Speaker 1: What's the deal with translation memories? Coming up. Hello and welcome back to the Freelanceverse. Thanks so much for coming back to the channel, nice to have you here. Today's topic are translation memories. Cut tools for translators are still a very mysterious topic for many of you out there. Are they really needed? What are the benefits? Do you need to pay hundreds of dollars just to work in a software as a translator, right? My short answer, if you ask me, yes. Look, as freelance translators, we are already very fortunate that we have very little overhead costs, right? We really need just a good laptop, a nice workspace and a few software, a few tools to work in. So these are real working tools. So it makes sense to invest a bit of money. And then when you start getting the jobs, you will immediately earn it back anyways. So yes, I would definitely say invest in one, especially since they are 100% tax deductible, at least in Belgium and I guess in most of the places, because they are your actually primary working tool, right? Anyways, I wanted to start a little series in collaboration with MemoQ to show you the different aspects of various cut tools. And today we start with translation memories. A translation memory, also referred to as TM, is basically a linguistic database that stores previously translated text. This text is also known as segments or translation units. They are mostly sentences or very short text units you're translating and whenever you're saving the segment, it goes into your translation memory. So it basically keeps track of the source text with the corresponding target text that you translated and it puts it into a basket. So the purpose of a translation memory is basically helping translators and reviewers to provide more context, to provide already translated text, so that when they encounter a segment that has already been worked on by someone else or by themselves, they don't have to work on it again. This obviously saves a lot of time, effort, money as well for the client, it ensures productivity, it also increases productivity, but it also ensures, I can't think of the word, consistency of course. It ensures consistency because, especially in very repetitive texts, when you're working on a, I see a screen here, so when you're working on a screen manual and you have ten times in this whole manual they talk about the on-off button, then you need to have the on-off button very consistently translated. You can't have it once, say, the on button, once the power button, once the switch, you know, so it gives you a sense of consistency, definitely. In addition, a translation memory can store also additional information such as metadata, who was the one confirming the previous segment, do you trust this translator or do you know they have kind of a bad reputation already, right, so you should double check it. This metadata information can sometimes be quite useful. You can also have a topic, for example, if you have a sports, a legal, a medical TM, you can have a topic in there so you know exactly where the match is from. So metadata can be useful, I don't really use it that much because I know the TMs that I have, I don't have that many, right, so I don't use metadata but I can see that if you have a bigger corporation or an agency it can be very important and useful. Now that we have the theoretical definition part out of the way, let's actually get to the screen and look at it from a practical side. Very quick intermission here to show you what MemoQ has just brought out, this amazing spelling quiz, I think you should all go and try and do it. It's very simple, you have 30 seconds to guess as many correctly spelled words as you can. It's quite hard, so try it out in the description, let me know in the comments how many points you got, so let's try. Idiosyncrasy, that's not correct. A quiz, that's not correct. Tomorrow, that's correct. Mnemonic, that is correct. I have no idea what that is, but I was wrong. Sherbet is correct. Pronunciation, that is correct. No, it's not. Ridiculous. That... Okay. Okay, I'm back. No. It's so hard. Ugh. I had 9 points. A typo trooper. You seem to know a thing or two about proofreading. To make sure that your translations are flawless, MemoQ Translator Pros automated quality checks can always spot errors for you. Make sure to check it down below and let me know how good you were. Alright, so a MemoQ project usually consists or always consists of a translation document, so the file you're actually working on, and the resources. In the resources you can add all kinds of things, amongst them also a translation memory, what we talk about today, but there can also be a glossary, a translation turn-base, a style guide, just basically whatever reference material that you would have to kind of collect from everywhere else, you can combine in a CAD tool. That's what it's for, basically. So when you create a new project, what we will do here in a second, you always have to add either an existing or a new translation memory. I'm working on two screens here, that's why it's always disappearing. So you have to do one or the other. You can't work on a MemoQ project without a translation memory, so there always has to be one. So let's just do here a test, test freelance. Yes, I will do English to German. I prepared a little document here from the European Commission. So when you want, you can add the client here, European Commission, project, domain. The more you add, the more you will eventually have in the TM as metadata. I go next, then translation documents. That's where you import the document that you want to work on. This is a document about the European Year of Skills. That's the official name of this year at the Commission. I just took this from the website. You can also do that. It's a very good practice because these are kind of texts you would