long-distance control

[Daniel Everett]

This also makes perfect sense to me, Gilles. I cite this research in my new book, Cognitive Fire: Language as a Cultural Tool, to appear late next year or early 2011.

This is similar to a point I make in a paper to appear in a volume on recursion being edited by Peggy Speas and Tom Roeper, where I argue that recursion is vital to thought and serves as a tool in language which can, among other things, increase the information rate of individual sentences. 

Even tests for embedding in grammars depend to a large degree on arbitrary distinctions between sentences vs. discourse. Intonation, for example, can take a single multiclausal sentence or several sentences (e.g. intonational paragraphs) in its scope.

Herb Simon's 1962 paper on the Architecture of Complexity discusses the importance of recursive handling of information across a range of tasks, including organizing a small watch repair business - the nonrecursive approach to watch repair will fail if competing with a recursive approach because it is less efficient, less able to take interruptions in construction of watches. 

Phil Lieberman has an interesting new paper to appear "The creative capacity of language, in what manner is it unique, and who had it? in which he argues for the importance of 'reiteration' as a more general case of which recursion is a subcase.

-- Dan

On 23 Dec 2009, at 13:41, Gilles Fauconnier wrote:

> 
> Interesting exchange between Dan and Tom.  In cognitive linguistics, we go perhaps one step further.  The powers of recursion, and more generally integration, lie in the human capacity to build vast networks of mental spaces with no discernable limits on the depth of embedding or the iteration of integrations.  Discourse, narratives, the cultural evolution of mathematics, religious thought, all display these powers.
> 
> What's remarkable about sentence syntax is actually how little recursion it has (as a formal combinatorial system), compared to meaning construction.  This is because language only needs to prompt for the
> recursive thought processes.  It does not reflect them directly.
> 
> Gilles
> _________
> 
> 
> Gilles Fauconnier
> Department of Cognitive Science
> University of California San Diego
> La Jolla CA 92093
> 
> E-mail    gfauconnier at ucsd.edu
> http://cogsci.ucsd.edu/~faucon/
>