A difficult translation often begins as a difficult source. The translator receives three “final” files, an acronym list from last year, screenshots that contain essential text, and sentences whose meaning depends on a meeting no one documented. Fluency cannot repair decisions the source team never made.
That upstream problem now has a timely reference point. The International Organization for Standardization published ISO 18968:2026, Translation-oriented writing—Text production and text evaluation on August 27, 2026. Its public abstract describes guidance for authoring, editing, evaluating, and handing over specialized source-language content intended for translation. The intended users include authors, editors, translators, language-service providers, and tool makers. Fiction, journalism, and advertising are outside its stated scope.
For publishers, medical writers, learning teams, and compliance groups, the practical question is not whether every project needs a new certification exercise. It is whether the source gives a translator enough stable, explicit information to preserve meaning across languages—and whether the production system keeps that information attached to the content.
The short answer
Freeze one authoritative source, resolve ambiguity before handoff, govern terms and non-translatable strings, separate meaning from presentation, deliver context with the file, and test the result with target-language reviewers. Translation readiness is a source-quality gate, not a last-minute vendor task.
Why source readiness matters now
ISO 18968 is the first edition of this standard, and NISO highlighted it in its September 2026 standards roundup. The public table of contents shows the breadth of the problem: formatting; terminology; grammar, syntax, and style; presentation; file formats; layout; reference material; terminology aids; locales; templates; lists and glossaries; source review; and evaluation methods. In other words, translation-oriented writing is not a synonym for short sentences. It connects editorial decisions to the handover and quality-control system around them.
That connection matters even when machine translation or generative tools are part of the workflow. Automation can process a string quickly, but it still needs to know whether “charge” means a fee, an electrical state, a criminal accusation, or an instruction to power a device. It needs a boundary around what should be translated, context for variables and labels, and a current term choice. Faster translation increases the value of preventing ambiguity upstream because the same unresolved source can propagate across many languages at once.
Translation-ready is more than plain language
Clear writing is part of the answer. The ISO 24495-1 plain-language standard applies broadly to text, including technical writing and legislative drafting, and is designed for most written languages. The EU Translation Centre likewise advises authors to make texts clear, concise, and consistent because translators are readers as well as writers.
But a source can be readable and still be unsafe to translate. A friendly instruction such as “do it as before” is concise but context-poor. A neat table may repeat “normal” without saying whether it describes a laboratory range, device state, or user permission. A familiar abbreviation may have several domain meanings. Translation readiness asks an additional question: can someone outside the source team recover the intended meaning, scope, relationships, and constraints without guessing?
Plain language also has to survive the journey. Canada’s accessible plain-language standard says that multilingual communication should be in plain language in each language and recommends working with translators familiar with plain language in the target language. It also calls for enough preparation context about terms, names, numbers, abbreviations, metaphors, and cultural references. That is a useful warning against treating the clearest English version as proof that every translated version will be clear.
Six translation-readiness gates for specialized content
Name the audience, purpose, locale, and source of truth
Record who will use the content, what action or decision it supports, which language and regional variants are required, and which file is authoritative. Include the owner, approval state, version, and effective date. If different audiences need materially different explanations, create intentional variants rather than asking one source to serve everyone. Stop parallel “final” files before translation begins.
Make meaning explicit where readers could reasonably diverge
Resolve pronouns with unclear antecedents, compressed noun strings, vague time references, unexplained comparisons, and instructions that depend on visual position alone. Expand an acronym on first use and state what a placeholder represents. Review headings, table cells, buttons, alt text, warnings, footnotes, and quiz feedback in context—not as isolated strings. The goal is not mechanical simplicity; it is controlled meaning.
Govern terminology, names, numbers, and reusable claims
Give each consequential term an approved source form, definition, permitted variants, prohibited alternatives, subject domain, and owner. Flag product names, citations, code, and other strings that must remain unchanged. Define rules for dates, units, decimal separators, dosage expressions, and legal or regulatory phrases. When content is reused, link the instance to the governed source instead of copying text into an unmanaged glossary.
