| name | languaging |
| description | Use when writing or rewriting ANY content for GeoVerity (homepage, services, insights, contact pages) before generating text - establishes mandatory register stratification with plain professional language (B2-C1) for service pages targeting administrators/project managers, and academic register (C1-C2) for Insights journal posts only |
Languaging: GeoVerity Register Stratification Framework
MANDATORY: Examine this skill before ALL writing and rewriting attempts.
Overview
Register stratification framework for GeoVerity content surfaces, grounded in systemic functional linguistics (Halliday) and audience design theory (Bell).
Core Principle: Register selection must be audience-driven and genre-appropriate, not uniformly applied.
Critical Rule: Service pages (Homepage, Services Hub, Pillar/Spoke pages) use plain professional language (B2-C1 CEFR) for administrators and project managers. Academic register (C1-C2) is ONLY for Insights Journal posts.
What "Languaging" Means in This Skill (Do Not Skip)
Operational Definition: Languaging is the active process of using language to do things — to build meaning with an audience, to frame expertise, to guide decisions, and to position yourself socially. It treats language as action, not just text. Under this view (Swain; Halliday; Vygotsky), wording is never neutral: every lexical choice, sentence shape, and tone choice performs work. It establishes authority, clarifies a problem, signals who the intended reader is, and marks who "belongs" in the conversation.
Why This Matters for GeoVerity: For GeoVerity, languaging is how we align voice, cognitive load, and social positioning with each audience segment. We do not just "write copy." We actively produce a social relationship with administrators, project managers, or researchers through register. That's why a Services page and an Insights Post cannot share the same style — because they are performing different social actions.
1. Languaging as Action (What are we doing with this text?)
In practice: Every page has a job. The language on that page must directly serve that job.
Homepage / Services Page Job
Goal: Reduce friction, build trust fast, and move a decision-maker toward contact.
Example (GOOD for Services):
"We evaluate your AI models across 120+ languages and tell you where they're risky."
Action performed: Reassures an administrator that we solve a concrete operational problem.
Example (BAD for Services):
"The epistemic instability introduced by multilingual generative systems demands institutional recalibration."
Action performed: Signals academic debate, not operational support. Violates B2-C1 service register.
Insights Post Job
Goal: Contribute to an expert conversation and shape thought leadership.
Example (GOOD for Insights Post):
"While multilingual LLMs expand institutional reach, they also destabilize traditional assumptions about authorship, assessment, and evidence. Institutions cannot outsource those judgments to detectors alone."
Action performed: Argues, theorizes, reframes policy. This belongs in C1-C2 Insights, not Services.
Enforcement Hook: Ask: "What is this page trying to cause in the reader?"
- If the answer is "book a call / request support" → Use Services register (B2-C1)
- If the answer is "rethink a policy or framework" → Use Insights Post register (C1-C2)
2. Languaging as Meaning-Making (Are we co-constructing understanding?)
Languaging assumes meaning is negotiated with the audience, not dumped on them. We must "speak in the reader's world first," then layer complexity only if allowed by that surface's register.
Services Page Pattern (B2-C1)
Structure:
- State the reader's problem in their terms
- Name what we do
- State the outcome
Example:
"Your faculty are worried about AI-written student work. We build integrity frameworks that acknowledge where AI is actually being used, instead of pretending it isn't. That lets you update policy without starting an arms race around detection tools."
Why this is compliant:
- Plain professional language
- Short clauses, main-clause first
- Directly names their world ("your faculty," "AI-written student work")
Insights Post Pattern (C1-C2)
We are allowed to interrogate assumptions and use theoretical framing.
Example:
"Faculty anxiety around AI text is not only about plagiarism; it reflects a deeper loss of epistemic trust in the act of submitting written work as evidence of learning."
Why this is compliant:
- Concept-building, not service positioning
- High lexical density ("epistemic trust," "evidence of learning")
- Acceptable only in Insights
Enforcement Hook:
- If the copy frames a shared operational scenario and walks the reader toward an outcome → Service register
- If the copy problematizes concepts and reframes the discourse → Insights register
3. Languaging as Identity Work (Who are we telling the reader they are?)
Languaging marks identity. The same company can position the reader as a decision-maker ("You set policy") or as a peer in a research community ("We, as a field, need to reconsider…"). That shift in identity is not allowed to drift across surfaces.
