User name:
Forgot User name / Password?

Register (Free)

Please take a minute to read our privacy policy.
Login or register
About | Contact | Blog | Site map


4. You want to align a source document and its translation to create a TM.

Perfect alignment will always require human verification, but you can get pretty close by doing the following:

Open the source document. Strip it of all extraneous formatting by saving it as txt. Close it and reopen. You should now be rid of all the unnecessary stuff that will not go in the TM anyway, like pictures, etc. Now, observe the document and determine how it should be segmented – which ESPs should be used (End of Segmentation Punctuation, like periods, question marks, exclamation points, columns…). Important. If the text contains periods that are not ESP (like the period after abbreviations, …, etc.), they should be replaced by a symbol that is not already in the document, like “µ” for instance.

Then, using F&R, replace all the ESPs by themselves + a paragraph mark. (I.e.: Find ? Replace by ?^p, without wildcards), then remove all double paragraph marks as explained before, and change it into a 1 column table based on the paragraph marks. (If you had to replace periods by a symbol, replace it back.) Make sure there are no empty lines or segments cut in 2 parts. Add an extra column to the right.

Do the same for the target document, but at the end, instead of creating an extra column, select and copy the target column, and paste it into the second column of the source document. From there, it’s alignment as usual, and there will probably be mistakes to correct, but the columns are already as closely matched as possible.

Now that your table is aligned, with 1 source and target segment per line, you can either create a TM or create a Wordfast/Trados segmented file.

To create a TM, add the missing columns (see “translations memories ” for the exact composition of a TM), and populate them. Then select the table, convert it to tab delimited, and save as a txt file.

To create a segmented file, add 1 column before, 1 column in the middle, and 1 column after. Populate the first column with {0>. Populate the middle column with <}100{>, and the last column with <0}. Select those 3 columns and apply the tw4winMark style (you can – and should –copy that style from a segmented document in your normal.dot). Also apply the “hidden” attribute (format/font-hidden, hidden is in the option cases). Select the source column and apply the “hidden” attribute too.

Make sure you DO NOT have empty lines. Now convert the table to text, delimited by a symbol which is not used in your document. Remove all those symbols with F&R and you have a perfectly normal Wordfast/Trados segmented file.

Note that I entered a match value of 100%, but you are free to enter any match value that you think is applicable, from 0 to 100%. However, since you just review it segment by segment, the matches should be pretty safe.

This is it for now. These ideas are not really meant to be followed to the letter, although they can be, but to present how different situations can be solved using basic functions of Word. Hope you find this useful.