Keep structure and localization instructions machine-readable
Do not flatten essential words into screenshots or encode meaning only through color, position, or manual formatting. Preserve headings, lists, table relationships, language tags, and text alternatives. For HTML and XML workflows, the W3C Internationalization Tag Set 2.0 defines interoperable data categories for translation flags, localization notes, terminology, language, directionality, space preservation, and quality issues. The exact implementation varies, but the principle is stable: downstream systems should not have to infer what the source could declare.
Hand over context, not just a file
Package the approved source with the audience brief, target locales, glossary, style guidance, cited references, screenshots or prototypes, character and layout constraints, non-translatable items, and a named route for questions. Explain variables and show complete sentences where software exposes only fragments. Include a change log so translators can distinguish new meaning from editorial cleanup and avoid reworking unchanged text.
Evaluate the delivered experience in each target language
Check semantic accuracy, terminology, completeness, local conventions, links, layout, reading order, and the real task a reader must complete. Use a qualified target-language reviewer and include representative users when comprehension or accessibility carries meaningful risk. Record defects by source cause as well as target symptom. A repeated mistranslation may reveal one ambiguous source sentence that should be fixed for every locale.
A practical pilot scorecard
Start with one bounded, consequential asset: a product instruction, clinical-training module, policy, help article, or submission component due for revision. Choose two target languages with different structural or layout demands if those languages are already in scope. Do not add languages merely to make the pilot look ambitious.
Score the source before handoff using observable checks: one approved version exists; audience and locale are named; critical terms have owners; references open; abbreviations are defined; visuals have translatable text alternatives; strings identify variables and constraints; and the handover includes context. Then record translator questions, source changes requested after handoff, target-language defects, turnaround time, and repeated issues. Counts are diagnostic, not a universal maturity score. A source with fewer words can still generate more consequential questions.
After the pilot, convert the recurring causes into editorial checks, template fields, or structured metadata. Keep the translator’s questions: they are evidence about where the source model hides meaning. Also preserve identifiers for citations and reusable passages. The same discipline used to keep citation status reviewable before publication helps a multilingual team prove which source and references informed each version.
What content teams should do now
Do not begin by rewriting the entire style guide around a standard’s title. Obtain and assess the full ISO 18968 text if formal alignment matters to your organization; the public abstract and contents are not a substitute for the standard. Meanwhile, convene one author, editor, translator or localization lead, and content-system owner. Walk a recently translated asset backward from target defects to source decisions and handover gaps.
Pick the three source defects that caused the most risk or rework. Assign each one a prevention point: an authoring prompt, editorial checklist, terminology field, schema rule, preflight test, or handover requirement. Run the same scorecard on the next version. The useful shift is small but fundamental: translation quality no longer begins when a vendor receives the file. It begins when the source team decides how meaning will be written, structured, and kept.
For editorial and publishing teams
Keep source structure intact before content travels.
Superscriptify helps teams clean citation-heavy Word and PowerPoint files, align references, and prepare structured content for dependable review and reuse.
Explore workflows for publishersSources and further reading
- ISO 18968:2026: Translation-oriented writing—Text production and text evaluation, published August 27, 2026
- NISO: New and Emerging Specifications and Standards, September 2026
- ISO 24495-1:2023: Plain language—Governing principles and guidelines
- Translation Centre for the Bodies of the EU: Clear writing and writing for translation
- Accessibility Standards Canada: CAN-ASC-3.1:2025 Plain Language, multilingual communication and evaluation provisions
- W3C Recommendation: Internationalization Tag Set 2.0
This article provides general editorial and localization information, not legal, contractual, certification, or standards-conformance advice. Consult the complete standards, applicable requirements, and qualified language professionals for decisions about a specific product, market, or regulated document.