Media type parameters for braille

I would be interested in receiving feedback from members of this list with
respect to my proposal for adding parameters to the braille media type for
designating (1) whether the output is to be embossed or presented on a
refreshable display; and (2) the dimensions of the braille device. The
form of the relevant media declaration would thus be as follows:

media="braille embossed [x] [y]"
or
media="braille displayed [x] [y]"
where the optional x and y values represent, respectively, the number of
cells per line and the number of lines per page available on the braille
device. The concept of a "page" is not directly applicable to braille
displays, since most of them can present only a single line of text at
once. However, larger displays have recently been developed, although they
are not as yet widely available and are the subject of ongoing research.

Rationale:

The principal reason for distinguishing between embossed and displayed
braille is that some stylistic features which are useful on paper, are not
applicable when reading a document with a braille display. These include
headers and footers, page numbering and block protection (to keep related
text on a single page), all of which may potentially be controlled by
style sheets. The formatting of tables may also differ depending upon
whether the document is being embossed or displayed dynamically by the
HTML user agent.

The justification for the optional x and y parameters that specify the
dimensions of the output device, is not so clear. The motivation for this
concept stemmed from consideration of the problems associated with the
formatting of tables in braille, and the need for tables to be presented
differently depending upon whether or not they exceed the line length of
the braille line. If the presentation of tables is to be controlled via
style sheets, then it may be desirable to create alternative styles, with
different parameters, for producing tables on devices with various line
lengths. For example, some braille embossers support narrow paper and thus
generate output of 32 cells per line, and braille displays vary in size
between 20 and 80 cells. The treatment of tables in braille is a complex
issue which in my opinion needs to be considered and discussed thoroughly.
I am aware that other subscribers to this list are much better qualified
in this area than I am, and would be interested to know whether a solution
has yet been proposed. The number of lines per page may also be a useful
parameter that would affect the formatting of tables, especially if they
were presented in paragraph form and there was a need to keep all of the
contents of a particular element in the table on a single output page. One
difficulty which is also relevant in this context is that the number of
lines per page varies depending upon whether six-dot or eight-dot braille
is being used. Although most braille codes available today employ a
six-dot system, eight-dot braille is becoming more common. For example, in
Denmark, it is being used to increase the correspondence between print and
braille character sets, thereby facilitating the deployment of
software-based translation schemes. For further details, see ch. 9 of Lars
Christensen's AITIIB research project report, details of which can be
found at http://www.sensus.dk/aitiib.htm. An eight-dot code is also
proposed by the Science Access Project at Oregon State University:
http://dots.physics.orst.edu/. Furthermore, it is theoretically possible
that six-dot and eight-dot codes may both occur within a single document,
thus resulting in different numbers of lines per output page in different
parts of the same text. The specification of page dimensions may not be so
important when reading a document with a braille display, since the
controls on the display, in conjunction with controlling software, should
provide ample means of traversing a document. However, in formatting
tables, it may still be desirable for style sheets to be able to ensure
that a certain item of text is presented on a single line, in which event
the line length parameter is once again crucial.

As is evident from the foregoing discussion, I have not reached firm
conclusions in relation to these issues. Nevertheless, I thought that it
would be appropriate to seek input from others rather than to continue
pondering these questions alone.

Regards,

Jason White.

Received on Sunday, 29 June 1997 21:33:47 UTC