On 28 October 2015 at 07:43, R A Brown <[log in to unmask]> wrote: > 2015-10-28 3:46 GMT+03:00 Granny Grammar: >> >> "How many parts of speech does a loglang need?" >>> >>> "Parts of speech" is a unit of analysis imposed by the >>> person looking at the language, >> >> Yes, at least in the sense that it is also true of taxonomy in the natural sciences. > not a property of the language. >>> >> Even for natlangs that's debatable, but for engelangs we generally consider that the description of the language is the definition of the language, tho I suppose that a simpler description that generates the same sentences might, if discovered, be preferred. How many parts of speech does a loglang per_se need? IIRC > both Jörg and I _suggested_ just two: lexical words ("full > words") ~ function words ("empty words") :) I was about to reply when I noticed that delayed smiley. As for how many parts of speech a loglang needs, my current thinking would be that there's no reasonably objective answer to converge on yet, tho further study might well reveal one. Firstly there are alternative candidates for grammatically minimal loglang grammar and secondly those are amenable to alternative analyses with different numbers of PoS. We already know that under certain analytical assumptions one can argue that a loglang has one PoS, but those analytical assumptions don't necessarily commend themselves. I guess my overall feeling about the PoS inventory question is that the notion 'PoS' gets not very interesting and not very applicable at this level of minimality and basicness. --And.