This is a spelling correction program for the Explorer.  It
implements two ZWEI commands: M-X Correct Spelling and M-$ (correct
region or word spelling).  To run the program, load the defsystem and do
a make-system on 'speller.  Both of the correction commands will be
available in ZWEI.  Spelling correction is quite fast --- instantaneous
if the word is correct, a few seconds if SPELLER has to search for
possible replacements.  


     M-$ will correct either a region or the word at the point if no
region is defined.  For a single word, it will display "OK" in the
mini-buffer if the word is correct.  If not, a menu will pop up.  The
bold menu entries are replacement suggestions.  Click on one of those to
replace the word in the buffer with the suggestion.  The italic entries
are commands to SPELLER.
  
     "Skip" will cause the speller to not correct this word.  
     "Type-in" will let you type in a replacement word (which will
               not be checked).  
     "Abort" ends spelling correction.  
     "Accept" tells SPELLER to assume this word is correct in all
              further occurrences in the document.

The speller can also prompt you for the same commands in the minibuffer.
People who like to keep their hands on the keyboard will prefer this.
It is also somewhat faster than popping up a menu.  Just set
SPELLER:*PROMPT-IN-MINIBUFFER* to t.  

M-X Correct Spelling will correct the spelling of the entire document.
It starts at the beginning of the document looking for words in error.
When one is found, it is highlighted with reverse video and the menu
described above is displayed.  Unless aborted, it will continue this
until the end of the document.  The point is restored to the same
location it had when the Correct Spelling command was entered.  M-$
behaves like M-X Correct Spelling when correcting a region.

     As of right now, there are no provisions for either user
dictionaries or document specific dictionaries.  (The Accept command
is specific to the editor buffer and thus fairly temporary.)  There is
also no provision for modifying the main dictionary.  Useful extensions
would include document specific and user specific dictionaries with the
ability to edit them.  If anyone would care to write such extensions,
send them to me and I will include them in future distributions.

     The dictionary is based on the Ackerman/Mattson/Gorin/Adler
dictionary that was used on some CPM systems.  While a little bit
lightweight by today's standards, it is about 40,000 words long and
fairly heavy on technical words.  It contains some place names (states
and a few cities) and some personal names.  Where there are two
acceptable variants for a word, e.g.  "labelled" and "labeled", only one
is represented in the dictionary.  The theory is that consistency is
more important than artistic license.  

     When the make-system is done, a process is spun off to load the
main dictionary.  This takes a couple of minutes, so if you try to use
one of correction commands immediately, you will get a message "Loading
main dictionary..." and will wait until the load completes.  If the
process loading the dictionary gets an error, it will quietly die and
the speller will retry the load when a spelling command is first used.
Any errors on that load will be visible and correctable by you.

     This code is made available with the usual hacker restrictions: It
is not to sold nor included in any package that is sold.  Otherwise, it
may be copied freely.  I would appreciate it if any modifications would
be sent to me for inclusion in the package.  I would also like to hear
of any bugs, although fixes are neither promised nor likely.  I can be
reached at JOHNSON@TI-CSL.CSNET-RELAY or

Douglas Johnson MS 238
Texas Instruments
P.O. Box 226015
Dallas, Texas 75226

 -- Doug