| Renato |
I’m Renato Beninatto. |
| Michael |
And I’m Michael Stevens. |
| Renato |
Today we’re having a conversation about translation technology. |
| Michael |
And does it matter? Does technology matter in translation? |
| Renato |
Absolutely, and we need to separate in this conversation the concept of translation management and translation memory, the concept of managing the language part of the process and managing the translation process itself. So there is an ecosystem around the language services business, and there are several players in that, and it’s very common, and we hear this from our listeners, they’re confused, and what creates this confusion? |
| Michael |
The confusion between the different types of systems? |
| Renato |
Yes. |
| Michael |
Well, I think one, there is a plethora of options out there. There’s so many companies working in this space right now; there tends to be an ebb and flow between consolidation and innovation in this space, and so we have many different new names being introduced all the time, and so it’s hard to keep up with what exactly is going on. |
| Renato |
Yes this is why I like to separate the two conversations, right, so at one level you have the business people, the project managers on the LSP side, the buyers, the localization managers and the people on the buyer side. These people deal with technologies that help manage the process. |
| Michael |
And by manage the process, you mean they’re primarily interested in moving files generally? |
| Renato |
Exactly, moving files and allocating files in the right boxes, in the right places and so… and you have… well technologies like the kind of standard is the SDL suite, you have the world server which is that a lot of companies in Silicon Valley use Idiom WorldServer that was acquired by SDL, and SDL calls it SDL WorldServer now. They have their TMS that Adolfo – |
| Michael |
That’s kind of the legacy product that is looking to come back. |
| Renato |
In our interview with the CEO of SDL he mentioned that they’re focusing some of the efforts in developing that. You have companies like Wordbee. You have systems like Across that is a closed system that includes the translation management system and the translation memory system inside a certain work… these are work flow management tools. |
| Michael |
And GlobaLink would be one of the next things. |
| Renato |
GlobaLink from translations.com, TransPerfect, so there’s a few players in this space, and these tools live on the client side. They are hosted by the client who’s managing their internal processes, so it’s about how these tools connect with their publishing systems, and the requirement for the LSP is to connect into these systems, talk to them using their client versions or working directly on an instance in the customer server. |
| Michael |
Yes, and in general these are on the client side for a few different reasons, one of them is the security aspect, that it’s actually tying to a string repository or somewhere where files are sitting, that they don’t want to just be giving outsiders access to the other places, the investment that they make, to make those connections. |
| Renato |
And that’s the interesting part because a lot of this technology was designed back in the day of the client/server, and the security where the cloud people were afraid from the cloud, and you have now these new technologies like XTM and Memsource, that are cloud native - |
| Michael |
Yes. |
| Renato |
WordBee also that I mentioned before. These are technologies that are more designed for the reality of today where the cloud is not anymore something that is scary and unsafe; it’s actually easier to hack a server hosted by…than something in the cloud, so the attitudes are changing around the use of the cloud and the big players. I mean today if you look at the… at Google, Amazon and Microsoft, a lot of the revenue comes from cloud-enabled solutions; cloud first, this is how companies are working, but we are still, in the translation world, we are still tied to this old paradigm of hosting security confidentiality, hosting things at home, and then you go to the legal department of these companies, and they have all their contracts and all their confidential stuff stored in the cloud, but translation people don’t want to talk about it, but that’s another story. |
| Michael |
But we’re talking about the technologies that are focused on the file, that these are management systems, really they are… who do they support the most? Who gets the most benefit of them? |
| Renato |
Well they support the client. |
| Michael |
The client. |
| Renato |
But there are also the management systems for the LSP, so things like Plunet, XTRF, Prototxt 3000, these LSP management tools where they set up projects; you have, a project that is 65 languages, you set one language up and then all the others are just automatically copied, and the work flows are built around that. |
| Michael |
So that’s helpful in, like, the task management in making sure people are getting things out of going to an FTP site and pulling a file down? |
| Renato |
