Feature requests
Martin Staffa
first of all, Laurence, thanks for AntConc; what it does, it does
really well.
However, there are some points I'd like to address that in my opinion
could be improved in future versions. Some of them have probably been
mentioned before, but I couldn't find a list or roadmap or anything
like that.
- make concordance plot data exportable (you addressed that in your
screencast, so I know you are aware of it). Actually it would be
already very helpful - if you could export the textual data (number of
hits) and - if the length was given in words (tokens), and not in
characters. This would be very helpful e.g. to plot frequencies over
time. Since you have it in the conc plot, couldn't you make it
available in a simple list that can be exported?
At the moment, I work with a corpus that does two (odd?) things:
First, meta data about the file is saved in a {tag} metadata {/tag}
structure, which AntConc does not understand. I don't know if this is
so common that it warrants - an option in antconc to exclude the
content of predefined tags, but yeah it would be nice. Second, antconc
counts contracted forms as two words, whereas my corpus doesn't. So a
nice to have option would be to count contractions as one word
(although this strikes me as odd, too, and I actually don't know what
the usual practice is).
I wonder, have you ever thought about going Open Source development
with AntConc? It's a great tool, and I think a lot of people would and
could contribute with more functions and plugins.
Laurence Anthony
Thank you for your interest in AntConc. Let me try and address each
point you make.
>
> first of all, Laurence, thanks for AntConc; what it does, it does
> really well.
Thank you!
> However, there are some points I'd like to address that in my opinion
> could be improved in future versions. Some of them have probably been
> mentioned before, but I couldn't find a list or roadmap or anything
> like that.
Having a public roadmap is a great idea. I will consider adding one
very shortly to the website. (I have it here on my computer!)
>
> - make concordance plot data exportable (you addressed that in your
> screencast, so I know you are aware of it). Actually it would be
> already very helpful - if you could export the textual data (number of
> hits) and - if the length was given in words (tokens), and not in
> characters. This would be very helpful e.g. to plot frequencies over
> time. Since you have it in the conc plot, couldn't you make it
> available in a simple list that can be exported?
For several years, my admin duties prevented me from working on
AntConc. But, I have more time now. My plan was/is to convert the code
to a more modern graphical toolkit that allows export of plots (as
well as other nice features like drag-and-drop, printing, ...).
However, I've also recently decided that while I recode everything, I
should also keep updating the current version to add new features that
are simpler to implement. Let me think about your suggestions and see
if they can be added to the current version as a minor update.
>
> At the moment, I work with a corpus that does two (odd?) things:
> First, meta data about the file is saved in a {tag} metadata {/tag}
> structure, which AntConc does not understand. I don't know if this is
> so common that it warrants - an option in antconc to exclude the
> content of predefined tags, but yeah it would be nice.
This is already possible within AntConc. Go to the Global preferences
and set your tag definition there.
> Second, antconc
> counts contracted forms as two words, whereas my corpus doesn't. So a
> nice to have option would be to count contractions as one word
> (although this strikes me as odd, too, and I actually don't know what
> the usual practice is).
The usual practice is to count contractions as two words. However, you
can edit the token definition in the global settings so that an
apostrophe is included as part of a word. The easiest way to do this
is by using the user-defined token class.
> I wonder, have you ever thought about going Open Source development
> with AntConc? It's a great tool, and I think a lot of people would and
> could contribute with more functions and plugins.
Yes! I certainly have. The most important reason for me is that an
open source license would allow users to check the code to make sure
that things are working as they should. But, as you say, it also
allows people to contribute. Right now, AntConc is not really designed
in a way that allows people to add plugins. When I upgrade AntConc to
the more modern graphical toolkit (see above), this will be part of
the new design.
You should see AntConc becoming open source in the near future. But,
right now I am in the middle of several changes that will allow the
current version to run natively in OS X. This is probably the number
one request I have right now (from Mac users). When I finish that,
I'll look at the open source issue.
Thank you again for the comments. I hope my reply answers all your questions.
Best regards,
Laurence.
Martin Staffa
It seems I should have read the documentation a little closer, so I
would have found some helpful options I thought were missing. I will
definitely look into the token settings. With the head tags however,
it seems you can only define one headtag, is that right? The corpus I
use is tagged like {title}...{/title}, {source}...{/source}, so I can
only exclude one at a time.
Your plans for AntConc sound really great, and I look forward to
seeing the new version!
