Learning in English Only Is a Decision. Usually a Bad One.

 A safety module for a plant in Tamil Nadu. Written in English. Delivered to supervisors and operators, some of whom read English comfortably, many of whom read it slowly and think in Tamil, and a few of whom nod politely at the completion screen having understood perhaps sixty percent of it.

The completion rate on that module was 94%. The comprehension rate was unmeasured, unmeasurable, and — I would bet a good deal of money — considerably lower.

This is one of the quietest failures in Indian corporate learning, and it persists for the least defensible reason imaginable: not because anyone decided it, but because nobody ever decided otherwise.

The default that nobody chose

Ask why the content is in English and you'll get a chain of perfectly reasonable answers.

The L&D team works in English. The licensed content library came in English. The subject-matter experts wrote their material in English. Translation was expensive, so it was deferred. And the completion numbers looked fine, so it never became urgent.

Each link in that chain is individually sensible. The chain as a whole encodes a decision that no one would defend if it were stated plainly: the employees who face our highest-consequence risks — the plant floor, the field, the frontline — will learn about them in a language they process with more effort and less confidence than head office does.

Written down like that, it's obviously wrong. It just never gets written down.

What it actually costs you

The damage isn't ideological. It's operational, and it shows up in three places.

Comprehension, especially under load. Reading a second language is a cognitive tax. It's manageable when you're relaxed and it degrades sharply when you're tired, rushed, or stressed — which is precisely the condition in which safety knowledge, escalation procedures, and compliance judgement actually get used. Content people "completed" in English is disproportionately content they cannot retrieve under pressure. For safety training, that isn't a learning problem. It's a risk-management one.

Engagement, immediately and visibly. A frontline employee opening a platform and finding everything in a language associated with head office learns something about who the system is for, in about four seconds. Every organization I've seen introduce serious vernacular content reports the same step-change in frontline engagement — not because the content improved, but because the platform finally stopped signalling this wasn't built for you.

Credibility of the whole programme. The dealer-network training, the product certification, the induction — if these arrive in a language the audience half-follows, the entire learning function acquires a reputation as a corporate ritual rather than a useful tool. That reputation is very hard to reverse, and it contaminates everything you launch afterwards.

The excuse that expired

For a long time, the objection to vernacular content was economic, and it was legitimate. Translating and maintaining a course in six languages meant six agency invoices, six review cycles, and six forks that would drift out of sync the moment the master changed. So companies translated the two or three highest-stakes modules and left the rest.

That constraint has substantially dissolved. AI translation and versioning have taken multilingual content production from an agency project to a review task — draft generated in minutes, then checked by someone who actually speaks the language and knows the plant. The bottleneck moved from production to review, and review is cheap in a company that has native speakers in every location, which every Indian enterprise does.

Which means the vernacular question is now, genuinely, a choice rather than a budget constraint. Modern learning management platforms increasingly treat multilingual delivery as a workflow rather than a project — one master, many versions, updated together, tracked in a single completion record. If your platform can't do that, you should be asking why; and if it can and you aren't using it, you should be asking yourself the harder question.

What "doing it properly" looks like

A few things I'd argue are non-negotiable once you decide to take this seriously.

Translate by consequence, not by convenience. The prioritization isn't "whatever's easiest to convert." It's: where does misunderstanding hurt most? Safety, machinery, compliance, customer-facing product knowledge. Start there and work outward.

Review with someone who does the job, not just someone who speaks the language. Machine translation gets the grammar right and the register wrong — technical terms, plant vocabulary, the way people actually name a component on the floor. The reviewer you want is the supervisor, not the language graduate.

Keep versions in sync or don't bother. The vernacular version that lags the English master by two policy updates is worse than no vernacular version, because it teaches the wrong thing confidently. Single-source it, publish together, expire together.

Don't sub-title everything and call it done. Subtitles help. They are not the same as a module built in the language, and on a small phone screen in bright sunlight they are frequently useless.

And measure comprehension, not completion. A short applied assessment in the same language tells you whether the message landed. Completion tells you the video reached the end, which is a fact about your CDN, not about your workforce.

The objection: "but our official language is English"

This comes up in almost every conversation, usually from someone senior, and it isn't stupid — it just answers a different question than the one being asked.

Yes, your corporate language is English. Your policies are in English, your systems are in English, and there are real arguments for a single working language in a company operating across a dozen linguistic regions. Nobody is proposing you run the board deck in six languages.

But "the language we conduct business in" and "the language people learn most effectively in" are separate questions, and conflating them is where the error lives.

Nobody thinks the shop-floor safety instruction painted on the wall should be in English because the corporate language is English. Everyone accepts, instinctively, that the wall should speak to the person standing in front of it. The learning platform is the same wall. It just happens to be administered by head office, which is why it inherited head office's language by default rather than by design.

There's also a practical asymmetry worth naming. An employee who reads English fluently loses almost nothing from a Tamil or Hindi version being available — they'll pick whichever they prefer, and most will pick English. An employee who reads English with effort gains enormously. The downside of offering the choice is close to zero; the downside of not offering it is concentrated entirely on the people with the least power to complain about it.

Which is, I suspect, the real reason it took this long.

The thing underneath

I've come to think that language is a proxy for a bigger question that runs through every part of corporate learning: who is this actually for?

The English-only default, the forty-minute module, the desktop-first design, the head-office pilot group — these aren't separate failures. They're the same failure, expressed four times. Each one is a design decision made by people at the centre, for people at the centre, then distributed to a workforce that mostly isn't at the centre, with genuine surprise when it doesn't land.

The fix in every case is the same, and it isn't a feature. It's the discipline of designing from the edge inward — from the plant supervisor, the field rep, the dealer's technician — and letting head office be the population that adapts, since head office is the population with the slack to do it.

Nobody wrote the English-only policy. That's exactly the problem. It's time somebody wrote the other one.