Go to the previous section.
Ispell is a program that helps you to correct typos in a file,
and to find the correct spelling of words. When presented with a
word that is not in the dictionary, ispell attempts to find
near misses that might include the word you meant.
This manual describes how to use ispell, as well as a little about its implementation.
The simplest emacs command for calling ispell is 'M-$' (meta-dollar. On some terminals, you must type ESC-$.) This checks the spelling of the word under the cursor. If the word is found in the dictionary, then a message is printed in the echo area. Otherwise, ISPELL attempts to generate near misses.
If any near misses are found, they are displayed in a separate window, each preceded by a digit. If one of these is the word you wanted, just type its digit, and it will replace the original word in your buffer.
If no near miss is right, or if none are displayed, you have four choices:
Insert the word in your private dictionary. Use this if you know that the word is spelled correctly.
Accept the word for the duration of this editing session, but do not put it in your private dictionary. Use this if you are not sure about the spelling of the word, but you do not want to look it up immediately. The next time you start ispell, it will have forgotten any accepted words. You can make it forget accepted words at any time by typing M-x reload-ispell.
Leave the word alone, and consider it misspelled if it is checked again.
Replace the word. This command prompts you for a string in the minibuffer. You may type more than one word, and each word you type is checked again, possibly finding other near misses. This command provides a handy way to close in on a word that you have no idea how to spell. You can keep trying different spellings until you find one that is close enough to get a near miss.
Lookup. Display words from the dictionary that contain a specified substring. The substring is a regular expression, which means it can contain special characters to be more selective about which words get displayed. See section `Regexps' in emacs.
If the only special character in the regular express is a leading
^, then a very fast binary search will be used, instead of
scanning the whole file.
Only a few matching words can be displayed in the ISPELL window.
If you want to see more, use the look program directly from
the shell.
Of course, you can also type ^G to stop the command without changing anything.
If you make a change that you don't like, just use emacs' normal undo feature See section `undo' in emacs.
If you want to check the spelling of all the words in a buffer, type the command M-x ispell. This command scans the file, and makes a list of all the misspelled words. When it is done, it moves the cursor to the first word on the list, and acts like you just typed M-$ See section Checking a single word.
When you finish with one word, the cursor is automatically moved to the next. If you want to stop in the middle of the list type Q or ^G. Later, you can pick up where you left off by typing C-X $.
You may check the words in the region with the command M-x ispell-region. See See section `mark' in emacs.
The commands available are the same as for checking a whole buffer.
Until ispell becomes part of the standard emacs distribution, you will have to explicitly request that it be loaded. Put the following lines in your emacs init file See section `init file' in emacs.
(autoload 'ispell "ispell" "Run ispell over buffer" t) (autoload 'ispell-region "ispell" "Run ispell over region" t) (autoload 'ispell-word "ispell" "Check word under cursor" t) (define-key esc-map "$" 'ispell-word)
(It will do no harm to have these lines in your init file even after ispell is installed by default.)
To check the words in a file, give the command ispell FILE. This
will present a screen of information, and accept commands for every word
that is not found in the dictionary.
The screen shows the offending word at the top, as well as two lines of context at the bottom. If any near misses are found, they are shown in the middle of the screen, each preceded by a digit.
You may use the same commands as inside of emacs to accept the word, place it in your private dictionary, select a near miss, or type a replacement See section Checking a single word. You may also choose from the following commands:
Print a help message.
If you type your interrupt character (usually ^C or DEL), then ispell will immediately enter its command loop. If ispell was generating near misses at the time, then all that it had found so far will be displayed, along with a message stating that there might be more, and that you can type RET to generate them. If it was scanning the file, it will display `(INTERRUPT)' where it would normally display a bad word, and the commands that change the file will be disabled.
The feature is handy if you have left out a space between words, and ispell is futilely looking up the 1000 potential near misses for a string that has twenty letters.
When ispell is run with no arguments, it reads words from the standard input. For each one, it prints a message telling whether it is in the dictionary. For any words not in the dictionary, near misses are computed, and any that are found are printed.
% ispell word: independant how about: independent word: xyzzy not found word: ^D
Whenever ispell is started the file `ispell.words' is read from your home directory (if it exists). This file contains a list of words, one per line, and neither the order nor the case of the words is important. Ispell will consider all of the words good, and will use them as possible near misses.
The I command adds words to `ispell.words', so normally you don't have to worry about the file. You may want to check it from time to time to make sure you have not accidentally inserted a misspelled word.
The `-u' flag tells ispell to be compatible with the traditional `spell' program. This flag is automatically turned on if the program is invoked by the name `spell'.
This flag causes the following behavior:
All of the files listed as arguments (or the standard input if none) are checked, and misspellings are printed on the standard output. The output is sorted, only one instance of each word appears (however, a word may appear more than once with different capitalizations.)
You may specify a file containing good words with `+filename'.
The troff commands `.so' and `.nx' (to include a file, or switch to a file, respectively) are obeyed, unless you give the flag `-i'.
The other `spell' flags `-v', `-b', `-x' and `-l' are ignored.
