Semantic layer

Semantic atoms are an engineering interlingua, not a theory of human cognition.

That sentence is normative and appears verbatim in the Atom Explorer UI.

Atoms

30–50 atoms, each {id, category, definition, epistemic_status, nsm_prime, exponents{en,te,hi}, related[]}. Atoms that are Natural Semantic Metalanguage primes (Wierzbicka 1996; Goddard & Wierzbicka 2014) carry nsm_prime: true and epistemic_status: "ESTABLISHED" — established as NSM primes, i.e. as a well-known research program's inventory, not as proven cognitive universals. All other atoms are ENGINEERING.

Categories: FOUNDATIONAL, RELATIONAL, ACTION, STATE, EMOTIONAL, SOCIAL. Relations between atoms: opposite, broader, narrower, co-occurs.

Atom structures

Concepts carry an AtomStructure: a small typed expression tree with operators STATE_OF, CAUSE, BECOME, NEG, HAVE, AGENT_OF, EVENT_OF, PLACE_OF, DEGREE, stored as nested JSON ({"op": "STATE_OF", "args": [{"atom": "atom:GOOD"}]}), never as a string. A pretty-printer renders STATE_OF(GOOD) etc.

Examples:

  • GOODNESS = STATE_OF(GOOD)
  • CHILDHOOD = STATE_OF(CHILD)
  • HOPELESSNESS = STATE_OF(NEG(HAVE(WANT(GOOD AFTER NOW)))) — coarse; tagged HYPOTHESIS
  • TEACHER = AGENT_OF(TEACH)
  • SIMPLIFY = CAUSE(BECOME(SMALL(DEGREE))) approximations are flagged HYPOTHESIS where coarse.

Decomposition granularity is honest: where a concept resists decomposition into ≤ 50 atoms, the structure is marked HYPOTHESIS and the explanation engine phrases it as an approximation ("can be approximated as…").

Affix functions

20–30 cross-lingual functions (fn:ST abstract state, fn:AG agent, fn:PRIV privative, …). A function is a functional label: members of its equivalence class differ in distribution, register, and productivity — every equivalence-class note says so (ENGINEERING).