Since you are recooding from scratch, I just add some quick ideas I
had about the interface. To me, it is kind of confusing that there are
Global Settings which are not really global in the sense of "every
instance of the program keeps these settings as a default on startup".
When you work a lot with corpora that are set up in the same fashion,
it would be nice if AntConc had a truly global settings that could be
overwritten on a per instance basis, without having to import a config
file on startup. Maybe this could be handled by a Project Manager,
that saves files, settings, and results and a separate configuration
manager that saves different sets of config options (tags, token
options). Included in this could also be some stuff that is now in the
tool settings, like word lists, reference corpora etc. So you could
define base folders for your corpus files, reference corpus, word
lists etc. and AntConc handily displays them in every instance, and
you don't have to import them everytime.
Then I would move the tool settings closer to the actual tool pane,
maybe as a sidebar. At the moment, some settings are at the bottom,
some in their own menu, some appear twice, and overall it's not really
clear why some are there and some not, e.g. I'd have treat all data as
lowercase for collocates at the bottom, since it is probably the
default for most uses. And especially nice for beginners woulr be an
interactive help, where you can get info for each function. I guess
this is something were the community could help you out a lot.
Best Regards,
Martin
Laurence Anthony
> Thank you for your quick answer.
You're welcome!
> It seems I should have read the documentation a little closer, so I
> would have found some helpful options I thought were missing. I will
> definitely look into the token settings. With the head tags however,
> it seems you can only define one headtag, is that right? The corpus I
> use is tagged like {title}...{/title}, {source}...{/source}, so I can
> only exclude one at a time.
I would think that a header is usually defined by a single tag...
hence the term *head*er. Have you thought of wrapping the tags,
"title", "source," etc within a single "header" tag? AntConc works
with the TEI specification, which is an international standard.
Perhaps you can look at that.
>
> Your plans for AntConc sound really great, and I look forward to
> seeing the new version!
I got busy this week, but I will try and get lots of work done on
AntConc next week, after the new semester settles down here.
> Since you are recooding from scratch, I just add some quick ideas I
> had about the interface. To me, it is kind of confusing that there are
> Global Settings which are not really global in the sense of "every
> instance of the program keeps these settings as a default on startup".
> When you work a lot with corpora that are set up in the same fashion,
> it would be nice if AntConc had a truly global settings that could be
> overwritten on a per instance basis, without having to import a config
> file on startup.
The settings can be exported to a file, and this will be imported
automatically on startup. So, you could create different workspaces
and just put multiple copies of AntConc in different folders with a
corresponding settings file. Most people don't find the quick "open
settings file" menu option too difficult. Do people really switch
settings so often in a single session?
>Maybe this could be handled by a Project Manager,
> that saves files, settings, and results and a separate configuration
> manager that saves different sets of config options (tags, token
> options).
> Included in this could also be some stuff that is now in the
> tool settings, like word lists, reference corpora etc. So you could
> define base folders for your corpus files, reference corpus, word
> lists etc. and AntConc handily displays them in every instance, and
> you don't have to import them everytime.
Hmm. It sounds more complicated this way. Have settings files with
obvious file names would seem easier. Also, I have the problem that I
want to keep AntConc completely self-contained. If I start adopting
this approach, AntConc would probably need to be **installed** into
the file system. This comes at a heavy price for users who don't have
install privileges (e.g. students in a lab).
> Then I would move the tool settings closer to the actual tool pane,
> maybe as a sidebar. At the moment, some settings are at the bottom,
> some in their own menu, some appear twice, and overall it's not really
> clear why some are there and some not, e.g. I'd have treat all data as
> lowercase for collocates at the bottom, since it is probably the
> default for most uses.
I think the "treat all data as lowercase" option is scattered around
as you say. But, I'm not sure about any others. Which others do you
find problematic?
>And especially nice for beginners woulr be an
> interactive help, where you can get info for each function. I guess
> this is something were the community could help you out a lot.
I agree with this. However, WordSmith has an interactive help feature,
but actually I find it quite annoying when the help messages pop up
everywhere even when I don't need them. Don't you think a separate
document describing the features is better? Perhaps, some kind of
right-click popup help could be introduced. But, the whole idea of
AntConc is that it should simple enough to not really need
documentation. I find the biggest problem for beginners is loading in
a corpus. But, a help system wouldn't help in this case. What
aspects/functions/buttons do you think beginners find difficulty with?
Thank you again for the feedback.
Laurence.