Babel, Pentecost, Language, and the Nations

Language sits near the center of Christian mission. Scripture itself crosses languages. The gospel moved through Aramaic, Greek, Latin, Syriac, Coptic, Armenian, Ge’ez, and countless later vernaculars. Missionaries have learned languages, translated Scripture, created writing systems, compiled dictionaries, taught literacy, and sometimes carried their own linguistic prejudices into the process.

Artificial intelligence enters this history with astonishing promise. Machine translation, speech recognition, text-to-speech, multimodal models, and low-resource language research may lower barriers that have constrained Christian communication for generations.

It is therefore tempting to frame AI translation through two biblical scenes: Babel and Pentecost. The temptation should be resisted long enough to read both carefully.

Genesis 11 tells of a human society sharing one language, building a city and tower, seeking to make a name, and resisting dispersion. God confuses their language and scatters them. Christian interpretation has emphasized different features of the story: pride, concentrated power, refusal of the creation mandate, human self-exaltation, and divine judgment. The text does not permit a simple equation in which linguistic diversity itself is merely a curse Christians should technologically undo.

Acts 2 presents another remarkable linguistic event. The Spirit enables the gathered disciples to speak in languages recognized by visitors from across the Mediterranean and Near Eastern world. The crowd hears the mighty works of God in its own languages.

Many Christians have described Pentecost as a reversal of Babel. There is genuine theological insight in the contrast: fragmentation does not prevent God’s redemptive purpose, and the gospel crosses linguistic boundaries by the Spirit’s power. But a simplistic “Babel created many languages; Pentecost returns humanity to one language” reading misses an obvious feature of Acts 2. The diversity remains. The miracle is intelligibility across difference.¹

Pentecost does not erase languages. It makes witness heard through them. This matters for mission. Christian unity does not require linguistic sameness. Translation is not a temporary concession until everyone reaches a single dominant language. The movement of the gospel into another language can become an affirmation that a community may receive, confess, teach, and theologize Christian truth without first abandoning its linguistic identity.

AI can serve this work when it helps people hear in languages that have historically received fewer resources. But a theology of Pentecost should not be turned into a prophecy of machine translation. The Spirit’s work in Acts is not reducible to overcoming a communication bottleneck. Pentecost inaugurates Spirit-empowered witness and the formation of a people. Translation technology may serve that witness. It is not the event’s fulfillment.

The same caution applies to Babel. A global AI platform that translates everything into everything does not automatically solve the moral dynamics Genesis 11 exposes. Indeed, centralized language infrastructure could reproduce some analogous concerns if a small number of institutions control the models, data, standards, and interfaces through which the world’s languages are mediated.

The Christian response should therefore hold together two goods that technological culture often separates: intelligibility and plurality.

Intelligibility matters because the gospel should be communicated clearly. Christians should want people to access Scripture, teaching, theological education, and ordinary communication in languages they understand deeply. There is no virtue in forcing people through unnecessary linguistic barriers.

Plurality matters because languages carry histories, identities, humor, metaphors, social relationships, and ways of naming experience. A system can preserve propositional content while flattening register or cultural resonance. A dominant language can function not only as a communication tool but as a gatekeeper to education, theology, and institutional power.

This is especially important in low-resource language AI. The goal should not be to make every language computationally identical to English. Some languages are primarily oral. Some have multiple competing orthographies. Some have dialect continua whose boundaries are social as well as linguistic. Some communities may value speech tools more than text generation. A technology strategy shaped only by the data easiest to collect may misread what the community actually needs.

The Church has a particular responsibility here because Christian organizations sometimes hold unusually rich linguistic resources: Bible translations, dictionaries, audio recordings, terminology databases, literacy materials, and translation notes. These resources can help build language technology. They can also create questions of consent and ownership. The fact that a mission organization possesses a corpus does not automatically mean every later machine-learning use is morally uncomplicated.

Pentecost also corrects missionary pride. The gospel crosses languages because God acts, not because the messenger achieves cultural mastery. Missionaries should pursue linguistic competence seriously, but competence itself does not create spiritual authority. AI can help workers learn faster, rehearse more often, and reduce avoidable mistakes. It should not encourage the illusion that translation is the same as understanding people.

A perfectly translated sentence can still be culturally foolish. A missionary may know the words and misread the relationship. A local Christian may speak the language natively and still disagree with a foreign translator’s theological assumptions. Language is a necessary dimension of contextualization, not the whole of it.

