Re: kill your darlings

[Marcin Borkowski <mbork@mbork.pl>]

On 2019-06-24, at 10:42, Richard Melville <6tricky9@gmail.com> wrote:

> On Mon, 24 Jun 2019 at 05:13, Emanuel Berg via help-gnu-emacs <
> help-gnu-emacs@gnu.org> wrote:
>
>> Everyone knows that everyone use their favorite
>> constructs in peach and in writing. E.g.,
>> I like to use "e.g.", and I like to end
>> sentences with something like "for sure" -
>> no doubt :)
>>
>> On a mailing list this is not really a problem.
>> But for e.g. in a relationship in can become
>> very enervating, without exaggerating :) (Okay,
>> you get it, I'll stop now. Or will I...)
>>
>> We can't (?) program our relationships with
>> Elisp, but I wonder if there is a tool or
>> method to detect "darlings" in a text.
>> For example, I'm writing a LaTeX text now - it
>> isn't even halfway done, but currently at
>> 1965 lines, I have used the word "emellertid"
>> 8 times (it means "however" but sounds more
>> stiff and old-fashioned) - and if I weren't
>> aware of it, it'd be a good idea if Emacs could
>> tell me I overused the word, so I could
>> consider removing some of them. And perhaps
>> (actually it is likely) there are other of my
>> "darlings" that I *am* unaware of!
>>
>> The kind of stuff I described first, with
>> sentence constructions and so on, I get it it
>> is probably very difficult for a computer
>> program to detect. But overuse of words could
>> be as simple as
>>
>> - count all words
>>
>> - see what words are the most common
>>
>> - are there word there that much longer than
>>   the others? warn the user about possible
>>   overuse
>>
>> - obviously, if one is writing a paper on the
>>   mating process of the Trigonosaurus, one
>>   would simply disregard the recommendation to
>>   not use that wierd word all the time
>>
>> - to compare the text to the Internet would be
>>   a possibility, but I don't really like it.
>>   It would mean the program would try to make
>>   you write like everyone else. That's not the
>>   point: the point is to make you aware of
>>   something, that you might be unaware of!
>>
>> Is there anything like that going on anywhere
>> in the Emacs world?
>>
>> --
>> underground experts united
>> http://user.it.uu.se/~embe8573
>> https://dataswamp.org/~incal
>
>
> Yes, it's called proofreading.

Wow, you made my day!

Seriously though, there _are_ things like that.  I never used proselint
(http://proselint.com/), but it seems to do a similar thing - though it
detects common mistakes (and things perceived as mistakes in our stupid
times, apparently), and Emanuel wanted something that could somehow
detect peculiarities of style of a particular person.  I'm pretty sure
it's doable, though I know only a very, very little about natural
language processing to provide any details.

Doing statistics might be more difficult in languages with a lot of
inflections (like Polish or Latin).  English is way simpler wrt that;
I have no idea about Swedish (though it's not easy to imagine it being
more difficult than Polish;-)).

Best,

--
Marcin Borkowski
http://mbork.pl