fbpx

Podcast: Where are you on the machine translation maturity model?

Today’s discussion

In this week’s episode of Globally Speaking, our hosts discuss the Machine Translation Maturity Model (MTMM) with Jordi Mon Companys, Global Product Marketing Director, and Valeria Cannavina, Senior Project Manager at Donnelley Language Solutions

Guest: Valeria Cannavina

Valeria is a Senior Project Manager at Donnelley Language Solutions. She has extensive experience with process improvement and intra-company workflow innovations.

Guest: Jordi Mon Companys

Jordi is the Global Product Marketing Manager at Donnelley Language Solutions. He regularly works with designers and coders with a clear focus on product strategy, launch and growth.

Where to listen

The full podcast can be downloaded and listened to on all major podcast platforms, or from the Globally Speaking website. If you prefer to read the full transcript, we have provided that below, as well!

About Globally Speaking

Globally Speaking Radio isn’t just about how we as language professionals can improve our skills. It’s also about building awareness of how important translation and localization services are in helping global brands succeed in foreign markets—no matter where their business takes them. Globally Speaking is an independent podcast produced by Burns360 that does not necessarily represent the views of Nimdzi Insights or any other sponsors.

Host: Renato Beninatto

Renato is the co-founder and CEO of Nimdzi Insights, one of the language industry’s leading analyst and consulting firms. He has over 28 years of executive-level experience in the localization industry. He has served on executive teams for some of the industry’s most prominent companies, and he co-founded the industry’s first research analyst firm. A dynamic speaker and communicator, Renato is a highly regarded thought leader in the language industry, and is known for creating innovative strategies that drive growth on a global scale. He has also served on the advisory board for Translators Without Borders.

Host: Michael Stevens

Michael has 10 years of experience in the localization and IT industries. He is the Growth Director for Moravia, where his primary role is to assist companies who are inspired to create global software that changes the world. A well-networked entrepreneur, Michael’s main interest is in connecting and bringing people together. He not only enjoys learning about a company’s exciting ideas and developments, he also has a keen ability to add value—and fire—to new and innovative thinking.

This is an episode of the Globally Speaking Radio Podcast. Globally Speaking Radio is sponsored by RWS Moravia and Nimdzi.