This is why AI-assisted language work should repeatedly return to human communities. Models can generate hypotheses about register, idiom, or cultural meaning. Native speakers can say whether the hypotheses are true. Local churches can decide whether Christian terminology is intelligible and faithful. Translation should move through community rather than treating community as the final quality-control test for a product designed elsewhere.

The Bible’s linguistic vision is ultimately richer than either uniformity or fragmentation. Revelation imagines a redeemed multitude from every nation, tribe, people, and language before God. The imagery does not require us to settle exactly what language looks like in the new creation. It does make one theological point difficult to miss: the nations are not redeemed by becoming culturally indistinguishable.

For Christian AI strategy, the implication is modest but powerful. Reduce unnecessary language barriers. Invest where technological resources are scarce. Build speech, translation, and accessibility tools that serve actual communities. Do not treat dominant-language convenience as the definition of progress.

The goal is not a world in which every language becomes functionally identical. It is a Church capable of bearing faithful witness across difference without demanding that difference disappear first.

Babel Without a Technology Myth

Pentecost and Intelligible Witness

Pentecost is sometimes invoked as a technological ideal: everyone understands. But Acts 2 also destabilizes control. The disciples speak through the Spirit’s initiative; the multilingual crowd hears; observers disagree about what the event means. Peter interprets. A community forms.

The event is not a frictionless communication protocol. This should make Christians cautious about technologies that promise perfectly controlled reception. Missionary communication always enters human freedom, culture, history, and interpretation. Clear language matters enormously. It does not guarantee a controlled response.

Language, Hospitality, and Power

Christian translation is sometimes described purely in terms of information transfer. A richer account sees translation as a form of hospitality: the effort to make space for another person’s language rather than requiring that person to enter mine before hearing something important.

This is one reason low-resource language technology has missionary significance beyond efficiency. When a theological school can teach only through a former colonial or global language, students may access content but still experience theology as something that arrives from elsewhere. Developing theological vocabulary and resources in local languages can strengthen the ability of churches to think, teach, pray, and debate within their own linguistic worlds.

AI can accelerate this work. It can also tempt organizations to believe that translation has occurred simply because sentences were converted.

Hospitality requires listening to how the guest actually receives what was offered. When one language dominates theological publishing, that language gains more than communication efficiency. Its categories, debates, metaphors, and institutional concerns become disproportionately available.

AI can deepen this pattern if low-resource language outputs are primarily translations of high-resource theological content. A multilingual system may increase language access while preserving intellectual dependency.

A stronger missionary vision includes local-language authorship, not only translation. AI should help writers research, edit, and publish in languages that historically lacked publishing infrastructure.

The aim is not merely more English theology in more languages. It is more theological agency across languages.

Language data raises unusual ownership questions because no individual owns a language, yet communities can have legitimate interests in recordings, stories, sacred material, and culturally significant expressions.

Mission organizations should resist the simplistic claim that because a corpus is legally available, every downstream AI use is morally settled.

Consent may need to be revisited when the use changes radically. Material recorded to produce a dictionary may not have been contributed with the expectation that it would train a commercial generative model. Scripture data may be legally licensed while local churches still have concerns about representation.

This is not an argument for impossible permission structures. It is an argument for asking who benefits and who gets to decide.

The Nations Without Erasure

Revelation’s vision of peoples and languages before God provides a final correction to technological uniformity. Christian hope does not imagine redemption as the victory of the most computationally convenient language.

This matters because technology often treats standardization as progress. Standardization can be useful. Unicode, shared protocols, and orthographic standards can dramatically improve access. But standardization should remain a servant of human communities.

If a minority language survives only by becoming computationally legible on terms set elsewhere, the technical success may contain cultural loss.

The Church can pursue a different goal: intelligibility without erasure, access without uniformity, translation without domination.

Christian hope resists measuring progress by the computational convenience of linguistic standardization. The long-term test is what grows around the tool. The ministry may gain speed while losing understanding, or it may use speed to create margin for better judgment and relationship. Success should therefore be evaluated over time through competence, accountability, resilience, and service rather than through the first impressive demonstration.

Footnotes

1. On Pentecost, linguistic difference, and Christian translation across cultures, see Keener, Acts: An Exegetical Commentary; Sanneh, Translating the Message: The Missionary Impact on Culture ↩