Critique of the Elicitation-Based Bible Translation Model

Section 1

Executive Summary

The elicitation-based Bible translation model (hereafter "the model") represents a sincere and principled response to a genuine problem at the heart of global Scripture access: how can the translation of sacred texts be anchored in the authentic voice of a language community when the translator is not a native speaker? By placing the native speaker at the center of the production process — asking them to supply target-language equivalents for each unit of source-language content — the model aspires to a level of mother-tongue fidelity that is difficult to achieve through secondary linguistic study alone. That aspiration deserves to be taken seriously.

However, a rigorous analysis of the model's structural mechanics reveals significant and compounding weaknesses across five critical domains. First, on feasibility grounds, the sheer word-count of the Protestant canon — approximately 783,000 words — combined with the logistical demands of sustained facilitator-speaker sessions, makes end-to-end elicitation an extraordinarily slow, fragile, and expensive pathway to a completed translation. Second, on scalability grounds, the model is essentially non-scalable: because it depends on a highly specialized facilitator-speaker dyad, the throughput per unit of organizational investment is vastly lower than either the mother-tongue translator (MTT) model or AI-assisted drafting pipelines. Third, at the level of linguistic accuracy, word-by-word elicitation is prone to well-documented artifacts including translationese, loss of discourse coherence, and register misalignment. Fourth, the ethical architecture raises unresolved questions about community ownership, data sovereignty, and theological accountability. Fifth, the model operates outside the technology ecosystem — particularly Paratext, PTXPrint, and emerging AI integration pipelines — that constitutes the effective standard of practice for the global translation movement.

This review does not argue that elicitation has no place in Bible translation. On the contrary, it identifies two domains where elicitation methodology offers genuine and important value: language documentation at the pre-translation stage, and community checking sessions during the revision phase. The constructive argument is that elicitation should be positioned as a targeted tool within a larger, integrated workflow — not as the primary end-to-end method of translation production.

The report concludes with a recommended hybrid framework combining pre-translation elicitation for language documentation, community-owned drafting by trained mother-tongue translators, LLM-assisted first-draft generation where appropriate, full integration within Paratext and PTXPrint workflows, and AI voice model deployment for audio Scripture production in oral-primary communities.

Key Findings at a Glance

Domain	Assessment	Severity
Feasibility	Multi-decade timelines per language; single-point-of-failure in speaker availability	High Concern
Scalability	Cannot address the need of millions of people in language communities at any realistic deployment level	Critical Concern
Linguistic Accuracy	Structural elicitation artifacts undermine naturalness and discourse coherence	High Concern
Ethics	Ownership, consent, data sovereignty, and power dynamics inadequately addressed	Significant Concern
Technology Integration	Outside Paratext/PTXPrint/AI ecosystem; creates downstream publication barriers	High Concern
Elicitation as Tool	Genuinely valuable in language documentation and community checking phases	Positive (Targeted)

Section 2

Background: The Global Scripture Access Challenge

As of 2024, approximately 7,396 known spoken or signed languages exist worldwide, representing approximately 7.45 billion people. Of these, 756 language groups now have access to the complete Bible, 1,726 have a completed New Testament, and a further 1,274 have some portions of Scripture. Translation work is actively underway in 3,526 or more languages across 173 or more countries, with Wycliffe International Alliance agencies engaged in at least 3,146 of these. The number of languages still needing Bible translation to begin stands at millions of people in language communities — down from higher figures the previous year, reflecting significant acceleration through global partnerships.

The organizations coordinating this work represent decades of accumulated institutional knowledge. SIL Global (established 1934), Wycliffe Bible Translators (established 1942), and Biblica (established 1809) collectively sustain the world's largest network of translation programs, supported by the United Bible Societies operating in more than 240 countries and territories.

The traditional translation pipeline proceeds through several interconnected phases. A translation team — ideally composed of trained mother-tongue speakers of the target language — begins with exegetical study of the source texts in their original languages (Hebrew, Aramaic, and Greek) or in close proximity to them via bridge-language resources. A draft translation is produced, followed by a rigorous back-translation step. Community checking sessions gather native speaker feedback on naturalness and comprehension. Finally, a qualified translation consultant reviews the entire project before publication.

