LINGUIST List 10.979

Thu Jun 24 1999

Disc: Dependency Trees

Editor for this issue: Karen Milligan <karenlinguistlist.org>

Directory

1. Marco Antonio Young Rabines, Dependency trees

Dear listers:

A week ago I posted a query to The Linguist asking for hints on where on the
net one can get information on the mathematics of dependency trees. As I
have only received one (though very useful) response, I would like to tackle
a discussion with the following (which I've aleady started to 'run' in
another list):

As I am no Dependency Grammar-literate (just someone who's scanned scarce
www literature on the matter), I would like you to let me know if there is
in DG any formal device analog to the one widely used in some mainstream"
constituent grammars (PP and HPSG) in order to deal with such phenomena as
correference, unbounded dependencies and binding (something like
c/o-command).

As for trees and the like, do you think there could be a unified treatment
for constituent and dependency trees regarding hierarchical relations, for
example? If these relations are important for DG (I don't know it), could
they receive a unified treatment, say, in graph theoretic terms? In the case
of constituent trees, dominance and c-command (and their variants among
different phrase structure theories) often seem to depend upon the set
theoretic notion of inclusion. Therefore, the constituent style of defining
hierarchical relations cannot be a candidate for a unified approach.

I know that the simple fact of considering the possibility of a unified
treatment is not justified per se (there isn't such a thing among
constituent grammar theories, not even in a single theoretic framework); but
it could be useful, in which case it should be demonstrable. I do not mean
that there ought to be the same hierarchical relations for all theories. But
if they are a component of a theory, it would be more economical to resort
to some common notions in order to define them, no matter the particular
theory we're dealing with.

And last but not least, do you think that constituency is the most important
feature of grammars that employ constituent trees? What's your opinion?

Best regards,

Marco Antonio Young Rabines
Departamento de Linguistica
Universidad Nacional Mayor de San Marcos
Lima