Print

Print


Practically, I would think a lookup table would be needed - where every expressive sign has a receptive sign assigned to it and perhaps a mapping adjustment.  Then the symbol coordinates would need to be mirrored and the adjustments and substitutions made.

Bill

On 10/2/2012 11:47 AM, Ingvild Roald wrote:
[log in to unmask]" type="cite">
Hi Oscar,

I'm sorry, but I do not think that mirroring the video is a solution. It would be like viewing a left-handed signer, and you would have to remember the other conversions.
Mathematically, the receptive/expressive signing spaces are symmetrically placed/ mirrored according to a vertical line halfway between the signer and the viewer.
If all symbols were following the same rule when swirched between left/right and forward/backward, this could be managed mathematically.
In SignWriting, some symbols change filling (hand symbols), as well as orientation; whereas other symbols just change direction (forward/backward) and others stay just the same.
It is possible to work through the whole set of symbols and make sets/ groups of symbols that behave in the same way, and write the rules for them. But it is a big job.

I still wish you good luck with your work, and I'm sure that one day we will get this program, when someone gets the funding,

Ingvild

> Date: Tue, 2 Oct 2012 17:17:37 +0200
> From: [log in to unmask]
> Subject: Re: annotation in receptive rather than expressive view point
> To: [log in to unmask]
>
> Just another idea:
>
> what about simply mirroring the video to be transcribed? That could be
> easily done! And then, the transcription is easily done in expressive form.
>
> Oscar.
>
> Am 02.10.2012 14:56, schrieb Charles Butler:
> > Thanks Ingvild,
> >
> > I am steadily going through the DEIT LIBRAS dictionary and putting it on
> > line (with permission). It gives me a chance to read the description,
> > look at both the illustration and the signwriting and transcribe it
> > again. It's more than 8000 signs, so by the time we get it all in, it
> > should rival the ASL dictionary in scope. I'm hoping to get permission
> > to work with INES (the sign language school in Rio de Janeiro) to add
> > the signs they have which are not currently in DEIT Libras.
> > Charles Butler
> > [log in to unmask]
> > 240-764-5748
> > Clear writing moves business forward.
> > ------------------------------------------------------------------------
> > *From:* Ingvild Roald <[log in to unmask]>
> > *To:* [log in to unmask]
> > *Sent:* Tuesday, October 2, 2012 7:02 AM
> > *Subject:* Re: annotation in receptive rather than expressive view point
> >
> > Hi Charles,
> >
> > I was answering to the 'translate in your head' problem - which I see as
> > easier if it is done manually, not just cognitively. I do understand the
> > differnce between machine translation and manual/ personal translation,
> > and I do see your point. I hope that someday w will have an automatic
> > translation from video, but that will be at least as hard as
> > voice-recognition systems. For natural langugaes, that is most often a
> > huge task, because of the many personal and dialectical differences. - I
> > think that making an avatar signing from what is written in expressive,
> > to be viewed in receptive mode, is an easier and therefore more
> > available sort of program. To acheive this, the automatic transformation
> > between expressive and receptive writing would be a step. But this needs
> > time, and funding. In teh meantime, we have to struggle along doing this
> > tranformation by body and mind, and making the dictionary puddles as
> > well as the litterature puddles and others seadily bigger and better.
> >
> > Ingvild
> >
> > ------------------------------------------------------------------------
> > Date: Tue, 2 Oct 2012 03:36:31 -0700
> > From: [log in to unmask]
> > Subject: Re: annotation in receptive rather than expressive view point
> > To: [log in to unmask]
> >
> > Ingvild, I understand, what I am talking about is machine translation,
> > and the machine has to be taught to read a video, as is, receptive, and
> > translate that into expressive. The point of machine translation is just
> > that, having a sufficiently large recognition program to translate a
> > third-party into expressive. Humans can internalize and write, a machine
> > has to be taught to do so.
> >
> > Sometimes watching a video I may get a hand wrong, or a twist in the
> > wrong direction, and if I write down what I see, I can then correct it
> > to the opposite. That's my bias, and my early history with SignWriting,
> > writing whatever I see, whether on my hands or someone else's.
> > Charles Butler
> > [log in to unmask]
> > 240-764-5748
> > Clear writing moves business forward.