encounter as a general translator. Then I go to the next one and there are the translation memories. Now, as you can see, the two ones are already here because I created them already in preparation for this video. If you don't have them yet, you can click down here, bottom left, create, use new, and then it opens another window and you can create a translation memory. If you want to use the one that I used from the European Commission, this has about 50,000 entries all around Commission vocabulary terminology. You can download this from the website. You can do this from many institutions. People that don't know this are always surprised by this. If you're working with official government institutions, not only, also big corporations sometimes, they provide translation memories or glossaries online where you can either look it up online or download the exported file. You could create new, then give it the name European Commission, and then import the downloaded file into this translation memory. I did that, so this is now filled with 50,000 terms. As a second TM, I have the Master Adrian TM. I get into that in a second. I click now both of these because I want to use both. Then, as you can see here, the type, I make it bigger. It's an important distinction between working and master. These are two different TM types. Usually, MemoQ assigns working and master to one TM. Me personally, I like to attach two TMs and then assign one as the working and one as the master. This is already called master, so I click on it and it becomes the master, and this is the working. No, wait, it's wrong. It's the opposite round. Set as working and the master, set as master, like this. Okay, what does this mean? A working TM is basically, how can you describe it? It's the one that you use in progress, right? If I start translating now and I start entering the segments, I'm confirming the segments, they go into the working TM, into the European Commission TM. When I update the segment and I confirm it again, it just overwrites the one in the working. It doesn't do anything in the master yet. When you're finished, completely finished, saved, approved, and you deliver it to the client, then the project gets closed. You can, this, the button up here, yeah, deliver here. When I click that and I send it to the client or I export it, then the project is closed, it's all approved, and the final polished translations go into the master Adrian TM, right? So there you only have the finished versions, the approved versions. That is especially good if you are building your own proprietary, I can never say that word, translation memories, because this is data, this is very valuable data, and it can be leveraged later on in negotiations with clients, right? So you have to be very careful with what is your data, what is the client's data. If you're collecting master TM data when you're translating, this can become very handy because you have a huge database of all you've previously translated. So whenever, like this is not my actual master TM, but if I do any job in MemoQ, I attach my master Adrian TM, and almost every segment something comes up, some kind of match, right? Often it's irrelevant, but it still gives you some indication, and that can be useful. You can export a TM and you can share it with friends or with colleagues or with very trusted clients, but be careful and treat it like your terminology bank account, basically, right? You're the only one to give access to people you really, really trust. There we go. We have these two set up, so we continue. I don't have a term base for this, so I click on finish and it gets imported and I can show you what it actually looks like. Now, usually you could just start immediately translating in here, but I want to show you where you'll find the TMs here on the left, this toggle translation memory. When I go in here, then you see again the two, and I could actually go into it now, the European Commission one. You see the two, you see which one is working, which one is mastered, and in the translations you see the document. So, let's get into it. I'm opening the document, and the first thing you will see, hopefully if it works, if everything is set up well, yes, it is. Okay. Top right is translation results. This is a very important panel. I usually make it a bit bigger. As you can see, the first segment's European years of skills. Every year at the Commission is a European year of something, right? So, it's very likely that every European year of something is in this TM, if it's updated. And it seems to be updated, because Europäische Siegertage Kompetenzen, so there is an official name for this. You can still verify that on the Commission page, maybe quickly do it if you're not sure if you're using this TM for the first time. But if you trust the TM, then it's a 101% match, that means it is context verified. Let me just tell you what that means. A 101% match is a so-called context match, that means that the source text in the segment text is exactly the same as the match from the translation memory. In addition, the context of the source text is also the same as the context that was stored in the translation memory. So, a 101% match you can almost always use without even checking, but you should still check. I'm going to show you in a second why. But this one you can definitely use, and you see this is the match number one. So, with the CTRL-1, it immediately inserts that if I'm active. So, Europäische Siegertage Kompetenzen, and I can click CTRL-Enter, and it's confirmed. And then in the next segment, as you can see, it's called Why a European Year of Skills? There's a big overlap, in fact 77%, and it immediately adds it because it knows that you will need this, right? And now you can just add Why a. Now, I mean, I do a direct translation now. Warum eine Europäische Siegertage Kompetenzen? It's