by Dan Crevier, Richard Northcott, Abbey Kazuya and Ayumi Kimura
UniDict program requires System 7.0 or later (MacOS 8.0 for some features), with KanjiTalk or the Japanese Language Kit. For full functionality, you should have AppleScript, Text-to-Speech, Claris XTND engine, Contextual Menus. To enable client-server remote dictionary searching AppleTalk and FileSharing will need to be installed as well.
The CD-ROM contains two installers: one for the Japanese localized version; the other for the English version of UniDict. The default setting is to install a number of recommeded dictionaries along with the application. If you do not want to use all the dictionaries or do not have enough space on your hard disk, you may selectively install dictionaries and their corresponding indices with the Custom Install feature.
You can also use UniDict straight from the CD-ROM by clicking on the alias. This will be slower, but you won't have to put anything on your hard disk. It is also possible to just copy the indicies for dictionaries and have the dictionary files remain on the CD-ROM or even on a server.
UniDict is set up to use various folders in the same directory at the application. Firstly let's look at these folders:
this is where UniDict places index files for the dictionaries it has imported. Only index files in this folder are searched. The index files include the location of the base dictionary. The Indices Folder also holds the "ElementBase" file that is required by the "Kanji Constructor" .
this is the place where you would usually keep your dictionary data. It is also the place where UniDict looks if a dictionary has been moved. UniDict does not automatically load files in the Dictionary Folder, you need to create an Index for it first.
Startup Items Folder:
applications or their aliases in this folder get launched automatically when UniDict is run. This is useful to setup background applications etc. that you want running at the same time. A good example is "DTalker" which speaks Japanese sent to it from UniDict via AppleEvents.
this folder is for UniDict plug-ins. These allow UniDict to create indices for new dictionary formats and to allow communication with other applications. This plug-in architecture makes UniDict the most flexible dictionary engine available.
If you haven't used Unidict before, the first step is to set up your dictionaries. On the CD-ROM UniDict comes with EDICT, ENAMDIC and KANJIDIC and their indices. EDICT is the main E<>J dictionary. ENAMDIC is the place and people name data moved to a separate file. KANJIDIC is used for the Kanji Search window (see below). Your absolute minimum install is the EDICT dictionary in the "Dictionaries" folder and its index in the "Indices" folder. Plus if you wish to use "Kanji Constructor" you should also have "ElementBase" in the "Indices" folder.
Other less important dictionaries are in the Support Folder without indices. To use these or add new dictionaries from other sources to UniDict you need to create an index for each of them. To do this you should first move the dictionary into the "Dictionaries" folder on your hard disk and then build an idex for them. There are two types of dictionaries that can be imported. They are "Text Dictionaries" and "Kanji Dictionaries". The latter is for KANJIDIC only and you can only have one kanji dictionary installed at one time. All other dictionaries use the first import method. For details of import options see the "Dictionary Index" section below.
The text dictionaries themselves should be in the EDICT format. It is basically as follows:
|kanji <space> [reading] <space> /definition 1/definition 2/<character return>|
|$@4A;z(J [$@$+$s$8(J] /chinese characters/ideographs/|
The format is described in more detail in the EDICT documentation on the CD-ROM. UniDict does support dictionaries in other formats, such as tab delimited, though. Japanese dictionaries can be in EUC or ShiftJIS (SJIS) format, with ShiftJIS giving slightly faster results. You can convert text dictionaries to ShiftJIS using the "EUC>SJIS" utility included in the "EDICT & KANJIDIC tools" folder. You can make your own dictionaries or modify existing ones with a text editor or UniDict's built-in Glossary Editor. Whenever you change a dictionary you will need to rebuild its index before using it. UniDict also lets you build dictionaries from scratch with the "New Glossary" function. Glossaries have their indices automatically updated for you. See the "Glossaries" section below.
If you wish to install a lot of dictionaries, you might want to change the amount of memory Unidict uses through the "Get Info..." command in the Finder. This will speed up your searches and let you handle more dictionaries and larger results lists. Also read about creating sets in "Sets" section below.
