Xref: utzoo comp.bugs.sys5:356 comp.bugs.4bsd:727
Path: utzoo!mnetor!uunet!nbires!isis!aburt
From: aburt@isis.UUCP (Andrew Burt)
Newsgroups: comp.bugs.sys5,comp.bugs.4bsd
Subject: Re: Non-word "accreditate" in /usr/dict/words
Message-ID: <2222@isis.UUCP>
Date: 11 Mar 88 05:54:24 GMT
References: <1693@desint.UUCP>
Reply-To: aburt@isis.UUCP (Andrew Burt)
Organization: Math/CS, University of Denver
Lines: 33

In article <1693@desint.UUCP> geoff@desint.UUCP (Geoff Kuenning) writes:
>...I stumbled across a non-word in my /usr/dict/words file.

Brings to mind a word that showed up on a list of five letter palindromes
(no, I wasn't bored, I was making a handout about regular expressions
with ^\(.\)\(.\).\2\1$ as an example):

	rever

Now, I admit I did find it in the OED.  But that was the only dictionary
of mine that listed it (of about a half dozen).

If I saw this in a document I'd assume it was a misspelling of "revert" or
"revere", etc.; and spell allows "revers" as a plural, which probably
should be "reverse".

This brings up an interesting question:  Should /usr/dict/words list
words that are technically allowable (listed in some notable dictionary)
but are (a) very uncommon and (b) very close to likely misspellings of
far more common words -- at the expense of not catching what are
probably typos?  To my mind, a spelling checker should flag words that
are correct over omitting incorrect words.

I can't see "but it makes the dictionary complete" argument being used
since many common words (in a Unix environment) are missing, such as
"filename", "pathname", "stdin",...  (Maybe a -u(nix) option to spell is
in order... :-)

-- 

Andrew Burt 				   			isis!aburt

              Fight Denver's pollution:  Don't Breathe and Drive.