At 19:08 +0100 2001/02/12, Frank Mittelbach wrote: > > -- Again, if there was a better parser at hand, one would not need to have > > any markup at all, because it would be able to see that `20' is the object > >give up, unless you want to wait for a parser which "understands" human >language (or write one but don't hand wave it) Writing a parser for English is clearly a research project; see for example http://lands.let.kun.nl/TSpublic/tosca/ One does not get very far with LALR(1) on this problem (parsing English), even less far when trying work with a grammar that depends on semantic information. Computer languages such as C and C++ are not strictly LALR(1), but can be made to parse in such chunks. >how do you do without markup in this case: > > The $a$ in the formula is a variable The usual remark on this: Can you parse it? :-) -- If you can parse it, it must be possible. Right? -- The general picture, though, is that the more general grammars the parser can handle, the less markup will be needed. >while contrieved i came across that particular problem in math when i tried to >understand an article in Hungarian (i think) about number theory and misstook >an "a" being text as part of a longer inline formula because it was >incorrectly coded (by you perhaps?) ie not identifiable easily as math not >text. Sorry, I don't know Hungarian. Hans Aberg