Central to this pipeline is Paratext, the global-standard software platform developed jointly by SIL Global and the United Bible Societies. Its companion tool, PTXPrint, converts Paratext projects into print-ready PDFs formatted for minority-language publications. Together, these tools represent an integrated, field-tested infrastructure that has enabled remarkable acceleration in global translation output.

The field is now entering a new phase characterized by the integration of large language models (LLMs) and AI voice technologies — developments that define the technological horizon against which any proposed translation methodology must be evaluated.

Section 3

The Elicitation Model Described

In the elicitation-based model as proposed, a trained facilitator — who is not a native speaker of the target language — sits in direct contact with a fluent native speaker of that language. The facilitator presents source-language content, typically from a bridge language such as English or from the biblical source texts themselves (Greek or Hebrew). The native speaker then supplies the target-language equivalent — the word, phrase, or rendering that seems natural and accurate in their language. The facilitator records and transcribes these responses, building up a corpus of elicited target-language material over the course of many sessions.

The fundamental characteristic distinguishing this model is that the primary production agency — the act of formulating target-language text — resides entirely with the native speaker, not the facilitator, and that the method of elicitation is the principal mechanism by which that text is generated.

It must be acknowledged that elicitation as a method has robust and respected precedent in descriptive linguistics and language documentation. Field linguists working in the tradition of Leonard Bloomfield, Kenneth Pike, and the Summer Institute of Linguistics have used elicitation techniques extensively to document previously unwritten or under-documented languages. In this limited and targeted context, elicitation is irreplaceable. The question this review addresses is not whether elicitation works as a linguistic tool — it does — but whether it is an appropriate and sufficient primary method for producing a complete, accurate, theologically faithful, and publishable Bible translation.

Review Synthesis

"The strength of the elicitation approach lies in its centring of the native speaker. Its vulnerability lies in the assumption that what is produced under elicitation conditions constitutes a natural, discourse-embedded, genre-sensitive translation rather than a sequence of metalinguistically mediated linguistic performances."
— Review synthesis of linguistic fieldwork literature

Section 4

Feasibility Analysis

4.1 Time and Labor Costs

The most immediate challenge confronting the elicitation model is straightforward arithmetic. The Protestant biblical canon contains approximately 783,000 words distributed across 66 books, 1,189 chapters, and 31,102 verses.

Scope Calculation

If a facilitator-speaker pair conducts daily sessions of 100 elicited items per day — an optimistic figure that assumes no re-elicitation, no extended discussion, no logistical interruption, and no fatigue — a single pass through the 783,000-word Protestant canon requires approximately 7,830 sessions. At one session per day, seven days per week, this represents approximately 21.5 years of uninterrupted daily work for a single pass of the source material alone. Multi-pass revision, back-translation, community checking, and consultant review would multiply this timeline further. Even at a generous session productivity of 300 items per day, a first pass requires approximately 7.2 years of daily sessions.

For purposes of comparison, the traditional translation pipeline — involving trained mother-tongue translators working in Paratext with consultant support — typically requires 8 to 12 years for a complete New Testament and 15 to 25 years for a full Bible. The elicitation model, rather than compressing these timelines, appears structurally likely to extend them — in some scenarios dramatically.

Beyond the raw word-count mathematics, the logistical demands of sustained elicitation impose additional time costs. Session scheduling must accommodate the speaker's agricultural, family, and community rhythms. In many communities where undocumented or under-translated languages are spoken, these rhythms are seasonal and inflexible.

4.2 Human Resource Constraints

The elicitation model as described is structurally dependent on a highly specific dyadic relationship between one facilitator and (typically) one primary native speaker. This creates a single-point-of-failure risk qualitatively different from the risks faced by multi-member translation teams using established methodologies.

