Explain how
I don't deny the usefulness of logic, and you know that. I just deny the infallibility of users of logic. That's hardly controversial.
I don't deny the existence of language, but I do deny the existence of language as something that can be measured apart from how it is used. If a person writes a dictionary, then that person is sampling the usage of language, either digging at their own navel (going by their own opinions of the meaning of words), or digging at a communal navel (going by an agreement of opinions on meanings of words of some peer group), or digging up actual usage (making a scientific sampling of usage).
In all cases there are lexicographic limitations: economy, first and foremost. How many different usage examples must be included? Which ones can be grouped together on the basis of similarity of parameters and meaning? Which idiomatic uses to include?
In the past, digging at own or communal navels was the predominant method. This renders all "Good Old Dictionaries" highly suspect, specifically in that the group of peers have never been representative of a majority of people using the language.
There is no language outside of the use of language. No hands waving, this is a cold fact.
