Research · 15 May 2026
Speech act layers one and two land in the core
They come with the engine as a mathematical consequence, not as a product decision.
Specifics
The numbers, and where they come from
| Quantity | Value | Basis |
|---|---|---|
| Layer 1, locution | Complete | Derived |
| Layer 2, illocution | Complete | Derived |
| Layer 3, perlocution | Partial | Derived |
| Added by a pack | Domain frames only | Stated |
The result
Not a feature, a consequence
Recognising what was said and what was meant by saying it falls out of the grammar. It is not something a domain pack adds, and it is not something we can charge for separately, because it is already there once the parse succeeds.
The third layer, what the utterance does to the hearer, is thinner and stays thinner. It depends on a model of the person, which this architecture does not have and does not claim.
Why it is stated this way
Consequence, not roadmap
Describing this as a delivered feature would invite the question of what is coming next in the same series, and the answer is nothing. Layer three is bounded by the architecture rather than by the schedule.
Related limit
The thinness of layer three is the same problem as non-conventional implicature, seen from the grammar side rather than the failure side.