Readme file for AntConc 2.4.0 (Windows)

Developed by Laurence Anthony, Ph.D.

Dept. of Information and ComputerEngineering
Faculty of Engineering
Okayama University of Science
1-1 Ridai-cho Okayama 700-0001
Japan
May 29th, 2003
anthony 'at' ice.ous.ac.jp


AntConc started out as a relatively simple concordance program, but has
been slowly progressing to become a rather useful text analsis tool. 
It is  written in Perl 5.6 using
ActiveState's excellent Komodo 1.2 development environment. The program
can be launched by simply double clicking on the .exe executable file, 
which can be downloaded from the Laurence Anthony laboratory website at
http://antpc1.ice.ous.ac.jp/. The program can run under any windows environment
including Win 98/Me/2000/NT and probably will run with no problems on Win XP.An
earlier version of the software  for Linux is also available. (If anyone wants the latest version
of AntConc ported to Linux please let me know).
If a user finds any problem launching the
program under a particular OS, please let me know using the email
address below.

Currently, AntConc can generate concordance lines either from a
single text or a set of texts stored in a single directory. A text or set of texts to be analysed
 can be chosen by selecting 'Open File' from the
'File' menu and then selecting the appropriate texts in the standard way. 
Similarly, a directory of texts can be chosen by selecting 'Open
Dir' from the 'File' menu. A list of the chosen texts  is displayed in
 the left frame of the main window, below the name of the chosen directory
(if appropriate). The output display can be saved either to the clipboard or a .txt file by selecting 'Save
Output' from the 'File' menu. The color used to highlight the search
term in the concordance lines can be changed from the default color
'blue' by selecting the 'Highlight Color' option in the 'Options'
menu. AntConc also has a sort feature, whereby the results
can be realligned in alphabetical order according to either the target
word, or words to the left or right of the target word. The words on
which to rearrange the concordance lines are chosen via the button bar at
the bottom of the main window.
The highlight color for the Level 1 and Level 2 sort can be changed via the settings menu.

To produce a set of concordance lines of text, a user needs to perform 5
actions:

1) Select either a single text or set of texts stored in a single
directory (see above)
2) Enter a search term on which to build
concordance lines in the entry box on the left of the button bar.
3) Choose the number of text characters to be outputted on either side of
the search term, using the increase and decrease buttons on the right of
the button bar (default value is 30 characters)
4) Click on the 'Start' button in the main window to start the concordance lines results
generation (Note: The concordance generation can be halted at any time
by clicking on the 'Stop' button in the main window)
5) Select a target word on which to rearrange the concordance lines. 0 is search word, 1L,
2L.. are words to the left of the target word, 1R, 2R .. are words to
the right of the target word. Note that two levels of sort are possible, with the
second level not-acivated when the software is first launched.
6) Click on the 'Sort' button to start the sorting process.

New since AntConc2.3.0 is the ability to display a plot of the location of the concordance
results for each file as a kind of barcode. The plot is produced automatically and can be
viewed via the "Concordance Search Term Plot" page from the main window.
New concordance search criteria can be entered on the Plot page, and then
pressing start will allow the user to see dynamically in what files and in what
positions the concordance results are generated. The plot can also be enlarged
or reduced in size using the 'Zoom' buttons in the button bar. Unfortunately, 
the Plot display cannot be saved at present.

Also new since AntConc2.3.0 is the ability to generate Wordlists for the files selected.
This feature is accessed via the "Wordlist" page from the main window. By selecting the
page, a wordlist will be generated automatically according to the criteria displayed in the
button bar at the bottom of the page. These criteria can be changed to produced
wordlists ordered by frequency or alphabetically, and with the frequencies displayed
or omitted, or presented in inverted order. At any time during the wordlist generation, the
user can stop the process.

A number of additions have been added since AntConc2.4.0. First, I have added the ability to see
how far along the processing of files is. This is useful when processing a large number of
texts as it tells the user what the time requirement will be. If the processing of files is
stopped for any reason, the progress bar will indicate this, and by clicking on the reprocess
buttons, a user can reset the system. Otherwise, the AntConc will only work with the
files that have been processed.

