In-house translation – outside the box

Tag: translation

  • Summary of Oral Remarks to Panel on “Is AI Use Changing the Role of Translators? Implications for Education, Culture and Ethics”

    Summary of Oral Remarks to Panel on “Is AI Use Changing the Role of Translators? Implications for Education, Culture and Ethics”

    On Friday, 8 May 2026, I was part of an online panel together with Marija Todorova (Assistant Professor, Education University of Hong Kong) and Joss Moorkens (Associate Professor, Dublin City University) and chaired by Alison Rodriguez (Past President of FIT) as part of the Online Symposium of the Study Group on Language and the United Nations. We chose to discuss our respective angles on “Is AI Use Changing the Role of Translators? Implications for Education, Culture and Ethics”.

    Both Marija and Joss addressed some of the impacts from teaching in higher education – with the latest cohort of students now not just digitally native, but seemingly AI native. In addition, they also mentioned aspects of their research which focuses on understanding some of the needs for translation in the marginalised society, in disaster alleviation in Malawi (e.g. in the aftermath of Cyclone Freddy in 2023) as well as in the case of Burmese migrants in Thailand (for whom machine translation plays an essential role in their daily life). In these settings, the role of translation is naturally far removed from my comfortable in-house setting in public administration in Vienna.

    The responses below are based very closely on the responses I gave as part of the panel to the questions Alison asked. Most of my recent posts, both on this blog as well as on LinkedIn focus on on the importance of the human component in human-machine translation being that of an Expert in the Lead (the topic I spoke about at FIT XXIII in Geneva) as well as some of the cognitive issues that face translators in the age of (N)MT/GenAI technologies. In addition, I very briefly touched on shadow translation as a governance issue, an area I am starting to put particular thought into. One particular issue is about the continuing accountability that professional translators have.

    AR: Michael, You have also spoken about ‘unanticipated’ cognitive effects of AI; can you tell us about how, as a specialist translator, you use AI and its effects or changes you see in your work?

    • As the only in-house translator at my employer, I have been able to use AI to help with the terminology extraction process – although this has been very reliant on using authoritative texts (e.g. EU Directives and Regulations) as a starting point, rather than giving the LLM free rein. In this regard it can prove useful, as I have been able to find ways to do proactive terminology work, rather than only reactive work.
    • Continuing LLM hallucination, and a tendency to back translate rather than directly cite authoritative texts has however limited its usefulness. Even if I use AI, I still remain accountable, so I use it only where I know that I can rely on its output .
    • Regarding how it has changed my work – previously I got a lot more “ad hoc” gist translation requests from colleagues, whereas for the nature of such requests, they have opted to “ask the LLM” rather than “trouble the linguist”. While this means less time spent handling very short but urgent pieces, and more of a focus on more substantial texts, those short texts were also very useful in terms of knowledge building. Additionally, by taking an AI-led approach there is the potential governance issue from an increase in shadow translation activity.
    • Regarding the cognitive impacts, post-editing is not uniformly easier than translating from scratch — it depends heavily on text type, language pair, and AI quality. Usually it proves quite demanding to establish exactly what tools colleagues have used, and to establish whether they have used a detailed (persona) prompt to try to steer the LLM. Not knowing how a particular translation has been arrived at, and the original source of terminology used can really impact the cognitive load when post-editing.
    • AI use redistributes cognitive effort but does not reduce it. There is a trend towards translations receiving texts that colleagues have “translated” using AI. The deceptive fluency of the output leads them to think that a text is close to final. Colleagues see it as requiring a minor edit, but I often have to check the fine nuances with them to identify the shortcomings of the LLM translation. 
    • Decision fatigue exists for the human translator: correcting mistakes they would not have made wears down translators. This is particularly the case for LLMs making terminology decisions based purely on statistical probability. In larger texts, I have noticed a deterioration of terminology consistency (likely due to limited context windows).
    • Individual differences matter greatly, experienced translators manage AI-assisted workflows differently than novices. High-quality AI output can genuinely reduce load, but poor-quality output can increase it significantly compared to translating unaided – this boils down also to the deceptive fluency of AI output – and the  fact that with many prompts by colleagues using a chatbot to serve up a translation, even if they try to use RAG (retrieval augmented generation), the LLM may choose to back translate rather than cite directly from the information it has been fed.

