discussion of text alignment tools

[mcthulhu]

http://www.proz.com/forum/translator_re ... ml#2678642 mentions several free tools for sentence-aligning a source text with its translation (e.g., to produce parallel texts as pairs of sentences). LF Aligner might be the most obvious choice for home use but several others are also mentioned.

[Iversen]

If you have to align a very long text then it may be relevant to use a special tool, but the bilingual texts I use for intensive are short so I just use MS Word (similar programs like LibraWriter can also be used).

I make a table with two columns, and then I put a page or so of text in the target language to the left and the translation to the left. Then I can align the texts simply by moving the vertical line in the middle to the left or to yhe right right, and I can do fine adjustments by changing the font size (or the font itself), for instance to get the texts to match nicely up at the bottom of each page. The tables don't have to end precisely at a page shift - once you have found a position of the central line that gives good results you can usually do several pages with that setup. If not, then you just start a new table beneath - just as you do if you switch to another source text.

Long ago I experimented with other layouts, including interlinear and side-by-side layouts where the sentences were exactly matched, but I concluded that they weren't more easy to work with than the side-by-side layout you get with the method described above, and they were much more difficult to produce.

If I wanted a bilingual edition of the Lost time of Proust or the collected works of J.K. Rowling then I would still be satisfied with a tool that just produced a simple side-by-side setup. If you really want to read whole books then you must be fairly good at reading already, and then it doesn't matter whether the translations are half a centimeter higher or lower than the original passages.

The big problem is to get translations that follow the original texts so closely that you actually CAN match sentences and paragraphs, Often you'll find that whole sections in humanmade translations are missing or based on another version of the original text or whatever, and I haven't heard about any software that can solve that problem.

[hedgehog.chess]

I use LF Aligner to create a Excel file in which the sentences more or less match. Then I go through it and cut/paste text or delete rows for the text to match exactly. Some examples of how it looks after processing can be found here: https://forum.language-learners.org/viewtopic.php?f=19&t=6438

[mcthulhu]

Iversen, I was a bit confused by your response, but I think we might be speaking two different languages here... Alignment in this context does not refer to physical page layout, or to whether translations are "half a centimeter higher or lower" than the original.

Alignment by a sentence aligner means that an algorithm is used to determine which sentence(s) in a translation most likely match which sentence(s) in the original text. The sentence pairs, whether in the form of spreadsheet columns or a TMX file or something else, could be reformatted for presentation in various ways, but that has nothing to do with the process of identifying the sentence pairs.

[MorkTheFiddle]

Aligned texts get frequent mention hereabouts. I started a different thread, Bilingual Texts Redux, about the location of some parallel texts here https://forum.language-learners.org/vie ... =19&t=6725

[Iversen]

Alignment by a sentence aligner means that an algorithm is used to determine which sentence(s) in a translation most likely match which sentence(s) in the original text.

Since I don't use sentence aligners I naively used the word alignment in its normal, purely geometrical sense. It is actually a surprise to me that it is used differently in sentence alignment software - I would have expected a word like 'matching' for the task you describe.