Automated SignSpelling Sequences

Wed Mar 3 17:05:37 UTC 2010

SignWriting List
March 3, 2010

Hello Kate!
Thank you for this message below.

Yes, we all love the idea of "automated" SignSpellings. And I am grateful to Steve for creating software that let's us all put in our own ideas about what the sequences should be, because there are several theories out there, and each one is based on logic that is logical ;-))

Meanwhile, I do have my own theory, based on asking people to find signs in a SignBank database that had around 3000 signs...and their reaction was different than I had expected...I too assumed that Location should be placed early in the SignSpelling Sequence, only to find that people found the signs quicker, if they looked up by beginning handshapes, movement, and ending handshapes, almost like a "configuration" that their eyes got used to..."hands-movement-hands"...and then the Location was put after that...the same hands-movement-hands at different locations... starting with neutral signing space, and then placing the same configuration in other locations from high to low...

So I respect the other theories too...we are all learning together and we need time to experiment...

Also, Steve needs time to program the automation which will also take time...

We have a temporary automated SignSpelling Sequence already, but it needs more special programming to make it fit any one theory properly...

So we are talking 2011, before any clean automation is accomplished...I think you should go ahead with your project, Kate, assuming there will not be automated SignSpellings to help you right now...You can see what we have at the moment, by clicking on the BSW analyze button underneath each sign entry in SignPuddle Online and then click on Sort in the BSW page...that gives you the current automated SignSpelling Sequence if the writer did not enter one themselves...

And you are right, that there is tremendous variation in the SignSpelling Sequences that are currently in SignPuddle Online...but that is not necessarily true that they are "mis-spelled" because there are several theories as to what the sequence should be, based on how fast deaf kids can look up signs in dictionaries that are sorted by sign-symbols...

That is great if the avatar programming does not need more detailed Location Markers, but if it does, it can be programmed too...it is just a matter of knowing what to do...so I look forward to seeing your work...

Val ;-)

-----------

On Mar 3, 2010, at 1:21 AM, Kate wrote:

> Hi Val
> 
> I really love the idea of "automated" SignSpelling, which I think is best for avatar work.
> There should be one standard way of writing SignSpelling sequence.
> The "automated" SignSpelling is/will be consisted, less confusion!
> 
> Some of the SignSpelling sequence on Signpuddle are 'mispelled' ( I think )
> 
> I'm not using Location Markers, Val. But once I've uploaded my project online,
>  then you can evaluate it.
> 
> When will the "automated" SignSpelling be available for download, I cant wait!
> 
> Kate
>  
> 
> 
> 
> On Sat, Feb 27, 2010 at 7:49 PM, Valerie Sutton <sutton at signwriting.org> wrote:
> SignWriting List
> February 26, 2010
> 
> Hello Rachel, Charles, Steve and everyone -
> 
> Steve has already programmed an "automated" SignSpelling Sequence. The automated version of the SignSpelling Sequence is not perfect, but it is very close to "one" theory on how to arrange the sequence...Based on the dominant hand first, and the second hand second, in the beginning handshapes...and so forth...
> 
> The automated SignSpelling Sequence has not been placed into the actual SignPuddle software yet, but right now, you can take a look at the automated SignSpelling Sequences by clicking on the BSW Analyze button under each sign in SignPuddle Online, and it will show you an "automated" SignSpelling Sequence, IF the writer did not already state what the SignSpelling Sequence should be...the good thing about this is that then, if someone forgets to put in the SignSpelling Sequence in their own way, then the "default" will kick in...
> 
> The default is not based on what symbol we wrote first...if we did that, then the Sequences would be a total mess! We would literally have to re-write every sign in SignPuddle, since no one has been asked to write in a certain sequence...For example, when I write a sign that is relating to the face...I write the face first, and then place the hand near the part of the face that the handshape is relating to...if we based the automation on what I wrote first, then the sign would begin with the facial expression, instead of the handshape...so that is why we programmed the automation differently...but it is working..I am really impressed with how close Steve has gotten the automation to work, based on the dominant hand that begins the sign...
> 
> The automated SignSpelling Sequences do not use the Detailed Location Markers in SymbolGroup 30, which I believe you might want for the SignTyp linguistic database...
> 
> The Detailed Location Markers are not written at all, for everyday use. They don't exist in the writing of the sign, because we can see the Location in a different way, when writing for everyday use. So the automated SignSpelling Sequences cannot include those symbols, since they are not written.
> 
> BUT...we can go in manually and place the Detailed Location Markers into any SignSpelling Sequence we want to...so this means that we, the transcribers, working on the SignTyp database, will have the job to go in and place those Detailed Location Markers into the SignSpelling Sequence data...It will be a fun job!
> 
> Thank you, Rachel and Charles, for these discussions, and for working with SignWriting and SignTyp - I am really looking forward to using SignTyp in the future...I know it will be a valuable database...;-)
> 
> Val ;-)
> 
> --------------
> 
> 
> 
> On Feb 26, 2010, at 5:17 PM, Rachel Channon wrote:
> 
> > Hello - I really like the idea of setting up a "default"
> > SignSpellingSequence.  I wonder if it could be done on the basis of the
> > order in which the person puts the symbols in the  SignWriting composition
> > area.  So for a sign like ASL WOMAN where the hand touches the chin and then
> > the chest, the writer might start by putting a symbol for the face in the
> > area.  Then they move the hand-orientation symbol to the chin.  Then they
> > add a second hand symbol at chest level.  Then the default sequence for hand
> > location would be hand-at-chin followed by hand-at-chest.
> >
> > (please forgive and correct any terminology errors here).
> > Rachel
> >
> > -----Original Message-----
> > From: sw-l-bounces at majordomo.valenciacc.edu
> > [mailto:sw-l-bounces at majordomo.valenciacc.edu] On Behalf Of SignWriting
> > Sent: Friday, February 26, 2010 10:26 AM
> > To: SignWriting List
> > Subject: Re: [sw-l] SWML site
> >
> > SignWriting List
> > February 25, 2010
> >
> > Hello Kate and Steve and Charles -
> > It is great to see how useful BSW is for programming...
> >
> > There is one interesting issue I would like to bring up...
> >
> > Since the SignSpelling Sequences are put in manually by the writer, and
> > since there are several different theories on what the Sequence should be,
> > it means that programmers who do not know sign language or SignWriting, if
> > they depend entirely on the sequences entered by others, may receive an
> > inconsistent group of sequences....
> >
> > The ASL Dictionary puddle online is a free and open database that has not
> > been edited much...Any writer is welcome to add signs and their own feelings
> > on what the SignSpelling Sequence should be...some people are still learning
> > to write signs, and others are new to the software and are learning how to
> > use it, and others are new to adding Sequences...so the inconsistency may be
> > confusing for your poor avatar! ;-))
> >
> > And I suspect that the SignSpelling Sequences used for programming an avatar
> > might be different than the sequences used for sorting dictionaries for deaf
> > children...because I would assume that the programming needs more detail to
> > really see the exact location for the avatar, but when deaf children are
> > looking up signs in dictionaries, my experiences so far show, that detailed
> > SignSpelling Sequences are not necessary for simple look-ups...however, my
> > experiences are not a real research study at all...just my guess...so a real
> > study does need to be done, to find out what theory on SignSpelling
> > Sequences works best for everyday use, and what is necessary for other
> > projects, such as programming for avatars...
> >
> > And Kate...you mentioned you thought it was automatic...Steve is working on
> > the idea of "automated" SignSpelling Sequences that could be the computer's
> > "best guess" at the sequence, based on the alphabetical order of the symbols
> > in the ISWA, and so forth, but then the writer can still go in and change
> > the automated sequences, when necessary...that will be a WONDERFUL step
> > forward, because oftentimes the automated sequences are pretty close to what
> > the writer would have chosen anyway...however...the automated Sequences have
> > not been programmed yet into SignPuddle, so right now we are required to do
> > the sequences manually every time we write a sign...
> >
> > You can find the automated and manually created SignSpelling Sequences for
> > each sign in SignPuddle right now, when you click on the BSW Analyze button
> > under each sign in SignPuddle...
> >
> > And when SignPuddle 2.0 is released, we will be able to have SignSpelling
> > Sequences in the signs inside documents too, so we will be able to search
> > documents for symbol frequency and so forth...I wish those could be
> > automated as well...
> >
> > so for your avatar work, I might do another sequence than I would for deaf
> > children -
> >
> > for your avatar work, I would put in more detailed Location Markers from
> > Group 30, in the sequence, so your avatar would know some details of
> > location for exact sign production...although I may be wrong...maybe your
> > animation can move just as well without it and if so that is super!
> >
> > Val ;-)
> >
> >
> >
> >
> > On Feb 25, 2010, at 6:58 PM, Steve Slevinski wrote:
> >
> >> Hi Kate,
> >>
> >> Yes, the <seq> tag represents a symbol in the SignSpelling Sequence.  If a
> > sign doesn't have any <seq> tags in the XML, then the sign entry does not
> > include a SignSpelling Sequence.  The sequence is manually entered.
> >>
> >> You can use the new SignText online.  I'll have the new SWIS downloads
> > ready in early March.
> >>
> >>
> >> Considering the sign in your email, an ASL puddle sign for deaf, we can
> > look at it together:
> >> http://signbank.org/signpuddle1.5/canvas.php?ui=1&sgn=4&sid=5206
> >
> 
> 
> 
> 
> 
> ____________________________________________
> 
> SW-L SignWriting List
> 
> Post Message
> SW-L at majordomo.valenciacc.edu
> 
> List Archives and Help
> http://www.signwriting.org/forums/swlist/
> 
> Change Email Settings
> http://majordomo.valenciacc.edu/mailman/listinfo/sw-l
> 
> 
> 
> 
> 
> ____________________________________________
> 
> SW-L SignWriting List
> 
> Post Message
> SW-L at majordomo.valenciacc.edu
> 
> List Archives and Help
> http://www.signwriting.org/forums/swlist/
> 
> Change Email Settings
> http://majordomo.valenciacc.edu/mailman/listinfo/sw-l

____________________________________________

SW-L SignWriting List

Post Message
SW-L at majordomo.valenciacc.edu

List Archives and Help
http://www.signwriting.org/forums/swlist/

Change Email Settings
http://majordomo.valenciacc.edu/mailman/listinfo/sw-l