debatable whether you would like to have this in a question format. It's not really common as a German title, so I would probably reword this in a way, but I'm not doing this now just to give you an indication of how it works. Segment three, this is actually a metadata from a logo, but probably since it's a placeholder, you don't need it. So, you can, for example, copy source, and then it takes the English over, and you don't really want to add this into the TM because it's an exception, right? It's something that stays in English, so you don't want to populate the translation memory with that. And this you can do with Ctrl-Shift-Enter instead of Ctrl-Enter, or you can right-click and then you see it. Ctrl-Enter is a normal confirm, and the Ctrl-Shift-Enter is a confirm without update. So, if you do that, then it gets confirmed, but it's not populated to the TM. It can be useful sometimes. A quick word to the colors up here. You can see different colors of translation matches. So, these can be different matches. I'm not sure of all the colors. I'm going to put up a screen. Where can you see it? Maybe here. Of all the colors, and I'm going to put the link down below without going into too much detail. I know that red is translation memory. Blue is turn-based, I think. Yellow is machine translation. Orange. I'm not sure what orange is, but make sure to check it out. It's just different forms of sources of matches. And I'm going to also put the link down below for the percentages because that's actually quite useful. Now, I quickly want to show you something. If I go down here to how, I figured I found this out when I prepared this video. This is just a simple segment that's called how, and there is a translation 100% match with V, which is the direct translation, right? And you would assume, I mean, it can't be different than V, right? How is V? So, you would just take it in there and confirm it. However, if I show you the original document, now this goes a bit further into cultural knowledge because I show you the original document here. This is it, and then I go down to this how. Now, if you look at this, the previous sentence, helping people get the right skills for quality jobs and helping companies, in particular small and medium enterprises, address skills shortages in the EU is what this year is all about. How? By showcasing skills. So, it's kind of a rhetorical question, right? If you just had here V, like it already proposes as a 101% match, so you assume it's correct, it would be so strange in German, right? You would never have that. This would be so strange. I would know exactly why this is there. It looks very translated. It's obviously an English rhetorical question, how? So, you would not have that here. You would have to rewrite that. You would probably have no question here and simply start by this one. So, 101% context match doesn't save you from thinking and taking into account cultural competencies, right? It's very easy to just go through and take term-based matches, take translation memory matches, but that's not how translation works. Luckily, it's not how it works because this is something that we are here for. This is something we can excel at, right? So, definitely keep that in mind when you're translating. There are some agencies predominantly that don't want to pay you for 100% and 101% matches. I would always reject that idea exactly for that reason, right? You still need to check it. You still need to come up with a solution for it. And if I had to now rethink this whole sentence, take off this V, somehow merge the segments so it's not an empty segment missing at the beginning of the sentence, and I don't get paid for any of this, that's pointless, right? So, be careful with that. And the last thing I want to mention that is very important when it comes to TMs is the concordance search. So, if I go up here, I think it's review, is it? Hang on, it's not review. It's translation. Sorry. If I go up here in the tab translation, you find concordance. Concordance basically, it even describes here, search for occurrences of a word or expressions in your project's translation memory and LiveDocs corpora. This function helps you translate recurring expressions consistently and often it will even give you likely translations. So, you can basically search. It opens in this window again. You can search your whole translation memories that you have attached for a certain term. So, let's say I want to know more about skills. What is in there about skills? I search and I get all of these matches, like very many matches actually, and you get the word that you were looking for, you get it in the middle, so this under text, and then you have a prefix and a suffix. What else do I have? Yes, in your settings, you can set whether the TM should allow more than one translation for a certain segment. That can be quite interesting, especially when you're working with across different translators on the same project, and then suddenly you see, oh, this person translated the same sentence differently than me, and he probably worked on it before you. So, it makes sense that you use the same translation as the person did to keep consistency. As I mentioned before, you can export a TM in a translation memory that is the format that you can use and send it to your peers if you want to. Just be careful with it. And as I said, once you finish the whole thing, approved, you send it. All of these matches go into your master translation memory that you can build and you should build throughout your career to make it a huge resource of your knowledge and your previous work. There you go. This is all the information I have on translation memory. Make sure to check out MemoQ in the description. Thank you for sponsoring the video. And you make sure to like and subscribe. And I see you next Monday with the next one. Bye-bye.