Since its debut in 1997, the International Association for Machine Translation (IAMT) Award of Honour has acknowledged the work of linguistic luminaries. The IAMT Executive Committee selects an organization member every odd-numbered year who particularly embodies the spirit of innovation that fuels the professional organization. 

The year 2025 was no exception, as the committee selected Mikel L. Forcada as its honoree. We connected with Forcada to ask him about his thoughts and feelings regarding the recognition as well as his view of the modern machine translation (MT) landscape. 

Congratulations! How does it feel to be the recipient of the IAMT Award of Honour?

Thank you very much. I am deeply honored, humbled, and grateful to be considered for this prestigious award. To see my name alongside the remarkable individuals who have received it before is truly overwhelming. They are figures I have long admired and continue to inspire me. Moreover, I feel incredibly fortunate to receive an award for simply doing what I love: researching, developing, and teaching MT, and then promoting it through the European Association for Machine Translation (EAMT), which, as you surely know, is one of the three sister associations that make up the IAMT. This award, in truth, also belongs to the wonderful colleagues who walked this path with me. Their contributions have been invaluable.

Aside from this award, what’s your relationship to the IAMT, and how has it contributed to your work and career?

I have been a member of the IAMT through the EAMT since the early 2000s. This has kept me in touch with what was going on globally in the field of MT. I have served as secretary and president of the IAMT; I was president from 2017 to 2019, as being the president of the EAMT is a rotating position. The IAMT, as the umbrella organization, takes care of the Machine Translation Summit, which is held every other year and brings together a community of top researchers, developers, vendors, and users of MT and other translation technologies. Being part of the IAMT keeps me abreast of the field’s development and offers a forum to make my research known to that diverse audience. By the way, I am currently the treasurer of the EAMT, the European branch of the IAMT.

Could you give us a brief overview of your career? What were some of the jobs and roles that led you to this moment?

I graduated and got my PhD in chemistry, and after a postdoctoral Fulbright fellowship at IBM (San José, California), I came back to Spain, crossed over to computer engineering, and was hired as a lecturer at the Universitat d’Alacant in 1994. I taught formal models of computation, compiler construction, and computing for translation majors and started doing research with neural networks and translation. We experimented with very simple translation tasks and rudimentary encoder–decoder architectures as early as 1997 and got into rule-based MT systems, but we lacked the formidable computing power available now. By 2002, I was a full professor, heading a small group of researchers and a number of projects in MT. I’d like to highlight Apertium, the free/open-source machine rule-based MT platform that we launched in 2005 and has been the basis for the development of many MT systems, some still in use at public institutions like the Valencia and the Spanish governments.

What are the greatest lessons you learned along the way that contribute to your current work and this latest recognition?

My involvement in MT has expanded my horizons: neural networks, finite-state and other formal models of computation, languages, and translation. I feel really fortunate to have been given the chance to orbit around the intersection of all those fields, because I loved languages and computers since I was a child. As a result, I have learned how important — and sometimes how difficult — it is to build good interdisciplinary teams and to help run interdisciplinary conferences. This is particularly important as, when it comes to the contribution of technology to society, a good summary is nihil de nobis sine nobis (”nothing about us without us”). Everyone affected by technology has to be as empowered as possible so that advances contribute to their well-being and prosperity. My contribution to bringing together such a diverse community may be the one of reasons why my colleagues have decided to give me this award.

Could you tell us a little about your role as a founding partner and chief research officer at Prompsit Language Engineering? 

The development of the Apertium MT platform was performed by a group of programmers and translation experts hired by a series of publicly funded projects. Project money was going to run out someday. So I was persuaded to start a company so this compact group of very skilled people could make longer-term professional plans. Originally, the idea was to offer services around Apertium, but Prompsit currently does many other things; in particular, we have acquired world-class expertise in parallel and monolingual corpus curation through our participation in European projects where my former university also took part. Being chief research officer of Prompsit flows naturally from my involvement in its creation and the close collaboration between my former university and the company.

What is your perspective on the current state of MT and your view of its advantages and disadvantages?

There is no denying that neural MT, particularly after the advent of transformers and the availability of specialized computing hardware, has brought about a giant leap in the usefulness of MT. Just to name an example, MT between my mother language, Basque, and its neighboring languages (French and Spanish) was particularly challenging until the advent of neural MT; now, it is just one more language, so that usefulness of the MT systems one can train depends basically on the availability of training corpora, like for most languages.

There are, however, important issues. Not everyone has access to the specialized, powerful computing equipment needed to train and run the systems, leading to the concentration of machine translating power and the subsequent dynamics of dependency. Building the equipment generates toxic and inert waste, and running it consumes a lot of energy and generates greenhouse gases that harm our environment — much more CO₂ is generated per translated word than in the past. Machine-translated text is often deceivingly fluent, and, requiring less but more careful post-editing, it is hard to know why a specific translation was produced. The unjust biases present in training corpora are often amplified. For many useful systems, one does not really have access to the corpora that were used to train them or to the actual training parameters. I find these issues of opaqueness, environmental harm, and centralization deeply worrying.

With the discussion and debate over AI and large language models (LLMs) dominating the industry, where do you believe MT fits into the technological ecosystem? 

Current LLMs have their basis in the transformers that were invented to do MT. One could say that MT is now one among the many “text continuation” tasks that language models can perform, benefiting from what they have learned when trained on large, multilingual, and in principle nonparallel corpora. This, in principle, gives them additional power that comes with a price: We run more general, computationally hungrier, more expensive models trained to do many other tasks, where a carefully trained, more frugal, task-focused neural MT system would produce results that would be equally useful in many applications. At Prompsit, however, we sometimes find it hard to convince our customers of this.

As a retired professor, could you tell us about your experiences and insights gained in academia? What role do you think education plays in the language space today?

Teaching translation technology is one thing I truly miss. I did that for more than 25 years, and getting in the shoes of people without a technical background has challenged me to find ways to explain how MT works and how it will affect future translation professionals; as a result, it has given me a wider view of translation technologies that takes into account everyone at stake (translators, developers, users), as we do in our international conferences. MT users — professional translators on the one hand and regular people on the other hand — need to be educated about MT. MT literacy and, more generally, generative AI literacy is a must, particularly in view of the issues I mentioned earlier. Translation professionals and the general public need to know what to expect and how to make informed choices.

Aside from work, what else occupies your time? What are some of the other things you’re passionate about?

When I was a teenager, I was about to become a licensed amateur radio operator but lost interest; five years ago, I retook the hobby and took the test to get the license. One of my favorite amateur radio activities involves hiking to contact other amateurs from mountain tops. As a language lover, I am also attending the public language school to improve my Italian and my Portuguese. I am one of the members of Softcatalà, a 27-year-old nonprofit that contributes to the presence of the Catalan language in the digital world. I am also on the executive board of a small political party and on the team supporting an elected member of the city council of Alacant. And, of course, I am happy to have more time now to be more involved with the company I founded, Prompsit.

Anything else you want to add?

Yes, I’d like to compliment MultiLingual for its contribution to bringing together the interdisciplinary community around translation technologies that also makes up the IAMT.