Method note · Research in development
From spoken words to defined observations.
Measurement is a starting point. Understanding what it means is a separate responsibility.
Start with a defined measurement.
Speech timing, word variety, grammatical structure, and the progression of language can be described using explicit measurement rules. Each describes an aspect of a recorded response; none, on its own, establishes a personality trait.
| Measurement | Definition |
|---|---|
| Speech timing | Speaking intervals and pauses, with defined duration requirements. |
| Word variety | Variation and repetition within specified lengths of text. |
| Language structure | Patterns in sentence construction, evaluated within each language. |
| Semantic progression | Relationships between successive blocks of language. |
Keep the evidence attached.
A summary should remain connected to the source material it describes. The reference feature work preserves source references, measurement definitions, and the conditions required for a result.
Missing evidence should remain missing. It should not become a confident zero.
Available and unavailable measurements are distinguished. Language boundaries, input quality, and minimum exposure remain part of the interpretation.
Validation is a separate responsibility.
The speech and text feature work does not include a trained personality prediction model. Connecting observations to personality requires separate evidence.
- Fresh evaluation data, separate from development.
- Checks across languages and contexts.
- Evidence that interpretations remain useful beyond one recorded response.
- Clear communication of uncertainty and limitations.