Do I give a cl*ck about memoQ?
Birdbrains in business and politics, and flashbacks to my medical device materials days
That little girl on the bench is a distant granddaughter of the three roosters I brought to live in a Lisbon park when the alternative was to butcher them to lessen the pressure on my own hens at home. All the intermediate generations seem to have maintained a memory of my feeding calls, and when I’m in town to teach, visit or whatever, I like to check in and see how the feathered family is doing.
It’s been quiet here for about a month as I indulged in a pre-election writing break and concerned myself with family health matters like the cataract surgery I’m chaperoning today, which gave me the opportunity to be with the birds in the park again. If anyone missed me in this morning’s office hours, I apologize for that; the tiring travel and shock of all the schoolboys in the US announcing “your body, my choice” to their female counterparts simply erased my awareness of plans and any temporospatial anchor points. I’ll be back on call next week Thursday from 10 a.m. to noon Lisbon time if anyone has questions that need live answers.
The last few weeks have not been without some interesting challenges and reminders that basics are not always what we might assume. Questions from that time could make an easy half dozen tip posts. And they underscore the need for some kind of memoQ basic instructions reboot camp. So I’ll make my first run at that in two weeks.
Registration for the “memoQ Basics Reboot Camp, First Draft”…
… is here. One of the foundational assumptions of that session is that your first encounter with memoQ might be an online project, or you might be an experienced memoQ user making your first forays into online projects from some client with a memoQ TMS server. I’ve been surprised to see some users with more than a decade of experience stumble here, and it made me aware that some of the options I take for granted in these cases might not be as generally known as I assumed.
We’ll talk about differences between web browser and desktop software translation interfaces, including resource management and share performance tips. Some performance tips are for things not under the direct control of translators, but awareness of them might enable you to work more effectively by sharing these tips with project managers who are able to make some helpful changes.
I’ve planned two hours for the tutorial on Thursday, November 28th, 2024 at 4 p.m. Central European Time. More details on the content will be shared next week. I can do a German version shortly afterward if there’s sufficient interest. Drop me a message if you or others in your circle want that.
The Curious Case of the Concordance, and Other Mysteries…
A couple of weeks ago, while I was busy moving and restacking our farm’s winter supply of firewood, I got a call from an experienced memoQ teacher and surviving freelance translator about an online project he had taken on, in which the concordance failed to return any results for his terminology searches. This is a guy who knows his way around rather well, but I ran him through the usual suspects gauntlet anyway and asked performance-relevant questions on the assumption that he might be dealing with timeouts.
How many translation memories and LiveDocs corpora are attached to the project? Nine TMs, all with 10k translation units or less. I think there were no corpora in this case, but the slower indexes of these can be a real performance issue in online projects in many ways if there’s a lot of data.
Is your project’s synchronization set to manual rather than the default of automatic synchronization? This question is utterly irrelevant to his Concordance problem, but I felt compelled to ask, because users are often unaware of these settings or simply forget, and this automatic synchronization really slows things down for many users in many projects, especially if their Internet connection speeds and bandwidth aren’t great. (See this nerdy post by tech guru Gergely Vandor on memoQ online project connection types for more tips on speeding things up!)
Do you have local copies of the project TMs? No in his case. This is an option for project managers while setting up projects, and it not only speeds things up in lousy bandwidth situations, but it enables offline work for those of us who live with frequent rural power outages or who have an adventurous life in some place frequented by hurricanes. (One participant at my last webinar was without Internet connectivity at his home for a month after one such storm.)
If a project manager has not enabled the automatic download of the online translation memories and term bases while creating the project, users may be able to make local copies themselves with the option Synchronize Offline in the context menu:
Without any particular evidence to support that assumption, I thought perhaps Concordance responses might have some kinship with Translation results in that increasing the configured delay might help, but that was unhelpful in this case.
An expert memoQ developer examined the relevant log and found NOTHING.
You’re probably thinking I’m going to offer a solution here. But that’s not the case. The LSP had “no problems” doing Concordance lookups and so dismissed the translator’s troubles. But the problem did get resolved, I suppose. After what he said was the fifth reboot, the Concordance lookups started working. WTF? I dunno. Some things are like that, and trying to solve all the world’s problems in a Facebook Messenger chat is worse than seeing through a glass darkly. Especially when getting relevant information is more difficult than extracting teeth without anesthesia.
A little more joy was had with…
… a review project a friend got stuck on. She was given XLIFF files which had some import issues.
This was another one of those cases with a developer who just doesn’t understand how to work with XLIFF specifications and thus does incredible bullshit that simply doesn’t work in the real default world of most CAT tools.
Children, children, children. Don’t make your target language the same as the source language. Even if for some reason you want to populate the target tag content with source text, indicate the target language you want in your trgLang declaration, please. I’ve seen worse stupidity than this; here a simple edit to change the designation to en was enough.
But problems like this come up all the time, especially when the exports for localization are done by inexperienced individual developers. I’ve probably seen at least a dozen cases of problems like this over the years. Never trust developers to get a specification right the first time. Or even the second time. Always, always test these localization formats before assuming that they work, and if there’s a problem, take some small comfort in the likelihood that the problem isn’t you. In an earlier time, most of these devs would be pushing a broom.
There were about six follow-up issues for that project which I’ll talk about later, because they get all icky regexy, but in a way that makes a useful set of lessons because the problems encountered are common.
Problems with the XLIFF file led to the use of a CSV file instead, and there the multilingual Excel and delimited text filter was used, and of course there were issues because each “cell” is treated as a segment even if it contains many paragraphs with many sentences and just gets f-ing huge. But no problem. Cascading an HTML filter for the import resolves that issue. But that’s also a subject for another day.
I have important consultations with my Trappist friends on the future of autocracy in ‘MurriKKKa.