Speaker availability: Native speakers of minority languages are subject to migration, illness, death, and the ordinary demands of livelihood. A project that has accumulated five or eight years of elicited material may find itself without its primary speaker, with no trained replacement capable of maintaining linguistic and contextual continuity.
Speaker expertise: Not all fluent native speakers are equally capable of the metalinguistic reflection that effective elicitation requires. Identifying and retaining a speaker who can navigate the semantically complex terrain of biblical poetry, prophecy, legal code, and epistolary prose requires linguistic sophistication that cannot be assumed.
Facilitator competency: The role demands an unusual combination of skills: source-language competency (ideally including Greek and Hebrew), fieldwork training adequate to recognize elicitation artifacts, and the interpersonal acumen to sustain a productive relationship across years of intensive collaboration. This skill combination is extraordinarily rare.
Institutional continuity: If the facilitator leaves the field for any reason, the project faces a significant continuity crisis. The accumulated elicitation records may be interpretable only by the original facilitator.

4.3 Recording, Transcription, and Versioning

A completed Bible translation is not merely a collection of words; it is a structured, versified, cross-referenced document that must be checkable at every level and ultimately publishable in both print and digital formats.

Critical Risk

Without integration into Paratext's USFM-structured project environment, elicitation transcriptions accumulate as unstructured or semi-structured text files. Across thousands of sessions spanning years, the risks of data corruption, inconsistency between sessions, loss of version history, and inability to track revisions are substantial. A single storage failure, without robust cloud-synchronized backup, could destroy years of irreplaceable work.

Paratext addresses these risks through built-in project versioning, cloud synchronization through the Paratext Registry, structured USFM markup that enforces canonical structure, and automated checking tools that catch mechanical errors before they propagate through the document.

Section 5

Scalability Analysis

5.1 Throughput Mathematics

The scalability problem of the elicitation model is, in some respects, even more serious than its feasibility problem. Feasibility concerns whether the model can work for a single language project; scalability concerns whether it can contribute meaningfully to the global translation challenge as a whole.

Consider a moderately optimistic scenario: one facilitator-speaker pair completes a full New Testament in 10 years and a full Bible in 25 years. To reach the millions of people in language groups currently identified as having an expressed or active translation need — using elicitation as the primary method — would require commensurate numbers of pairs operating in parallel, each sustained for a quarter-century or more.

By contrast, emerging AI-assisted translation workflows offer a qualitatively different throughput profile. LLM-based drafting tools can generate a first-draft New Testament within days — a draft requiring extensive revision and checking, but providing a structured, Paratext-compatible starting point far ahead of where elicitation begins.

Methodology	Estimated NT Timeline	Scalability	Technology Integration
Traditional MTT + Paratext	8–12 years	Moderate (multi-member teams)	Full (Paratext, PTXPrint)
LLM-Assisted Draft + MTT Checking	3–6 years (projected)	High (draft generation is rapid)	Full and native
Elicitation-Based (End-to-End)	15–30+ years (estimated)	Very Low (dyadic bottleneck)	None (requires post-hoc conversion)
Elicitation (Language Documentation Phase Only)	6–18 months per language	High (targeted and time-limited)	Compatible (feeds into MTT pipeline)

5.2 Training Pipeline Bottleneck

Even if an organization committed to the elicitation model as its primary methodology, it would face an immediate and severe training pipeline bottleneck. Producing a facilitator with the requisite combination of biblical-language competency, fieldwork methodology, transcription skill, and cross-cultural relational capacity is a multi-year training investment. Graduate-level biblical-language training alone — sufficient proficiency in Koiné Greek and Biblical Hebrew — requires two to four years of serious academic study.

The mother-tongue translator (MTT) model offers an instructive contrast. Community members who are already native speakers of the target language are trained in translation principles, exegetical tools, and Paratext software — a training investment typically measured in weeks to months rather than years, because the deep linguistic competency already exists.

5.3 Geographic Reach

A significant proportion of the language groups currently identified as having active translation needs are spoken in regions characterized by difficult logistics: the highlands and island groups of Papua New Guinea (which alone contains approximately 840 languages, many with small speaker populations in remote mountainous communities), Sub-Saharan Africa's rural hinterlands, the Amazonian rainforest basin, and hill-tribe communities across South and Southeast Asia.

Geographic Context

Papua New Guinea alone contains approximately 840 distinct languages, representing one of the world's most concentrated areas of linguistic diversity. Many speaker communities number in the hundreds or low thousands, are accessible only by small aircraft or multi-day foot travel, and experience frequent infrastructure disruption. A decade-long elicitation project in such a context depends on a degree of sustained logistical stability that cannot be assumed.

