Na AI Go Save Our Language?

A language dies in the long quiet years when the children stop answering in it. The grandmother speaks. The child understands but replies in the bigger language of the bigger market. The grammar survives one more generation as comprehension, then disappears as speech. By the time a linguist records the last fluent speaker, the language has usually been dead for thirty years in which the linguist is left to capture a specimen.

Nigeria has more than 500 languages, depending on where one draws the line between a language and a dialect. Ethnologue counts 520 living indigenous languages and twelve extinct. Hausa, Yoruba, and Igbo each have tens of millions of speakers; Pidgin has tens of millions more. A long tail of smaller languages are down to a few hundred speakers.

Before the eulogies, a harder question: does it matter, and if it does, can the technology, Artificial Intelligence, now reorganising everything else be turned toward this before it's too late?

It matters, for reasons older than Nigeria. Writing preserves memory which shapes survival. When Europe's vernaculars began appearing in print in the fifteenth century, books told their speakers that their language was fit for serious thought too, as only Latin had been. Without books, a language is cut off from the myths and stories a people uses to explain who they are.

A colonial lie lingers nearby with a proclamation that Africa had no history because it had no writing. But West Africa Timbuktu's libraries and the Sokoto Caliphate produced substantial written literatures. Many Nigerian languages never made the crossing into print because conquest broke transmission chains and the economy attached opportunity to English and the three largest languages, so a parent's love became, rationally, a reason to raise a child away from the smaller one.

The Nigerian state made this more urgent in 2025, when the government cancelled the National Language Policy and made English the main medium of instruction. It reversed the 2022 policy which had required six years of mother-tongue instruction. Faced with teaching children in Nigerian languages, though, the country chose English. That is the setting for the machines.

The promise is real. Modern language technology can transcribe recordings, help build a dictionary and grammar from a corpus that once took a career to assemble. It replaces neither linguists nor fluent speakers. But for a tongue with a few hundred speakers and no written tradition, the floor for what's possible has significantly dropped.

There is already an African model for doing this. Masakhane is a research network founded in 2019 with hundreds of researchers across the continent. Native speakers are co-authors with openly released models. In 2026, Masakhane joined Microsoft's AI for Good Lab, the Gates Foundation, and Google.org to launch LINGUA Africa, an open call funding community-grounded language work, on the principle that a language belongs to its speakers.

Profit is another force. Most of the world's easily available text is already absorbed into large models; low-resource languages are among the few bodies left. The incentive is obvious: harvest whatever Nigerian-language speech and text exists, fold it into systems owned elsewhere, and sell it back to the communities it came from. The language survives in the model, dies in the mouth, the most modern way a language has ever died.

If the machines are to rescue rather than harvest, the work cannot be left to the market or to volunteers alone. It requires the state to treat documentation of endangered languages as national infrastructure. That means funding the unglamorous work, recording, orthography, corpus-building, training computational linguists accountable to the communities they serve, and giving the work status.

It would also take Nigerian Pidgin seriously. Pidgin is the largest language in Nigeria the state refuses to fully acknowledge, spoken by tens of millions across every ethnic line, the nearest thing Nigeria has to a genuinely national tongue built inside the country rather than inherited. It has been creolising for generations; for many Nigerians it is now a first language, with no settled orthography and no place in the school system its reach warrants.

Professor Ben Elugbe, who died in 2025, spent much of his career arguing that this was an error the country could afford to correct. With Augusta Omamor, he wrote the foundational study Nigerian Pidgin: Background and Prospects (1991). He rejected writing Pidgin as broken English, favouring a system closer to Nigeria's indigenous orthographies. Pidgin is Nigerian and should look like one on the page. The standardisation he called for would do for Pidgin what print did for the European vernaculars.

Both the dying languages and the unacknowledged one reflect the same habit: the state deciding which Nigerian speech counts as serious. New technology makes that cost harder to ignore. What's uncertain is whether anyone with power thinks it's worth doing.

Can AI save Nigerian languages from extinction? Not by itself. It is a tool, and tools follow the intentions of whoever controls them. Whether it happens is a political question.

What's true is that the door is open. With a few thousand recorded sentences and a community treated as co-authors rather than subjects, a language headed toward specimen status can instead become archived, teachable, searchable, alive enough to be handed to a next generation. The first great chance for most of these tongues was the printing press, and they missed it through no fault of their own.

It would be the bitterest kind of repetition to miss the second one, while the means to take it sit in every phone in the country, waiting for someone with authority to decide that a grandmother's sentences are worth as much as an advertisement.

They are worth much more. Someone should say so with a budget.