Topology of DTRS - reference sought

[Ivan A. Sag]

Hi Steve,

> I have a strong recollection of reading a paper which proposed changing the
> topology of the HPSG DAUGHTERS attribute to something like the following
>
> HEAD-DTR  [1]
> NON-HD-DTRS [2]
> DRTS [1] o [2]
>
> I am unfortunately unable to recall the details of the source of this
> proposal. Can anyone help?

I'm not sure I've ever seen that one. My 1997 relative clause paper (Journal
of Linguistics) uses NON-HD-DTRS to replace the proliferation
of features: SUBJ-DTRS, SPR-DTRS, COMP-DTRS, etc. But that paper doesn't use
a DTRS feature.

In Ginzburg and Sag (2000) (essentially following the way our grammars are
implemented in the LKB system at CSLI), we declared features as follows:

phrase: [DTRS list(sign)]
hd-phrase: [HD-DTR sign]

This way, the redundant features are still avoided but the feature NON-HD-DTRS
is eliminated. The value of HD-DTR is identified with one member of the DTRS
list. This seemed like a cleaner way to set things up.

Best,
Ivan

-------------------------------------------------------------------------
Ivan A. Sag
Professor of Linguistics
WWW: http://www-csli.stanford.edu/~sag/sag.html
Stanford University - Stanford, CA  94305  USA
-------------------------------------------------------------------------