And there is the vendor management element—the translators are identified, their productivity is tracked and that invoices can be generated, the payments can be made, etcetera, etcetera, so this element of the translation management system for the LSP has to do with getting the projects distributed and done, and the best ones are connected to the client TMS, so you get the translation, you leverage your translation memories and everything, and then you put it directly into the client’s translation management system. |
| Michael |
So the big benefit is these files get to the people who need to work on |
| Renato |
Exactly, but here’s the thing—when we talk about translation technology, we tend to mix all these things into the same basket. |
| Michael |
Yes, what’s the other part of the basket? |
| Renato |
Well, the other part is the translation memory systems which are extremely important for the individual translators. Now here we’re talking about memoQ, Trados, Wordfast—those are license-based tools-Across is part of that, but then you will have MateCat, smartCAT, Google Translator Toolkit—these are free tools, and now the old IBM, TM2 also was made open source, so these are free tools that perform exactly like the paid tools that are part of this ecosystem of translation memories. What they do is essentially leverage existing translations and theoretically what I believe is the function of a good translation memory system is to increase the productivity of the freelance translator, or the in-house translator; the individual who is closest to the |
| Michael |
So compared to the translation management systems, instead of being concerned with the file, they are more concerned with the word? |
| Renato |
Yes, I think that the unit is different; it’s a segment, to be more accurate, it’s not the word itself, but the different segments. But you see the problem with this is that… and we might need to have a little shift in the way we think about technology in the language business, because we’re still trying to define, which is really, really silly, after over 30 years I still hear people talking and asking about who owns a translation memory? Who manages and what… so the way things are done today is that, essentially, let’s screw the translator, and the LSP gets leverage and the final client gets the benefit of that leverage. Sometimes this final client is acting as an LSP, giving out existing translation memories and trying to reap the benefits of having the supply chain leverage translations that were done before. And what I find sometimes is that nobody really stops to think about it, and sometimes this whole process ends up costing more than if people really looked at the benefits that technology can bring, with a more holistic view of the process, instead of thinking about cents per word, or one hundred percent matches and 90% matches – |
| Michael |
Percentage of a cent. |
| Renato |
- yes, percentage of a cent. We should look at the whole picture and the productivity that can be leveraged from that, and ultimately allow the person that is closest to the document, which is the freelance translator, to work with whatever tool they want to use. Because, you know, do you know what is the best translation memory tool in the market? |
| Michael |
Are we endorsing products now, is that what you’re going to do? |
| Renato |
I’m going to endorse all of them, because the best translation memory system in the world is the one that you know how to use. |
| Michael |
Yes, yes that’s where you gain the most efficiency; that’s where the least amount of errors are made. I mean you look at that, is the particular type of email I use, the most efficient email? For me it is, because I’m used to using it; I know quick keys, I know how to process through it quickly, it’s the same for translators. |
| Renato |
Exactly. And then you find the situations where vendor managers, and I find it funny, in 2010 I made a presentation at a conference, and I said that in the next five years, translation memory would become totally free or irrelevant. Somebody called me out on that a few weeks ago, so we’re in 2017, you said – |
| Michael |
Where is this free stuff you were talking about? |
| Renato |
- there is, I mean MateCat, smartCAT and GTT, or Google Translator Toolkit, are very good examples of free technology. You can go to a browser, log in and start translating now, leveraging translation memories with machine translation, so it is free. What hasn’t made it irrelevant is this structure, this construct, that everybody’s tied to that there is a right way to do it, is that trying to essentially avoid paying for somebody’s work. And then you get these ridiculous situations where you have, and I’ve interviewed, I’ve done project management workshops and vendor management workshops, and you ask a vendor manager the job of an LSP is to find the best, most qualified translator to do a job. |
| Michael |
Right, which is good in theory; it sounds good. |
| Renato |
Exactly. But then you find this perfect guy with a PhD in the domain that you want, you need a translator, he’s the perfect person to do the job, but he only has Trados 3.9.4, and the requirement is Trados 7.3.2. |
| Michael |
He has the wrong version, and therefore he’s stuck. |
| Renato |
He has the wrong version; he cannot do the job; I’ll go to the next one, and then they end up finding somebody that is not necessarily perfectly fit for the linguistic part of the job, but they happen to have the right tool. |
| Michael |