By the way, ispell seems to be about three times faster than traditional spell.
Commands valid in both modes:
DIGIT Select a near miss I Insert into private dictionary A Accept for this session SPACE Skip this time R Replace with one or more words L Lookup: search the dictionary using a regular expression
Standalone only:
Q Accept rest of file X Abandon changes ! Shell escape ? Help ^Z Suspend ^L Redraw screen ^C Give up generating near misses, or show position in file
Emacs only:
M-$ Check word M-x ispell Check buffer M-x ispell-region Check region M-x reload-ispell Reread private dictionary M-x kill-ispell Kill current subprocess, and start a new one next time ^G When in M-x ispell, stop working on current bad word list ^X $ Resume working on bad word list.
Two words are near each other if they can be made identical with one of the following changes to one of the words:
Interchange two adjacent letters. Change one letter. Delete one letter. Add one letter.
Someday, perhaps ispell will be extended so that words that sound alike would also be considered near misses. If you would like to implement this, see Knuth, Volume 3, page 392 for a description of the Soundex algorithm which might apply.
Ispell's arguments are parsed by getopt(3). Therefore, there is considerable flexibility about where to put spaces between arguments. The way to be safe is to give only one flag per dash, and put a space between a flag and its argument.
If ispell is run with no arguments, it enters `ask' mode See section Using ispell to look up individual words. With one or more file name arguments, it interactively checks each one.
-p privname
-d dictname
-l
-u
-a
-S
-D
-E
Ispell can be used as a subprocess communicating through a pipe. Two interfaces are available:
To use this interface, start ispell with the '-S' flag. Ispell will print a version number and greeting message that looks like:
(1 "ISPELL V4.0")=
The number is the version number of the protocol to be spoken over the pipe. The string is a message possibly of interest to the user.
All messages from ispell end in an equal sign, and ispell guarantees not to print an equal sign except to end a message. Therefore, if you do not want to deal with the greeting, just throw away characters until you get to an equals.
Ispell then reads one line commands from the standard input, and writes responses on the standard output.
If a command does not start with a colon, then it is considered a
single word. The word is looked up in the dictionary, and if it is
found, the response is t. If the word is not in the
dictionary, and no near misses can be found, then the response is
nil. If there are near misses, the response is a line containing
a list of strings in lisp form. For example:
INPUT OUTPUT the t xxx nil teh ("tea" "ten" "the")
The near miss response is suitable for passing directly to the lisp
read function, but it can also be parsed simply in C. In
particular, ispell promises that the list will appear all on one line,
and that the structure will not change. A parser that reads the whole
line, then treats the parentheses and quotes as whitespace will work fine.
The list will contain a maximum of ten strings, and each string will be no longer than 40 characters. Also, the capitalization of the near misses is the same as the input word.
If the input line starts with a colon, then it is one of the following commands:
:file filename
Run the word checker over the named filename. The response is zero or
more lines each containing a number. The numbers are file offsets of
words that do not appear in the dictionary. Since the near miss
checker is not run, this is fairly fast.
After the last number, there will be a line containing either t if
the checker got to the end of the file, or nil if it received an
interrupt. If ispell ignores any interrupts received except while
scanning a file.
:insert word
Place word in the private dictionary.
:accept word
Do not complain about word for the rest of the session.
:dump
Write the private dictionary.
:reload
Reread the private dictionary.
:tex
Enable the tex parser for future :file commands.
:troff
Enable the tex parser for future :file commands.
:generic
Disable any text formatter parsers for future :file commands.
To use this interface, start ispell with the '-a' flag. Ispell will read words from its standard input (one per line), and write a one line of output for each one.
If the first character of the line is *, then word was found in the
dictionary. (Other versions of ispell made a distinction between words
that were found directly, and words that were found after suffix
removal. These lines began with +, followed by a space, then
followed by the root word. To remain compatible with these version,
treat + and * the same.)
If the line starts with &, then the input word was not found, but
some near misses were found. They are listed on the output line
separated by spaces. Also, the output words will have the same
capitalization as the input.
Finally, if the line starts with #, then the word was not in the
dictionaries, and no near misses were found.
INPUT OUTPUT the * xxx # teh & tea ten the
This section is excerpted from the ITS spell.info file.
Words in SPELL's main dictionary (but not the private dictionary) may have flags associated with them to indicate the legality of suffixes without the need to keep the full suffixed words in the dictionary. The flags have "names" consisting of single letters. Their meaning is as follows:
Let # and @ be "variables" that can stand for any letter. Upper case letters are constants. "..." stands for any string of zero or more letters, but note that no word may exist in the dictionary which is not at least 2 letters long, so, for example, FLY may not be produced by placing the "Y" flag on "F". Also, no flag is effective unless the word that it creates is at least 4 letters long, so, for example, WED may not be produced by placing the "D" flag on "WE".
