Beginner's first steps with MemoQ and Wordfast

I am still very much a beginner in Translation Memories (TM) but I have accumulated quite a large "base" of translated documents. So I am especially interested in the alignment function of TM programs.

MemoQ (www.kilgray.com) offers its 2.0 release as freeware if you send them a postcard, hence the name "Postcardware". I decided to give it a try to translate (from German to French) a 16 page long catalogue for plasma cleaning machines. The deadline was generous. I had previously translated two 38 page long instruction manuals for such machines.

MemoQ needs MS .NET Framework 2 to run (Framework reminds me of an integrated office suite of good ole DOS time). Installation is easy (45 MB with Getting started manual). To start, you create a project and import a pair of documents to align. My first pair was refused because the original had been partly written with Word 97 - Word 2000 with SR2 is required at least. The second one was accepted, aligned and I opened it to edit the alignment: original on the left, a center column showing eventual links (colored differently if aligned automatically or manually), translation on the right. BUT the program played havoc with the French accented characters.

I tried Support, wondering if I could get any help - for freeware! - and the response was tremendous: frequent exchanges, prompt and friendly answers. It turned out that my translation (made with Word 2003) was mixing non-Unicode characters with an incorrect code page setting which could be ignored by Word but not by MemoQ. Support cured it by unifying all fonts to Arial, saving the document into a DOCX file and saving it as RTF in a Unicode-compliant file format.

I got it for alignment in a new project with sound French accents. MemoQ is a pleasure to align. Click a cell, it turns blue. Click another cell in the opposite column which turns blue, the first cell keeps a light blue hue, waiting for an align/unlink/join/split command (right click or shortcut - my favorite!) and the affected cells realign, blank cells being inserted if needed. No need to Control-click: convenient and user-friendly. Support was again needed to export my alignment to a TM, as there is no explicit instruction in Help or Getting Started.

I inserted my plasma catalogue for pre-translating and faced another shock: Tags! I had almost more tags than characters - the catalogue having been converted from pdf to Word by Solid Converter. Proz.com is replete with warnings not to accept pdf originals - but I had, and once the first step is taken, there's no going back.

I tamed my file by suppressing all variations of kerning / spacing / sizing Solid Converter is fond to add and got rid of most tags, opened it for translation and was stunned to get zero suggestions.

MemoQ sports now the 2.2 release, claiming many improvements. One can upgrade to it for 198 € or buy the full version for 390 € and get all upgrades forever for free.

I turned to Wordfast (familiar among Prozians) because I just wanted to have a glimpse of how it handles alignment. I applied the Extract and Align tabs from the +Tools on the same instruction manual file pair mentioned above. The alignment looks like a common Word table with specific shortcuts added. At first sight, unlinking and realigning cells requires much care and many cut and paste - not really my cup of tea.

These tests nudged me to buy Similis (Freelance + one year support) and to run the same test. Similis accepted the original file pair (not the Unicode doctored file) without grousing or maiming a single French character. Pre-translating the plasma catalogue (the oven fresh file spat by Solid Converter) displayed without tags but did not offer any suggestion either.

The deadline for my catalogue drawing near, I went back almost ruefully to my familiar forceps pre-translating method, Word Frequency Counter Advanced plus Useful File Utilities' Batch Replacer (see my article Beginner's first steps with Similis).

Moral of the test: Similar topic and terminology is not enough to build a usable translation memory. I still have a long way to go, looking forward for new lessons from experience and expanded know how in using these fascinating, but not all powerful tools.