Context-freeness and the computer processing of human languages

  • Authors:
  • Geoffrey K. Pullum

  • Affiliations:
  • University of California, Santa Cruz, Santa Cruz, California

  • Venue:
  • ACL '83 Proceedings of the 21st annual meeting on Association for Computational Linguistics
  • Year:
  • 1983

Quantified Score

Hi-index 0.00

Visualization

Abstract

Context-free grammars, far from having insufficient expressive power for the description of human languages, may be overly powerful, along three dimensions; (i) weak generative capacity: there exists an interesting proper subset of the CFL's, the profligate CFL's, within which no human language appears to fall; (2) strong generative capacity: human languages can be appropriately described in terms of a proper subset of the CF-PSG's, namely those with the ECPO property; (3) time complexity: the recent controversy about the importance of a low deterministic polynomial time bound on the recognition problem for human languages is misdirected, since an appropriately restrictive theory would guarantee even more, namely a linear bound.