Karl Dubost wrote:
Le 7 juil. 2007 à 15:53, Philip TAYLOR a écrit :
I know. But that is the same argument that is used to
defend the lack of emphasis placed on grammar in many
schools today -- "people will understand you, so it
doesn't matter whether you speak correctly or not".
The one who speaks is the "author" of the voice, aka authoring tools and
author.
The one who listen is the "parser" of the voice, aka browsers.
When the former needs to be strict as much as possible (strong grammar
rules), the latter one needs to be fault tolerant to make the discussion
meaningful.
In a class context, it matters. Then the "parser" can be also a "teacher".
In daily life, it doesn't work. Fault tolerance makes the communication
possible. Do you always express yourself in correct English?
No, but I try to.
Do you always fix the English of people you are communicating with?
No, unless I am in a situation (such as TEFL, or when dealing with
non-native speakers whom I already know are receptive to suggestions
for improvement) in which it would be "appropriate behaviour", or
occasionally just for fun. Never on a list such as this, where it
would be both insulting and counter-productive.
For example, I have never seen you fixing one of my emails for its poor
syntax or spelling. But I know that I do mistakes in English.
See above. Please feel free to correct me any time I attempt
a reply in French :-)
I think the main target for strictness should be authoring tools. We are
not very good at doing that so far.
The primary target for strictness should be the formal specification of
the language (e.g., the syntax might /require/ or /reject/ quotation marks
around attribute values, but should not allow both unless a sub-rule clearly
states (e.g.,) "... are required only when the attribute value contains
a special character (see Appendix S).".
The secondary target should be authoring tools, which should be required
to honour all requirements of the formal specification.
The tertiary target should be authors who are creating web pages
/ex nihilo/.
If there is to be a quaternary target, and if it is to be user agents,
a.k.a. "browsers", then the requirement should be that a "conforming
browser" should have an easily-accessible "strict mode" such that
any "conforming browser" can immediately become a validator, thereby
reducing the load in the existing W3C validator and providing an
immediately available tool for validating hand-written (or otherwise
crafted) web pages.
Philip Taylor
