[developers] processing of lexical rules

Ann Copestake Ann.Copestake at cl.cam.ac.uk
Tue May 3 01:27:15 CEST 2005


I `tested' your active parser changes on a couple of ERG sentences and hereby
ask Ben to test with test suites etc ...

orthographemicp - seems good - but I think we will still need the flexibility
to override how this gets set as a user-fn, since I envisage some grammars
using alternative morphophonemics.

I didn't allow for recursive application in irregular forms but I think the
changes needed are very minor.  prefixation in MWEs - not too difficult to
allow for, but not entirely trivial.  At the moment the assumption is that it
will apply at the same place as suffixation and if we have two positions I
would prefer to make it generally settable.  We have to work out how to create
a partial-tree by merging two sets of rule specifications since I assume there
might also be suffixes.  If we're going to do this, we might as well allow for
multiple positions for any affix, unless we know this never happens in any
language!  I take it these changes are not really urgent ... 

Ann



More information about the developers mailing list