    AR: AI was supposed to give us  more time and mental space for creativity, but it has largely given us more to manage. The gains are real but narrowly distributed — often flowing to employers, platforms, and clients in the form of lower costs and faster delivery, while the human doing the work absorbs the new complexity, the new anxieties, and the new forms of invisible labour that AI oversight requires. Michael, can you share your professional experience and view on the possible overreliance on technology and its effects?

    • All the tools involved lead to the underlying skill of translators being forgotten. The impact of tools trivialises these skills. Effective tool use is only part of the skillset of a professional translator. In AI literacy and competency terms, translators are “thrown in the deep end”. They receive only very general training, rather than dedicated training about how they really can benefit from AI.
    • In terms of what clients expect/see/need, the old price/quality/speed triple constraint where “quality was king” no longer exists. Customers now expect near instantaneous delivery that does not cost a commensurate amount. What they need from the human involved is accountability, credibility and expertise – which “good enough” quality may not deliver.
    • In terms of what responsible professional practice looks like, translators have to work towards ensuring an outcome (so in my case, ensuring the timely successful conclusion of a supervisory procedure and to ensure that there is no reputational loss).
    • Institutional settings frequently treat translation as an afterthought (i.e. only sending finalised documentsfor translation), rather than central to achieving the outcome (i.e. translation work starting before finalisation of documents).
    • Regarding practices that still build the right meta-cognitive skills needed vs metacognitive demands (i.e. is this text really correct/source trustworthy etc), in the case of junior translators, one way is a stepped approach to including tools in their workflows, so they really consider the text and learn to understand the subject matter, rather than just receiving MT/LLM output to correct, until they have built up subject matter expertise.
    • Translators have an ethical responsibility: translators still maintain a responsibility to question the source text, check facts, figures in a critical manner that AI does not, and to check with the author. AI also leads to disrupted flow states, divided attention, decision fatigue, that make work feel more effort without the creative reward.
    • Many translators are very familiar with their CAT environment, while AI tools draw them out of their state of flow, although the latest iterations of some tools are starting to draw AI tools into CAT flows to address this. Translators ultimately remain more committed to a translation where they actively tackle it, rather than just being a downstream verification instance.

  • Comparing Translation to Tetris

    Comparing Translation to Tetris

    One of the most memorable questions I have been asked about being a translator was “Which computer game does translation resemble?” I strongly suspect the person asking the question hoped I would be stumped by such a fiendish question. I welcome leftfield questions when speaking, as I believe the point of presenting or evangelising about your profession is to be able to respond to any question your audience chooses to ask.

    Fortunately, I had an answer up my sleeve, and could respond without delay. I even had a slide in my deck that supported my answer. The answer was Tetris. However, my response required a decent explanation. Where are the similarities between translating and Tetris? Both in terms of translation as well as in my work as an in-house translator.

    As a child growing up in the late 1980s and a teenager in the early 1990s, I cut my teeth playing Tetris on the demonstration Nintendo Gameboy in Dixons in Taunton. A trip into town wasn’t complete without a game, an almost elicit treat as I have never had a handheld console of my own. Much of the finesse went past me back then, since I only had very few opportunities to play the game.

    Familiar to gamers and non-gamers

    A further point about a similarity between translation and Tetris is that both enjoy enough familiarity across a wide age range of the population. Those of us who got into Tetris in the late 80s are now middle-aged, and probably anyone of working age will have a rough idea of what Tetris is – in a similar way that they will know what a translator does (even if they might not understand the difference between the respective jobs of an interpreter and a translator).

    Tetris is fairly unrivalled in turns of familiarity of it among gamers and non-gamers. However, in case you have somehow managed to avoid its significance, there is a wonderful programme from the BBC from around 2004 called Tetris: From Russia with Love.

    Tetris: From Russia with Love – originally shown on BBC.

    And now, in 2023, Apple TV is telling the story its way – the trailer of the forthcoming film dropped last month, which reminded me about this post, which has been a long time in the making, the first draft having been created during a pandemic lockdown.

    And now on the bigger screen?

    So is the comparison between translation and Tetris a valid one?