Second, a new feature has been added that enables a user to view a file in it's original
form, with search termes (if any) higlighted. The location of search terms in a particular
file can also be navigated using buttons at the bottom of the window.

Finally, I have had many requests for the ability to generate not only wordlists, but also
Keyword lists for a particular set of files. This Keyword generation feature has now
been added to AntConc2.4.0. To generate keywords, the user must first select a
reference corpus of texts which the system compares with the target files. Usually,
this reference corpus will contain standard English, but any corpus can be used. The
system then compares the frequencies of words in the target files with those in the
reference corpus and identifies usually frequent (or infrequent words) using a choice
of two statistical methods. The most effective is proably Log-Likelihood (the default
option), but the popular Chi-Squared method can also be used. The choice of a reference
corpus, statistical method, and keyword threshold cut-off point etc, can be changed via
the Keyword Preference menu.

Many small bugs have also been identified and corrected in AntConc2.4.0. Please
let me know if you find a bug in the program and I will correct it in an updated version.

AntConc can be used freely for individual use for non-profit research purposes, and
freely distributed on the condition that this readme file is attached in
an unaltered state. If the software is planned to be used in a group environement,
please let me know how you plan to use the software, and I will then give you permission.
The software comes on an 'as is' basis, and the
author will accept no liability for any damage that may result from
using the software. Although there is no support available for AntConc, the author
 would be happy to hear comments and/or suggestions from
anyone using the software, and will try to fix any bugs etc. when time
is available.

Known Issues

1. When the file or directory selection dialog boxes are accessed, if they are dragged across
the top of the AncConc main window, they leave 'ghost' traces. These traces are removed
when the dialog boxes are closed. This appears to be a problem with the perl modules I  use
to access the Windows API. I have yet to find a way to solve this problem. Any help or
suggestions would be grateful.

2. When a large number of concordance lines are generated (or words or keywords), the scrollbar
becomes sensitive to where on the bar the user clicks and drags to view lower down entries.
Sometimes this results in a user not being able to view the last lines unless the the cursor is
repositioned on the scrollbar. The is an annoying bug in the scrollbar subroutine (not mine!) and 
I am waiting for someone to fix this.

Revision History

2.4.0 A major upgrade since 2.3.0
First, progress indicators were added to'pages' of AntConc. 
Second, a new file view feature was added to view target files in their original state.
Third, a keyword generation feature has been added using log-likelihood and chi-squared methods.
Finally several bugs were found, in particular, bugs centered around the wordlist generation
feature. This feature of the software should work much quicker now.
Also, the user can interupt the processing of files in any 'page' of the software.

2.3.0 A major upgrade since 2.2.3
First, the abiliy to view concordance search results as a barcode plot graph and 
a feature to produce wordlist according to different criteria were added. Numerous
bugs centered around the way the software entered a 'Busy' mode were corrected.
The main core of the software was also updated resulting in a quicker, 'cleaner' processing
of the data. Performance improvements should be noticed as a result.

2.2.3 Updated file and directory selection dialog boxes to run smoothly in a Windows environment.
Also, changed the default colors for sort highlighting, and search window frame size. A number
of small bugs were also corrected

2.2.2 Corrected critical fault with compiler than caused program to expire when evaluation version 
of ActiveState Perl Development Kit expired. Sorry folks!! I didn't realize this would happen!!

2.2.1 Corrected bug which prevented new concordance lines being generated if the search term
was left the same and then new files were selected. Port to Linux also completed.

2.2 Designed new subroutines for selecting directories and files to solve 
rendering of dialog windows problems. This also enables an easier port to Linux.

2.1
Added a second level of sort.
Added ability to restrict searches to full-words only, case sensitive
Added ability to search using full Perl implemented regular expressions
Added ability to save results either to a file or the clipboard

2.0	Added new sort feature, for rearranging concordance lines.
Tidied up the interface.
Made the system more robust for novice users. (Now bad input will not cause the
system to crash so easily).

1.1	Added binding to allow return key to
launch concordance search.
	Also, recompiled software so that no
console is required.

1.0	First version

Laurence Anthony
anthony 'at' ice.ous.ac.jp