> > ------------------------------------------------------------------------
> > *From:* Ingvild Roald <[log in to unmask]>
> > *To:* [log in to unmask]
> > *Sent:* Tuesday, October 2, 2012 6:09 AM
> > *Subject:* Re: annotation in receptive rather than expressive view point
> >
> > As I see it, the easier way to convert from receptive to expressive for
> > writing, is to view the video and copy the sign you see by doing it
> > yourself. Then you write what you do, the sign as you are making it from
> > viewing the video. That is, do the translation /transistion manually
> > rather than 'in your head'.
> >
> > All the best,
> >
> > Ingvild
> >
> > > Date: Mon, 1 Oct 2012 15:02:50 +0200
> > > From: [log in to unmask]
> > > Subject: annotation in receptive rather than expressive view point
> > > To: [log in to unmask]
> > >
> > > Hi Charles,
> > >
> > > thanks for your answer.
> > >
> > > > Oscar, as you have an assistant annotating video tapes, trying to
> > > > translate in your head to an expressive point of view may be driving
> > > > you crazy.
> > >
> > > Yes, I have been thinking about that too. And it would be preferable to
> > > do annotation from the receptive view point. However, I need the
> > > transcriptions in an expressive view point in order to match all the
> > > other entries in SignPuddle (as I use them to initialise my system). The
> > > manual annotations are intended to serve as evaluation of the
> > > initialized system, thus they need to match.
> > >
> > > If there was an automatic (mathematical) way of converting receptive
> > > into expressive view points, then we could do the "easier" annotation.
> > > But I learnt from Steve Slevinski, that this has not been implemented
> > > and to me it doesn't seem trivial to implement it.
> > >
> > > Or does anybody think differently?
> > >
> > > Regards, Oscar.
> > >
> > >
> > > Am 28.09.2012 13:51, schrieb Charles Butler:
> > > > Oscar, as you have an assistant annotating video tapes, trying to
> > > > translate in your head to an expressive point of view may be
> > driving you
> > > > crazy. One project in Belo Horizonte is using receptive SignWriting
> > > > specifically when annotating video tapes so that you see parallel
> > > > movements, not mirror movements when you look at them side by side. You
> > > > write what you see on the videotape, not reverse it to your own hands.
> > > >
> > > > What this means is that the videotaped person's left hand is on your
> > > > right, and the videotaped person's right hand is on your left. You have
> > > > to remember that you are writing another person's hands, not your own,
> > > > so when you look in a dictionary like Delegs or any of the current
> > > > SignPuddles, you will not find what you see on a videotape, but its
> > > > expressive equivalent.
> > > > Charles Butler
> > > > [log in to unmask]
> > > > 240-764-5748
> > > > Clear writing moves business forward.
> > > >
> > ------------------------------------------------------------------------
> > > > *From:* Charles Butler <[log in to unmask]>
> > > > *To:* [log in to unmask]
> > > > *Sent:* Friday, September 28, 2012 7:29 AM
> > > > *Subject:* Re: help with signwriting
> > > >
> > > > Oscar,
> > > >
> > > > Reply, in the first sign, you are using a "both hands" arrow when the
> > > > hands are moving separately. If you are bringing the hands back toward
> > > > yourself, you need two arrows toward yourself, put them next to each
> > > > hand rather than in the middle. This is a common mistake as I'd be able
> > > > to read it, but the hands are not moving in a common path. This is a
> > > > common mistake, a single arrow is only used when both hands are
> > actually
> > > > together.
> > > >
> > > > In the second sign, your left hand is pointed downward, but you are
> > > > using a right hand arrow moving twice. Use a left hand arrow or a right
> > > > hand, not a mix. You could move your right hand in this fashion, but
> > > > your hand would be twisted outward rather awkwardly, unlikely that this
> > > > is what you mean.
> > > > Charles Butler
> > > > [log in to unmask]
> > > > 240-764-5748
> > > > Clear writing moves business forward.
> > > >
> > ------------------------------------------------------------------------
> > > > *From:* Oscar Koller <[log in to unmask]>
> > > > *To:* [log in to unmask]
> > > > *Sent:* Friday, September 28, 2012 6:06 AM
> > > > *Subject:* help with signwriting
> > > >
> > > > Hello everybody,
> > > >
> > > > I added following appended 4 entries to the German Sign Puddle. The
> > > > editors noted in each case that the writing is not correct. Could
> > > > anybody explain to me, what should be changed?
> > > >
> > > > Thanks
> > > > Oscar.
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > >
> >
> >
> >
> >
>