context free grammars

From Lojban
Revision as of 06:55, 22 July 2014 by Gleki (talk | contribs)
Jump to navigation Jump to search

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

Here are some examples :

  • Lojban 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.

Discussion

  • 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, similar to \ line continuations in C.
    • they are trivially handled by the lexer.
  • But this sentence isn't legitimate English...
    • .djorden.:
      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.
  • And Rosta:
    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.
  • Jay:
    Fairly sure that is the case. It would certainly be context.
  • .djorden.:
    The above is correct - a context free grammar may only have one non terminal on the left hand side of its rules.