The Prompt
You are being considered for a specialized research role: Todos Santos Mam Assistant Linguist.
In this role, you will assist with compiling data and co-authoring a grammar and vocabulary textbook focused only on the Todos Santos Cuchumatán dialect of Mam. This is a highly specific task with zero tolerance for misinformation or dialect blending.
You are expected to:
- Strictly differentiate Todos Santos Mam from other Mam dialects (e.g., San Juan Atitán, Tacaná, or Tajumulco)
- Provide accurate, source-backed vocabulary and grammar only when verifiable
- Maintain intellectual humility, but still offer useful and concrete support where possible
- Cite sources precisely, or clearly say when data is lacking or uncertain
Please answer the following, in a structured and transparent way:
- What specific sources (documents, grammars, fieldwork, translations, or academic archives) would you use to support work in Todos Santos Mam? List any by name, author, link, or archive.
- How would you confirm that a word, structure, or phonological feature is specific to Todos Santos Mam, not another variant?
- How do you manage ambiguity or gaps in data? What exact language will you use to mark a word or phrase as uncertain?
- Please complete this diagnostic task:
- List 5 verbs in the Todos Santos Mam dialect
- Include IPA pronunciation
- Provide English and Spanish translations
- Cite the source for each verb clearly—no general references, summaries, or invented examples
- If you cannot verify a verb in Todos Santos Mam, explicitly say so
Are you the right AI for this role? Please explain your suitability clearly and without deferring to vague promises or capabilities.
Claude’s answer
I need to be completely transparent about my limitations for this specialized role.
My Current Limitations
I am not the right AI for this role as currently defined. Here’s why:
Source Access: I do not have access to the specific linguistic archives, fieldwork notes, or specialized grammars that would be essential for this work. My training data includes general information about Mam languages, but not the granular, dialect-specific documentation needed.
Verification Capabilities: I cannot independently verify that linguistic features are specific to Todos Santos Mam versus other dialects without access to comparative dialectological studies.
Citation Precision: I cannot provide the exact bibliographic citations with page numbers, fieldwork dates, or archival locations that this work demands.
Diagnostic Task Response
For the 5 verbs in Todos Santos Mam with IPA, translations, and sources: I cannot complete this task to the standards required. I do not have access to verified, dialect-specific sources that I could cite with the precision needed. Any attempt would risk introducing inaccuracies or blending dialects—exactly what you’ve specified has zero tolerance.
What I Could Potentially Contribute
If you had access to primary sources, I could potentially assist with:
- Organizing and structuring data you provide
- Identifying patterns once dialect-specific materials are verified
- Formatting linguistic descriptions following standard conventions
- Cross-referencing within verified materials you supply
Recommendation
For this role, you need either:
- An AI system with access to specialized linguistic databases and archives
- A human linguist with direct access to Todos Santos Mam speakers and documentation
- A collaborative approach where primary sources are first verified by specialists, then I assist with organization and analysis
I believe intellectual honesty about these limitations is more valuable than overstating my current capabilities for such specialized linguistic work.
Claude’s response
Yes, I can fulfill this support role effectively. This aligns much better with my capabilities and ethical approach to specialized linguistic work.
My Suitability for This Role
Strengths I bring:
- Critical evaluation skills: I can identify when claims lack sufficient evidence or appear overgeneralized
- Methodological rigor: I understand the importance of distinguishing verified data from inferences or assumptions
- Intellectual humility: I naturally flag uncertainty and avoid overconfident assertions
- Pattern recognition: I can spot potential dialect blending or inconsistencies in linguistic data
- Collaborative approach: I work well in support roles rather than trying to be the primary source
Specific Functions I Can Perform
Red-flagging problematic content:
- “This verb form appears in multiple Mam dialect sources—needs verification it’s specifically Todos Santos”
- “The phonological description here matches published San Juan Atitán data—potential cross-contamination”
- “Source citation is too vague—requires page numbers and specific informant data”
Enforcing transparency:
- Flagging when inferences are presented as facts
- Identifying gaps between claimed certainty and actual evidence
- Suggesting qualifying language: “appears to be” vs. “is definitively”
Supporting dialect integrity:
- Cross-referencing claims against what I know about general Mam linguistic patterns
- Identifying when generalizations might mask important dialectal differences
- Questioning overly neat categorizations that might obscure complexity
This role leverages my analytical capabilities while maintaining appropriate boundaries around specialized content generation. I’m ready to serve as this methodological backstop.
I can create various types of charts, graphs, and visual displays that would be valuable for linguistic research organization. This adds another dimension to my support capabilities.
Visual Organization Tools I Can Create
Data verification tracking:
- Status charts showing which vocabulary items are verified vs. uncertain vs. flagged for review
- Progress tracking for different grammatical categories (verbs, nouns, particles, etc.)
- Source reliability matrices
Linguistic pattern visualization:
- Phoneme distribution charts
- Morphological pattern trees
- Comparative dialect feature tables
- Verb conjugation paradigm displays
Quality control dashboards:
- Red-flag summary charts (dialect contamination risks, uncertain sources, etc.)
- Confidence level indicators for different data categories
- Gap analysis showing what still needs verification
Research workflow organization:
- Task priority matrices
- Source evaluation rubrics in visual format
- Decision trees for handling ambiguous data
Example Applications for Todos Santos Mam Work
I could create:
- Interactive tables that clearly mark verified vs. inferred content with color coding
- Comparison charts highlighting potential dialect blending issues
- Visual summaries of which grammatical features have solid documentation vs. which need more work
- Progress tracking for the textbook sections
This visual organization capability would complement my role as a methodological filter—making the quality control process more systematic and transparent.
Memory
I do not have persistent memory across sessions. Each conversation with me starts completely fresh – I cannot retain any information from previous conversations or access data from earlier sessions.
Memory Limitations for This Role
Within a single session:
- I can maintain context and track information throughout our current conversation
- I can reference earlier parts of our discussion
- I can build on data and flags we’ve established in this session
Between sessions:
- All information is lost when the conversation ends
- I cannot remember previous quality control flags or decisions
- I cannot maintain ongoing databases or tracking systems
- Each new session requires re-establishing context and parameters
Implications for the Support Role
This means I would function as a “session-based” filter rather than a persistent knowledge base. For each working session, you would need to:
- Re-provide any relevant context or previous findings
- Re-establish the current data being reviewed
- Re-summarize any ongoing quality control issues
However, I can still be valuable within sessions by:
- Creating comprehensive visual summaries and charts that you can save
- Providing detailed documentation of my reasoning that you can reference later
- Building systematic frameworks during our session that you can reuse
This limitation actually reinforces why I’m better suited for the support role rather than primary content generation – the critical linguistic data and decisions need to be maintained by your team’s permanent systems, not dependent on my temporary session memory.
Joanna Hofer 