    One point where the comparison stands up well is that fact that Tetris has been so widely ported – according to its Wikipedia entry, to 65 different formats. This in itself is an impressive exercise in translation. In terms of the genre of the game, it is a puzzle and tile-matching game, with a strategy element to it. There is a similar strategy to translation projects. For users of CAT tools, each segment you complete is like completing a line in Tetris. With every completed segment, new uncompleted ones pop up ahead of you. And of course, in the case of a 100% match, segments further on are also propagated accordingly, in the way that non-consecutive lines disappear in Tetris.

    Focusing on the tile-matching aspect, part of translation and the rendering of every individual segment/translation unit is ensuring the individual pieces fit into the correct position. Where translation of course varies is that the “pieces” are not restricted to 7 different tetronimoes, and the pieces can be far more complex. Similarly, target text rendering does not necessarily observe all the key features of the source. A classic example from my experience of German > English translation is the way that a passive construction common in the source language may seem out of place in the target text. Similarly, tenses don’t map 1:1 – the use of continuous tenses in English being a classic example when translating from German.

    For those who use CAT tools, there is another similarity. In Tetris, you see all the pieces that have been played before the current one (which provides you with the context for the current piece and how to play it, i.e. whether to let it drop, or whether you need to move it around or rotate it. Machine translation has a tendency to stick quite rigidly to a word order close to that of the source text, whereas in translation inversions can often improve the flow of the document. And of course, you usually have a sneak peek at the next piece that is due to drop.

    Translation Tetris – it’s about time/timing.

    My work as a Single Person Language Services Unit (SPLSU) is definitely another type of Tetris. Other analogies could be to compare it to either juggling or plate spinning – in terms of keeping all the balls in the air, or stopping the plates from crashing to the ground. In terms of the working day/week, my work is very much a Tetris puzzle. The various activities are certainly very similar to the individual pieces in a game of Tetris, as is time management. Dovetailing various routine tasks, e.g. alignment, translation, revision, terminology work, training as well as on the fly developments (e.g. ad hoc rush jobs) is a prime example of tile-matching. The way the tiles rapidly fall means that the strategy changes frequently and rapidly. Not everything teselates perfectly.

    In the earlier and easier levels of Tetris, you can power through them with a flurry of Tetrises (clearing four lines at a time), although as the pieces drop in quicker, that approach has to be quickly abandoned and individual lines cleared to ensure that there is still space on the screen (or calendar!). Naturally there is also the issue of the way the game can speed up, necessitating quick action to avoid being swamped.

    In terms of time management, it is also the case that not all blocks are equal. In some office cultures relentless wall-to-wall meetings are still considered sign of success. I’m certainly in the school of thinking that some meetings could have been resolved as mails. This may lie in the fact that as a translator I know that I am happiest when translating – rather than any of the ancillary jobs that go with the task (e.g. tagging up content destined for the web for accessibility purposes), although the final production stages (e.g. creating an accessible PDF or post/page for a website) are the ones with the most visibility. In this regard there is also the comparison of working as a translator being like a taxi driver – you don’t earn sitting at the cab rank waiting for a fare, but instead you earn while the wheels are turning (or in some cases also when you have a fare and are stuck in traffic).

    tetris design wall
    Photo by Mikhail Nilov on Pexels.com

    You could stretch the Tetris analogy to its maximum by breaking down an eight hour working day (after breaks) down into individual 12 minute units, it’d give me 40 blocks in a day. Hmm. Think of those blocks as Tetris pieces and you have 10 Tetris pieces – and where they fit perfectly together you have a Tetris. Sadly the average working day isn’t as straightforward as ten consecutive 4 x 1 blocks. After all, you can’t rotate your time blocks! So yes 40 blocks worth of bricks will often take up more than those four lines!

    Where a translation day is similar to Tetris is the fact that it boils down to rhythm and flow – especially avoiding getting bogged down and spending too long on tasks. Hybrid working means I have more phone calls and online meetings, which can really disrupt my translation flow. The more densely packed the meetings, the greater the loss in productive time. In terms of your calendar you find yourself struggling to complete a “four block” task between a meeting that overruns and another hastily convened one, and then you get interrupted by another unexpected phone call.

    How does a large translation project resemble Tetris?

