Recently approved MA Thesis on SignWriting

Bill Reese wreese01 at TAMPABAY.RR.COM
Wed May 11 19:28:07 UTC 2011


When you say we may need a more relative layout strategy using polar 
coordinates, that reminds me of the days just before computers when we 
had powerful calculators.   I programmed a utility on my calculator for 
traversing, a surveying and civil engineering method of getting from 
point to point over land.  It uses bearings and distances, which is 
basically a polar method, using the last point of a leg as the first 
point of of the next leg.  The output of the program was the X,Y 
coordinates of each point.

Months ago I wrote to the list an idea about using shapes derived from 
traversing.   What I mean by that is that if you plot traverse paths for 
each sign in a large enough set of signs in a database, a set of paths, 
traverse shapes, may be determined.   These paths could become an 
alphabet of shapes, so that if you were to define a signwriting string 
utilizing these shapes, you would be fetching a predetermined path and 
applying the symbols of the sign to that path.   It would function 
similarly to how the Kanji alphabet is applied to Characters in Japanese 
language to guide how the Characters should be pronounced.

Like your idea of face symbols as diacritics of the head circle,  
perhaps we can define placement similarly to what is already used in 
word processing:  A system similar to sub-scripts and super-scripts.   
In this method, imagine horizontal lines through a sign, each 
representing first the baseline then a series of super-script lines 
above it and a series of sub-script lines below it.  You'd also need to 
define "pre-script" and "post-script" as locations of placement for 
symbols before and after the first symbol. This is close to a Cartesian 
system but it's simplified to a predetermined set of locations that 
would build upon existing methods (and terminology) of word processing.  
This could tend to cause signs to be more iconized and simplified, which 
is a natural progression of any written language.   Eventually, just how 
a box in Japanese is recognized for the word "sun," which was originally 
written (in Chinese) as a circle, simplified and iconized signs would 
come to be recognized with their associated concepts, rather than as a 
"picture" of how the sign is made.

Obviously, these two ideas can be combined so that you could have the 
shape alphabet defining placement of symbols ("characters" or "text") on 
the script locations.   Restricting to predefined script locations could 
allow a smaller set of shapes, which could make for a manageable alphabet.

Obviously, it's all theoretical, and I'm just hoping my perspective and 
ideas can help somehow


