Maintaining a positive mental attitude...

It's not altogether unusual for me to mess up the first run at aligning source and translation documents, but between last night and this morning, it became clear that what I have been given is a translation and a heavily edited source file. In other words, the fragile balance between the structure of the source and translation—fragile because not only does a translation not match a source on a word-for-word basis, but there are times when sentences must be split, or combined—was disrupted time after time.

What is needed is a smart utility that will scroll two documents in such a way as to display source in one window and the corresponding translation in the other. Word has a feature along this line (scroll side-by-side) but (a) the feature does not take account of differences in formatting, and (b) I've found that getting it to work is not always a given. (It's pretty capricious about reformatting windows, for example.)

Given that often, I can jerry-rig something that'll get most of my desired behavior at a fraction of the effort required for a proper go, I need to get smart and figure out how to go about this. The files I am dealing with are not small.

Cheers...

Alignment summary...

I ran through the alignment software offered by MemoQ, Trados (2007 version), and both Wordfast Pro and Wordfast PlusTools—in pretty much that order—and just as one finds what is lost in the last place one looks for it, it was the PlusTools that enabled me to align two files that weighed in at about 75,000 words each.

Besides its clunky interface, the MemoQ aligner did not do a good job of segmenting the files, and while the Trados aligner exhibited superior segmentation, it ran slowly on my old Windows 7 machine and exhibited some of the same interface issues as MemoQ (i.e., an inability to delete stuff).

The Wordfast Pro aligner is not very intuitive, and does not appear to have the ability to join or split segments, so I gave that up as an alternative pretty quickly.

The PlusTools aligner did a very good job of segmenting the files, and even has a feature that helps undo some of the damage done by abbreviations (such as "г." and "гг." in Russian, which are used whenever a year or year range is stated, or in front of any city name).

A good indicator of how well PlusTools addressed my needs is the fact that I was able to make good progress rather quickly.

In an ideal world, one would review both source and translation as one aligns, but this is not an ideal world, so one uses all sorts of speed-reading tricks to check the alignment and move through the file. Numbers, names, dates, and the beginnings and ends of sentences are useful checkpoints. Sometimes, I would risk skipping a few paragraphs if they were preceded and followed by what looked like a "good" alignment pair

Yet despite all the tricks, I was not able to progress much faster than about 12,000 words per hour.

In the end, I managed to create a translation memory file with just short of 3,000 pairs (called "translation units). Given the number of abbreviations and names (both of persons and settlements) in the first part of what is on my plate, I think creating this TM will be worth the time invested.

Oh? Is it Friday night?

It completely slipped my mind. :^)

Cheers...