What if labels could speak everyone's language?
Translating institutional trust into social trust through culturally-aware information design
"These numbers mean nothing to me. I trust what my neighbor tells me, not what some label says."
— Ms. Zhang Wei, 58, Jiaxian County, China
Standard Information Design Excludes Non-WEIRD Populations
WEIRD Assumptions
Western, Educated, Industrialized, Rich, Democratic populations represent 12% of the world but inform 95% of design decisions.
Trust Deficit
Numerical labels assume institutional trust. For many populations, trust flows through social networks, not government agencies.
Numbers Don't Translate
"12g sugar" means nothing without cultural anchors. Abstract units lack contextual meaning outside WEIRD frameworks.
Not Just Language
Translating text isn't enough. The conceptual models behind information must be culturally adapted.
The Core Question:
How do we translate information design from institutional trust systems (WEIRD) to social trust systems (non-WEIRD) without losing accuracy or patronizing users?
Ethnographic Research Across Two Contexts
16 interviews and contextual inquiry across rural China and Eastern Washington
Ms. Zhang Wei
58 • Retired Factory Worker
Jiaxian County, Henan Province, China
"These numbers mean nothing to me. I trust what my neighbor tells me, not what some label says."
Trust Pathway
Pain Points
- •Cannot understand numerical nutrition labels
- •Distrusts institutional information sources
Maria Rodriguez
43 • Farmworker
Yakima County, Eastern Washington
"I don't read English well, but I know this brand because my cousin uses it. That's how I decide."
Trust Pathway
Pain Points
- •English language barriers on detailed labels
- •Unfamiliar units (ounces, grams, percentages)
Research Methodology
📋 Data Collection
- •Semi-structured interviews (n=8 per context, 16 total)
- •Contextual inquiry in grocery stores and local markets
- •Usability testing with prototype labels (n=16)
- •Field documentation through photos and ethnographic notes
🔍 Analysis Methods
- •Thematic analysis of interview transcripts
- •Affinity mapping to identify patterns across contexts
- •Comparative comprehension testing (standard vs. visual labels)
- •Trust pathway mapping from participant mental models
👥 Participant Recruitment
Participants were recruited through community organizations and local networks in rural Henan Province, China, and Yakima County, Washington. Inclusion criteria: primary household food purchasers, limited English proficiency (Eastern WA cohort), and self-reported difficulty understanding nutrition labels.
Key Research Insights
Trust Deficit, Not Comprehension Deficit
Rural consumers can understand information when presented appropriately, the issue is they don't trust institutional sources.
Numbers Don't Convey Meaning
Abstract numbers (grams, percentages) lack contextual anchors. Users need relatable reference points.
Visual Comparison Drives Understanding
Users process relative comparisons faster than absolute values. "More than X" beats "50% of daily value".
Mental Models Are Culturally Specific
How people measure, compare, and judge varies by cultural context. There is no "universal" representation.
Language Barriers Are Secondary to Conceptual Barriers
Even when information is translated linguistically, it fails if the underlying concepts aren't culturally relevant.
Serving Sizes Are Arbitrary and Confusing
"1 serving" means nothing without cultural context. People eat until satisfied, not until they've consumed exactly 28 grams.
Context-Aware Translation Engine
The same product information, translated into four different cultural trust languages
Coca-Cola Classic
Same product, four different trust languages
Standard WEIRD
Rural Chinese
Eastern WA
Low-Literacy
⚡ Comprehension Speed Comparison
Limitations & Open Questions
🔍 Scope Limitations
- •Domain specificity: This prototype addresses nutrition labels but doesn't solve medication instructions, financial documents, or legal forms, other critical information types that affect non-WEIRD populations.
- •Cultural oversimplification: Real cultural variation is far more nuanced than four categories. Within "Rural Chinese" and "Eastern WA" contexts exist tremendous diversity.
- •Small sample size: With n=16 participants across two contexts, findings may not generalize to all non-WEIRD populations globally.
❓ Unresolved Design Questions
- •Trust indicators for isolated individuals: The "Neighbor Verified" badge assumes local trust networks exist, how does this work for people without strong community ties?
- •Delivery mechanism: We tested with printed labels attached to products in our demo. But what's the best way to deliver this information at scale? Mobile apps? In-store digital displays? QR codes? Each has accessibility trade-offs.
- •Dynamic translation accuracy: Can visual metaphors (sugar cubes, walking minutes) maintain accuracy across thousands of products? Does "3 sugar cubes" work for both soda and granola bars?
⚙️ Technical & Scalability Challenges
- •Database architecture: How would this work with 1000+ products? Real-time translation requires robust backend infrastructure and content management systems.
- •Maintaining cultural accuracy: Who validates that trust indicators and visual metaphors remain culturally appropriate? This requires ongoing collaboration with community stakeholders.
- •Regulatory compliance: How do culturally-adapted labels satisfy FDA/USDA legal requirements while remaining accessible?
What I'm still learning: These limitations don't undermine the core insight, that information design must respect different epistemologies, but they highlight that this is a starting point, not a complete solution.
This is just the beginning.
Lens demonstrates that accessibility isn't about "simplifying" for others, it's about respecting different ways of knowing.