Publishing Partner: Cambridge University Press CUP Extra Publisher Login

Discussion Details

Title: Re: Disc: Starling Study: Recursion
Submitter: John Limber
Description: The sentiments of Geoffrey Pullum and others (see Discussion item are similar to my own
expressed to a different discussion group in regard to Gary F. Marcus'
(Nature 2006 440:1117) overview to Genter et al (2006).

Professor Marcus is quite, shall I say generous, in his attribution of
recursion to these birds. My own bet is that they’ve cleverly learned to
count up to four and match to successive samples -- a capability long
suspected in corvids and other species.

(This, by the way, is a non-trivial accomplishment— humans probably can’t
do very much better without conscious counting. See Koehler, O. (1956).
The ability of birds to 'count'. In J. R. Newman (Ed.), The world of
mathematics (pp. 489-496). New York: Simon and Schuster.)

For example, if presented aabb, they see aa and respond positively if the
following stimulus, bb, is of the same number, two. Marcus certainly knows
that recursion is a theoretical concept enabling generation of an infinite
set from a finite set of rules — not something one can see.

Yet in this study only a small subset of 'sentence' strings is actually
used. Why then would anyone want to 'generalize' from performance on a
very small n, up to four, to the infinitely large number implied by

The motivation for ascribing recursion to English is based on a number of
considerations — including the reasonable but theoretical idea that any
limit on 'depth' of recursion (magnitude of n in AnBn below) is a matter of
memory and not a fact of English. Another assumption is that the same
lexical, phonological, syntactic and semantic processes are used in clauses
whether n=1, 2, 3, 4,....43... 1176... In any case, attribution of
recursion in human language is hardly based on observing an infinity of

Getting back to the starlings — none of these sentence processing issues
ever arise and we seem to have a possible pre-existing alternative
explanation of their behavior -- successive match-to-sample. Perhaps we
should be less generous in our interpretation of these data as
demonstrating recursion?

Koehler himself thought birds' numerical skills might be relevant to the
understanding of language evolution; Genter et al (2006)still might have
something useful to contribute on this issue but I don't see it yet.

(This piece of research provides an amusing counterpoint to the recent
claim (which I haven't read yet) that the Amazonian Piraha language is not
recursive! The notion of 'recursion' seems to have gone beyond its
mathematical or linguistic significance into the realm of language politics.)

The experience Jackendoff, Pullem et al had with Nature only supports my
belief on this latter point.
John Limber
Durham, NH
Date Posted: 21-May-2006
Linguistic Field(s): Philosophy of Language
Language Acquisition
LL Issue: 17.1553
Posted: 21-May-2006

Search Again

Back to Discussions Index