Re: [OT] character collection for an international keyboard layout (ISO/IEC 9995-2 and 9995-3)

From: Asmus Freytag (
Date: Thu Dec 20 2007 - 03:35:27 CST

  • Next message: Hans Aberg: "Re: CLDR Usage of Gregorian Calendar Era Terms: BC and AD -- Can we please have "CE" and "BCE" ?"

    On 12/19/2007 11:46 AM, Karl Pentzlin wrote:
    > As a (new) member of DIN NA 043-01-35-01 GAK, a German group
    > related to ISO/IEC JTC 1/SC 35/WG 1, I am concerned with the
    > ISO/IEC 9995 standard, Keyboard layouts for text and office systems.
    > Before I do any detailed statements in the standard group, I want to
    > discuss my general ideas in the public.
    > Any comments or hints are welcome.
    > In the current version of Part 2 of the aforementioned ISO 9995 is stated:
    > For the input of graphic character repertoire of collection 281
    > (titled MES-1) as specified in amendment 1 to ISO/IEC 10646:1-2000,
    > a Common Secondary Group Layout (to be used as group 2) is specified
    > in ISO/IEC 9995-3...
    > In my opinion, this character collection is [ill-]suited as base
    > for a standardized keyboard layout, for the following reasons:
    > ...
    > Thus, I propose to use the more complete set of collection 282
    > (MES-2) of ISO 10646, with some modifications as enumerated below.
    > It is of course a somewhat complicated task to put these characters into
    > a concise keyboard design, but the "Europatastatur" (European Keyboard
    > as shown on , in German; an English
    > presentation is found on )
    > shows that such things can be done.
    > The collection 282 (MES-2) is:
    > ...
    > I propose to use a set based on collection 282, without Greek and
    > Cyrillic (i.e. Latin script only), without block graphics (as their use
    > is very limited nowadays), and without any legacy ballast, namely
    > - Karl Pentzlin

    these are some interesting ideas. The problem with keyboards that
    command a very large set of characters is that typists have problems
    remembering the key-bindings. For people that speak a given language,
    learning to type the (limited set of) additional characters for that
    language is a manageable task, and the convenience of having a
    particular "universal" keyboard layout available on any workstation in a
    large organization (and not just on their own desk) can be a benefit.

    However, when it comes to typing mathematics, it's not clear that this
    is the correct approach. The number of symbols needed is too large and
    their use is too varied for most users to get the necessary
    reinforcement that would allow them to learn arbitrary key-bindings. For
    such needs, some form of input method that translates a mnemonic string,
    such as \sum into the proper symbol is likely to be a better match.
    There are other schemes for quick symbol selection as well.

    Moreover, mathematics cannot be typed without Greek, so your decision to
    leave out Greek immediately means that whatever scheme you put into
    place has little to offer to the serious user - as a result, you are
    better off simplifying the design to what is needed for natural
    languages (including the needs of commercial/legal documents).

    (For a list of characters used for mathematics, see for example, UTR#25
    at )

    For users that need to enter names and other information in native
    spelling, i.e. in languages that they are not familiar with, the problem
    is that locating a key on a keyboard (even a virtual keyboard) can be
    more cumbersome than using a well-designed pick-list, for example. To
    address the problem of the Dutch secretary who suddenly has an Azeri
    boss and needs to frequently use some special character for that one
    name, a system would need to provide a combined on-screen and keyboard
    oriented approach, whereby the user can locate an unfamiliar character
    in a well-designed list and if the character is used frequently enough,
    can easily find out and learn the keyboard shortcut for it.

    A well-designed list would be set up to help users with quick access to
    characters and give enough information to help avoid misidentification
    among similar characters. A keyboard layout cannot do that (look no
    further than the scrambled letters of Basic Latin to see what I mean).

    I think this is an area where the existing designs can all be improved upon.

    Hope you find some of these thoughts useful as you begin your active
    contribution to the standardization process.


    PS: an aside on MES-1 and MES-2. The criteria by which these were
    designed have always seemed less than satisfactory. But apart from that,
    the real issue with these collections is that, at best, they are sets
    for "systems coverage", that is, sets that a system is supposed to
    support for input, processing and display - but from that it doesn't
    follow that they are the proper sets to consider for _keyboard_ input at
    the exclusion of other forms of input methods.

    This archive was generated by hypermail 2.1.5 : Thu Dec 20 2007 - 03:38:05 CST