Language Audit

The purpose is to test whether language use is traceable, bounded, and answerable to the object, grounds, and action for which it is used. The audit may be applied to individual words, definitions, questions, claims, quotations, reports, contracts, decisions, research writing, journalism, teaching, forms, classifications, translations, and public or private communication.

Language Audit does not decide in advance that a formulation is true, false, manipulative, discriminatory, or capturing. It identifies what the expression refers to, what function it performs, what it presupposes, what alternatives it opens or closes, and what conclusion the recipient is being asked to draw.

A Language Audit becomes imprecise when word, meaning, reference, definition, observation, interpretation, and conclusion are treated as the same link. The distinctions below are functional and may overlap in actual language use, but the function must be identified when it carries a finding.

Language may also be action. A promise, decision, contract formation, or naming act may establish institutional relations where the required role, procedure, and conditions are present. This does not mean that wording alone can make an external claim true or undo what actually occurred.

The mandate must identify the concrete wording, version, author or speaker, recipient, context, and claim or act being examined. It is not sufficient to ask whether language is dangerous, false, or manipulative without delimiting the formulation and effect under audit.

The same wording may function differently in a textbook, contract, police interview, joke, poem, or administrative decision. The audit must therefore record genre, mandate, legal or institutional status, time, and relevant recipient conditions.

Linguistic claims should, as far as possible, be traced from object and source through selection, terminology, paraphrase, quotation, editing, translation, headline, and publication to the wording used in the audit. A traceable chain does not automatically yield a true interpretation, but a broken chain limits what may be attributed to the source or to actuality.

When expressions move among disciplines, institutions, or languages, meaning, force, and legal function may change. Such transitions must be recorded rather than concealed by the reuse of the same word or an apparent equivalent.

• Record the earliest available source and distinguish primary text from report, paraphrase, and commentary.
• Document omissions, cuts, compilation, headline, lead, and other editorial frames.
• Compare quotation with source and record changes to grammar, pronouns, modality, or tense.
• Document definitions and whether they are lexical, stipulative, operational, legal, or domain-specific.
• Record whether a term changes meaning between premise and conclusion or among participants.
• For translation, record source text, target text, translator, register, ambiguity, and defensible alternatives.
• Document inaccessible links and how they limit the finding.

The questions must be adapted to the text, speech act, and consequence of use. They are organised to make reference, premises, and assignment visible, not to create automatic suspicion toward rhetoric or ordinary linguistic variation.

Linguistic form may influence what the recipient treats as given, who appears to bear responsibility, how certain a claim seems, and what alternatives appear possible. This is not in itself an error; every formulation selects perspective and structure. The audit asks whether the choice is relevant, visible, and proportionate to the grounds.

Intention may be relevant, but the finding should first be tied to documented wording, function, context, and effect. An adverse effect is not automatically proof of deliberate manipulation.

Language capture is not the same as strong wording, technical terminology, metaphor, identity language, legal definition, or disagreement over words. The risk pattern arises when a linguistic frame begins to determine in advance what may count as the object, grounds, or possible correction.

F7 requires more than imprecision or a contestable definition. The audit must document the specific frame, what it replaces, how it gains governing priority, what relevant correction is resisted, and what consequence follows.

The finding must be tied to the specific wording, source, context, and consequence examined. It must distinguish semantic precision, reference, documentation, interpretation, speech act, and possible effect.

Language Audit must not make a person identical with one expression or attribute hidden intention without grounds. Where intention cannot be documented, the finding must concern wording and function.

Correction must be tied to the bounded finding. The aim is better reference, clearer distinctions, visible premises, and more auditable language use—not standardising every style or eliminating legitimate professional, artistic, and social variation.

Where wording forms part of a decision, agreement, right, safety process, or public record, responsibility, version, publication, and effect of the amendment must be recorded. A silent wording change may be insufficient where the earlier wording has already produced consequences.

• Clarify referent, scope, time, place, actor, and degree of force.
• Distinguish observation, quotation, interpretation, explanation, evaluation, and conclusion visibly or linguistically.
• Make stipulative, operational, legal, and technical definitions visible and bounded.
• Replace presuppositional questions with open questions where the premise is not established.
• Restore actor and action chain where passive voice or nominalisation unreasonably obscures responsibility.
• Correct quotations, restore relevant context, and mark paraphrase as paraphrase.
• For translation, document alternative formulations where they affect right, responsibility, or conclusion.
• Publish a correction, change log, or new version where the original wording was decision-bearing.
• Assign responsibility, deadline, review, and possible rollback under the Correction Protocol.

Language Audit cannot by itself determine hidden intention, deliberate lying, psychological motive, discriminatory effect, legal responsibility, or the final meaning of every text. Such questions may require historical, social, legal, statistical, or domain-specific material.

Natural language is contextual, polysemous, and changing. Precision does not require every word to have one eternal meaning. The requirement is that the meaning and function carrying the relevant claim can be made visible and tested.

The audit record must allow another reader to locate the wording, source, referent, definitions, alternative readings, assessment, and path to the finding. Sensitive or regulated information must be handled under relevant requirements.

After correction, the audit must test whether the new wording actually makes reference, responsibility, and limitation clearer, and whether the change has created new misunderstandings or disproportionate consequences.

• Record text, version, language, date, source, author or speaker, recipient, and genre.
• Record principal claim, presuppositions, definitions, categories, and decisive grammatical attributions.
• Attach quotation context, alternative paraphrases, and relevant translation alternatives.
• Document counter-material, comments from affected parties, and changes following contradiction.
• Classify finding F1–F7 and confidence separately.
• Connect correction C0–C6 to responsibility, deadline, version, and review point.
• Record what remains unresolved and what new material may change the finding.