> ## Content Index
> Fetch the complete content index at: https://thepatrologist.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# The Future of Language Learning
- URL: https://thepatrologist.com/the-future-of-language-learning/
- Published: 2026-11-21T01:24:18.000Z
- Updated: 2026-11-21T03:45:58.000Z
- Author: Seumas Macdonald

About once a week one I either receive an email about a new project/resource/app or I chance upon one in the wild. Many of these are good, sincere, well-intentioned, and even well-done projects. But there is a proliferation of them as well, which means it's harder and harder for me to cast my eye over them all.

I have spent a good chunk of my life on languages. A lot of that on Ancient Greek, but plenty on other languages too. And for at least the last two decades I have had a strong and active interest in pedagogy, Second Language Acquisition, and how we can *do* learn languages and how we can *best* facilitate that. 

I don't think we are living in a golden age, but things are changing, and the question that many are asking is "will AI take my job?" In this post I want to consider X facets to the above topics

### The Rise of the Apps

One of the things that first got me interested in digital texts and reading platforms and the like is that I could see the potential for graded readers with helps, tailored to the learner (ideally, me!) that would help smooth the gradient of what "i+1" looks like at any point. I still think that would be the ideal, a perfectly customised reading platform with a wealth of texts.

That said, there are now more and more platforms and projects, and many of them do something like this or it's part of their package. And some of these are vibe-coded, which also means that any body with enough AI credits can really throw together what they want such an app to be or do, especially since we now live in an age where there are open-access data sets and digital text repositories for huge swathes of ancient literature. That's a big win. Built on the backs of years of digital humanities labour though.

Probably the best example of this is actually [Legentibus](https://legentibus.com/?ref=thepatrologist.com). I think that is a well-designed app that curates Latin texts, has tons of material, audio, the right helps, and is genuinely helpful to learners. For Greek, there is no Legentibus-level reader at present. I have a fondness for Kevilex, I'm aware of some other things out there. 

At the end of the day, though, no app makes you fluent in a language. I think the failure of Duolingo should probably tell us this. It's... not enough. What a good app can do is (a) give you lots of comprehensible input, especially reading, (b) give you language *practice*, i.e. drills, which have a very minimal effect on actual language acquisition, (c) gamify the process to trigger your dopamine reward systems.

But I still think there are considerable gaps, which I am not yet convinced that apps can bridge. 

What I will say is that building truly competitive and quality apps and learning platforms takes a level of resources that I don't have, so you won't see a SeumasU app anytime (probably ever). That's not a resource race I can win, let alone enter, so I am not really spending any time trying to be in the field.

### The Fruit of Communicative Methods

We are probably now at the stage where there just are a number of people who have *incredibly* good active skills in Ancient Greek (I'm not one of them, by the way). They've done the work, gone to all the courses, spent time in immersion settings, read the canon and beyond, and are now... amazing. 

I think that is the biggest advertisement for what is possible and achievable via communicative methods. A big shout-out to the teachers at the [Institutum Antiquitatis](https://www.institutumantiquitatis.com/?ref=thepatrologist.com), having watched them speak and interact online and in recordings, I'm truly impressed. They are not the only ones. 

It takes a lot of effort, and time, to get to that kind of level, but it reminds us all that it's possible, and that in the face of the nay-sayers, who have spent their years in traditional institutions doubling-down on grammar and poo-poohing the idea that you could actually become meaningfully competent in this sense and that it would be of any use (rather than just an amusing past-time, as others seem to think).

### The State of the Market

At the same time, I think this has also created a somewhat unforeseen consequence. There are more and more teachers doing something like what I am doing - communicative spoken Ancient Greek. Not that many, in absolute terms, but slowly the number is increasing. And there is demand, obviously, as more people realise that this is a better way to acquire language and want to learn this way. That demand is still relatively niche though, and it doesn't have much sway in traditional institutions, which are mostly still doing the same old thing with the same old results, with declining numbers in most historical language programs anway.

But it does make it hard. It's hard to get minimum numbers to run courses. I feel that, and I know other similar teachers experience the same. I don't think the market is saturated or anything, but it's a small market with a number of very good teachers, doing a variety of fascinating things, but struggling to get enough students to make classes viable, let alone a living. And I am not sure that will change any time soon.

### AI: its Potentials and Pitfalls

Enter LLMs. I think I have a moderate but critical stance towards AI in general. I don't use it for writing, and I am opposed to AI artwork. I think it's abilities are overrated and oversold, but it has some uses, and it can be very good at certain things. 

AI translation: Is surprisingly good and getting better. It's much better at translating into English than into ancient languages, but even that is improving. I don't think AI translation can or should replace human translation, partly because (i) it erases, steals, and exploits the labour of all the human translators that have gone before, especially of the work being translated; (ii) I think translation is an act of creation and meaning-generation that is a fundamentally human-authored endeavour, and so I think AI translation is at about the same level of regard as AI-authored materials; (iii) the nature of translation is not, I believe, reducible to machine translation (apologies to Star Trek's Universal Translator). 

That isn't to say that there are no places for AI translation, but I don't think it solves that many problems, and the rush to deploy it is... ill-conceived.

Perhaps one of its biggest problems is that it suggests, subtly, that translation can replace language learning. Why learn Latin if you can just AI-translate any Latin text? Well, because of two things: (i) it's not the same thing to read a document in the language of its authoring; (ii) you cannot evaluate a translation unless you know the language well enough to use it.

I think AI translation holds out the temptation that it can replace language learning, and that is a poisoned apple.

AI authorship: Can LLMs write decent Ancient Greek? No. Can they write meaningful, grammatically accurate, learner-helpful texts? Yes. 

I haven't done any recent testing, but my general observations are that LLM outputs are improving in this regard, but they suffer two considerable flaws. Firstly, like English authored by AI, it's AI-slop. It has all those flaws built into it. Secondly, AI in general has the problem that *it never knows when it's wrong*. Which means you need to know when it's wrong. And you can't know it's wrong unless you already know the answers you're looking for.

Can AI help us in language learning and teaching? Yes. But with lots of caveats.

### Fraught Forecasting

I don't really have any great future forecasting ability here. I think these trends will mostly continue along these trajectories. AI will improve, but it won't overcome its inherent flaws. But its uptake will continue. And all the associated problems with it, including data centres, will come with that.

I think the future of historical language continues to lie more and more outside traditional institutions. There are all sorts of difficulties and challenges besetting the academy. I *don't* think alternate institutions are going to have any easy path though - financially it's very hard to set up something that can meaningfully create a humanities education with all that it should have, while also supporting the faculty to do that. But at least in this niche, the niche of those wanting to genuinely acquire historical languages, the best paths now all appear outside traditional credit-hour subjects.

For myself, I continue to spend time reflecting on the 'best way forward', and 'what's worth spending my limited time on'. I think that continues to be authoring texts, which is why LGPSI and some other related or similar projects are at the top of my priority lists. I have some other ideas that I think are worth actioning, but I'll leave them to become more real before talking about them.

What I think we will realise is that all the apps, reading platforms, etc., are valuable tools but they don't bridge the gaps that we need filled: a broad and vast range of materials for learners at different levels, from zero up (which classical corpora just can't fill because most of our classical texts are too hard), better audio resources, and the need for live communication: the negotiation of meaning in real time between sentient humans. These are the things I'm most interested in pursuing.