Services Page Identity Move
Example:
"You are the person responsible for protecting academic integrity at your institution. We give you a defensible policy you can stand behind when you're challenged."
Reader position: Accountable leader who needs immediate, defensible, auditable solutions. Register: B2-C1 (correct for administrators/PMs)
Insights Post Identity Move
Example:
"Our current models of authorship assume linguistic stability, but multilingual fine-tuning has already eroded that assumption."
Reader position: Co-analyst of system-level change. Register: C1-C2 (scholarly stance, must not leak into Services)
Enforcement Hook: Ask: "Am I speaking to them as an operations owner, or as a fellow theorist?"
- Operations owner → Services register (B2-C1)
- Fellow theorist → Insights register only (C1-C2)
4. Languaging as Cognitive Tool (Are we letting the reader think through the problem?)
In applied linguistics, "languaging" refers to using language to think through complex problems (externalized reasoning, self-explanation). That maps to two different behaviors in GeoVerity:
On a Services Page: We do the reasoning for them so they don't have to
Example (GOOD for Services):
"AI detectors generate false accusations. We train faculty to assess process, not just output, so you reduce conflict and keep trust in the classroom."
Why compliant: We surface the reasoning as a finished, usable policy move.
Anti-example (BAD for Services):
"Because authorship verification remains epistemically unstable in multilingual assessment contexts, universities must reconceptualize evaluation itself."
Why wrong: This forces them to theorize. That belongs in an Insights Post.
In an Insights Post: We invite them into the reasoning process
Example (GOOD for Insights):
"If we admit that authorship can no longer be verified by output inspection alone, then assessment must shift toward supervised process evidence — drafting traces, revision logs, oral defenses. That shift has legal and labor implications."
Enforcement Hook:
- If you're making them think through institutional redesign → Insights mode (C1-C2)
- If you're promising them an implementable fix → Services mode (B2-C1)
5. Languaging as Multimodal Meaning (How do we handle examples, definitions, and jargon density per surface?)
Languaging is not just about words; it's about how explanation, definition, and framing are delivered for the specific audience.
Services (B2-C1)
We are allowed to use technical terms like "LLM fine-tuning," "governance," "integrity review," but we immediately ground them in effect.
Example:
"We audit your LLM fine-tuning pipeline to surface bias before deployment."
We do not unpack theory unless it supports a decision.
Insights Posts (C1-C2)
We are allowed (and expected) to introduce theoretical constructs without immediate operationalization.
Example:
"Institutional trust in assessment cannot survive if epistemic warrant is outsourced to probabilistic detectors."
This matches the existing register table on lexical density, passive voice tolerance, hedging, and nominalization: those aren't just style preferences, they are different languaging acts.
Enforcement Hook:
- If you define a term in plain language and tie it to an outcome → Services
- If you elaborate a term to situate it in a field-level debate → Insights
When to Use
Triggering Conditions - Use this skill when:
- ✅ Writing homepage hero copy
- ✅ Creating service descriptions (pillar/spoke pages)
- ✅ Drafting Insights journal posts
- ✅ Writing contact page microcopy
- ✅ Reviewing or editing ANY user-facing text
- ✅ Translating English content to Spanish (maintain register parity)
- ✅ About to use subordinate-initial clauses in service copy
- ✅ About to write "Given X, Y..." or "While X, Y..." in marketing pages
Symptoms You Need This Skill:
- 🚨 Service page reads like academic paper
- 🚨 Using complex subordination for administrators
- 🚨 Starting sentences with "Because...", "While...", "Given that..." on service pages
- 🚨 Heavy nominalization ("the verification of data quality via methodological frameworks")
- 🚨 Passive voice >15% on non-Insights pages
- 🚨 Graduate-level vocabulary on homepage/services pages
- 🚨 Jargon without definitions for non-specialist audiences
When NOT to use:
- ❌ Internal documentation (not user-facing)
- ❌ Code comments or technical specs
- ❌ Git commit messages
Quick Reference: Register by Content Surface
| Surface | CEFR Level | Avg Sentence Length | Subordinate-Initial | Nominalization | Passive Voice | Target Audience |
|---|---|---|---|---|---|---|
| Homepage | B1-B2 | 12-18 words | ❌ Never | Minimal | <10% | All segments |
| Services Hub/Pillar/Spoke | B2-C1 | 15-25 words | ⚠️ Rare | Low-Moderate | <15% | Administrators, PMs |
| Insights Hub | B2-C1 | 18-25 words | ⚠️ Rare | Moderate | <20% | All segments (scannable) |
| Insights Posts | C1-C2 | 20-35 words | ✅ Common | High | 30-40% | Researchers, scholars |
| Contact Page | A2-B1 | 8-15 words | ❌ Never | None | 0% | All segments (max clarity) |
Target Audience Profiles
Segment 1: Higher Education Administrators
Roles: Deans, Associate Provosts, Program Directors, IRB Chairs
Language Expectations:
- Standard Academic English WITHOUT technical AI/NLP jargon
- Active voice, SVO order, main clause initial
- Concrete examples over abstract theorization
- Problem → Solution → Outcome structure
- Avoid: Dense nominalization, subordinate-clause-initial sentences, discipline-specific jargon
Example:
- ✅ "GeoVerity helps institutions maintain epistemic integrity."
