[matrix] some type descriptions in tdl file

[Emily M. Bender]

Dear Altaf,

First, my apologies for misspelling your name.  I think I
have it right now.

> 1. What is the type 'predsort' and why it is used?

Open class lexical items generally have strings as the value of
PRED inside their relations.  For closed-class items (as well as the
semantic contribution of syntactic constructions) it is sometimes
useful to create a small type hierarchy of these predicate names.
'predsort' is the top of such a type hierarchy.  The term 'sort'
reflects the fact that predicate names don't have features 'inside'
of them.

> 2. Why '+' and '-' are used as a sub-type of boolean? What does they mean?

These are used for binary features.  That is, features which only
need to contrast two values.  An example is the feature AUX (for
'auxiliary') in English.  Features with '+' and '-' as their possible
value are declared to take values of type 'bool'.

> Thanks. I think I still have lots of these 'basic' questions in future and
> need to post here.

Please feel free. 

For general background on the syntactic framework, I recommend Sag,
Wasow and Bender 2003.  The Matrix does not adhere exactly to the
theory as presented there (nor is the formalism the same), but it
should give you the general idea.  You should also have Copestake 2002
for information about the LKB.


-- Emily