On Tue, Jun 28, 2022 at 6:25 AM John Rose <johnr...@polyplexic.com> wrote:
> ... > On Saturday, June 25, 2022, at 6:58 AM, Rob Freeman wrote: > > If all the above is true, the key question should be: what method could > directly group hierarchies of elements in language which share predictions? > > > Is this just intuition? > If you're asking whether it is just intuition that the grammar learned might be different for each sentence, I would say it's more of a hypothesis. I think there is evidence formal grammars for natural language cannot be complete. So for me that motivates the hypothesis beyond the point I would call it intuition. As support for the hypothesis I cite the history of linguistics going back to: 1) Chomsky's rejection of learning procedures and assertion any formal system for natural language grammar must be innate, not least because what is observed, contradicts: Part of the discussion of phonology in ’LBLT’ is directed towards showing that the conditions that were supposed to define a phonemic representation ... were inconsistent or incoherent in some cases and led to (or at least allowed) absurd analyses in others." Frederick J. Newmeyer, Generative Linguistics a historical perspective, Routledge 1996. 2) Sydney Lamb's counter that an alternative explanation was that the problem was "the criterion of linearity". 3) More recent work, including that done by OpenCog's own Linas Vepstas, that: Vepstas, “Mereology”, 2020: "In the remaining chapters, the sheaf construction will be used as a tool to create A(G)I representations of reality. Whether the constructed network is an accurate representation of reality is undecidable, and this is true even in a narrow, formal, sense." 4) Bob Coecke's work on "togetherness", and a "quantum" quality to grammar: Coecke: "we argue for a paradigmatic shift from `reductionism' to `togetherness'. In particular, we show how interaction between systems in quantum theory naturally carries over to modelling how word meanings interact in natural language." And more simply, my own experience, that when learning grammar (by distributional analysis), there is typically more than one, contradictory, way to do it (which is just a power of sets, and to be excluded, more than justified.) ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T5d6fde768988cb74-M385b2f935978c6bbd0e8d353 Delivery options: https://agi.topicbox.com/groups/agi/subscription