Subject: n-ary comparison

( 1 ) very strongly support alexis mr 's message on n-ary comparison being stronger than binary comparison . as real minds work , it is the pragmatics of not going down false trails which is at issue . i am puzzled only by what he mentions as an exceptional case , that is , how ternary comparison could * * * ever * * * be worse than binary . even if a third language is included which is more divergent , has more loan vocabulary , or whatever , and therefore poses extra problems or barriers . that is not a consequence of the ternary status in the comparison , because if the same language were included as one member of a binary comparison , the same problems would presumably be there . ( 2 ) i did not respond earlier only because of lack of time . the preference for n-ary comparison , the claim that n-ary is only a multiple of binary , is obviously false , just as the notion that one should not be engaging in comparison if one needs to use a dictionary is obviously false ( alice faber answered that one , that if that restriction had been enforced , our knowledge would be far less than it now securely is ) . both of these supposed restrictions are for the convenience of the linguist analyst , artificially restricting the data available so as to make it seem as if the analyst 's knowledge is more complete for the task than it is , and give a greater ( but artificial ) sense of psychological security . it goes along with the statement by some historical linguists that they " never want to make a mistake " . of course none of us do , but some are willing to in order to advance knowledge , and even those who do not want to cannot make themselves immune . the very act of disregarding data which is " too far afield " can sometimes lead to mistakes . always possible when trying to * discover * answers when we do not know the answers in advance . rather , we are all human , and gigantic problems will be beyond any of us to completely solve ( attempts at partial solutions do not imply guilt or wrongdoing ) . this in no way implies we should not attempt to solve those problems . we should be tolerant of others ' efforts and use the best part of them , always . ( 3 ) j " org knappen 's suggestion of " families , classes , and orders " a la the biologists does seem useful . can we have suggestions as to clues when we should use one or the other of such terms ? presumably some more sophisticated measure of degree of shared genetic material would be most analogous to the biological usages , rather than a simple lexicostatistical measure ? lloyd anderson