- ❌ "Given the epistemological challenges posed by generative AI in pedagogical contexts, institutions require..."
Segment 2: Enterprise ML/AI Project Managers
Roles: ML Product Managers, Data Science Team Leads, AI Ethics Officers
Language Expectations:
- Industry-standard terminology (LLM, RAG, fine-tuning) without excessive academic framing
- Actionable language ("Deploy," "Configure," "Evaluate")
- Quantitative precision (numbers, metrics, benchmarks)
- Avoid: Theoretical background without application, academic citation styles
Example:
- ✅ "We evaluate LLM performance across 120+ languages."
- ❌ "The phenomenological dimensions of LLM-generated text in cross-linguistic contexts..."
Segment 3: Academic Researchers & Thought Leaders
Roles: Faculty researchers, graduate students, policy scholars
Language Expectations (Insights Posts ONLY):
- Discipline-specific terminology, theoretical frameworks, citations
- Complex syntax with embedding, hedging, epistemic modality
- Argumentation structure (claim → evidence → warrant → counterargument)
- Expected: Complex syntax, nominalization, disciplinary vocabulary
Register Specifications by Content Surface
1. Homepage (/ and /es/)
Register: B1-B2 (Intermediate-Upper Intermediate) Genre: Promotional landing page Tenor: Professional-to-peer (B2B marketing)
| Feature | Specification | Example |
|---|---|---|
| Lexical Density | 40-50% (conversational-professional) | "GeoVerity provides verifiable AI training data across 120+ languages" |
| Sentence Length | 12-18 words average | Short, punchy sentences |
| Syntactic Complexity | Simple + compound (minimal subordination) | "We verify data quality. You build trustworthy AI." |
| Clause Structure | Main clause initial, SVO order | "GeoVerity helps institutions maintain epistemic integrity" ✅ "Epistemic integrity, which institutions must maintain, is supported by GeoVerity" ❌ |
| Voice | Active voice (>90%) | "We verify" not "Data is verified" |
| Nominalization | Minimal (prefer verbs) | "We verify data" ✅ "Data verification processes" ⚠️ |
| Jargon Tolerance | Low (define technical terms) | "AI training data (the text, images, and code used to teach AI systems)" |
Prohibited Structures:
- ❌ Subordinate clause initial: "Because AI systems require verified data, GeoVerity..."
- ❌ Heavy nominalization: "The verification of data quality via methodological frameworks..."
- ❌ Passive + abstract agent: "Data quality is ensured through processes..."
- ❌ Academic hedging: "Our services arguably contribute to..."
Approved Structures:
- ✅ Main clause initial: "GeoVerity verifies AI training data across 120+ languages."
- ✅ Active voice + concrete agent: "Our linguists verify data quality."
- ✅ Parallel structure: "Verify data. Build trust. Deploy confidently."
