I confess myself somewhat thrown for a loop by Randy J. LaPolla’s short paper “On categorization: Stick to the facts of the languages” (to appear in Linguistic Typology); he says sensible things about labeling linguistic categories based on resemblances, then continues thus:
My own view (LaPolla 1997, 2003, 2015, 2016), developed from my experiences with languages and communication over many years, departs radically from the Structuralist paradigm: I argue that there is no coding-decoding in communication, and no shared code among speakers of the same language; communication is achieved simply by ostension and abductive inference, regardless of whether linguistic forms are involved or not. The communicator does something (the ostensive act) with the purpose of the addressee inferring the intention to communicate and the reason for the ostensive act. By doing this, the addressee creates a meaning in their mind, which the communicator hopes will be similar to the meaning the communicator intended the addressee to create. That is, there is no meaning in the ostensive act (be it linguistic or not); the addressee creates a meaning based on the communicator’s use of the particular ostensive act in the particular context by creating a context of interpretation, out of the overall context of assumptions available to him or her at that moment, in which the ostensive act “makes sense”. As it is based on abductive inference, though, the outcome is non-deterministic.
Language use is one type of ostensive act.
Obviously I’d have a better idea what he’s talking about if I read his other work, but just in bare outline, it sounds reductive and hard to apply in practice; at any rate, I’ll be glad to hear other people’s thoughts on it. I certainly have no quarrel when he says in his conclusion that “combing hundreds of grammars (of varying quality) and extracting forms that one thinks might fit one’s comparative categories (regardless of what the author of the grammar might have said) is very problematic. It is much better to concentrate on languages one has a good knowledge of and contribute to typology by expanding our understanding of what is found and how we might understand it, including its historical origins.” (Thanks to John Cowan for sending me the link.)