On Wed, 2007-01-03 at 07:36 +1030, Tim wrote:
>> I'd go along with that. I am a native English speaker, and I've found
>> word processing grammar checking tools leave a lot to be desired. Some
>> of the results, I've seen, give you the impression that a 12 year old
>> wrote the document. They might find some glaring errors, but more
>> subtle ones go un-noticed.
> I think you have unreasonable expectations.
I don't agree. I don't really expect them to be able to do any better
than they do. Human language is a rather complex thing, with some quite
large regional variations. But computer programmers like to try and
reduce things down to the lowest common denominator. The two are
(Currently testing FC5, but still running FC4, if that's important.)
Don't send private replies to my address, the mailbox is ignored.
I read messages from the public lists.
I have some experience in translating complex data, which is somewhat
related to the issue of grammar checking. What occurs is event sequence
and context related, along with the baggage that we humans carry, called
memories. Thus two people conversing often have differing ideas about
the topic, based upon their own experience and memories, which is then
modified by the event sequence that lead to the
statement/paragraph/topic under discussion. This in turn is further
refined by the event sequence of the particular conversation and topic,
queried and rectified by questions and answers or statements by the
party(ies) involved. This also occurs in complex programs, where the
programmer has prior experience (or art as some like to call it), along
with personal coding styles, experience in fixing problems (or working
around them) which is particular to the view of the program and the
familiarity the programmer has with the hardware, software libraries and
tools at his disposal. Thus if five different programmers are asked to
create a table ot data encoded by color and location, you would get 5
different programs to accomplish that task. They could all be
equivalent in input and output, but the internalization of the data, the
procedures followed and the methods used to achieve that result would be
different. Also the appearance of the initial result from each
programmer would likely be different as well, in things like table
representation, dividers, lines or no lines, labels or no lables etc.
etc. Further it is unlikely that any of the outputs would match your
exact expectations on the first pass.
Thus even though the topic was understood, the concepts clear and the
desired output known, the output was somewhat non-deterministic. This
is true of Grammar as well.
If, for instance one were to make a speach about Gettysburg in modern
language, one might say:
Eighty seven years ago, people established America to grant liberty and
justice to everyone equally.
Does that say the same thing as:
Hey man, my great grandpa started this country so I could say this
Four score and seven years ago, our fathers brought forth on this
continent a new nation, conceived in liberty and dedicated to the
proposition that all men are created equal. Abrahm Lincoln.
Perhpas each has the same feeling to the speaker, and even is correct
for the audience, yet the choice of words would be very hard to justify
in a programming context without a lot of background information that
simply is not part of the words being spoken. The historical context,
the nature of the speaker, the nature and background of the audience,
and the requirements of settling an issue divisive to the nation. A
great task for a man, but one that cannot come (yet) within the
capability of a computer.
Grammar checkers can do some interesting things, and they can improve
clarity, but they do not make one accomplished in language. However,
their value for a non-native speaker is considerable. I have an ex-wife
who is Korean. It would have been wonderful to have had some better
tools to help us in our relationship.
Expectations are simple. getting something to meet those expectations
is much like correcting grammar, however, Tim, you can certainly add
your skill and knowledge to those attempting this very difficult and
challenging task. Dr. Kurzweil and others have been working on this for
decades. Check out the history of Dragon systems.