In a traditional multi-member translation team using Paratext with cloud synchronization, interruptions are disruptive but recoverable: team members can continue working independently and synchronize when connectivity is restored. In a dyadic elicitation relationship, where the facilitator must be physically present with the speaker, such interruptions bring the entire project to a halt.

Section 6

Linguistic Accuracy Concerns

6.1 Elicitation Artifacts

One of the most well-established findings in descriptive linguistics is that language produced under elicitation conditions differs systematically from language produced in natural communicative contexts. When a speaker is asked to provide a target-language equivalent for a source-language word or phrase in an elicitation session, they engage a mode of metalinguistic reflection — a conscious deliberation about language rather than an immediate communicative act.

Research in metalinguistic awareness and bilingual language processing consistently demonstrates that this reflective mode produces language that is more formal, more consciously constructed, and more influenced by the structural features of the source language than natural speech would be. The result is a well-documented phenomenon known as translationese: unnatural, non-idiomatic literalisms of source-language syntax, and overly formal register that does not reflect how the community would naturally express the underlying concepts.

6.2 Register and Genre

The Bible is not a single genre. It is a library of literary forms spanning ancient Near Eastern legal code, Hebrew poetry of extraordinary sophistication (the Psalms, Song of Songs, Lamentations), prophetic oratory, wisdom literature, historical narrative, apocalyptic vision, Hellenistic biography (the Gospels), theological epistle, and eschatological poetry. Each of these genres demands different idioms in the target language — different levels of formality, different sentence structures, different vocabularies, different rhythmic patterns.

A target-language speaker participating in an elicitation session has no reliable way to calibrate idiom appropriately when presented with isolated source-language units stripped of their genre context. The likely result is a translation that defaults to a single, relatively neutral idiom across all genres — losing precisely the literary richness and variety that characterize the biblical text.

6.3 Semantic Accuracy and Back-Translation

The industry-standard back-translation procedure is one of the most important quality-assurance mechanisms in Bible translation practice. The elicitation model lacks a native workflow for this critical step.

Theological Risk

Semantic drift in key theological vocabulary — particularly terms central to soteriology (salvation, justification, reconciliation) and Christology — represents not merely a translation quality problem but a theological one. Inconsistent or inaccurate renderings of these terms across a long elicitation project could result in a translation that conveys a subtly or significantly different theological message from the source text, with potentially serious consequences for the communities that will rely on the translation as their primary access point to Scripture.

6.4 Discourse Coherence

Translation quality is not reducible to word-level accuracy. Biblical narratives build across chapters; Paul's theological arguments in Romans or Galatians develop across multiple paragraphs and must be internally consistent to be comprehensible; the Psalms return to recurring images and motifs that give them their emotional and theological resonance.

Elicitation sessions separated by days, weeks, or months are structurally ill-suited to maintaining discourse coherence. A speaker who rendered a key image in one way in session forty may not remember that choice in session one hundred and forty. The narrator's voice — the distinctive way in which a given speaker would naturally tell a story — tends to fragment across hundreds of isolated sessions in a way that would not occur if the speaker were given the opportunity to produce extended, contextually embedded discourse.

Section 7

Ethical Considerations

7.1 Informed Consent and Community Ownership

The question of who owns an elicitation-based translation is a foundational ethical issue with significant legal, theological, and missional dimensions. When an outside facilitator elicits linguistic material from a native speaker over many years and assembles it into a translation document, the lines of intellectual authorship are genuinely unclear.

The global Bible translation movement has, over the past two decades, moved decisively toward models that affirm community ownership of translated Scripture. The elicitation model, in which an outside facilitator is the primary architect of the translation project, risks replicating patterns that the field has consciously sought to move beyond: patterns associated with extractive colonial linguistics, in which linguistic data was gathered from communities without meaningful consent, compensation, or attribution.

7.2 Speaker Welfare and Exploitation Risk

Long-term elicitation relationships create power asymmetries that require careful ethical attention. In many minority-language communities, outside facilitators — particularly those representing well-funded Western organizations — carry significant social capital. Speakers may feel socially obligated to continue participation even when it is burdensome, and may not fully understand how their linguistic output will be used, published, or attributed.

