Computer Laboratory

Technical reports

Computational models for first language acquisition

Paula J. Buttery

November 2006, 176 pages

This technical report is based on a dissertation submitted March 2006 by the author for the degree of Doctor of Philosophy to the University of Cambridge, Churchill College.

Abstract

This work investigates a computational model of first language acquisition; the Categorial Grammar Learner or CGL. The model builds on the work of Villavicenio, who created a parametric Categorial Grammar learner that organises its parameters into an inheritance hierarchy, and also on the work of Buszkowski and Kanazawa, who demonstrated the learnability of a k-valued Classic Categorial Grammar (which uses only the rules of function application) from strings. The CGL is able to learn a k-valued General Categorial Grammar (which uses the rules of function application, function composition and Generalised Weak Permutation). The novel concept of Sentence Objects (simple strings, augmented strings, unlabelled structures and functor-argument structures) are presented as potential points from which learning may commence. Augmented strings (which are strings augmented with some basic syntactic information) are suggested as a sensible input to the CGL as they are cognitively plausible objects and have greater information content than strings alone. Building on the work of Siskind, a method for constructing augmented strings from unordered logic forms is detailed and it is suggested that augmented strings are simply a representation of the constraints placed on the space of possible parses due to a strings associated semantic content. The CGL makes crucial use of a statistical Memory Module (constructed from a Type Memory and Word Order Memory) that is used to both constrain hypotheses and handle data which is noisy or parametrically ambiguous. A consequence of the Memory Module is that the CGL learns in an incremental fashion. This echoes real child learning as documented in Browns Stages of Language Development and also as alluded to by an included corpus study of child speech. Furthermore, the CGL learns faster when initially presented with simpler linguistic data; a further corpus study of child-directed speech suggests that this echos the input provided to children. The CGL is demonstrated to learn from real data. It is evaluated against previous parametric learners (the Triggering Learning Algorithm of Gibson and Wexler and the Structural Triggers Learner of Fodor and Sakas) and is found to be more efficient.

Full text

PDF (1.3 MB)

BibTeX record

@TechReport{UCAM-CL-TR-675,
  author =	 {Buttery, Paula J.},
  title = 	 {{Computational models for first language acquisition}},
  year = 	 2006,
  month = 	 nov,
  url = 	 {http://www.cl.cam.ac.uk/techreports/UCAM-CL-TR-675.pdf},
  institution =  {University of Cambridge, Computer Laboratory},
  number = 	 {UCAM-CL-TR-675}
}