Goals for the language (in no particular order and subject to community review):
- expressiveness: the language should be flexible enough to communicate many different meanings and support many different uses,
- unambiguity: utterances should have a single meaning (though the meaning may be vague, in which case it will bear markers for the ways in which it is vague),
- composability: grammatical marks should available for combination and reuse,
- clear semantics: the meanings of morphosemantic operations should be rigorously specified,
- separation of concerns: clearly distinguish between e.g. discourse level and logical level of language,
- parsimony: expressions of the language should be concise,
Additionally we will be guided by two Perl praxemes:
- There is more than one way to do it: We aren't trying to make users think/speak one way or another. Instead, we provide the tools to make explicit which model a speaker is using.
- Make the easy things easy and the hard things possible: Common expressions should be concise and require few inflections. As a correlate, vagueness should be the default, while allowing for the specific marking of exactness.
@porpoiseless says:
- This should be a language primarily for thinking about thinking.
- Since thinking can have a variety of objects, this language must have ways of discussing many areas of human interest.
- It should be rooted in phenomenology and methodology:
- /What is it like/ to think about thinking?
- /How do we go about/ thinking about thinking?
- "Thinking about thinking" is already social. It requires at least two participants: the investigator and the investigated.
- "Thinking about thinking" requires language (c.f. above and the private language argument).
- Language or something equivalent. See how semiotics is devoured from within by linguistics. It seems like the latter should be a subset of the former, but in order to describe a symbol system we must turn it into language.