[developers] divergences between `old' and `new' LKB branches

Stephan Oepen oe at csli.Stanford.EDU
Mon Jun 20 16:06:36 CEST 2005


hi again, ann,

montse agreed to forwarding her grammar.  i attach the archive below.
as explained in my previous message, with a ISO-8859-1 locale, i can
parse using this grammar (after reformatting `es_inflr.tdl', see the
earlier message).  not in the new branch, though: i get the effects
documented in the log below.

  - the input that parses in the old code (`el niño lloró') fails to
    generate a full result.  the derivation i expect is:

    (10 NSUBJH 0.0 0 3
     (5 HSPEC 0.0 0 2 (2 EL_D2 0.0 0 1 ("el" 0 1))
      (3 MASC-SING-NOM_INFL_RULE 0.0 1 2 (4 NIÑO_N1 0.0 1 2 ("niño" 1 2))))
     (8 1CONJ-3RD-SING-PPAST-IND-VERB_INFL_RULE 0.0 2 3
      (9 LLORAR_V1 0.0 2 3 ("lloró" 2 3))))

    manual unification of edges 19 plus 36 into HSPEC succeeds.

  - what are the edges numbered 18 -- 31?  i suspect they correspond to
    distinct hypotheses about how to inflect `el'?  if so, it is worth
    packing these (minimally, identical lexical entries with a common
    prefix of required orthographemic derivation should be collapsible
    into a single edge)?  edges would then have to be decorated with a
    disjunctive set of partial trees.

the latter might be a pure efficiency thing and possibly not worth it?

                                                  all the best  -  oe

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+++ Universitetet i Oslo (ILN); Boks 1102 Blindern; 0317 Oslo; (+47) 2285 7989
+++     CSLI Stanford; Ventura Hall; Stanford, CA 94305; (+1 650) 723 0515
+++       --- oe at csli.stanford.edu; oe at hf.uio.no; stephan at oepen.net ---
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

-------------- next part --------------
A non-text attachment was scrubbed...
Name: es_iula-v0.2.tar.gz
Type: application/x-gunzip
Size: 77178 bytes
Desc: not available
URL: <http://lists.delph-in.net/archives/developers/attachments/20050620/c627f04a/attachment.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: es.log
Type: application/octet-stream
Size: 42044 bytes
Desc: not available
URL: <http://lists.delph-in.net/archives/developers/attachments/20050620/c627f04a/attachment.obj>


More information about the developers mailing list