context free grammars

A grammar which rules do not depend on the context during parsing. I know, this is a circular definition.

Here are some examples :

  • Lojban's grammar, and more generally any LR or LL grammar, are context-free.

  • The C language is supposed to be context-free, except on the topic of type definition (if "typedef" preceeds a definition, the symbol defined becomes a type rather than a variable : semantics change depending on the context, given the same lexical grammar).
    • C is fairly good about it, compared to many other languages.

  • English is not context-free. The way we parse/understand sentences depends on the context, either past or future. Consider this text :

you if sense makes sentence this, please read the first six words backwards.

  • mi'e cein. Doesn't Lojban have si/sa/su, which must be "understood" by the parser in the same sense as the English example above in order to be correctly parsed?
    • maybe this doesn't address what you mean — but si/sa/su can be implemented below actual language parsing, simsa things such as \ line continuations in C.
    • they are trivially handled by the lexer.
  • But this sentence isn't legitimate English...
    • arguable; but the point still stands — english isn't context free. A phrase structure grammar for english would be hideously large and type 0-1, if a complete one were ever made, which is unlikely. --mi'e .djorden.

I thought a CFG was one where the left side of the rewrite rule is unconditional. E.g. "A -> B C" is context free, but "A -> B C, when A is preceded by D" ("D A -> B C") is not context free. --And

Fairly sure that is the case. It would certainly be context. --jay

the above is correct — a context free grammar may only have one non terminal on the left hand side of its rules. --mi'e .djorden.

Created by admin. Last Modification: Tuesday 10 of June, 2014 04:21:22 GMT by mukti.