The Transcript

M I’m Michael Stevens.
R I am Renato Beninatto.
M And today on Globally Speaking, we are looking at the Machine Translation Maturity Model—the MTMM as it’s being referred to.
R I don’t know about you, Michael, but when I think about maturity, I don’t think I have enough maturity to talk about it.
M Much less develop a model for such a thing or be a model of maturity, or whatever the case may be.
M Today we get to have a conversation about a new maturity model that has been developed about our industry, about part of our industry.
R And maturity models are great as an assessment tool. It’s for you to check out your status and how well you are in a certain…it creates a framework for analysis. Let’s listen to our guests.
J Hi, my name is Jordi Mon [Companys]. I’m the Global Product Marketing Manager at Donnelley Language Solutions, and I’m in love with language technology products: AI, machine learning, and natural language processing.
V I’m Valeria Cannavina, and I work as a Project Manager for Donnelley Language Solutions. I’m obsessed about process improvement and how to introduce new workflows to the companies I work for. And this is the main reason why I’m here today with Globally Speaking Radio to present my new white paper.
R So, Valeria, we’ve known each other for several years. And the first time I met you, you were working on a thesis around the Localization Maturity Model that we had published at Common Sense Advisory. So, what caught your attention to these maturity models? Because you spent quite a good time building this Machine Translation Maturity Model that we’re going to talk about today.
M Yeah. And Valeria, maybe fill out a little bit what that original Localization Maturity Model was about and why it struck you.
R So what attracted you to this concept of the maturity model, and what drove you to develop the Machine Translation Maturity Model?
V As part of my personality, I’ve always been very curious about implementing new processes and bringing new ideas to the business. And I’ve got a genuine interest in machine translation and all the technology related to that. So, I put this thing together for my crew inside, and then the relationship that I have with my colleagues here between technology, vendor management, and linguists, and so on. And I found out that there was a gap between the knowledge of machine translation of many clients and what we can offer.
  So, I saw that there was the need of filling this gap by coming up with a maturity model that would serve to guide our clients towards a path to maturity. And I saw that the point, the best idea, even my knowledge of the Localization Maturity Model was to start with exactly that one. So basically, taking the five levels of maturity, and based on that, develop a Machine Translation Maturity Model.
  So the maturity models are almost the same, but of course the key process areas are different. And they give you an overview of what I consider as being the key elements of a good machine translation.
M We’d love to hear in detail about the depth of the Machine Translation Maturity Model, but before we go there, why was the Localization Maturity Model important for you? What did that teach you, fundamentally?
V The Localization Maturity Model is important, first of all, because many clients know it, and it describes very well the path to maturity and the way to organize. Every level is divided into organization, process, and governance. And every process it describes there gives you the exact task that needs to be accomplished before you go to the next level.
  And it’s not just the maturity path. The Localization Maturity Model suggests a path to continuous improvement, which is what every company should enact. And it’s basically the limitation of the ISO standards that we have at the moment.
M Yeah, when we think about the localization process, oftentimes there’s so many factors it can become overwhelming. And when you look at the Localization Maturity Model, it’s simplified—almost down to a checklist—where you can kind of rate where you fit on a scale with certain things, and come up with a number and go, “Oh, wow, we have a good benchmark now.”
R Michael, maturity is a very interesting concept because it has nothing to do with the size of the company, with the type of company that it is…maturity has to do with the type of engagement that the organizations have with a certain type of technology. So, the mother of all maturity models is the Capability Maturity Model that was developed for software. And what it gives you is a map and a self-evaluation tool. You can go there, and you see, “Well, I am …” It’s like your Cosmo test in the magazine: “Does he love me, or does he not?” Or something like that.
M “Is the relationship hot or not?”
R Yes. So, the value here is it’s a tool. But I think that there are some important concepts that we need to take into consideration. And Valeria, I’d love to hear your take on this.
V As you just pointed out, Renato, I think that one of the points of strength, not only of this white paper, but what you were suggesting also about the Localization Maturity Model, is the flexibility, which is important. This white paper is not just for big organizations, small organizations. When I wrote this, I was thinking about all our clients: small clients, huge clients, multinational companies—whatever company can apply this.
  Because there is no limitation, because it’s flexible. Because you can choose what you want to improve: a task that you will see probably at level number two, while you’re a company that is working on some process that belongs ideally to level number three. So, this is one of the key elements of the Machine Translation Maturity Model—its flexibility and the fact that it’s for everyone.
  So even though the least of the tasks is not in depth, I think the Localization Maturity Model, by the way, the intention, the reasoning behind this is to make it as flexible as possible and to allow us to customize our services based on the client’s request around the machine translation models that we are presenting.
M In maturity models, the fifth level I think most people equate with large companies. Have you found that to be the case?
V Yeah, that is the case. What we define as mature companies, ideally at the fifth level, these are companies who have their own department of machine translation, meaning they have a team of engineers, of content reviewers; they have the technology to connect to our TMS or…a company really advanced, I think should be at level five. And I think that is the case.
M I mean, I think I would disagree with that a bit. Because there are companies out there who are less mature but have a global vision and have invested strategically in this area, though they may not be as big as a Microsoft, or an Apple, or a Google, or someone like that. They have set up a mature program with all of those capabilities in-house.