2. Services Hub & Service Pillar Pages
Register: B2-C1 (Upper Intermediate-Advanced) Genre: Service description (informational-promotional) Tenor: Professional consultant-to-client
| Feature | Specification | Example |
|---|---|---|
| Lexical Density | 50-60% (professional) | "Our multilingual data infrastructure supports 120+ languages with verified native-speaker annotations" |
| Sentence Length | 15-22 words average | Moderate complexity |
| Syntactic Complexity | Compound + moderate subordination | "We verify data quality so you can deploy models confidently" |
| Clause Structure | Main clause initial, purpose clauses acceptable | "GeoVerity provides verified training data [main] to ensure model trustworthiness [purpose]" ✅ |
| Voice | Active voice (>85%) | "Our team evaluates models" not "Models are evaluated" |
| Nominalization | Low-moderate (only for established terms) | "model evaluation" ✅, "the evaluation of model performance metrics" ❌ |
| Jargon Tolerance | Moderate (industry-standard terms OK) | "LLM fine-tuning" ✅, "decontextualized lemma frequency distributions" ❌ |
Problem-Solution-Outcome Structure:
1. **The Problem** (1-2 paragraphs, B2 register)
- State the challenge administrators/PMs face
- Use concrete scenarios, not abstract theorization
- "Graduate programs face declining trust in student work due to undetectable AI authorship."
2. **Our Approach** (2-3 paragraphs, B2-C1 register)
- Describe GeoVerity's solution
- Use active voice, process verbs
- "We partner with institutions to establish epistemic integrity frameworks."
3. **What We Offer** (Bulleted list, B2 register)
- Service deliverables in scannable format
- "✓ Faculty training on AI detection limitations"
4. **Why This Matters** (1 paragraph, B2 register)
- Value proposition, outcomes-focused
- "Institutions maintain accreditation standards while adapting to AI realities."
Prohibited Structures:
- ❌ Academic subordination: "Given the epistemological challenges posed by generative AI in pedagogical contexts, institutions require..."
- ❌ Excessive nominalization: "The implementation of verification processes through methodological rigor..."
- ❌ Passive + vague agent: "Data quality is ensured through processes conducted by teams..."
Approved Structures:
- ✅ Problem-first: "Graduate programs struggle with AI detection. We provide training on epistemic integrity frameworks."
- ✅ Process verbs: "We train faculty, evaluate models, and verify data quality."
- ✅ Concrete outcomes: "Institutions maintain accreditation while adopting AI tools responsibly."
3. Service Spoke Pages
Register: B2-C1 Genre: Detailed service specification Tenor: Consultant-to-informed-client
| Feature | Specification |
|---|---|
| Lexical Density | 55-65% |
| Sentence Length | 18-25 words average |
| Syntactic Complexity | Compound-complex (controlled subordination) |
| Voice | Active voice (>80%) |
| Nominalization | Moderate (technical terms) |
| Jargon Tolerance | Moderate-high (domain-specific) |
Feature-Benefit Structure:
**Key Features**
- Feature 1: [What it is] → [Why it matters]
- Feature 2: [What it is] → [Why it matters]
**Who This Serves**
- Administrators responsible for [X]
- Teams managing [Y]
4. Insights Hub & Category Pages
Register: B2-C1 (Hub), C1-C2 (Posts) Genre: Thought leadership portal Tenor: Professional-to-professional
CRITICAL: Hub uses B2-C1 (scannable), Posts use C1-C2 (scholarly)
Hub Page Register:
- Scannable post previews
- Category descriptions in B2-C1 register
- Post titles in plain language (avoid jargon-heavy titles)
- "Exploring graduate education, epistemic responsibility, and AI use in academia." ✅
- NOT: "Investigating the phenomenological dimensions of generative AI's impact on epistemic warrant in pedagogical praxis." ❌
5. Insights Journal Posts (Individual Articles)
Register: C1-C2 (Advanced-Proficient) Genre: Scholarly argumentation / thought leadership essay Target Audience: Academic Researchers & Thought Leaders ONLY
| Feature | Specification | Example |
|---|---|---|
| Lexical Density | 65-75% (academic prose) | "The epistemic collapse induced by LLM-generated text in graduate pedagogy necessitates institutional recalibration of authorship verification frameworks." |
| Sentence Length | 20-35 words average | Complex ideas require complex syntax |
| Syntactic Complexity | High (embedding, subordination, nominalization) | "While detection tools claim accuracy [concessive], empirical studies reveal failure rates exceeding 40% [main], suggesting institutions must adopt alternative frameworks [result]." |
| Clause Structure | Subordination acceptable (argument-driven) | Thematic progression, given-new structure |
| Voice | Mixed (passive acceptable for academic hedging) | "It has been argued..." "The data suggest..." |
| Nominalization | High (disciplinary norms) | "epistemic warrant" "authorship verification" "institutional recalibration" |
| Jargon Tolerance | High (discipline-specific terminology) | "phenomenological," "deontological," "hermeneutic," "corpus-driven," "fine-tuning," "RLHF" |
| Hedging | High (epistemic modality) | "may," "suggests," "arguably," "potentially," "appears to" |
Approved Structures (Insights Posts ONLY):
- ✅ Subordinate clause initial: "While AI detection tools proliferate, their empirical accuracy remains contested."
