People who know me and my work know that I have a very low opinion of machine translation use in most language service situations. Even in the best scenarios, it offers no value to me in my routine work as a translator of scientific and intellectual property texts (patent filings and litigation mostly). So why am I totally excited about the new e-book by my friend and colleague Marek Pawelec? For several reasons.
- MT discussions bore the crap out of me. But when Marek asked me to review a pre-release copy, I was actually entertained by his clear, concise writing and the superb way he explained basic concepts of resource management in general that most memoQ users still don't master. I was shocked at how much fun I had reading about a subject I hate!
- He talks about more than just how to configure memoQ to use DeepL, Gargle Trashlate or some other MT engine. He details strategies and best practices for effective use that many people might not be aware of. He talks about how to circumvent prohibitions on MT use and how to catch people who do that. And more. I didn't learn something on every page, but it's probably not an exaggeration to say I did on every other one.
- Pseudo-translation using a special plug-in for the Pre-translation step is covered in wonderful detail. This technique has been very important to my work for nearly 20 years now. I use it to identify hard-coded interface strings in software I translate and to check and quote large documents that might have paragraphs or whole pages scanned and inserted as graphics that look like editable text - or charts whose text can be selected in the document but never show up on the memoQ working grid after import. Marek also discusses other uses of pseudo-translation I never thought of (layout checks, for example) which could have saved me a lot of grief over the years.
The only complaint I have about this book is that it's too cheap. The author teaches me more in its 36 pages than most can in 200 pages, and the learning is worth a Hell of a lot more to my business than fifty cents per page. Anyone else would probably have written much more and communicated far less of value, but that's a special gift that Marek has. Long ago, his talk at a memoQ Fest was the first time that regular expressions (regex) made any sense to me (as a casual programmer for about 40 years at the time I had approached the topic many times and mostly just found confusion). There aren't many people in this world who can take complex topics and make them seem simple and interesting to nearly anyone. Marek can. Richard Feynman could. I can't name many more on that list.
So... all I can really add is to tell you to go spend €18 here: https://payhip.com/b/tF62
The value you'll receive as a memoQ user at any level, even if you never use machine translation, is a large multiple of that price.
Update 2022-05-05
There are now editions available in Dutch and French:
Dutch: https://payhip.com/b/XaZSU
I wonder... if one completely agrees with your first paragraph and therefore rightly so as a professional stays clear of all that smells MT, why should one still need to buy and read this book? What value can one then get from it?
ReplyDeleteAs we discussed in the Twitter thread, Herman, the information about pseudo-translation (third bullet point) is of enormous value for many types of projects. Although it uses the machine translation infrastructure, it has nothing to do with MT.
DeleteMy opinion about MT, and PEMT, changed a lot nowadays: considering it is the future of our business, whether one likes it or not, and certainly not for all fields, but for many, I think that any smart language provider must learn to ride the wave, because the other way is just sinking
ReplyDeleteNot mentioning that, an increasing number of direct clients (i.e. industries) use Deepl and other MTs by theirselves, then send the pretranslated papers to language providers, so "il dado รจ tratto"
The future? Hardly. You do have breathless wankers on LinkedIn talking about a future without translator corpora such as translation memories with everything being provided by some idiotic algorithm, but that claim more of a promotional strategy for their sleazy business models, not how real translators will serve the needs of clients twenty years from now.
DeleteAs for industrial clients using MT, thinking in their monolingual, subliterate way that it need only be touched up a bit, shrug and offer to "review" at the full translation price and ignore the garbage they send you :-)