Language Search Window
UniDict's Language Search Window is simple enough. Just type a word into the search field and click the "Search" button. UniDict will display the results of the search in the section below. Selecting any text in the results section will place that text into the search field ready to search again. Of course you can paste text into the search field as well, or use the "AutoPaste" preference.
The UniDict search field supports input of any kind --English, kanji or kana pronunciation. It will display all the matching entries in all the dictionaries indices installed. This can lead to a very large list. The matching portion of each entry will be displayed in red on color screens.
The "Compounds" menu lets you set the depth of searches on kanji characters. Compounds (jukugo) are words created by combinations of kanji. The "All" setting lets you find all of the compounds (jukugo) containing a kanji. This is the setting you would normally use. The "None" setting will give returns of only the single kanji or characters in the search field. "Kanji is First" will return only compounds that begin with the search character. "Kanji is not First" will return all compounds that contain the character but do not use it as the first one.
Other search options allow you to narrow or widen general search parameters.
"Exact Match" set searches to return only text that exactly match the search field text. With this option off, if you search for "hand," you will also get things like "handball." Turning "Exact Match" on lets you narrow down long match returns.
The "Fuzzy Match" option sets 'kana' searches to ignore long vowels and "$@!<(J". This is useful if you are not sure of the correct pronunciation of a word. Setting it would return "$@$H$&$-$g$&(J" even if you only searched for "$@$H$-$g(J". Alternatively you would also receive a return of "$@$H$&$-$g$/(J" from a search on "$@$H$&$-$g$&(J". This option should be used sparingly.
The "Deinflect" option tries to unconjugate Japanese verbs. In other words UniDict would return for you "$@$7$k(J" even though you searched for "$@$7$j$^$7$?(J". (Deinflect only supports the common verb conjugations.) If you search a lot of Japanese text you might consider using this option until you get used to handling Japanese verb forms.
Go Back contains a history of recent searches. You can use it to call back search results after you have searched for something else. Clicking on it takes you back to the previous search. Holding it down opens a pop-up menu of recent searches. Items in italics are no longer in memory and will be searched from the index again.
After a language search in done the results returned from the dictionaries and plug-ins are assembled in the area below the search input field and search option checkboxes. You can save the entire window or copy and paste out portions of it. As with all parts of UniDict, this area supports drag and drop export. To quickly make another search on part of the results displayed you can just select it and it will automatically be placed in the search field. Typing Return or pressing the "Search" button will then search on your selection.
Kanji Search Window
UniDict has another window for displaying information about individual kanji characters. You can search from the Language Search Window with the "Do Kanji Search" menu command. It also supports direct input in its own search field.
The Kanji Search Window has a search field similar to the Language Search Window. It supports English meanings, kanji characters or their kana readings. You can also search by another category by selecting it from the "by :" pop-up menu.
The results of your searches are displayed in the field on the bottom right. Clicking on one of those characters will show an enlargement and detailed information about that character. You can sort the kanji result list by a number categories including Stoke count or Radical.
UniDict uses "KANJIDIC" for kanji searches. KANJIDIC has many different fields including readings, textbook and reference code numbers, and more are added periodically. The fields supported by each character will vary. UniDict support them all in a flexible way with the character's details being placed into a scrollable list on the left. You can select fields in the list and copy them directly. For the standard fields that are not in the scrolling list, you need to select the information you wish to copy from a pop-up button to the right of the list. There is also a menu command in the Edit menu for copying the displayed kanji.
You can select whether you wish to see the readings in kana or romanized by changing the preference in the Preferences dialog.
The Kanji Search Window has a "Go Back" button that is similar to the one the Language Search Window. It will return results for previous kanji searches.
The "Radicals" button displays a visual list of the major radicals and their variants. The listings are in loosely associated groupings and do not use any formal radical classifications. Clicking on any of the radicals will call up a list of characters that use parts that "look like" the radical. When using the Radical button you do not need to select the "by:" search method.
