philosophy

Language learners and educators often have one main question: How do we achieve fluency effectively? But as AI becomes increasingly capable of producing language, that question starts to feel incomplete. The better question might be: Fluent for what?

What is the reason for language study and multilingualism in a world where AI can write essays, explain concepts, generate ideas, entertain, translate, and converse?

While I don’t have the answers, I do have positions that I’ve developed over 15 years of working in writing and language education. Here are a few of them.

1. Fluency does not always have to be the goal.

As a metric it’s empty unless tied to a purpose, making function more valuable than the pursuit of “fluency.” Some learners need conversational capacity, others wish to read novels and never bother with speaking, while still others need a working knowledge of business jargon to participate in meetings. The question is: What do I need to be able to do repeatedly and without friction? Not: Am I fluent?

2. Many parts of the language field are stuck in spectacle rather than substance.

We’ve reduced language content to challenges, streak tracking, one-upmanship, and fraud hunts. At a certain point, as AI becomes more reliable, we need to move past spectacle into new substantive uses.

3. Language performance is not knowledge.

It’s behavior under pressure. It’s possible to “know” a grammar rule but never be able to accurately use it in real time. Language, then, is more about trained reflexes and less about understanding structures.

4. Automaticity is built through repetition, not novelty.

n classrooms around the world, educators focus on making activities fun and new, but new reflexes require the slow, mechanical buildup of patterns that only comes through repeated exposure to the same or similar input.

5. Input is crucial; not all input is equal.

Authentic input is necessary, as is input that is comprehensible enough to process and structured enough to reinforce patterns. And to be clear, Krashen never argued that passive input alone was sufficient for language mastery.

6. Forced communication too early can do harm.

Learners who are pushed to provide output before they have appropriate, grammatical reflexes do not build fluency. They build workarounds, and those workarounds fossilize.

7. Modern language teaching should move beyond delivering content or providing feedback into curating the cognitive environment.

A language teacher’s job is no longer about explaining, assigning, and entertaining. It’s also not about being the native speaker in the room. It is about designing and managing appropriate input, sequencing exposure, stabilizing patterns, and scaffolding output and retrieval.

8. Fluency and writing in the AI era must provide new justifications.

If a language learner can fake fluency behind a camera, and an aspiring writer can produce a passable essay through prompting, why invest in learning to write and speak at all? The answers lie in cognition, identity, independence from mediation, and depth of access.