I am reviewing contraction and possessive recognition techniques and all of the suggestions I have received have some flaw in them.
Good spellcheckers are hard to build in the 80's.
There is now available a spelling checker program. This program takes an English text file and outputs the file with all the misspelled words and their line numbers at the end. There are over 47000 words in the dictionary. This is a lot but not enough to guarantee that your favorite words are in there. So, When you find correctly spelled words marked as not in the dictionary please send them to me and I will update the dictionary. To use this program put the following command in your login.com file: $ WEB*STER :== $DOC:WEBSTR 'P1 Typing WEBSTER file.ext will run the program and process the mentioned file. The default input extension is .MEM. Webster will create a file of the same name with the extension .CRF Please forward any problems and comments to me. I will update the dictionary every couple of weeks. If your favorite words do not show up after you send them to me please be patient.
I have been informed that the 'p1 in the command line of my previous message is incorrect and should be removed. The correct line should be: $ web*ster :== $ doc:webstr This should work fine. If there any problems let me know.
It has come to my attention that contractions (can't, didn't, etc) don't work in the webster program. The Dictionary has these words and the program will recognize them if I define the apostrophe (') as an alphabetic character. I have done this and contractions now work. There is, however, a price for this feature. The dictionary does not have all of the possible possessive forms of words. Since the program is able to recognize the word "didn't", it will not find the word "witch's" in the dictionary. I use contractions more then I use possessives. If there is a problem with this I would like to hear about it.
Webster can now handle almost all cases of contractions and possessives. If it can't find a word in the dictionary it scans the word and truncates the word at any apostrophe that may be there. For example: "witch's" will become "witch". The dictionary is consulted again with the new word. This leave the only problem being a misspelling like "can'tj". the 'tj is removed and the word "can" is in the dictionary. So, no mention will be made of the word in the listing. If you don't like this feature I will set it back to not handling possessives at all.
I guess I was unclear. Webster will only strip after the apostrophe if the word is not found in the dictionary. Words like can't and didn't will be found in the dictionary so they will not be flaged as wrong. The only inaccuracy is the "can'tj" case mentioned earlier as the word "can'tj" is not in the dictionary. I am reviewing contraction and possessive recognition techniques and all of the suggestions I have received have some flaw in them. Therefore, I am regrouping and will release a new version over the weekend.
This is the final word on contractions vs. possessives. Webster first looks up a word in the dictionary. This includes any apostrophes that may be in the word. Most of the common contractions are listed in the dictionary and will be found this way. If the word is not in the dictionary Webster checks do see if the word ends in S'. If it does, the ' is removed and a dictionary lookup is done on the root word. If the root is not found in the dictionary the original word with the apostrophe is added to the misspelled list. If the word ends in 'S the 'S is removed and a spelling check is done on the root word. Once again if the root is not in the dictionary the whole word is placed in the misspelled list. This system will catch almost all cases of contractions and possessives. There is one exception, however. A singular word ending in S is, by the rules of usage, made into a possessive by adding 'S. So, the possessive of abacus is abacus's. If you have this correct in your text Webster will not complain. However, A common mistake for this case is to treat the word as if it was plural. ( abacus') This is incorrect but Webster will not flag it as so. This is not a major problem. I have found only a very few nouns that end in S. It is difficult to say words like abacus's so people usually put these words in a prepositional phrase ( of the abacus). The ackwardness of these words has caused most of them to fall into disuse. Spending 10 minutes looking in the dictionary only uncovered a few. ( abacus, lotus, mass, and marquis) I sure there are more but not many.
The file [MAHAR]NUWORDS.DOC lists the new words that have been added to Websters dictionary.
Webster is no longer in the disk area DOC:. It is now in the area DOK:. You should change your login.com file to reflect this change.
Jan 26, 1984