Tageditor so slowwwwwww... (and what to do with it)

This short message is intended for the users of TagEditor, which is a part of older releases of Trados (I use Trados 2007).

You may have already encountered documents that make TagEditor run terribly slowly. This might be the case especially of long documents with complex formatting (many tables etc.). In my case, it was a document in the OpenOffice format (.odt).

Just sitting and waiting several long seconds for each segment to open is one option, not a particularly pleasant one.

Better option is this one: extract all the text from the TTX file, translate it in MS Word and then automatically translate the TTX file using the same memory. It works and it's much faster, although you spend some extra time with the exports and imports. Here's how to do it:

(1) Analyse the document (Translator's Workbench - Tools - Analyse).
(2) In the Analyse dialogue, enter 99 in the text field next to the "Export Unknown Segments" button and click the button.
(3) Change the "Save as type" to "Word .rtf", enter the desired name and click the "Save" button.
(4) Open the RTF file in MS Word and translate it using the Translator's Workbench, but not TagEditor. It runs much faster, doesn't it?
(5) After finishing the translation, use the Translate function of the Translator's Workbench (Translator's Workbench - Tools - Translate): add your TTX file in the "Files to translate" area, enter 100 in the "% or higher match value" field, and click "Translate".
(6) Now, you should check the TTX file whether all the segments have been translated. You can either save the translated file in the original format (if you can open it) – this is the best option for proofreading. Or you can use the following procedure: create a new TM, perform a cleanup of a copy of the TTX file using the new TM, and analyse the uncleaned TTX file using the new TM (containing the segments from the cleanup) - if the analyse shows anything except 100% matches, they are probably untranslated segments. You can export them as described above, so that you can locate them easily.

It's not that difficult and you can really save a lot of time.

P.S. Once you have all the segments translated, you should proofread the whole document. Again, you can save the translated file in the original format (if you can open it) – that is the most comfortable solution. Another option is to copy and paste all the content of the TagEditor file in MS Word. You can then proofread the translated file in MS Word. If any change is needed, make it in TagEditor (I assume that the TTX file is what you have to send to your client.).

My professional life from 15 September 2009 to 14 March 2010

In 2008, I developed my own system for recording my jobs/assignments. Since that time, I have been adding new features and improving the existing functionalities. In July 2009, I added a Time Management feature, which allows recording my exact working time. Each record includes date, time (from–to), job code, activity code. In this manner, I can precisely calculate how much I earn per hour for each job and I can also extract some summary statistics. So, what was I doing from 15 September 2009 to 14 March 2010?

Working time (total): 945 h 20 min
– Administration tasks: 91 h 57 min (9.73 %)
– Preparatory tasks: 69 h 43 min (7.37 %)
– Translating: 522 h 16 min (55.25 %)
– Proofreading: 225 h 28 min (23.85 %)
– Job completion tasks: 8 h 27 min (0.89 %)
– Quality assurance tasks: 10 h 26 min (1.10 %)
– Interpreting: 2 h 40 min (0.28 %)
– Transport to interpreting: 14 h 23 min (1.52 %)

Explanation:
Administration tasks – reading/answering e-mail messages (in relation to translation jobs), entering new jobs in the system, invoicing, going to post office, programming the system, updating website, writing this blog etc. This is the only type of activity which is not assigned a particular job code;
Preparatory tasks – creating/converting translation memories, preparation of files (conversion using OCR, formatting), etc.;
Job completion tasks – everything you have to do after the text is complete (sending by e-mail, using FTP etc.);
Quality assurance tasks – this is a special set of defined tasks relating to a certain class of jobs only. The client has provided certain tools, which have to be utilized for this purpose.

When was I working?
My most productive time of day is from 8:00 to 12:00 (33 % of all my working time). The period from 12:00 to 18:00 (although it’s longer by 2 hours) accounts for the same percentage – 33 %. 15 % of my working time fall in the period from 21:00 to midnight. The remaining periods of a day account for the remaining 19 % (for example only 3 % of my working time fall in the period from 2:00 to 6:00).

Are you a freelancer? Do you have any similar data in relation to your work? I would be very happy if you can provide any feedback and/or comparable data in a comment or you can send me e-mail message.

Proofreading Text Translated Using Trados (2)

(This article is based on my experience with SDL Trados 2007 and MS Word 2003.)

Last week, we learned some basic methods that can be utilized by proofreaders working with a text translated using Trados. We focused particularly on the case when the client provides translation memory (TM) and requests that the 100% matches should be left unchanged.