So it becomes a… for the LSP involved, an act of efficiency that they already have groups of translators who know particular technologies, and that’s a limiting factor. You may not be getting the best translator, but it is the easiest person to pick because they meet the requirements. I get the question of… like clients say “how knowledgeable with technology does my LSP need to be?” “I want great engineering, I want great development”, how important is that? |
| Renato |
Okay; so from my perspective what really matters to the end client is that they have good content in a timely manner, in the languages that they need, right? So it’s like the sausage; they need the sausage. They don’t need to know how the sausage is made. And within a certain budget; I agree budget is important. But the role a good LSP would be an LSP that is technology agnostic. |
| Michael |
Oh, tell me more about that term because I hate that term, but tell me more about it. I’ve worked for two companies; the current company I work for uses that same phrase, and I tend to get squeamish. Some of it is because of my theological background, but agnostic is… I don’t believe in it. |
| Renato |
Exactly. |
| Michael |
So I don’t believe…these companies don’t believe in technology? |
| Renato |
So should we say technology atheist? |
| Michael |
Perhaps that’s a little weird. |
| Renato |
I don’t know. Agnostic is that they don’t know. |
| Michael |
They don’t know. |
| Renato |
And the atheist is that they don’t believe. |
| Michael |
So wow, that’s beautiful that the theologian is getting corrected but, yes, you’re right! |
| Renato |
No, no but let’s address that point; it’s a very good point. |
| Michael |
No, yes this point though is that there is a sense of they should be passionate about technologies and what is the most appropriate fit or the thing they’re most passionate about, rather than… |
| Renato |
Let’s talk about the world that we live in. People say that we live in an API economy, and APIs are software that connects one type of software to another one, right? And it lets different pieces of technology to talk to each other, it’s like application – |
| Michael |
Program interface? |
| Renato |
- programming interface; that’s what it means. So theoretically an API is a translator of different types of content, and you have an API connecting a translation memory system to a content management system, to a translation management system, to a database, to whatever you want to… I don’t know, an exchange converter, a currency converter, whatever that is, the API allows two different systems to talk to each other. So the value that I believe modern LSPs bring is the ability to understand and to allow systems to talk to each other. So the easy way to address a business problem is to just do it the way we have always done it, and not think about it, right? So you have inefficiencies in that process. If you have to manage 22 different translation memory systems, you’re essentially managing inefficiency. But if you have smart people inside your organization, smart engineers that know how to deal with APIs, it will allow, theoretically, a translator to work with any system that they want, that has an API, and most of them do, the LSP will work in one system of their preference, that is going to connect to this different… so take BlueNet or XTRF, they have connectors to the different translation memories, so those translation memories will feed into the TMS and they will be ready for delivery. So the next step is to make sure that this BlueNet, TMS or XTM or Memsource or whatever, talks to the final client’s need, and this is where being technology agnostic is smart, because you cannot change… imagine a large company, I don’t know, let’s pick one, see it here we have McDonalds french fries in front of us, so let’s pick up McDonalds. |
| Michael |
Yes. |
| Renato |
So you cannot get McDonalds to change their translation memory system overnight. Why? Because they invested… they are a huge company, lots of people were involved in approving changes and servers and things like that, it’s not that easy, but it’s easy for an LSP to adapt to the technology of McDonalds, Bank of America, Ikea, whatever company you want to connect to, so I think that today, the smart positioning for an LSP is to say “I will deliver translation in any way that my client wants it to be delivered”, and the smart thing for an LSP to negotiate is not to accept the way the client dictates how translation needs to be done. I want translations to be delivered in Trados 14.72; it shouldn’t matter. |
| Michael |
Because then the client is taking on the responsibility of those conversions in the integrity of the file. |
| Renato |
A smart way for us to work, and this is the other word that we haven’t used much, is the concept of interoperability. So we have a good degree of interoperability today. You have a standard, you have XLIF, you have… what’s the TMS standard there? |
| Michael |
TMX? |
| Renato |
TMX, thank you. |
| Michael |
We get there eventually. |
| Renato |
We’ll get there, getting old. So in one way or another you can easily get a cube ray file to talk to a Trados file to talk to a MateCat file and so on, so theoretically this is where the intermediaries should be investing time, it’s getting the content and harmonizing it in order to deliver it to the final client, who should get it in a standard format—not a specific format. |