"V" flag: ...E --> ...IVE as in CREATE --> CREATIVE if # .ne. E, ...# --> ...#IVE as in PREVENT --> PREVENTIVE
"N" flag: ...E --> ...ION as in CREATE --> CREATION ...Y --> ...ICATION as in MULTIPLY --> MULTIPLICATION if # .ne. E or Y, ...# --> ...#EN as in FALL --> FALLEN
"X" flag: ...E --> ...IONS as in CREATE --> CREATIONS ...Y --> ...ICATIONS as in MULTIPLY --> MULTIPLICATIONS if # .ne. E or Y, ...# --> ...#ENS as in WEAK --> WEAKENS
"H" flag: ...Y --> ...IETH as in TWENTY --> TWENTIETH if # .ne. Y, ...# --> ...#TH as in HUNDRED --> HUNDREDTH
"Y" FLAG: ... --> ...LY as in QUICK --> QUICKLY
"G" FLAG: ...E --> ...ING as in FILE --> FILING if # .ne. E, ...# --> ...#ING as in CROSS --> CROSSING
"J" FLAG" ...E --> ...INGS as in FILE --> FILINGS if # .ne. E, ...# --> ...#INGS as in CROSS --> CROSSINGS
"D" FLAG: ...E --> ...ED as in CREATE --> CREATED if @ .ne. A, E, I, O, or U, ...@Y --> ...@IED as in IMPLY --> IMPLIED if # .ne. E or Y, or (# = Y and @ = A, E, I, O, or U) ...@# --> ...@#ED as in CROSS --> CROSSED or CONVEY --> CONVEYED
"T" FLAG: ...E --> ...EST as in LATE --> LATEST if @ .ne. A, E, I, O, or U, ...@Y --> ...@IEST as in DIRTY --> DIRTIEST if # .ne. E or Y, or (# = Y and @ = A, E, I, O, or U) ...@# --> ...@#EST as in SMALL --> SMALLEST or GRAY --> GRAYEST
"R" FLAG: ...E --> ...ER as in SKATE --> SKATER if @ .ne. A, E, I, O, or U, ...@Y --> ...@IER as in MULTIPLY --> MULTIPLIER if # .ne. E or Y, or (# = Y and @ = A, E, I, O, or U) ...@# --> ...@#ER as in BUILD --> BUILDER or CONVEY --> CONVEYER
"Z FLAG: ...E --> ...ERS as in SKATE --> SKATERS if @ .ne. A, E, I, O, or U, ...@Y --> ...@IERS as in MULTIPLY --> MULTIPLIERS if # .ne. E or Y, or (# = Y and @ = A, E, I, O, or U) ...@# --> ...@#ERS as in BUILD --> BUILDERS or SLAY --> SLAYERS
"S" FLAG: if @ .ne. A, E, I, O, or U, ...@Y --> ...@IES as in IMPLY --> IMPLIES if # .eq. S, X, Z, or H, ...# --> ...#ES as in FIX --> FIXES if # .ne. S, X, Z, H, or Y, or (# = Y and @ = A, E, I, O, or U) ...# --> ...#S as in BAT --> BATS or CONVEY --> CONVEYS
"P" FLAG: if @ .ne. A, E, I, O, or U, ...@Y --> ...@INESS as in CLOUDY --> CLOUDINESS if # .ne. Y, or @ = A, E, I, O, or U, ...@# --> ...@#NESS as in LATE --> LATENESS or GRAY --> GRAYNESS
"M" FLAG: ... --> ...'S as in DOG --> DOG'S
Note: The existence of a flag on a root word in the directory is not by itself sufficient to cause SPELL to recognize the indicated word ending. If there is more than one root for which a flag will indicate a given word, only one of the roots is the correct one for which the flag is effective; generally it is the longest root. For example, the "D" rule implies that either PASS or PASSE, with a "D" flag, will yield PASSED. The flag must be on PASSE; it will be ineffective on PASS. This is because, when SPELL encounters the word PASSED and fails to find it in its dictionary, it strips off the "D" and looks up PASSE. Upon finding PASSE, it then accepts PASSED if and only if PASSE has the "D" flag. Only if the word PASSE is not in the main dictionary at all does the program strip off the "E" and search for PASS.
Therefore, never install a flag by hand. Instead, just add complete new words to the dictionary file, then use the build program with the options '-a -r' to replace as many roots with flags as possible.
I first came across ispell on TOPS-20 systems at MIT. I tracked it down to ITS where I found the PDP-10 assembly program. It appeared that it had been in use at the MIT-AI lab since at least the late 1970's. I think it case from California before then.
I wrote the first C implementation in the spring of 1983, mostly working from the ITS INFO file.
The present version was created in early 1988, and was motivated by the desire to make it run on 80286's, and to provide a better interface for GNU EMACS.
There is another widely distributed version of ispell, which was forked from my 1983 version and has a different set of features and clever extensions. It is available from the directory /u/public/ispell at celray.cs.yale.edu.
People who have contributed to various versions of ispell include: Walt Buehring, Mark Davies, Geoff Kuenning, Rober McQueer, Ashwin Ram, Greg Schaffer, Perry Smith, Ken Stevens, and Andrew Vignaux.
Pace Willisson
pace@ai.mit.edu pace@hx.lcs.mit.edu
(617) 625--3452
Go to the previous section.