![[Deutsch]](../common/de.gif){kind=small}
![[Hrvatski]](../common/hr.gif){kind=small}
|
|
ESpell |
It looks for certain "particles" in the text and tries to replace them with other "particles". It tries out all the variations and checks them all with the spelling checker. Only replacements that happen to be O.K. are actually done.
You must first go to 
and choose the profile.
Here you can also decide how ESpell will behave when there are more than one appropriate replacement for
the word we are checking. If we have chosen 
, that's what we'll
see in such situations:
This is how we choose the right word:
and 
) and click 
or press EnterWith Add to exclusions dictionary! it is possible to add a word to the intern exclusions dictionary of the particular profile.
IMPORTANT: Open "Tools/Options/Spelling and Grammar" and check the settings there. What the spelling checker ignores, ESpell can't process.
You start ESpell with
that button. Do not select anything. Just put the cursor where you want to start. ESpell will stop at the end
of the "story". "Stories" are body, header, footer,
footnotes. Keep in mind that, if you want to use ESpell on different "stories", you must do it
separately - you must put the cursor in the body and start ESpell, put the cursor in the footnotes and start
ESpell again etc. You can use 
to insert
the stop sign in the document on the place where you want ESpell to stop. For it to work, you must activate
the option 
in . If it's necessary, you can always break the program
execution with "Ctrl + Pause".
KEEP IN MIND: Blue words will be ignored!
ESpell is relatively slow. It's well
programmed, but all programs that work in this environment (Microsoft Word, VBA) are slow. If the document is
rather long, you might want to "take a walk". Activate the option 
in . ESpell will change
the color of all dubious words into red. When you later press ,
it will process only red words which will be considerably faster.
does the same
as but with only one word. Select the word or
put the cursor in it or on it's beginning.
deletes the color
attribute of the selected text. If you want to use it on just one word, you don't have to select it. Just put
the cursor in it or on it's beginning.
is a simple
auxiliary program. It works on the same principle as ESpell but without consulting the spelling checker. It
acts on the selected text or, if nothing is selected, from the cursor to the end of the "story".
The profiles you don't want you can delete.
After you have finished correcting text with ESpell you should spell-check it. You will maybe find words that ESpell should have changed but couldn't find the proper replacement for them in the spelling dictionary. You should change these words manually and than add them to the user dictionary (Right click/Add) so that ESpell would be able to correct them in the future.
OCR programs can learn. But they can't learn everything. It's, for example, almost impossible to tell the difference between the letter O and the number 0. In such cases you must use the context. And this is exactly what ESpell does. It tries to replace the zero with O and checks to see if it makes sense.
The other typical errors may vary depending of the font, paper and print quality etc. It is usually hard to tell the difference between "rn" and "m", between the letter "l", the number "1" (and sometimes the letter "I" and "i" too). These errors are part of the OCR profile. But you should create your own profiles for different kinds of materials you scan.
OCR programs don't (at least Abbyy - the only OCR program I've tested doesn't) make difference between
"-" like in "error-handling" or "above-mentioned" and hyphens that divide words
into syllables at the ends of rows. ESpell can do it but there is one trick: Replace all "-" with
some letter that does not appear in your text, "³" for example (of course, only if your text
is not in Polish language). You can use to do this. Then use ESpell. It will try to replace "³"
with "" (with nothing) wherever it makes sense. At the end you need to change the remaining "³"
back to "-". Again, you can use .
In some German texts we have ae, oe, ue and ss, instead of ä, ö, ü and ß. To correct that, you can not simply turn all "oe" back into "ö" etc. because there are words like "Goethe" that should actually stay that way. The solution: ESpell.
There is a similar problem in many other languages. To be precise: there was. But the folks are still afraid to use the extended characters, which are part of their national alphabets, on the Internet. As a result, there is still a lot of transcribed texts in Italian, Croatian, Serbian, Polish etc. that need to be transformed back into their original form. ESpell can do it all.
By now, profiles exist only for German and Croatian. (The profiles for Croatian can be used for Serbian but
it seems that there is no spelling dictionary for Serbian. If one of you knows it exists and where to find it,
please let me know.)
Profiles have extension .ini but they don't have the internal structure of standard windows ini files.
We will see how they are built using "Umgeschriebenes Deutsch.ini" as an example:
The file "Umgeschriebenes Deutsch.exc":
|
row No. |
content of the row |
explanation: There are two kinds of exceptions: whole words and word parts |
| 1 | 0 | 0 because 0 rows with whole words follow |
| 2 | 0 | Word parts - 0 too. You should look at the file "Srpski.exc". There you can see more. |
If you create your own profiles put the corresponding"*.ini" and "*.exc" files in the folder ESpellProfiles. Avoid using names of type "*.#.#.ini" (for example OCR.1.3.ini) because they are reserved for future versions of the program.
IMPORTANT: Both files must exist! If .exc is empty than it must contain zeros as shown below.
Structure: The first row reads #TRUE# if we want the replace operation to be case sensitive. Otherwise it reads #FALSE#. All other rows tells us what to replace and with what. Unicode character code numbers can stand in place of characters. It's nevertheless possible only when we have "particles" that consist of only one character.
If you create your own profiles put the corresponding"*.ini" file in the folder ZConvProfiles. Avoid using names of type "*.#.#.ini" (for example OCR.1.3.ini) because they are reserved for future versions of the program.
ESpell is free for noncommercial use. If you are a commercial user or you simply like the program, you are encouraged to send $7 × number of computers it is installed on. But you can also send more or less. For now it is only possible to send cash. You can send it on the address:
Zoran Vucic
Ljubijska 45
10040 Zagreb
Croatia
If you want to use parts of the source code in your program, contact me.
What are your experiences in working with ESpell?