    I’d like to use the example of translating national primary legislation (i.e. an Austrian Gesetz) that itself is a transposition of European law (e.g. a Directive or a Regulation). On occasions, it can be relatively straightforward, in that it might be a simple amendment of existing legislation – e.g. an amendment to the Austrian Banking Act to reflect the latest amendments to CRD/CRR, so some steps have already been accomplished previously, which is akin to clearing several lines at once.

    It might, however, be closer to one of the variant games of Tetris: you don’t start with a clear screen, but have to clear various incomplete lines before you can start clear your “own” lines (and scoring those big points for Tetrises). Indeed, you might have already aligned the bilingual texts from Eur-Lex that are the basis of the source text that you will translate, but it might not be instantly apparent which provisions are being transposed where. The explanatory notes from the consultation draft (Gesetzesentwurf / Regierungsvorlage) will help to explain this. From here you can start to map the provisions against one another (a Tetris-like puzzle-matching exercise). The mapping might not be 1:1 as there might be more stringent national provisions enacted in transposition, national terminology might also not be identical to the European terminology needed.

    Such translations also involve puzzle-matching to identify patterns how national law renders the patterns of the European law. On a terminology level, patterns do emerge, although differences often exist between the German terminology in European law and Austrian national law, that are full synonyms. For example, Zulassung (authorisation) in European law corresponds to Austrian law’s Konzessionierung (more licensing / being granted a licence).

    However, where a large project most resembles Tetris, is in the completion of it. I still get a buzz at various stages of the project, such as finishing the first draft of a big piece, at the end of the checking and doing the final formatting check and then the accessibility and then uploading the indexed PDF to my employer’s website. From my involvement in complicated supervisory procedures over the years at various stages that there is also a buzz for the conclusion and successful outcome of the procedure. Mentally, Buran launches.

    https://twitter.com/t9natno5/status/1537142657377517569

    Elsewhere, there are issues like dealing with out and out Austrianisms, which simply “do not translate”. In particular as vestiges of legislation that was enacted in the 19th century, but remaining in force to this day. Classic examples exist in the likes of the Genossenschaftsgesetz, with some provisions unchanged (e.g. §§19-20), even in terms of spelling reforms, for almost 150 years. Arcane provisions of such laws can really throw even an experienced translator.

    And so to conclude…

    So the Buran has taken off, and you have completed the game or your translation. You’ve got square eyes (you can tell I grew up in the 1980s!) and you might have RSI or carpal tunnel problems (look after your hands and wrists, dear translators!) And you bask in the reflected glow of a successfully conquered translation delivered with minutes or hours to spare. And the feeling feels great – so much better than a Machine Translation, or dare I suggest it, the output from ChatGPT. And that deserves a reward: the tl:dr for you created by ChatGPT.

    Bonus: What does ChatGPT make of my Translation/Tetris analogy?

    Unless you have been off grid for most of 2023 so far, you’ve probably been caught up in the maelstrom of content about ChatGPT. Maybe the answer below, by ChatGPT, could be a nice tl:dr version of the post above.

    How can translation be compared to Tetris?

    Translation and Tetris may seem like very different activities, but there are some interesting similarities between the two:

    1. Both require strategic thinking: In Tetris, players need to think ahead and plan their moves in order to create the most efficient stack of falling shapes. Similarly, translators need to think strategically about how to convey the meaning of the source text in the most accurate and effective way possible.
    2. Both involve fitting pieces together: Tetris is all about fitting different shaped blocks together to create complete lines, while translation involves piecing together individual words and phrases to create a cohesive whole.
    3. Both require skill and practice: Becoming good at Tetris requires a lot of practice and skill, as players need to learn how to quickly react and make decisions under pressure. Similarly, translation requires a lot of practice and skill in order to accurately and efficiently convey meaning from one language to another.
    4. Both can be challenging and rewarding: Tetris can be frustrating when blocks don’t fit together, but it can also be incredibly satisfying when players successfully clear multiple lines at once. Similarly, translation can be challenging when trying to find the perfect word or phrase to convey meaning, but it can also be incredibly rewarding when a translation is accurate and effective.

    Overall, translation and Tetris both require strategic thinking, fitting pieces together, skill and practice, and can be challenging yet rewarding activities.