[HarfBuzz] Text layout advice for written sign language

Steve Slevinski slevin at signpuddle.net
Thu Feb 3 19:04:16 PST 2011


Greetings,

I'm looking for advice on text layout for SignWriting. SignWriting is 
used for sign languages. It is an unusual script with unusual requirements.

The glyphs of SignWriting are related to sight, rather than sound. While 
the sounds of a voiced language occur one after the other, the sights of 
sign language can occur at the same time. The spatial relation is 
written rather then the temporal order.

There are a large number of glyphs: 37,811. I have a file system based 
font released under the open font license. I have a Unicode proposal 
that uses 674 codepoints on plane 1. 3 proposed Unicode characters 
identify a symbol. I have renamed each glyph SVG according to the 
proposed Unicode character sequence. An automated import should 
automatically build the font.
http://www.signbank.org/iswa - ISWA 2010 HTML Reference
http://www.signpuddle.net/iswa - ISWA 2010 Font Reference
http://www.signpuddle.net/plaintext - Symbol Encoding with characters

Each word occurs in a separate 2-dimensional Cartesian coordinate space. 
Each word has a defined center at the point (0,0). The words are aligned 
based on their center. The words can be aligned top to bottom, or left 
to right.

The basic idea of the script is that for each word, the writer chooses 
the symbols and the writer chooses the symbol positions. Standardized 
rules determine the center. There is no internal structure to the 
positions. Symbols are positioned in relation to the center rather than 
relative to other symbols. The coordinates for each symbol represent the 
top, left of the symbol glyph.

As a formal language, each word can be divided into the plain text of 
symbols and the rich text of spatial layout and temporal annotation. The 
plain text of symbols will be represented with proposed Unicode 
characters. The rich text of spatial layout and temporal annotation will 
be represented with a lite markup using ASCII characters and number 
digits. A regular expression can identify Formal SignWriting words.

There are several issues that complicate layout. I have been solving 
these solution on the server side with PHP or on the client side with 
JavaScript. Would it be possible to use HarfBuzz to work with a lite 
markup and plane 1 characters to display SignWriting text?

Here's an example:

A𝡌𝪎𝪓𝦉𝪌𝪜𝨊𝪌𝪒M27x37𝡌𝪎𝪓n28x11𝦉𝪌𝪜n2x7𝨊𝪌𝪒n19xn38𝨾𝪌𝪒n19xn38


I believe I am nearing a stable open standard for SignWriting.  The
Internet Draft I released earlier this year is very close.  Based on feedback I've
received on the I-D, I have a revision planned that clarifies Formal SignWriting as
plain text and rich text.  The next revision should be stable for the foreseeable future.
http://tools.ietf.org/html/draft-slevinski-iswa-2010-00 (Jan 2011 draft)

I would like to solve this at the appropriate level.  It appears HarfBuzz may be the right level.

Any advice or ideas?

Thanks for your time,
-Steve




More information about the HarfBuzz mailing list