Now, let us have a look at some more sophisticated methods. The first one is very easy. It is, however, necessary to prepare the job for the translator before he/she starts to work on the job.

Analyse the files to be translated (Translator’s Workbench: Tools >> Analyse). In the Analyse window, use the button labelled “Export Unknown Segments”. The button is greyed out unless you have just completed the analysis. When you click on it, you will be prompted to enter the name of the file (use any name you like) and to select the file type – use Word (*.rtf). Before clicking the button, you can even decide what segments should be exported, based on the matching criteria. Be careful: the expression “100% or lower match value” means “lower than 100%”.

In this way, you can create RTF file containing all the segments differing from the current TM. The translator should be now given the exported RTF file (to be translated by him/her) as well as the original file for reference purposes. Remember that this is quite inconvenient for him/her.

What is convenient? It’s convenient to have the reference segments (the 100% matches and repetitions) right in the translated file. For this purpose, use the Translate function (Translator’s Workbench: Tools >> Translate). Leave all the values and checkboxes in the default setting and you can get a pre-translated file with all the 100% matches pre-translated.

You can either hand over the pre-translated file to the translator or – if necessary – you can even “lock” the pre-translated segments. What is meant by “if necessary”? If you have a large document where almost all segments are already in the TM, it may take some time to just go through it. If the client does not want to pay for the 100% matches (and there are such clients), it is better to do more “automated” preparations, so that the translator spends less time with the files.

First, create a character style (not paragraph style, this is important) and name it e.g. “skipthis”. Change the style of all the pre-translated text into “skipthis”. For this purpose, use the Find/Replace function. “Use wildcards” option must be checked. Find: “\{0\>*\<0\}” Replace by: “^&” (type everything that is between the quotation marks in the respective fields, but not the quotation marks) formatting on replace: style=skipthis. (The segments changed in the skipthis style may lose some formatting, but it is not a problem – you can create the final version of the translated file from a clean source document, if all the segments are translated in the TM.) For better clarity, you can set that the skipthis style is in grey colour.

The final step: tell Trados to skip any text in the skipthis style. Translator’s Workbench: File>>Setup>>Non-Translatable Text. Type skipthis in the line on the left side, click External and push the Add button. Now the translator should be able to see everything what he/she should see, while Trados opens only those segments, which have not been pre-translated.

Proofreading Text Translated Using Trados (1)

(This article is based on my experience with SDL Trados 2007.)

Generally, Trados and similar CAT tools make the work of a translator easier. The same applies for the work of a proofreader. If not for anything else, it is quite convenient to open the text segments one after another and see the source text in one box and the target text in another one. There are, however, some special issues, which require special attention and maybe also certain know-how.

Some customers provide their own translation memories (TM) and require that the 100% matches should not be altered (quite often they are not paid as well). In such a case, the translator’s work is easy – if a 100% match is encountered, he/she just presses the necessary combination of keys and continues with the following segment. What can do a proofreader, however, working on such a job? How does he/she recognize the 100% matches provided by the client from the other parts of the text? Can he/she check whether the translator observed the customer's requirement not to alter the 100% matches?

First of all, if you obtain a TM from the customer, keep one untouched copy aside. Check whether the translation units in the TM contain user ID’s of their creators which can be clearly recognized. Export the TM as TMX. If necessary, open the file in Notepad or similar application and change the values of the parameter "creationid" to something like "original_TM" or so. Then import this modified TM in a newly created translation memory. (Even if the "Created by" parameter does not need to be modified, we will use the exported TMX file later.)

Make sure your translator is properly using the ID parameter (Menu Settings >> User ID in Translator's Workbench). You can change this parameter only when no TM is opened. The same applies to you – every person touching the translation should have his/her own User ID, even if they work on the same machine with the same Trados application.

Now you can go through the text and do your work. If you see a segment where the parameter "Created by" shows the customer's ID, you can skip it (unless the parameter "Changed by" shows anything but empty value – that would mean that somebody has been doing something with it).

When you finish, it's time to check once more that your translator did not change the 100% matches with the TM provided by the customer. Import the original memory (saved as TMX at the beginning) into the current memory containing all the new translations. In the Import menu in the section "Existing translation units" check the option "Overwrite".

Go to the beginning of the text, open the first segment and press Alt-*. Trados will go through the document, stopping in the segments, where the TM differs from the text in the actual file. Typically, this would indicate a 100% match changed by the translator. However, this does not have to be always the case, so be careful and check each individual instance thoroughly.

Now you have completed your task.

You may ask: "OK, and what about 99% matches differing by formatting only etc.?" Yes... you are right. This topic has not yet been explored completely. See you next week.