[erg] missing EP in a qeq relation

Xuchen Yao xuchen at coli.uni-saarland.de
Fri Aug 13 23:44:43 CEST 2010

Hi Dan,

Thanks for correcting the bug. I confirm it has gone away with the 
latest update of ERG.


On 08/13/2010 11:12 PM, Dan Flickinger wrote:
> Hi Xuchen -
> You're right that the ERG had a faulty definition for the generic noun type that we use for the token |1700s|, such that the HOOK.LTOP value was not identified with the relation's LBL.  I have fixed this error, and also corrected an unrelated issue with an entry for "with", and checked these changes in to the `trunk' version of the ERG.  (The `1004' version is now frozen/stable.)  So I would recommend that you try out the trunk version, at least to see if this problem has been resolved.
> Cheers,
>   Dan
> ----- Original Message -----
> From: "Xuchen Yao"<xuchen at coli.uni-saarland.de>
> To: erg at delph-in.net
> Sent: Friday, August 13, 2010 9:24:53 PM
> Subject: [erg] missing EP in a qeq relation
> Hello,
> My code broke with the following sentence, which might indicate a bug in
> ERG:
> there was a problem with the laws in the 1700s in England.
> I used LKB to parse this sentence. Lots of parses were output. But
> starting from the 2nd parse tree (I checked until the 4th), the MRS is
> malformed. As can be seen from the attached picture, there's a "h30 qeq
> h38" relation but "h38" can't be found from the MRS.
> I'm using ERG 1004 and the LKB version is: Date: 2010-04-28 14:29:38
> -0700 (Wed, 28 Apr 2010)
> Xuchen

More information about the erg mailing list