  So, I don’t think when people look at these models, they think, “Well, just because a company is very large, that equals very mature.” Sometimes they’re very mature small companies, and sometimes they’re very immature large companies. That scale of one to five doesn’t necessarily equal size of company.
J I do agree with you, Michael, I think that it’s irrespective of the size of the company. It’s a matter of culture, of company culture and how they embrace technology, and especially how they approach it, how they invest resources in it. And I do also agree with Valeria that usually big companies have enough resources to dedicate to this, but the spark of this investment comes from an open mind and from a research perspective to actually explore and discover this particular area—in this case machine translation—and how to step-by-step grow. And yeah, you can find a level-five- / level-four-maturity company that is definitely not huge, so I do agree with you, Michael.
R And at the same time you can find a huge company that is very immature. They’re not ready to look at this. And I’ll just play here with traditional examples, let’s say a company like Walmart, that is not necessarily… It’s a huge company; they have a lot of money; it’s global, but it’s not in their DNA. They’re in retail; they’re not really investing in this kind of stuff. And you will have another company like Amazon, which is a huge conglomerate, where you have pockets inside the company that are very advanced.
M Amazon Web Services for instance; they’ve been very global and mature from the beginning.
R And you have other parts of the company that have no idea of how global this could be. So, the beauty of a maturity model, and this applies also to several of our listeners that are LSPs, because as service providers, they need to be aware of what is going on, and a Machine Translation Maturity Model is a self-evaluation tool that they can use to see where they are and where they want to be.
  And Valeria, do you have to grow sequentially? Do you have to go from level one, to level two, to level three, to level four, to level five? Or can you skip around?
V The Machine Translation Maturity Model is based on flexibility and on the non-sequentiality of the taskS. So, there might be some tasks at level number two or level number three that can be skipped and go to the next level, or there are some of the tasks that can be implemented regardless of the level a company is at.
  For example, in the white paper, we talk about, I think it’s at level four, having a machine translation department. Again, the idea of having a machine translation department is a fluid idea, which means that a customer might want to outsource only the terminology part, while keeping internally other stuff that he thinks he can contain; while, for example, asking his LSP to develop a controlled language handoff for a language combination, which would be something I would imagine at a later stage of the maturity path, or something like that. So, yes, I think flexibility is, again, one of the key aspects here.
M The five levels as outlined by the maturity model are:
1. Initial
2. Repeatable
3. Defined
4. Managed, and
5. Optimized.
  Valeria’s going to share with us details related to each one of these five levels.
V Okay, the five stages of the Machine Translation Maturity Model resemble the five levels of maturity that you can find also in the Localization Maturity Model.
  Basically, it has a first level, which is the Initial one, or can be called also the reactive one. At this stage, I see a company that requires machine translation only when absolutely necessary. Those are the companies that have not a lot of awareness of machine translation, and they don’t even have a budget for that.
  The second stage is the Repeatable one. We talk about companies that have very little knowledge of machine translation, but they’re, let’s say, making their mind up to it. So, probably they’re starting to document some of their tasks, and they can repeat some of their tasks.
  At the third level, the Defined level, we find companies that have a clear idea about machine translation, and they’re already looking into how to integrate machine translation into their business. So, companies at this level have a different awareness from the companies at levels one and two about machine translation. They see it as a strategic part of their business to go into globalization.
  The companies that we find at level number four, which is the Managed one, are companies that are starting to build their own machine translation department, and they’re taking on challenges to align machine translation to the corporate goal. Machine translation for them is part of the daily production processes. They work closely between engineers, content writers, and people who will review the translated content, and so on.
  While the companies that are at the fifth level, which is the Optimized level, means that they have a full team of engineers, terminologists, internal reviewers, project managers, and they have rules in place to aim at continuous improvement. So machine translation at this stage is not only a daily part of the production, but they actually are taking on steps to automate as much as possible all the previous steps to machine translation.
M How do you expect or how do you hope to see localization buyers use this model?
V That’s a very interesting question by the way, because we are not officially, but we are already applying this model with some of our clients, and it’s a process of discovery with them. So, we will be measuring the results of the maturity model within the next year, and at that point we will be ready to present the results. Probably we’re thinking of expanding the Machine Translation Maturity Model with a business case or case study to prove the value of what we are saying here.
R Jordi, is there anything else you’d like to mention or say based on our conversation?
J I think this initiative is good because the machine translation industry is not new, but it’s not as collaborative as we all would love to. The same happens with the TMS industry, and I can relate to the TAPICC initiative and how we welcome it from many players in the industry. So, we have released this to spark the debate, to have more information shared, and not only with our clients, but as an industry. We hopefully will talk next year and see the results of what this conversation that we just started has brought back.

Stay up to date as Nimdzi publishes new insights.
We will keep you posted as each new report is published so that you are sure not to miss anything.

Related posts