After an overview of the project, of its team, rationale and expected results, delivered by the coordinator Stavros Assimakopoulos, Jelena Gledić presented the survey of curricula. The survey encompassed programmes of more than 120 BA and MA degrees in language and linguistics from all over Europe. Results showed a general underrepresentation of the research, data acquisition and data handling skills that other components of the needs analysis identified as crucial in today’s language professions.

Silvia Bernardini then went on with a presentation on the main insights emerging from industry surveys, institutional position papers and academic works on required skills and competences in the language sectors. The take-home message was that as Artificial Intelligence is driving more and more profound changes in the world of work, education should focus on skills and tasks that until a few years ago fell outside the realm of language and linguistics proper, or tended not to be explicitly taught. These include technology-related skills such as data management and digital marketing, but also higher-level competences like critical thinking, problem solving and creativity.

It was then Adriano Ferraresi’s turn, who introduced the second major component of the needs analysis, devoted to the market’s perspective. He presented an analysis of a corpus of nearly 200 job posts addressed to experts in language and linguistics, collected from specialised and non-specialised job platforms. The set of skills, competences and tasks identified in this phase pertain to 4 main areas: data and research, technologies, disciplinary knowledge and communication/organisational skills. The data gathered from the corpus made the basis for the subsequent two steps in the needs analysis.

In her second presentation of the day, Jelena Gledić reported on the results of an online questionnaire targeting 70 employers from digital and data-intensive sectors, who do not necessarily identify language and linguistics graduates as a talent pool of employees for their business. Among the skills that are deemed crucial and at the same time most in need of improvement according to the respondents, problem solving, technical skills and organizational skills feature prominently.

In the last presentation devoted to company insights from the needs analysis, Michela Vella reported on the interviews carried out with 11 companies from various sectors, from language service providers and language technology services to the automotive and insurance industry. Like the previous components of the needs analysis, results show that there is a growing need for graduates in linguistics and language-related disciplines, even if existing curricula need to better prepare them for the goal-oriented character of industrial work.

The UPSKILLS profile

In his speech, he referred to the experience of DGT to illustrate the impact that AI has had on a “traditional” language-related profession like that of the translator at EU institutions. In 2021, the demand for translations at DGT has exceeded 100 million pages, and the figure only refers to the first six months of the year. Mr. Gren pointed out that the only way in which the institution is able to process such huge amounts of text is by relying on the EU Machine Translation (MT) system. This would not be sufficient, however: the human-in-loop remains a central factor for translation workflows to be successful, and both younger and more senior EU translators have taken on the challenge posed by MT, especially by updating their competences in the technological field.

The revolution brought about by AI, he continued, has not only led to changes in existing professions, but it has also created new opportunities, in the form of new positions. Among the diverse profiles at DGT open to language and linguistics graduates, he mentioned that of computational linguist, workflow designer, language technology coordinator, terminology specialist and data curator.

The talk concluded with a message of encouragement: provided that future graduates are ready to be embrace the change and get on “the digital train”, as Mr. Gren called it, they can look at the future with confidence.

Nikola Ljubešić, from Jožef Stefan Institute (Ljubljana), illustrated tasks revolving around various AI-based systems, including systems for part-of-speech tagging, Machine Translation, and hate speech detection. For most modern implementations of these systems to work effectively, human input is required at several stages. Mr. Ljubešić mentioned three tasks in particular for which “properly trained linguists” – as the title of his presentation goes – are needed. They can be asked to annotate the input data which are fed into the systems, to perform analyses of the errors in the output, or to build benchmarks against which the performance of AI can be compared. What are then the skills of a linguist properly trained for these tasks? According to Mr. Ljubešić, a strong theoretical background in linguistics and great flexibility in adjusting to different theorical frameworks, the ability to strictly follow annotation guidelines, and a good understanding of how AI works.