| Michael |
So the client can ask, instead of saying “I need you to use this particular tool” or whatever, they can provide the actual file to a LSP and say “this is what we’re going to be giving you on a regular basis, sure we’ll automate the delivery through a management system and all of that, but I need this back and I need it to be useable”. |
| Renato |
Exactly. |
| Michael |
And that’s a pilot, that’s something that can be verified pretty easily. |
| Renato |
Is there going to be loss? There probably… every conversion has a loss. Is that going to be huge, is that going to be relevant? You know, I believe that the benefits of the productivity outweigh completely the cost of this loss that you have in conversions. I’m speaking in theory – |
| Michael |
That could be tested. |
| Renato |
It can be tested, it’s something that… what we need in this process is that we have different levels… we have three stakeholders involved in this process, and the maturity level of the three of them, is really uneven, so sometimes the LSP is more mature, sometimes the buyer is the more mature, and very, very often the translator is very mature. I talk to translators all over the world who use all these tools, and they can tell you “well I prefer to work on this one”, like I said in the beginning, “the best one is the one that I know”, because they are trained in it. So a person that is working on memoQ every day, and is forced to switch and work on Trados, they are going to have a loss in productivity, so why are we penalizing the expert, why are we penalizing the person who is doing the job on a day to day basis, and we have them lose productivity if we can compensate for that in other stages in the process? That’s what I think; that’s my gut feeling about it. |
| Michael |
Yes, so you’re certainly siding, you’re putting the emphasis on the language service providers to honor the integrity of the translator’s work. |
| Renato |
And on the buyers who listen to our podcast, to stop thinking so much about…and I hate this expression, their linguistic assets. These are not assets; these are just files. |
| Michael |
Because you can’t sell them. |
| Renato |
Yes, well for me the concept is that an asset is something that has value that you attach to it, and for me, translation memory is like a coupon, right, you get a coupon for the pharmacy, you get “if you buy more shampoo you get a $5 discount” or something like that, and this is what translation memories afford a buyer, it’s like a discount, it’s like a coupon, it’s not a real asset that you can depreciate or you can re-evaluate and put in your balance sheet; it doesn’t appear in the financial statements of the company, so it’s not a real asset in the accounting concept sense of that. So one last thing though. |
| Michael |
Yes? |
| Renato |
If you are in a position where you have an opportunity to review your cycles and review your processes, this is a great time to do that, taking into account, I think that the concept that I want to use, the words that I want to use here, is future-proofing your investment, right? So disregard all those words of wisdom that you hear about security confidentiality; that is a given. That is something that is part of every system. No system is designed to be broken into and so on; all the vendors in the market have security confidentiality; your security and your confidentiality in their mind. So think about cloud-based systems, think about API, think about integrations, think about working in an ecosystem where the elements might be changed in the future, as long as you have good connectors and good systems in place to make that happen. |
| Michael |
And what are the steps a client can take practically, like in if they’re looking to invest or they’re looking to make a decision here, if they want to take products for a test run, is that available, like what should they be doing there? |
| Renato |
Everybody’s eager in the market to sell; everybody’s eager to talk. The power definitely is on the buyer’s side, so the buyer can essentially set up the rules, create comparisons and definitely talk to current users of the technology that is being offered, because… and it’s not asking for references, because the references are always going to be excellent. It’s essentially go to the market and try to find out who is using the technology that you are considering, and what they like and what |
| Michael |
Yes, and it sounds like it’s a good time to be a buyer in this space, because you – |
| Renato |
Absolutely. |
| Michael |
- there’s a plethora of technologies, I mean we covered I think the alphabet of companies, the list we have of just companies, and we haven’t named them all in this podcast, and that there is a chance that you can actually find the best fit for what you’re current needs are, and technologies that can grow along with you as your program matures or scales or grows. |
| Renato |
Very good. I think that we covered a lot of ground on this conversation, and if you have any questions or ideas, you know where to find us. Go to our website www.globallyspeakingradio.com. |
| Michael |
Or use another technology, like Twitter, and let us know what you’re thinking, |
| Renato |
LinkedIn, Twitter, Facebook, whatever. We will listen. |
| Michael |
We will. |
