LINGUIST List 2.111

Tuesday, 2 Apr 1991

Disc: Gallistel, Shoebox, Elgin, WordPerfect,

Editor for this issue: <>

Directory

1. Vicki Fromkin, Re: Conferences: SAMLA, Southeast Asian, SPP
2. Thomas E Payne, Shoebox
3. , Ozark English
4. Wayne Cowart, WordPerfect and special characters
5. "ACAD3A::FFJAL1", WP fonts

To: Bloom et al organizers of SPP -- Randy Gallistel is at UC, Los Angeles
(UCLA) NOT UC, Berkeley. We claim him as our own and want all the world
to know it. Please correct in future announcements.
Vicki Fromkin

Before everyone gets excited about Shoebox, let me add a note of caution.
I have to date implemented Shoebox on a 2670 clause corpus, and it is currently
being used in a field methods course here. So far we have encountered several
bugs and a few conceptual problems. I won't dwell on the bugs, as these should
we worked out as new versions emerge. I will mention a couple of conceptual 
problems.
 First, since the program is geared to generating dictionaries, it 
automatically alphabetizes all input by the first field in each record. You
can imagine what happens to text when the sentences come out in alphabetical
order, rather than in the natural order in which they were uttered! We had
to contrive a "dummy" field that fooled the program into keeping the text in
its natural order -- and still there are some record ordering problems that
I can't decipher.
 Second, there is no way to "jump" to, say, record number 1897, and have
access to that record and its context. You can goto 1897, but the "previous
record" command does not take you to 1896. Rather you go back to the record
prior to the record you issued the "goto" command from. If this sounds con-
fusing, the result is this: if you want to start counting or reading text
sequentially starting somewhere in the middle, you have to manually "browse"
through the entire database to get to the point you want. Fortunately, you
can start at the end or the beginning. But on my AT machine it takes about
10 minutes of hitting "PageDown" to get to the middle of my database.
 These conceptual problems plus the bugs make me hesitant to recommend
Shoebox wholeheartedly to the linguist interested in using it for interlinear
text work. I will say that if you are working primarily with dictionaries,
and/or smaller databases, it may be more useful to you than it has been to me.
 I will also say that I have appreciated all the hard work that John
Wimbish has put into this, and that it is a quantum leap beyond IT in terms
of the range of database operations that it can perform (or at least is
designed to perform). Like most non-commercial software, however, it is 
rather tricky and requires somme tinkering to make it work the way you want
it to. For those who want to just "use" computers, and who don't appreciate
spending hours pulling their hair out over uncooperative software, I would
say either wait a year or so, or get someone else to implement the program
for you.
 Respectfully, Tom Payne

Regarding Pamela Munro's posting (107-2): Suzette Haden Elgin lives in
Huntsville, Arkansas -- not Alabama.

Bill Reynolds

Several items lately have noted that WordPerfect (5.0 & 5.1)
allows the user to print virtually any special character on
any printer. This is true, and it is a valuable feature.
But anyone contemplating spending anything significant to
get this capability might want to proceed with a little
caution. When WP prints one of those 1500 or so characters
that is in CHARACTER.DOC or CHARMAP.TST but that is *not*
supported by your printer, it prints the character in
graphics mode. This works fine, but with many possible
computer/printer combinations it can be excruciatingly slow.
Even rather modest documents with lots of special characters
could turn out to be overnight jobs. Some realistic testing
with your kind of document on the specific hardware
configuration you plan to use may be in order. A good easy
test is to have WP print CHARMAP.TST (this is distributed
with WP and should be in the directory where you find WP).
Be sure to try this with Graphics Quality set to High.
If this is unacceptably slow, you might want to think about
a printer that can handle soft fonts (and a soft font with
the characters you need), a laser printer (with adequate
memory) or other options.

Wayne Cowart

>From Jeff Leer:

 My problem with the WordPerfect fonts is that to get all the
combinations of characters and diacritics you are forced either to
use Compose or else code in a complicated sequence of Advance commands.
The problem with Compose is that it is IMPOSSIBLE to search for a given
Compose combination, which means it is impossible to do a global
search and replace for a given Compose combination if you want to change
it to something else. I even wrote a letter to WordPerfect Corp. a couple
years back, and they didn't know of any way around that limitation.
 Anybody know of anything like Superfrench for MS-DOS machines?
They use that on the MacIntosh, and it's really nice. It counts diacritics
as ordinary ascii characters, e.g. accented a is stored as "a;" or the like.
You can pile on as many diacritics as you please, and search and replace
at will.

[End Linguist List, Vol. 2, No. 111]