The EDICT Dictionary File
[This page is quite out-of-date. More up-to-date documentation is available at
https://www.edrdg.org/wiki/index.php/JMdict-EDICT_Dictionary_Project.]
Welcome to the Home Page of the EDICT file within the JMdict/EDICT Project.
This page has been written by Jim Breen
(hereafter "I" or "me") and is intended as
an overview of the file, with links to more detail elsewhere.
Background
Way back in 1991 I began to experiment with handling Japanese
text in computer files, and decided to try writing a dictionary
search program in Turbo C under DOS, which used a simple dictionary file
contained in the
MOKE (Mark's Own Kanji Editor) package. To make this program more
useful, I began to expand the file itself. One thing led to another,
until I ended up running a fairly major project which has taken over a
large portion of my life.
I must acknowledge that the EDICT project has depended on many people who
have provided material and editorial assistance. A significant
proportion of the
compilation process has been carried out using electronic mail and
file transfers, and indeed the project would never have occurred
without the services provided by the Internet.
What is EDICT?
EDICT is a Japanese-English Dictionary file. For the full details, see
the
full documentation,
or the
old documentation.
It is a plain text document in EUC-JP coding, with its own format (which
has become known as "EDICT-format").
Originally it was compiled and
edited in this format, but from 1999 it has been generated as a legacy
file from an expanded database, along with the related
JMdict
(Japanese-Multilingual Dictionary) project. JMdict is an expanded file,
containing French, German, Russian, etc. translation, and is in XML format
and UTF-8 coding. In 2010 the maintenance was moved to an online database
system.
There are now two EDICT versions:
- the plain EDICT file. This is the original format, where there is
only one kanji form and one reading per entry. I regard this as a legacy
format, and only provide it for older applications. PLEASE do not use
this format for new applications, as I would like to withdraw it one
day.
- the enhanced "EDICT2" format. This can have multiple kanji forms and
readings in an entry, and also has other information such as
cross-references, restrictions, etc. and also uses kanji from the
extensions in the JIS X 0212 standard. It has almost all the information
in the full JMdict format. This form should be used for all new
applications.
The EDICT2 file currently has about 175,000 entries, and the legacy
EDICT format has nearly 200,000 entries (many of which are duplicates
as all the permutations of
kanji and readings generate distinct entries.)
A short overview of the EDICT project as
parallel English/Japanese text
is available.
Download
You can download the files in various formats:
edict2.gz,
edict.gz,
edict.zip
These are all on the
Monash ftp site.
You can also use EDICT2 online via my
WWWJDIC server.
Is it Public Domain?
EDICT can be freely used provided satisfactory acknowledgement is made
in any software product, server, etc. that uses it. There are a few other
conditions relating to distributing copies of EDICT with or without
modification. Copyright is vested in the
EDRG
(Electronic Dictionary Research Group) with the
file available under a Creative Commons Attribution-ShareAlike Licence (V3.0).
You can see the specific
licence statement
at the Group's site.
Other Dictionary Files
A number of other dictionary files have been compiled by me and others
as adjuncts or spin-offs from the EDICT file. I will list the major of
these below. Another summary can be found be found in the
documentation
of my WWWJDIC server.
- the KANJIDIC kanji information file.
(overview)
(download)
(documentation)
This file has an entry for each of the 6,355 kanji in the JIS
X 0208-1990 standard.
(The KANJIDIC file was cited as a
reference in the New Nelson character dictionary published in 1997.)
- A second file, KANJD212, which covers the 5,801 kanji in the JIS X
0212-1990 standard has been assembled, and
was released early in 1996.
(documentation)
(download)
- the ENAMDICT/JMnedict files of proper names. These now have over 720,000 names. Downloads:
(enamdict.gz)
(JMnedict.xml.gz)
or see the
documentation.
- the COMPDIC file of computing and
(tele)communications terminology. Has over 12,000 entries.
(documentation)
(download)
- the EDICLSD3 (Japanese-English Life Science dictionary), which is
the EDICT-format version of a major file produced at Kyoto University
by a
project group
coordinated by Professor Shuji Kaneko..
Software for using the EDICT files
-
WWW
There are several WWW options:
- my own
WWWJDIC
server, which has a number of mirrors in Canada, Japan, the US, etc.