The mother-tongue translator model offers a sharply contrasting ethical profile. In that model, community members are trained, employed with fair compensation, credited as the translators of their community's Scripture, and retained as the ongoing stewards of the translation. The power relationship is inverted: the outside consultant serves the community's translators, rather than the community's speaker serving the outside facilitator.

7.3 Data Sovereignty

Audio recordings and written transcriptions of elicited speech constitute a form of linguistic data that belongs, by any reasonable ethical standard, to the speaker and to the language community. Without a clear and explicit data governance framework — specifying who stores the recordings and transcriptions, who may access them, under what conditions they may be shared with third parties, and what happens to them if the project is discontinued — the community has no meaningful control over a significant corpus of their linguistic heritage.

This concern is heightened by the prospect of AI integration. Audio elicitation recordings that capture natural speech and extended discourse could constitute valuable training data for voice models and language technology applications. If communities are not explicitly informed of this potential use at the time of elicitation, their consent to the original recording cannot be interpreted as consent to these downstream uses.

7.4 Theological Accountability

Biblical translation carries theological stakes that distinguish it from other forms of linguistic documentation and translation. The established translation pipeline addresses this concern through multiple layers of accountability: trained exegetes in the translation team, consultant checkers with advanced biblical-language training, community theological review, and final publication approval by recognized ecclesiastical or missiological authorities.

The elicitation model's accountability chain is substantially less defined. The native speaker who supplies target-language renderings is a linguistic authority but is not necessarily a theological one, and may not have the biblical-language training to evaluate whether their rendering accurately conveys the source text's theological intent.

Section 8

Comparison to Paratext / PTXPrint Workflows

8.1 Paratext as the Global Standard

Paratext, jointly developed by SIL Global and the United Bible Societies, is currently used to translate the Bible in over 3,000 languages worldwide — making it the global standard platform for Bible translation work. It is not merely a word processor adapted for translation; it is a purpose-built environment that embeds translation best practices into the software architecture itself. Its USFM (Unified Standard Format Markers) specification, standardized in 2002, provides a universal markup language for Scripture that is compatible across all participating organizations.

Paratext's key features relevant to translation quality include: versification checking, spelling consistency checking, character inventory management, parallel reference checking, and a full consultant annotation and review workflow. These features are not optional enhancements — they are quality-assurance infrastructure that catches errors the human eye misses across projects of biblical scale.

8.2 PTXPrint and Publication Readiness

PTXPrint transforms Paratext projects into publication-ready PDFs formatted for minority-language print Scriptures — enabling translation teams in remote communities to produce professionally formatted Scripture without access to professional typesetters or commercial publishing infrastructure.

An elicitation-based translation that has not been structured within Paratext faces a substantial additional burden before reaching this point: the accumulated transcriptions must be converted to USFM, imported into Paratext, verified for structural integrity, and subjected to Paratext's automated checking suite before PTXPrint can process them.

8.3 Quality Assurance Gaps

Paratext's built-in automated checking tools represent accumulated translation wisdom encoded as software logic. Verse-completeness checks ensure that no verses have been inadvertently omitted. Punctuation consistency tools identify irregularities in quotation mark usage. Reference checking tools verify that cross-references point to existing content. Collectively, these tools catch hundreds of mechanical errors before the text ever reaches a human consultant — dramatically reducing the time and cost of consultant checking.

Elicitation transcriptions lack any equivalent automated checking infrastructure. Quality assurance in an elicitation-based project depends entirely on human review — a process that is both more expensive and less comprehensive than software-mediated checking at this scale.

8.4 LLM and Voice Model Integration

The integration of large language models and AI voice technologies into the Bible translation pipeline represents the most significant methodological development in the field since the introduction of Paratext itself. LLM-based tools trained on extensive biblical corpora and related-language data are demonstrating the ability to generate first-draft translations for language families where sufficient data exists — providing structured, Paratext-compatible draft text that mother-tongue translators and consultants can then evaluate, revise, and refine.

