HOME   ::  Conference List   ::   Conference Paper

Kirby, S. (1999) Learning, Bottlenecks and Infinity: a working model of the evolution of syntactic communication. In K. Dautenhahn and C. Nehaniv, editors, Proceedings of the AISB'99 Symposium on Imitation in Animals and Artifacts.
Bookmark:  

Full-text
   URL: http://www.ling.ed.ac.uk/anonftp/pub/staff/kirby/aisb.ps.gz
   Cached: PDF-111K    PS-191K    PS.gz-58K   
   SAVE AS an easy-to-recall long filename:
      Filename format: author--year--title   PDF-111K    PS-58K    :: About GZip'd PS
      Filename format: author--year--title--journal|proceedings|...--pages   PDF-111K    PS-58K   

Related links
   Source: http://archive.ling.ed.ac.uk/documents/d.../00/00/22/index.html
   CiteSeer: http://citeseer.ist.psu.edu/79033.html
  Web search: Google Web Search   ::   Google Scholar
  Within this site: Cited by (5)    References (19)

Paper at a Glance

Learning, Bottlenecks and Infinity: a working model of the
evolution of syntactic communication
Simon Kirby
Language Evolution and Computation Research Unit,
Department of Linguistics, University of Edinburgh
simon@ling.ed.ac.uk
Abstract Human language is unique in having a learned, arbitrary mapping between meanings and signals that is compositional and recursive. This paper presents a new approach to understanding its origins and evolution. Rather than turning to natural selection for an explanation, it is argued that general properties of the transmission of learned behaviour are sufficient to explain the particular properties of language. A computational model of linguistic transmission is described in which complex structured languages spontaneously emerge in populations of learners, even though the populations have no language initially, and are not subject to any equivalent of biological change. These results are claimed to be general and are explained in terms of properties of mappings. Essentially, as mappings are passed down through generations of imitators, syntactic ones are intrinsically better at surviving through the learning ``bottleneck''.
1 Introduction Why does human language have certain properties and not others? This is the central question for linguistic ex­ planation. Of particular interest to linguists are the prop­ erties that human language has that appear to make it unique among communication systems in the natural world. One such property is syntax: the mapping between mean­ ings and signals in human languages is uniquely compo­ sitional and recursive. That is, the meaning of a signal is composed from the meanings of parts of that signal, and furthermore, arbitrarily complex meanings can be ex­ pressed by embedding signals inside signals. A common approach in linguistics to the explanation of these basic properties of syntax is to appeal to innately given properties of our language faculty. The Chomskyan approach (e.g. Chomsky, 1986),
...
BibTex
@inproceedings{kirby99learningBottlenecks,
  author={S. Kirby},
  title={Learning, Bottlenecks and Infinity: a working model of the evolution of syntactic communication},
  year={1999},
  editor={K. Dautenhahn and C. Nehaniv},
  booktitle={Proceedings of the AISB'99 Symposium on Imitation in Animals and Artifacts},
  url={http://www.isrl.uiuc.edu/~amag/langev/paper/kirby99learningBottlenecks.html}
}


 HOME   ::  Conference List   ::   Conference Paper Comments to: junwang4 you-know-at gmail.com Last update: 2/2/08