[YG Conlang Archives] > [jboske group] > messages [Date Index] [Thread Index] >
pc: > a.rosta@hidden.email writes: > << > Technically, the BNF 'grammar' is more like a grammaticality-checker > than a true grammar. That is, it will tell you whether or not a > string is well-formed Lojban, but it won't tell you what it means. > >> > though I shouldn't say so, this is a hairsplitting technicality in > this discussion. OK, The BNF 'grammmar' (but isn't it in fact the > yacced grammar) is the final authority on syntax. The further move > to connection with meaning -- something which exists for no langauge > that I know of (including machine languages) -- is what jboske is > largely about. And making that move is greatly aided by having the > syntax fixed (and the vocab, too, of course). What you say is reasonable enough. But I think you'd be hard pressed to find a 'grammar' or syntactic description of a natural language that does not take into account meaning, and as syntactic theory matures, syntax and semantics increasingly merge into the same thing. So from a syntactician's point of view, the Lojban formal grammar is vacuous -- so what if the grammar rules some strings out and some strings in, and assigns such and such a structure to those strings? Unless the ruled-in strings are assigned meanings, and unless the syntactic structures assigned to them are also meaningful, they are empty artefacts of the analysis, not part of the language-grammar (that maps between sentence sounds and sentence meanings). Put it this way: if a linguist came along to do fieldwork on Lojban (and didn't have access to the blueprint documents), they would not come up with anything resembling the formal grammar, and if you then showed them the formal grammar they would find it very valuable. I know you disagree with some of this, because we've discussed it before. All I can say is I'm a syntactician, and with an open mind I'm trying to call it how I see it. --And.