Strategic Observation

The elicitation model, because it produces unstructured transcriptions outside the Paratext ecosystem, cannot readily integrate with LLM-assisted drafting pipelines or AI voice model workflows. This places it outside the trajectory of the field's most promising acceleration strategies at precisely the moment when those strategies are beginning to demonstrate real-world impact.

Section 9

Implications for 7,000-Language Global Scripture Access

9.1 Opportunity Cost

Every organizational resource committed to the elicitation model as an end-to-end translation methodology represents a direct opportunity cost relative to more scalable approaches. A facilitator who spends twenty-five years eliciting a single full-Bible translation could, in the same period, have trained and mentored a dozen mother-tongue translation teams — producing, in aggregate, twelve completed translations rather than one.

At the current trajectory — without AI acceleration — the global translation movement has estimated that reaching all remaining language communities could take more than 150 additional years. Methodological choices that extend per-project timelines rather than compressing them make this already sobering projection worse.

9.2 Where Elicitation Has Genuine Value

Positive Applications

Language documentation phase: Before formal translation work begins in an under-documented language, elicitation is the primary tool for building the lexical, phonological, and grammatical foundation that the translation team will need. A robust language documentation phase significantly accelerates and improves the subsequent translation process.
Community checking sessions: Eliciting native speaker reactions to draft passages — asking whether a rendering sounds natural, whether it conveys the intended meaning, whether a particular word choice carries unintended connotations — is standard and valuable practice in the community checking phase.
Oral Scripture seed data: Elicitation of extended natural narrative can produce valuable training data for AI voice models aimed at generating audio Scripture for oral-primary communities.
Theological vocabulary development: In the early stages of a translation project, elicitation can be used to explore how a community would naturally express key theological concepts — testing candidate terms for "covenant," "grace," "sacrifice," and "kingdom" against native-speaker intuitions.

9.3 Recommended Hybrid Framework

The constructive conclusion of this review is a recommended hybrid framework that preserves the genuine strengths of the elicitation approach while embedding them within a workflow that is feasible, scalable, technologically integrated, and ethically sound.

Pre-translation elicitation and language documentation (Months 1–18): Deploy skilled field linguists to conduct systematic language documentation using elicitation methodology. Build a community lexicon, establish phonology and orthography, document grammatical structures, and record extended natural discourse.
Mother-tongue translator recruitment and training (Months 12–24): Identify and recruit community members to serve as trained mother-tongue translators. Provide training in translation principles, exegetical tools, and Paratext software.
LLM-assisted first-draft generation (where applicable): Where sufficient related-language data exists, deploy LLM-based drafting tools to generate a Paratext-compatible first draft. Community translators evaluate this draft using their linguistic intuitions.
Paratext-integrated drafting and revision: All translation work is conducted within Paratext, leveraging its versification checking, spelling consistency, cross-reference tools, and consultant annotation workflow. PTXPrint is used for intermediate print-format review and final publication preparation.
Community checking with targeted elicitation: Draft passages are shared with extended community members through facilitated community checking sessions. Elicitation methodology is used systematically in this phase, where it delivers its highest value.
Audio Scripture production via AI voice models: Completed or draft passages are processed through AI voice models to produce audio Scripture for community distribution, with community translators serving as voice quality reviewers.

Framework Outcome

This hybrid framework honors the genuine values that motivate the elicitation model — native-speaker fidelity, community centering, cultural authenticity — while achieving them through mechanisms that are feasible, scalable, technologically integrated, ethically sound, and compatible with the most promising acceleration strategies available to the global translation movement today.

Section 10

Conclusion

The elicitation-based Bible translation model is, at its best, an expression of a deeply important and correct instinct: that the translation of Scripture into a language must be anchored in the living voice of that language's community, and that no amount of academic linguistic study by an outsider can fully substitute for the native speaker's intuitive knowledge of what their language can do and how it naturally moves. This instinct has driven some of the most important methodological advances in Bible translation over the past century, and it deserves to be affirmed without qualification.