(Please note that the WWWJDIC program is
not
available for download. There is no PC version.)
- the excellent Jisho.org server
- Jeffrey Friedl's server at sites in
Canada (site 1),
(site 2),
the USA,
etc.
There are many other WWW-based methods, and a larger list can be found
on my
online dictionaries
page.
A very useful site is
Rikai,
which massages WWW pages, placing popup translations from EDICT behind
the Japanese text. As well there is a Rikai-based
Mozilla Plugin
that achieves the same without going to the server. Needs Firefox
0.8.
-
Windows
(This section is quite out of date.)
While I do not
have a lot of direct experience (I don't use Windows much), the
following appear to be the options:
- use the
JquickTrans
program, also available from the
Monash ftp site.
Despite its name, it is a dictionary client.
- use the old
WinJDic
program, also available from the Monash ftp
site. It has the limitation of not being able to handle more than one
dictionary file.
-
use the
JWPce
freeware wordprocessor, also available from the Monash ftp site.
It has a good built-in dictionary function. The author, Glenn Rosenthal,
has promised a stand-alone dictionary version soon. The older JWP
wordprocessor, written by Stephen Chung, is also popular.
- another WP which uses the EDICT file is
NJSTAR.
NJSTAR comes with an early copy of EDICT. If
you want to use a more recent copy, you'll need to create special index
files. I think the utilities for this are in the DOS archive of NJSTAR,
but I cannot confirm this.
- the Roboword program from
Technocraft.
- use the DOS JDIC mentioned below.
All of the above work with just "English" versions of Windows.
-
WindowsCE/Windows Mobile
-
Unix/Linux (X-Windows)
- My own
xjdic
(V2.4) which is available from the
Monash ftp site.
It needs to run in a kterm window, and has been used successfully on
virtually every type of Unix & Linux system.
- the
Gjiten
package, which is very nice, and has its own flexible GUI. (Gnome)
- the new
gWaei,
which aims to be a "dictionary program for the Gnome desktop with support
for regular expressions."
- the fast and light-weight
Kiten
(KDE).
- for Emacs/XEmacs users there is
edict.el.
I don't know much about it, but I think it is included in the XEmacs
rpm.
On the Monash ftp site I have some
information
and the latest
release.
-
Smartphones
For Android phones there are two main options:
- the
AEdict app.
This excellent app uses copies of the dictionary files downloaded to the phone, so it
works well offline.
-
the
WWWJDIC for Android app.
This uses the WWWJDIC server via its API, and needs network access. It
has the advantage of always being up-to-date as the dictionary is
expanded and corrected.
For Apple iPhones two options are:
- the
Imawa app.
(formerly Kotoba)
which is very highly regarded. It is similar to AEdict, but uses the
JMdict data and hence provides all the information from the dictionary.
- the newer
EDICT with Grammar
app. This uses the common words subset in the old EDICT format.
WWWJDIC itself has a
simple mobile phone
interface, which I developed for Japanese keitai many years ago.
-
Macintosh
(This section is quite out of date.)
Mac users have a number of options if they have Japanese support with their
OS (I think the support is standard for later versions):
-
DOS
The two main main programs for DOS are:
- the
JDIC
program which provides a Japanese-English and English-Japanese dictionary function.
- the
JREADER
text reader,
which includes an integrated dictionary look-up function.
-
Others
There are also programs for Amiga, BeoS, Palm Pilots, etc. Most can be
obtained from the Monash ftp site.
There is a
Jabber bot
that does local EDICT lookups too.
Romaji?
None of the files in the EDICT project use romanized Japanese.
I get many requests for a romaji version of EDICT, however as I do not like
romaji and do not want to encourage its use, I will not be producing
romaji versions. There is
a romaji version dating from 1997 on the Monash ftp site. This was
prepared for a blind person who was using a non-Japanese Braille
interface. I (foolishly) placed it on my ftp site, and I have had a lot
of problems since it was not in step with the main file. That file is
now withdrawn, and I am asking all sites carrying copies to withdraw it.
Publications
If you like, you can collect some papers I have written about the project:
Other useful links can be found on my
Japanese Page.
Jim Breen
December 2013
July 2017