- ✅ Heavy nominalization: "The institutionalization of epistemic integrity frameworks requires faculty buy-in."
- ✅ Passive + hedging: "It has been argued that generative AI undermines traditional authorship models."
- ✅ Discipline-specific jargon: "Bakhtinian dialogism," "Vygotskian ZPD," "transformer architectures"
Argumentation Structure:
1. **Introduction** (2-3 paragraphs)
- Contextualize the problem (field, significance)
- State thesis/claim
- Preview argumentation structure
2. **Background/Literature Review** (2-4 paragraphs)
- Engage with scholarly literature
- Establish theoretical framework
3. **Analysis/Argument** (3-6 paragraphs)
- Present evidence (data, case studies, citations)
- Address counterarguments
4. **Implications** (1-2 paragraphs)
- Practical applications
- Policy recommendations
5. **Conclusion** (1 paragraph)
- Restate thesis
- Link to related GeoVerity service (if applicable)
6. Contact Page
Register: A2-B1 (Elementary-Intermediate) Genre: Transactional (form-based) Maximum accessibility
| Feature | Specification | Example |
|---|---|---|
| Lexical Density | 35-45% (instructional clarity) | "Tell us about your needs. We'll respond within 24 hours." |
| Sentence Length | 8-15 words average | Short, direct |
| Syntactic Complexity | Simple sentences | Imperative mood |
| Voice | Active voice (100%) | "Contact us" not "We may be contacted" |
| Jargon Tolerance | Zero | Plain language only |
Syntactic Feature Matrix
| Feature | Homepage | Services | Spokes | Insights Hub | Insights Posts | Contact |
|---|---|---|---|---|---|---|
| Subordinate-initial clauses | ❌ Never | ⚠️ Rare | ⚠️ Rare | ⚠️ Rare | ✅ Common | ❌ Never |
| Nominalization density | Low | Low-Mod | Moderate | Moderate | High | Minimal |
| Passive voice % | <10% | <15% | <20% | <20% | 30-40% | 0% |
| Average sentence length | 12-18 | 15-22 | 18-25 | 18-25 | 20-35 | 8-15 |
| Lexical density | 40-50% | 50-60% | 55-65% | 55-65% | 65-75% | 35-45% |
| Embedding depth | 0-1 | 1-2 | 1-2 | 1-2 | 2-4 | 0 |
| Technical jargon | Minimal | Moderate | Moderate-High | Moderate | High | None |
| Hedging | Minimal | Low | Low-Mod | Moderate | High | None |
Lexical Stratification Guidelines
Tier 1: Universal Vocabulary (All Surfaces)
Criteria: General Service List (GSL) 2000 most frequent English words Examples: "data," "quality," "verify," "trust," "help," "service," "system" Usage: Homepage, Services, Contact
Tier 2: Professional Vocabulary (Services + Insights)
Criteria: Academic Word List (AWL) + Industry-standard terms Examples: "methodology," "framework," "evaluation," "governance," "compliance," "integrity" Usage: Services Hub, Pillar/Spoke pages, Insights Hub
Tier 3: Technical Vocabulary (Spokes + Insights Posts)
Criteria: Domain-specific terminology (AI/ML, Education, Policy) Examples: "fine-tuning," "RLHF," "epistemic warrant," "IRB compliance," "transformer architecture" Usage: Service Spoke pages, Insights Posts (defined on first use in Services)
Tier 4: Disciplinary Jargon (Insights Posts ONLY)
Criteria: Specialized scholarly terminology Examples: "phenomenological," "Bakhtinian dialogism," "deontological," "hermeneutic," "sociolinguistic variation" Usage: Insights Journal Posts ONLY (not defined, assumes expert audience)
Information Structure Principles
Given-New Contract (Halliday)
All content surfaces should follow given-before-new information structure:
- Place known/contextual information in theme position (sentence-initial)
- Place new/focal information in rheme position (sentence-final)
Example (Services Page):
- ✅ "Graduate programs [given] face new challenges from AI authorship [new]. These challenges [given] require updated integrity frameworks [new]."