The "Fuzzy Match" checkbox in the Kanji Search Window is slightly different from the Language Search Window in that it does not affect long vowels, but it does work with small "ya,yu,yo" readings and SKIP codes.
The "X-Ref" button will become active only when a selected character has a cross-reference listing. Clicking on this will bring up a list of associated or non-standard "itaji" variants.
The "Sort by:" menu sets how characters in the search list are displayed. The default is number of strokes.
UniDict supports only one KANJIDIC file. It supports only the standard JIS Level 1 & 2 sets. It does not yet support the JIS X 0212-1990 set.
The Kanji Constructor uses a database of kanji parts and radicals. You need to have the "Element Base" file in the Indices folder for the menu item to become active. Opening Kanji Constructor is a window of elements often used in kanji characters. Clicking on one of the elements displays a list on the right of all the characters that incorporate it as either a radical or a part. Clicking on another element will reduce the list to only characters that include both elements, etc.
It also lets you construct kanji by clicking on multiple elements. The idea is to look be able to up kanji that you do not know the reading for by clicking on bits that you can recognize until the list becomes small enough for you to find the character you are hunting for.
Check boxes let you reduce searches to either JIS Level 1 or 2. This will reduce the time it takes to display the list.
Buttons let you clear the list or copy the selected character or lookup the kanji character in the Kanji Search Window. Double-clicking on a character has the same effect as pressing the "Lookup" button. If you have clicked on a kanji, you can put it into the clipboard by hitting Command-Shift-C.
"New" and "Open..." refer to UniDict's built-in text editor. It supports styled text, imbedded pictures or sounds and files over 32KB in size. You will find it useful to edit small dictionary files or editing search results for saving. UniDict's built-in editor holds all files in memory, so you will need to assign a great deal of memory to UniDict to edit large files.
"Show Search Window" opens the Language Search Window if it has been closed or brings it to the front if it is not.
"Import File" and "Export File" use the Claris XTND engine to open or save files in formats or encodings not normally supported by UniDict.
The "Copy As" menu item lets you copy the selected text and convert the contents of the Clipboard into various encodings. It supports JIS, EUC and Unicode.
The "Convert to $@$+$J(J" and "Convert to romaji" menu items convert the search field to those formats. This will help you when inputting kana and let you what readings are if you can't remember the kana pronunciation.
The "Speak Selection" function in the edit menu uses the Apple's Speech Manager Text-to-Speech engine, "MacinTalk Pro," to read aloud the selected text. If "MacinTalk Pro" is installed, you can select the voice used in the Preference dialog. If you are running MacOS 8 with Japanese Speech Manager, "MacinTalk$@$+$J(J", installed you can select which voice is used for the kana text portions and UniDict will read each for you in 'native' voices. If Japanese Text-to-Speech is not installed only the Roman portion of selected text will be read aloud.
"Speak with DTalker" will send the selected text to Catena's DTalker application via AppleEvents. DTalker will read kanji and kana text for you. DTalker will need to be running at the same time as UniDict for this menu to become active.
"Enlarge Search Field" shows you an enlarged view of the text in the search field.
This dialog is accessed from the "Edit" menu. It lets you control the behaviour of UniDict and set fonts for text in various places in the program.
"Auto Paste" is a feature that will pull the contents of the clipboard into a search field, in either the main search window or kanji search window, whenever one is brought to the front. This is very useful if you keep UniDict in the background while handling text in another application. A quick Copy and click on UniDict will set you up for a search. After UniDict comes to the front, all you need to do is hit return.
UniDict will automatically translate the readings for kanji dictionary entries from hiragana to romaji in the Kanji Search Window. If you are not yet confident with hiragana, you might set this to romaji.