On 5/11/2011 8:58 AM, Stuart Thiessen wrote:
> I think we will need to move to a more relative layout strategy. As I 
> mention in the thesis, there is some natural organization of the 
> symbols based on the layout of the body. There is also some natural 
> relationships between symbols such that certain types of symbols only 
> relate to other types of symbols. So, those relationships can also 
> help us in finding ways to organize the symbol relationships in how 
> the layout is encoded. I think that there would be no specific need to 
> mark the location of the first symbol in a sign box. It could be 
> understood to be placed in the center of the sign box. Then everything 
> else is positioned either relative to the previous symbol or relative 
> to the body. There is more detail in the last two chapters that 
> describe some of my preliminary conclusions. At the least, I think we 
> should use polar coordinates rather than cartesian coordinates.
> I wrote a computer program to help me do some of the analysis. Due to 
> timing constraints, I didn't tell the computer, "If they have the same 
> location or if the rectangles representing the two symbols overlap, 
> then calculate distances between symbols in a different way." My 
> program just made the assumption that no "outside pixels" of the 
> symbols overlapped or were superimposed upon each other. For a 
> preliminary analysis, this covers most of the symbols, but I knew a 
> more detailed analysis would need to cover superimposed or overlapping 
> symbols. It's just a matter of recognizing the situation and planning 
> different calculations.
> Marking the superimposed symbols as -1 would certainly work. Or one 
> could assume that no positioning information implies it is 
> superimposed. In other writing systems, it is my understanding that 
> ligatures (like æ) are actually handled as separate symbols in 
> Unicode. I don't know if that would work for SignWriting. For example, 
> for head circles, one option could be to have one head circle symbol + 
> a series of face diacritics. The renderer could then superimpose the 
> face diacritics onto a head circle symbol. Then if it runs out of 
> room, it could add additional head circles and add the remaining face 
> diacritics to those head circles. But that is not the approach that we 
> are currently using. Our current approach of superimposing the various 
> head circles also works, but that approach requires us to think about 
> opacity in the definition of our symbols which is not usually used in 
> most writing systems (to my knowledge).
> At this point, we have some time to explore some different approaches. 
> Hopefully, we will find one that will be well-accepted by the Unicode 
> Consortium.
> On Mon, May 9, 2011 at 1:16 PM, Bill Reese <wreese01 at 
> <mailto:wreese01 at>> wrote:
>     Great!  I'm glad they're working together on it.  I hope great
>     things come out of the collaboration.
>     I know I brought this up before but I'm wondering, Stuart, if a
>     concept of relative coordinate systems was discussed in your
>     thesis?   I did a quick scan so I'm not sure if it was.  What I
>     mean by "relative" is a coordinate system that's related to a
>     previous symbol in a sign according to that sign's signspelling
>     sequence.
>     The coordinate for the first symbol would be in absolute
>     coordinates according to the signbox, then the second symbol would
>     relate to the first symbol according to a coordinate system using
>     a point of the first symbol as the origin.
>     Doing it that way may allow establishing matrices of symbol
>     pairing in a sign.   I would imagine this to be similar to
>     "kerning" and possibly define distances according to the pairs
>     rotation of not only themselves but to each other.  Similar to
>     what you were saying about establishing minimum distances.
>     About the overlap of symbols that you mention.  I was wondering if
>     it couldn't also be solved by a matrix of symbol pairing so that a
>     particular matrix value would indicate overlap - say, a value of
>     -1.   On the other hand, do you think it would be possible to
>     create totally different symbols that are overlaps of two
>     symbols?  I ask this as that's what's done in other languages when
>     there's an overlap.  For instance, "æ" which looks like "a" and
>     "e" overlapped but is it's own symbol.  I would hazard a guess
>     that separate symbols are only possible when there's only a few.
>     Bill
>     On 5/9/2011 1:06 PM, Valerie Sutton wrote:
>>     SignWriting List
>>     May 9, 2011
>>     Hello Bill -
>>     Just want you to know that we have a group of
>>     Unicode-knowledgeable people working together on our SignWriting
>>     proposals that will be presented to the Unicode-related meetings
>>     over a period of years, and Steve and Stuart are both in the
>>     group, along with others as well - so we are all working
>>     together...The proposals have been separated into proposing the
>>     encoding of the symbols, or characters, first, (of the
>>     International SignWriting Alphabet 2010) and then once the
>>     symbols have been encoded, we will present a second proposal
>>     related to layout and symbol placement issues - so that second
>>     area is where different theories will be discussed until we can
>>     come up with a final decision for a second proposal - so we are
>>     taking this one step at at a time...
>>     An exciting time for all of us - smile -
>>     Val ;-)
>>     ----------
>>     On May 9, 2011, at 8:56 AM, Bill Reese wrote:
>>>     Stuart,
>>>     Wow, that was a lot of work!  I do have one question.  How would
>>>     the most recent work in Unicode and, more particularly, what
>>>     Steve Slevinski has written to the list affect the portion where
>>>     you talk about what may be needed for successful Unicode
>>>     acceptance?   From what it appears, it's well on it's way to
>>>     acceptance with what Steve and Michael Everson have done.
>>>     Bill
>>>     On 5/8/2011 1:02 AM, Stuart Thiessen wrote:
>>>>     Hello, all! I know it's been a long time, no see. I wanted to
>>>>     let you know that I have completed my MA thesis on SignWriting.
>>>>     For those of you interested in reading it, you can download a
>>>>     PDF from the University website. Just so you know, the PDF
>>>>     itself is about 22MB.
>>>>     _
>>>>     If you have any questions about it, just let me know.
>>>>     Thanks,
>>>>     Stuart

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the Sw-l mailing list