[SHOEBOX] Null Morphemes

At 19:32 21.7.2000 -0400, Jeff Sheets wrote:

>I have tried everything I can think of to get Null Morphemes to work, and I
>think we've probably found something Shoebox doesn't handle gracefully.  I
>know that I would have no interest in adding very many alternate forms  per
>noun and pronoun into a lexicon.  :)  Variables don't seem to be interpreted
>unless reduplication is used, so I think that is out.  The only alternative I
>can see to the inelegant solution of multiple alternate forms would be to
>leave it out and assume that it is there... or add -0 like David has been
>doing.  I have emailed SIL about it to see if they have a better solution.
I think it can be done.  Add whatever symbol you like for "null" to your
alphabet in the Language Encoding Properties dialog box, taking care to add
a Variable "Null" to the Variables tab.(I would advise against using the
digit 0, simply because in most fonts it is too similar to the letters
oO.  The uppercase Danish letter Ø (Oslash) might do if you haven't got øØ
in your orthography.  I use -- generally, not yet in Shoebox -- the
international currency symbol ¤, since it is similar to a crossed-out
circle, which some scandinavians use as an empty set symbol.)  Then you
must add your null symbol to the lexicon.  Look up "Context
Sensitivity—Preventing Incorrect Parses" and "Underlying Forms" in the help
file for suggestions how it may be done.  Note sepecially in "Underlying
Forms" how the polysemy of _hit_ is handled.
In David's specific case I think the lexicon entry for _el-_ might look
like this:
   \lx el-
   \u el-
   \u el-¤
   \ps MOOD
   \ge assertive
which means that you do have a choice between two underlying analyses for
each occurrence of _el-_, but you do not have to make lexicon entries for
all surface combinations of _el-_ with the various suffixes, and the
interlinear parse will look *much* nicer! :-)  I assume you will have to
add the null morpheme (or several null morphemes with different PoS and
glosses to the lexicon, as e.g.:
    \lx ¤
    \a
    \u ¤
    \ps AUX
    \ge is
I'm not sure you really can leave an empty \a field and get the right
result.  It will have to be tried.  If not, delete the empty \a field and
hope Shoebox won't mind that the null morpheme(s) never actually occur on
the surface!


/BP 8^)>
--
  B.Philip Jonsson mailto:bpX@netg.se mailto:melrochX@mail.com (delete X)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"Truth, Sir, is a cow which will give [skeptics] no more milk,
and so they are gone to milk the bull."
                                    -- Sam. Johnson (no rel. ;)