The voice menus will only appear when you have the Apple Text-to-Speech Manager, "MacinTalk", installed. If you have kana voices installed you will have another menu appear to let you set which voice speaks kana text for you. If you do not have MacinTalk and "Speech Manager" installed neither menu will appear.
UniDict lets you set the font, style, color, and size of text displayed in various parts of the application. This lets you set up display to your liking. The default setting displays search hits in red, Japanese text in Osaka, and Roman text in Geneva. It assumes you have the default fonts for a Japanese System or JLK. But you can set these to any font, style or size you wish. Complex Kanji can be hard to read with screen generations of outline fonts. If you have a large size bitmap font, might be better off setting fonts in the Kanji Search window to it. Since UniDict supports dictionaries with European text, it is best to keep your roman font settings to a non-ShiftJIS font.
The "Create New Index" function will create an index for a text dictionary and place it in the "Indices" folder. The index name will then be displayed in the "Dictionaries" menu. The "Create Index" function supports three different kinds of import. They dictate the size of index file and kind of words supported by searches. The most common and largest index type is the "Japanese <->English" type. This will create and index with both sides of the dictionary's text lines being searchable. The "Japanese -> English" and "English -> Japanese" types create index for only half of the text file. "Import All Kanji" includes all kanji in compounds in the file (this is the recommended format). Not selecting it will result in only the first character being indexed. "First word only" restricts index to the first word or character of an entry. This negates the "Import all Kanji" option. This should only used for special cases where you require one way searching such as from a ZIP number to address. "Use Japanese Script" is for files that include Japanese text. This sets the search result to the correct font. You would not select this option for a text dictionary that was English to German for example.
The index file includes alias information as to the location of the text dictionary it belongs to. If you created an index for a dictionary on a remote disk or a volume you have unmounted, UniDict will try to mount that volume. If it can not find the dictionary it will look in the local "Dictionary" folder. If it can not find it there, UniDict will ask you to locate it or delete the index.
Create Kanji Index...
The Kanji Search Window uses a different kind of dictionary format and index. To create an index for a new version of KANJIDIC use the "Create Kanji Index..." command. Only one kanji index can be used at one time. You must make sure you do not use this command with anything other than KANJIDIC. For more information about KANJIDIC and kanji searches see the section on the Kanji Search Window.
Sets are used to easily switch between different line-ups of dictionaries to be used in language searches. Sets are manipulated through the heirachical "Sets" menu found in the "Dictionaries" menu. This is useful if you have a large number of dictionaries or use special purpose files. Reducing the number of dictionaries searched will speed up UniDict's response time. Unidict will remember the set you used last and return to it next time it starts up. When new dictionaries are imported, they are automatically added to the standard set, but not to the other sets.
Sets are created with the "New..." command in the "Sets" submenu of the "Dictionaries" menu. UniDict prompts for the name of the new set, and adds it to the "Sets" submenu. Dictionaries can be checked or unchecked from the dictionary menu to specify whether or not they will be used for searches when using the set.
To switch to another set, simply select it from the set menu. The dictionaries in the checked "Dictionaries" menu change to reflect the new dictionary setting. UniDict also assigns command key shortcuts for fast switching.
Sets can be deleted by choosing the "Delete Set" command from the "Sets" submenu. This deletes the current set. The Standard set cannot be deleted.
UniDict comes with an interface for creating your own dictionary files from scratch. These are called Glossaries and are really just containers for text dictionary data in the EDICT format. Using the Glossary Editor removes the chance of making mistakes when editing text files by hand. When the "Include in Dictionary Menu" option is selected, UniDict will automatically create and keep updated an index for your glossary and include it in the "Dictionaries" menu ready for searching. "Import Glossary..." will open an existing glossary or a small EDICT text file. The Glossary Editor also supports copy and paste of text in the EDICT format. Glossary files are edited in memory, so you should restrict editing to small files unless you assign a large amount of memory to UniDict.
