[sw-l] Frustrations of SignWriter DOS users...

[Sandy Fleming]

Hi Stefan!

I hope you don't mind me playing devil's advocate and saying that SW-DOS was
more silver than gold!

I do think the SW-DOS keyboard got most things right and that as far as
keying goes I hope future SW word processors are very similar to it - not on
the screen, but on the keyboard.

One thing that people seem to like that I don't think is such a good thing
is developing a dependency on typing glosses to pull signs from dictionaries
as a way of typing text. I think this is bound to encourage bad habits such
as not bothering with variations in facial expression, not bothering to use
lanes, and not bothering with degrees of meaning. The idea of using
dictionary glossing for fast input means that we're depending on the fact
that written oral languages are ahead of SW in this respect, when we should
actually be concentrating on how to bring pure SW input methods up to speed
so that they're on a level with oral languages instead of dependent on them.

Of course, this means that SW writers have to know all the spellings in
their head, just as oral language writers can spell almost everything almost
without thinking.

So we need to find ways of speeding up input (as well as bringing education
to the point where people can spell well).

I'm going to suggest some ways that input might be speeded up.

Automatic positioning - there are the most usual positions for the head and
hands, but also sometimes the hand position can be related to a contact
symbol or the curve that marks the part of the head that the hand is placed
near. It won't be possible to make this 100% reliable bur it would reduce
the amount of repositioning required after a sign is typed. One of the best
things about SW-DOS is that it already does this quite well, so it's
probably a "must have" for future word processors).

Automatic insertion of non-dominant hand. A universal in sign languages is
that the non-dominant hand often mimics the dominant hand in either a cyclic
or simultaneous motion. Once the typist has put in the dominant hand and its
motion, it should be possible to type a keypress to insert the non-dominant
hand and its motion moving simultaneously, and another keypress (perhaps the
same keypress, but shifted), to insert the non-dominant hand moving
cyclically.

Quick access to the universal handshapes. A universal in sign languages is
that four handshapes (flat hand, "5" hand, index and fist) are used much
more frequently than any other handshapes. A word processor should offer
quick, maybe even single-key, access to each of those handshapes.

Improved selection of orientation. Currently, SW input programs typically
allow us to cycle between the six orientations of any displayed handshape. I
think it would make for faster typing if there were one key to flip between
floor and wall plane, and another to cycle through the three handshapes in
that plane. Possibly (keypress) to cycle the three orientations, and
shift-(keypress) to flip planes.

Language statistics. Once we have large dictionaries built up in SWML (eg
the Puddle dictionaries), it should be possible to analyse the SWML in a
database to find what sort of things appear very frequently and what rarely
occurs, either in a given language or as sign language universals. Word
processors could be designed to make more frequent patterns more readily
available.

Use of dictionaries without glossing. Obviously there's spellchecking, but
also there's the possibility that predictive input might be more effective
for sign languages than it is for oral languages. So we might be able to
type just some of the symbols until the software recognises what we're doing
as a unique sign from the dictionary (this does encourage some of the bad
havits of the glossing approach though - there's probably no real substitute
for knowing how to spell!).

What else?

Sandy