Lorenzo Musetti, from expert.ai, also focused on the role of linguists in developing AI technologies, especially in the field of Natural Language Understanding. The company regularly hires linguists both with and without a computational background, employing them in three main teams. Within the Research and Development team, linguists occupy positions as knowledge engineers who are in charge of creating knowledge graphs, annotating data, doing (some) programming and publishing articles on the team’s work. In the Professional Services team, linguists are mainly involved in project management, and liaise with technical teams in order to deliver customised solutions for customers. Finally, employees in the Academy team are responsible for all training and dissemination activities, where excellent public speaking skills are a must. In summing up the competences that a good job candidate should possess, Mr. Musetti highlighted knowledge of (computational) linguistics, of at least one foreign language and, importantly, curiosity and creativity.

Alice Cipressi presented then the perspective of Acolad, a large company in the language services industry. Most of its staff is composed of linguists, often with a background in translation, and this also applies to teams that are not strictly involved with linguistic tasks, like desktop publishing or quality assurance. The role of project manager was mentioned as one of the most requested ones: in this position, linguists/translators coordinate all actors involved in the translation process, from receiving requests to delivering the final product. For these tasks, excellent communication and problem solving skills are essential, but technical skills related to Computer-Assisted Translation (CAT) technologies are also seen as a very welcome plus. On the more technical side, Ms. Cipressi mentioned the profession of Machine Translation specialist, a role that has only recently been introduced in the company. The MT specialist has an all-round view of whether and how MT should be integrated in translation workflows: for example, they carry out feasibility studies, and create and test customised MT systems.

The final talk of the session was by Ilire Hasani-Mavriqi, from Graz University of Technology and Know-Center, who presented the profession of data steward. This is a very recent job profile introduced in Austrian Universities, whose main role to is to support data-driven research by playing an advisory role throughout the data lifecycle. Data stewards thus manage data and data infrastructures from the initial phases of research projects to data preparation, data archiving and reuse. Even though the profile does not target linguists specifically – and in fact the data managed by data stewards may come from all fields of research – this is a job where the communication and technical skills acquired by linguists are greatly valued.

Recordings from the Old jobs with new twists session

Ivana Kruijff-Korbayová introduced the European Masters in Language & Communication Technologies (LCT). The Masters focuses on NLP at large, and welcomes students both with a computer science background, and from language and linguistics, including theoretical linguistics, philology, translation and literary studies. After a brief presentation of the contents of the Masters, Ms. Kruijff-Korbayová discussed some of the areas where students tend to experience most difficulties, especially at the beginning of their studies. In particular, she mentioned mathematics and statistics, Python programming, machine learning and data structures. But most of all, she encouraged students to embrace the kind of “practical approach”, as she called it, that will let them deal with complex issues. Instead of expecting to learn how to do things properly, and then apply what they have learnt, she encouraged students to always try, and not be scared of making mistakes.

Gaetana Ruggiero then presented the point of view of an LCT student with a purely linguistics background. Her message was that it would be important for linguistics degrees to introduce computational concepts from the BA level, as this would broaden the students’ horizons and open up new opportunities for them.

The Masters in Technology for Translation and Interpreting (TTI) was presented by Sara Može. Its main aim is that of educating future developers of translation and interpreting technologies, and, in view of its strong research focus, making a solid springboard for PhD studies. The ideal applicant for this Masters has a BA in translation/interpreting, computer science or computational linguistics, is fully proficient in English, and knows at least another language. As for the desirables, they should have some experience as professionals or in research, and demonstrate strong IT skills with CAT tools or programming.

The TTI student Marina Tonkopeeva confirmed that IT skills are crucial in the Masters, but that soft skills and some knowledge of entrepreneurship definitely help in advancing one’s career both as a student and as soon-to-be professional.

The meme used to introduce the networking sessions

Concluding remarks

Many points of view were expressed during the day, but everybody agreed on a central point. AI and language technologies in general are taking giant steps forward every day, but they should not be seen as a threat by graduates in language and linguistics. In fact, they are creating new opportunities, and the need for linguists in the language industry is likely to stay the same – or even increase – in the foreseeable future.

It will take some effort from all parties involved: more connections will have to be established between industry and universities, and both university degrees and their students will need to be ready to adjust and take on the challenges of this incredibly rapid technological evolution. But the direction is the right one, and we hope that UPSKILLS will be able to contribute to this process.

Posted by Adriano Ferraresi (UNIBO team)