- ❌ "Updated integrity frameworks are required by challenges that graduate programs face."
Thematic Progression
Homepage: Constant theme (GeoVerity as repeated subject)
- "GeoVerity provides... GeoVerity verifies... GeoVerity helps..."
Services Pages: Linear theme (previous rheme becomes next theme)
- "AI systems require verified data [rheme]. Verified data [theme] enables trustworthy models [rheme]. Trustworthy models [theme] build institutional confidence."
Insights Posts: Split theme (complex thematic development)
- Academic argumentation allows non-linear thematic progression
Code-Switching & Bilingual Parity
English-Spanish Register Alignment
Critical: Spanish translations must match the register of English source text.
Register-Appropriate Translation:
| English (Services Page, B2-C1) | Spanish (Same Register) |
|---|---|
| "GeoVerity helps institutions maintain epistemic integrity." | "GeoVerity ayuda a las instituciones a mantener la integridad epistémica." |
| NOT: "GeoVerity auxilia a instituciones en el mantenimiento de la integridad epistémica." (too formal) |
| English (Insights Post, C1-C2) | Spanish (Same Register) |
|---|---|
| "The epistemic collapse induced by LLM-generated text necessitates institutional recalibration." | "El colapso epistémico inducido por texto generado por LLM necesita una recalibración institucional." |
Register Calibration by Variety:
- Latin American Spanish: Prefer slightly more direct/informal register than Peninsular Spanish
- Peninsular Spanish: Acceptable for formal Insights Posts
- Avoid: Overly formal constructions in service pages ("se ruega," "a la mayor brevedad posible")
Common Mistakes
❌ Mistake 1: Academic Register on Service Pages
Symptom: Subordinate-initial clauses, heavy nominalization, passive voice on Homepage/Services pages
Wrong:
"Given the epistemological challenges posed by generative AI in pedagogical contexts, institutions require comprehensive frameworks for the maintenance of epistemic integrity through methodological rigor."
Right:
"Graduate programs face new challenges from AI-generated student work. GeoVerity helps institutions maintain academic integrity with proven frameworks."
Why: Administrators and project managers need clear, actionable language, not academic prose.
❌ Mistake 2: Plain Language in Insights Posts
Symptom: Oversimplified syntax, no disciplinary terminology in scholarly articles
Wrong (for Insights Post):
"AI tools are changing how students write. This is a problem for universities. We need new ways to check student work."
Right (for Insights Post):
"While generative AI proliferates across pedagogical contexts, its epistemic implications for authorship verification remain contested. Institutional frameworks must recalibrate beyond detection-based models toward process-oriented integrity assessment."
Why: Academic audiences expect scholarly argumentation with complex syntax and disciplinary terminology.
❌ Mistake 3: Register Mismatch in Spanish Translations
Symptom: Spanish translation is more formal than English source
Wrong:
- English (B2): "We help you build trustworthy AI."
- Spanish (C1): "Auxiliamos en la construcción de inteligencia artificial fidedigna."
Right:
- English (B2): "We help you build trustworthy AI."
- Spanish (B2): "Te ayudamos a construir IA confiable."
Why: Register must match across languages for bilingual parity.
❌ Mistake 4: Jargon Without Context
Symptom: Technical terms undefined on service pages for non-specialist audiences
Wrong:
"Our RLHF pipelines optimize decontextualized lemma frequency distributions across polyglot corpora."
Right:
"We optimize language model training using human feedback and multilingual datasets."
Why: Administrators/PMs need industry-standard terms, not research jargon.
❌ Mistake 5: Subordinate-Initial Clauses on Service Pages
Symptom: Starting sentences with "Because...", "While...", "Given that..." on Homepage/Services pages
Wrong:
"Because AI detection tools fail 40% of the time, institutions need alternative approaches to academic integrity."
Right:
"AI detection tools fail 40% of the time. Institutions need alternative approaches to academic integrity."
Why: Main-clause-initial structure is clearer for busy professionals scanning content.
Register Swap Test (Mandatory Self-Check)
Before publishing ANY content, run this diagnostic:
If Drafting Homepage / Services Content
STOP and revise if you hear yourself doing ANY of this:
❌ Starting with subordinate clauses:
- "While X...", "Because X...", "Given that X...", "Although X..."
