Source material: corresponding translations or side-by-side texts[]
Most parallel texts you'll find will be side-by-side ones. The name is more or less self-explaining, though is should be said that it is normally whole paragraphs that are juxtaposed rather than single lines:
language1language1language1language1 . . language2language2language2
language1language1language1language1 . . language2language2language2
language1language1language1 . . . . . . . . . language2language2language2language2
You can make these in a word processor using either 'sheet' columns or columns in a table. To make the sections fit you can put in empty lines, but it looks prettier if you force to the sections to have the same number of lines by regulating the width of the columns.
Producing your own interlaced texts[]
An interlaced text has one line in the target language, then one line in the base language. Ideally each line would contain one sentence (or another natural unit), but sometimes you have to cut in the middle of something or you have to leave half a line empty.
They can be produced using a word processor (Word, OpenOffice writer, WordPerfect) plus a spreadsheet (Excel, OO Calc, QuattroPro). You take a text and its translation into a better-known language. Put each of them in its own word processor window and put one window above the other on your screen.
Then you start subdividing the texts in parallel chunks of about 2/3 of the width of the screen, each chunk on its own line. Take care that each chunk in one language as far as possible corresponds to one chunk in the other. It is not as difficult as you might think if the translation is faithful to the original, but in a few rare cases you may have to move something away from its context - do this in the base language version, not in the target language version as there may be some syntactical reason for the changed order of the components.
Then open a spreadsheet (for instance Excel) and put one text in column C, the other in column F. You should fill column A and D with numbers 1,2.... , and afterwards fill column B and E with respectively a's and b's. At this point check that the two versions of the text really correspond line for line to each other. You should also give each language is own color and font for easy recognition later. Then you cut out the content of the three columns D,E,F and place it below the content of columns A.B,C. The idea behind this arrangement is that you now can sort the whole thing according to columns A and B.
The final result[]
When you've done the steps described previously, your text will look like this:
1 a blahblah (language 1)
1 b blohblohbloh (language 2)
2 a blahblah (language 1)
2 b blohblohbloh (language 2)
...
The last thing to do is to copy the content of column C (i.e. the interlaced texts) to the word processor, where it appears as a table whose dividing lines can in principle be removed without harming the content of the cells, - but it is not really necessary to do it. That's all.
Using interlaced texts[]
Even with this techique it will take a very long time to prepare a whole novel. Interlaced texts are best suited for intensive reading, where you analyze every single sentence and learn every single word - a few pages per session are enough. With a better known language you can understand most of the text and most of the words, and then there is no longer any reason to use interlaced texts, and you can settle for side-by-side texts (which eventually also become irrelevant).