The UniDict CD-ROM comes with a Contextual Menu module for MacOS 8.x Systems. This is a very useful feature. If you install the Internet Address Detectors 1.01 (also on the CD), it includes an enabler that allows you to activate Contextual Menus from any application that supports text. The result is that any piece of text that you can select, you can hold down the Control key and a menu will appear allowing you to initiate UniDict searches from anyway -your Web Browser or whatever. If UniDict is not running, it will be launched and the search begun. You can select between the Language and Kanji Search from within the menu.
UniDict includes an AppleEvent suite to let you look up words from another application. You can either send an text string and just receive the result text back or run searches in the application just like the Contextual Menu feature above. UniDict also supports remote events over a network. Included on the UniDict CD-ROM, is a "Scripting" folder which has samples and details about scripting with various programs or AppleScript. It also has the source code you need to wire your own applications to converse with UniDict. AppleEvents and AppleScript will allow UniDict to grow and become more powerful as more applications support it.
UniDict supports plug-ins that enable it to support new dictionary types, send out AppleEvents to other applications or even to do remote searches via other copies of UniDict on your local network.
UniDict uses a plug-in architecture to support new types of dictionary formats, such as multimedia content or non-Macintosh CD-ROMs. Keep an eye on Enfour's UniDict Web page for new plug-ins. Some maybe bundled with commercial dictionaries.
UniDict has the ablity to be both a client and a server to enable AppleTalk LAN-wide remote searching. This enables you to keep rarely used dictionaries, CD-ROMs or other media supported by plug-ins, possibly even your entire dictionary collection in one central location. This is useful for company LANs or school language labs.
To set up for client-server searching, you need to set up a Macintosh elsewhere on your AppleTalk network as a server. Install on it UniDict and the dictionaries you wish to search remotely. Run the remote copy of UniDict and set up the current set (see "Sets" above). Make sure that "Application linking" is enabled in the "File Sharing Settings" control panel. You will need to give all the people who wish to access the server adequate file access privileges. It is easiest to set up to allow guest access, but check with your LAN administrator if you are not sure.
On your local machine (the client), drag the "Client Plug-in" from the "LAN Client" folder (inside the "UniDict Extras" folder on the UniDict CD-ROM) into the "Plug-ins" folder of your client copy of UniDict. Run the client copy of UniDict and you will be asked to locate the server. Select the zone (if you have more than one), the name of the UniDict server machine, and the name of the server UniDict. Make sure that you do not select any other application or your searches will fail. On your first search, you will be asked for your access password. Every time you quit the client UniDict the password will be forgotten, but the server will be remembered.
Any time the server is down or a name is changed, you will be asked if you wish to look for the server or not and if you do, the setup dialog will appear again. If you wish to force a change of server settings, hold down the Option key while launching UniDict. To temporarily disable remote searching, hold down the Shift key while launching UniDict.
There is no point in having the same dictionary searched in both places so either remove your local copy or turn off that dictionary in the Sets menu on the server machine.
Although fast, especially with EtherTalk, this style of searching is not as fast as local dictionaries either on a hard disk or read directly from the UniDict CD-ROM. If you do have room on your local drives we recommend that you install your dictionaries locally. You might create sets with your commonly used dictionaries stored locally and remote searching only kept as a backup.
*For commerical dictionaries, you should read their licencing agreement before installing dictionaries as a server.
UniDict is designed to be the ultimate "user's dictionary" engine. You input is valued and UniDict will grow to fit everybody's needs. Drop us a line if you have any particular suggestions. EDICT and the FreeWare dictionary community is always looking for more input, so if there is something you think you can add, give it a try.
For more information and all the latest news, check out this Web Page:
The latest version of EDICT, KANJIDIC and the other EDICT format files are at:
The information page for Japanese at Monash University is:
Jim Breen can be reached at:
Jim Breen <email@example.com>
Plus there are a great number of Japanese dictionary files on NIFTYServe
You can access these through CompuServe as well
To contact Enfour:
fax: (Japan) -81-3-5474-8934
Back to the UniDict top page
Enfour Group Top page