❌ Using academic terminology without operational grounding:
- "epistemic instability," "hermeneutic framing," "phenomenological pressure," "deontological imperatives"
❌ Asking the reader to rethink policy foundations instead of telling them what we do:
- "Institutions must reconceptualize..."
- "We need to problematize..."
- "Traditional frameworks require epistemological revision..."
❌ Performing identity work as "fellow theorist" instead of "operational partner":
- "As a field, we must reconsider..."
- "Our shared disciplinary assumptions..."
Diagnosis: You are doing Insights languaging. Stop and rewrite in plain professional language (B2-C1).
Fix checklist:
- Rewrite with main clause first
- Replace academic terms with industry-standard terms
- Frame as operational problem → GeoVerity solution → outcome
- Position reader as decision-maker, not co-theorist
If Drafting Insights Post
STOP and revise if you hear yourself doing ANY of this:
❌ Promising operational outcomes directly:
- "We help you..."
- "This lets you implement..."
- "You can deploy this framework next week..."
❌ Avoiding theoretical terms because you think they're "too academic":
- Writing "problem with checking" instead of "epistemic warrant"
- Writing "power issues" instead of "deontological constraints"
- Refusing to engage with disciplinary literature
❌ Writing only in short main-clause-first sentences:
- Refusing to use subordination for argumentation
- Avoiding complex syntax even when ideas require it
❌ Performing identity work as "service provider" instead of "intellectual peer":
- "We can solve this for you..."
- "Our clients need..."
Diagnosis: You are doing Services languaging. Stop and escalate to academic register (C1-C2).
Fix checklist:
- Rewrite to argue/theorize/reframe, not promise outcomes
- Use disciplinary terminology without immediate operationalization
- Use complex syntax to match complex ideas
- Position reader as co-analyst, not client
Swap Test Summary (Copy-Paste Diagnostic)
Services content should:
- Answer: "What problem do you have? What do we do? What outcome do you get?"
- Use main-clause-first sentences
- Ground technical terms in operational effect
- Position reader as decision-maker
Insights content should:
- Answer: "What assumptions are we interrogating? What evidence challenges them? What reframing do we propose?"
- Use subordination for argumentation
- Introduce theoretical constructs without immediate application
- Position reader as fellow theorist
If content does the opposite of its surface type, you have violated register stratification.
Quality Assurance Checklist
Before publishing ANY content, verify:
Register Compliance
- Identified content surface type (Homepage, Services, Insights, etc.)
- Applied correct CEFR level (B1-B2 for Homepage, B2-C1 for Services, C1-C2 for Insights Posts)
- Verified sentence length matches target range
- Checked passive voice percentage
- Confirmed no subordinate-initial clauses on service pages
Syntactic Rules
- Main clause initial on Homepage/Services pages (no "Because X, Y..." or "While X, Y...")
- Active voice >85% on service pages
- Minimal nominalization on service pages
- Technical terms defined on first use (service pages)
Audience Alignment
- Language matches target audience (administrators, PMs, researchers)
- Problem-Solution-Outcome structure for service pages
- Claim-Evidence-Warrant structure for Insights Posts
Bilingual Parity
- Spanish translation matches English register level
- No register shift between languages (e.g., B2 EN → C1 ES)
Integration with Other Skills
MANDATORY SKILL COMBINATIONS:
- Always combine with
building-pagesfor accessibility/performance compliance - Use with
templating-pagesfor Astro-specific implementation - Reference
brandingfor voice/tone alignment (branding specifies brand voice, languaging specifies register) - Follow
making-skill-decisionsfor skill discovery workflows - CRITICAL: Always run
checking-crappy-writingv1.3.0 after generating content- This skill checks register stratification
checking-crappy-writingv1.3.0 AUTO-FIXES AI artifacts (hallucinated citations, puffery, chatbot meta-language, formatting leaks, anti-detection evasion)- User reviews FIXES (not violations) and updates provenance
Execution Order (v1.3.0 AUTO-FIX workflow):
languaging→ Generate register-compliant contentchecking-crappy-writing→ AUTO-FIX violations- Claude Code assistant reports fixes to user (structured format)
- User reviews auto-fixes (accepts/rejects/edits)
- User updates provenance to "human-edited"
- User sets _meta.contentStatus to "approved"
- Iterate until PASS
IMPORTANT: Content you generate will be automatically scanned and fixed by checking-crappy-writing. User will review FIXES, not violations. See .claude/skills/checking-crappy-writing/SKILL.md for auto-fix output format.
