro[Call to this topic]

You can also contact site staff by submitting a support request »

Translation - what process do you use?

Advanced search


Translation news





Wordfast Pro
Translation Memory Software for Any Platform

Exclusive discount for ProZ.com users! Save over 13% when purchasing Wordfast Pro through ProZ.com. Wordfast is the world's #1 provider of platform-independent Translation Memory software. Consistently ranked the most user-friendly and highest value

More info »
Anycount & Translation Office 3000
Translation Office 3000

Translation Office 3000 is an advanced accounting tool for freelance translators and small agencies. TO3000 easily and seamlessly integrates with the business life of professional freelance translators.

More info »



All of ProZ.com
  • All of ProZ.com
  • Term search
  • Jobs