What this review argues is that the elicitation model, as a primary end-to-end translation methodology, fails to translate this correct instinct into a feasible, scalable, or adequately accountable practice. The arithmetic of the Bible's word count, combined with the structural dependencies and logistical demands of sustained elicitation, produces timelines and throughput figures that cannot address the global Scripture access challenge at the scale and urgency it requires. The linguistic artifacts endemic to elicitation conditions — translationese, idiom flattening, discourse fragmentation — undermine the very native-speaker fidelity that the elicitation model aspires to achieve. The ethical architecture leaves unresolved questions about ownership, consent, and power that the field has worked hard to address through community-centered methodologies. And the model's incompatibility with the Paratext/PTXPrint/AI ecosystem places it outside the most productive current trajectory of the translation movement.

None of these concerns require us to dismiss elicitation as a tool. On the contrary, this report has argued that elicitation deserves a prominent and valued place in the translation workflow — as the engine of pre-translation language documentation, as a quality-assurance mechanism in community checking, and as a source of oral data for AI voice model development. The argument is specifically against elicitation as the primary production mechanism for a complete translation — a role for which its structural characteristics make it poorly suited.

The global Scripture access challenge stands at a remarkable inflection point. The convergence of mother-tongue translator training networks, Paratext's established infrastructure, and emerging AI-assisted drafting and voice technologies creates an unprecedented opportunity to accelerate Scripture access for the world's remaining language communities within a timeframe that was unimaginable a generation ago. The millions of people in language communities still waiting for Scripture deserve our most rigorous thinking, our most scalable strategies, and our most faithful stewardship of the resources entrusted to the global translation movement for exactly this purpose.

Appendix A

Glossary

Back-Translation: A quality-assurance technique in which a draft translation is rendered back into a language accessible to the translation consultant (typically English or French) by a speaker who was not involved in the original translation. Discrepancies between the back-translation and the source text reveal semantic drift, mistranslations, and unintended theological shifts.
Community Checking: A phase in the translation workflow in which draft passages are reviewed by extended community members — beyond the core translation team — for naturalness, comprehension, and cultural appropriateness. Community checking validates that the translation communicates effectively to its intended audience.
Consultant Checker: A highly trained translation specialist — typically holding advanced degrees in biblical languages and translation studies — who reviews a complete translation project for theological accuracy, linguistic quality, and adherence to translation principles. Consultant checking is the final quality-assurance step before a translation is approved for publication.
Elicitation: A linguistic fieldwork methodology in which a researcher or facilitator presents stimuli (words, concepts, sentences, or source-language content) to a native speaker and records their verbal responses. Used in descriptive linguistics for language documentation, lexicon building, and grammatical analysis. In the context of this review, it refers specifically to the use of elicitation as a primary mechanism for generating target-language Bible translation drafts.
Large Language Model (LLM): A class of artificial intelligence system trained on large corpora of text to generate, translate, summarize, and analyze natural language. In Bible translation contexts, LLMs are being explored as tools for generating first-draft translations based on related-language data and biblical corpora, which mother-tongue translators then evaluate and refine.
Mother-Tongue Translator (MTT): A native speaker of a target language who is trained in translation principles, biblical exegesis, and translation software (typically Paratext) and serves as the primary drafter of a translation into their own language. The MTT model is the dominant community-centered approach in contemporary Bible translation practice.
Oral Scripture: Audio recordings or AI-generated voice renderings of biblical text in a target language, designed for communities in which oral communication is primary and literacy rates may be low. Oral Scripture may include dramatized readings, natural voice recordings, or AI voice model outputs.
Paratext: The global-standard software platform for Bible translation, jointly developed by SIL Global and the United Bible Societies. Paratext supports the full translation lifecycle — drafting, checking, revision, consultant review, and publication preparation — using the USFM markup standard. Currently used in translation projects for over 3,000 languages worldwide.
PTXPrint: A companion tool to Paratext that converts Paratext USFM projects into publication-ready PDFs formatted for minority-language print Scriptures. PTXPrint enables translation teams in remote communities to produce professionally formatted print Scripture without access to professional typesetters or commercial publishing infrastructure.
USFM (Unified Standard Format Markers): The standardized markup language used by Paratext to structure Scripture texts for Bible translation projects. USFM markers identify the structural elements of Scripture — books, chapters, verses, headings, footnotes, and other components — in a format that is portable, checkable, and compatible with the full range of Paratext's quality-assurance and publication tools. Standardized by the United Bible Societies in 2002.