Pre-Publication Register Audit
Automated Checks
- Sentence length analysis: Flag sentences >30 words on Homepage/Services pages
- Passive voice detection: Flag >15% passive on non-Insights pages
- Lexical density calculation: Flag mismatches with target range
- Readability scores: Flesch-Kincaid Grade Level, CEFR alignment
Human Review
- Linguistic review: PhD-level linguist reviews register appropriateness
- Audience testing: Segment representatives review drafts (administrators, PMs, researchers)
- Cross-linguistic review: Native Spanish speaker reviews register parity
Glossary of Linguistic Terms
CEFR (Common European Framework of Reference for Languages): Standardized scale of language proficiency (A1-C2)
Lexical Density: Ratio of content words (nouns, verbs, adjectives, adverbs) to total words; higher density = more information-packed
Nominalization: Converting verbs/adjectives into nouns (e.g., "verify" → "verification"); increases abstraction and formality
Embedding Depth: Number of subordinate clauses nested within a sentence; higher depth = greater syntactic complexity
Thematic Structure: Division of clause into theme (sentence-initial, given information) and rheme (new information)
Subordinate Clause Initial: Sentence structure where dependent clause precedes main clause (e.g., "While X, Y...")
Hedging: Use of epistemic modality to express uncertainty or tentativeness (e.g., "may," "suggests," "arguably")
Register: Contextual variety of language associated with particular situations, audiences, and purposes
Tenor: Social relationship between discourse participants (formal ↔ informal, expert ↔ novice)
Field: Subject matter or domain of discourse (technical, academic, everyday)
Mode: Channel of communication (spoken, written, digital)
References (Linguistic Framework)
- Halliday, M.A.K., & Matthiessen, C.M.I.M. (2014). Halliday's Introduction to Functional Grammar (4th ed.). Routledge.
- Bell, A. (1984). Language style as audience design. Language in Society, 13(2), 145-204.
- Bernstein, B. (1971). Class, Codes and Control, Volume 1: Theoretical Studies Towards a Sociology of Language. Routledge.
- Biber, D., & Conrad, S. (2009). Register, Genre, and Style. Cambridge University Press.
- Martin, J.R., & Rose, D. (2008). Genre Relations: Mapping Culture. Equinox.
- Council of Europe (2001). Common European Framework of Reference for Languages. Cambridge University Press.
Document Control:
- Version: 1.0.0
- Date: 2025-10-27
- Owner: GeoVerity Content Strategy + Linguistics PhD Stakeholder
- Full Language Register Plan: See docs/Language-Register-Plan.md for complete specifications
Red Flags - STOP Before Publishing
If you catch yourself doing ANY of these, STOP and revise:
Languaging Violations (Action/Identity)
- 🚨 Services page does academic languaging: Arguing/theorizing instead of solving/promising
- 🚨 Insights post does service languaging: Promising outcomes instead of interrogating assumptions
- 🚨 Wrong identity positioning: Services page treats reader as co-theorist; Insights post treats reader as client
Syntactic Violations
- 🚨 Using subordinate-initial clauses on Homepage/Services pages
- 🚨 Starting sentences with "Because...", "While...", "Given that..." on marketing pages
- 🚨 Heavy nominalization on service pages ("the implementation of verification processes...")
- 🚨 Passive voice >15% on service pages
- 🚨 Only short simple sentences in Insights posts (refusing complex syntax for complex ideas)
Lexical Violations
- 🚨 Academic jargon undefined on service pages ("epistemic warrant" without grounding in effect)
- 🚨 Plain language in Insights journal posts (oversimplified for academic audience)
- 🚨 Industry-standard terms avoided in Insights because "too technical"
Cross-Linguistic Violations
- 🚨 Spanish translation more formal than English source
- 🚨 Register shift between EN → ES (B2 English becomes C1 Spanish)
Swap Test Failures
- 🚨 Services content asks: "What assumptions are we interrogating?" (Should ask: "What problem do you have?")
- 🚨 Insights content promises: "We help you implement X" (Should argue: "Traditional models require reconsideration")
All of these mean: Revise before publishing. No exceptions.
Quick diagnostic: Run the Register Swap Test above before finalizing any content.