[matrix] matrix - general: no parse of words

[Emily M. Bender]

Dear Gertrud,

Thank you for these questions.  I will try to provide helpful
responses!

> this is the first of a number of questions on the type hierarchy of the
> grammar matrix, each we will post separately, for they touch different
> issues.

And thanks for this too --- it makes it easier to reply.

> This question is about parsing of stand-alone utterances with the
> original downloaded matrix.
> 
> The downloaded root file describes that stand-alone utterances could be
> parsed as well with the following type descr:
> 
> root := word-or-lexrule.
> 
> However, the example lexicon sets the type of e.g. a noun as:
> "common-noun-lex". We did only find noun-lex which is a subtype of
> word-or-lexrule, so we added:
> 
> common-noun-lex := noun-lex .

I'm not sure where common-noun-lex comes from.  Have you added
it already?  The grammar I get from putting your choices file
into the customization system only has noun-lex.  At any rate,
this orthogonal to the issue you are asking about.  (And positing
a type common-noun-lex is a perfectly sensible thing to do!)

> Loading the grammar went ok, however, the single noun of our lexicon was
> still not parsed. Can you please give us a hint?

The trouble is that the grammar we provided to you failed to
fully take account of the lex-root type.  That is, in order to define
an alternate root condition, you have to do two things:

1) define it (with a distinct identifier) in roots.tdl.
This was done properly.

2) add it to the list that is the value of *start-symbol*
in globals.lsp.  This is what we failed to do.

To solve this, edit the file globals.lsp, and change the
definition of *start-symbol* so that it looks like this:

(defparameter *start-symbol* '(root lex-root)
  "specifing valid parses")

Please let me know if this works.  I will try to answer your
other questions soon.

Emily