The Risks Digest

The RISKS Digest

Forum on Risks to the Public in Computers and Related Systems

ACM Committee on Computers and Public Policy, Peter G. Neumann, moderator

Volume 15 Issue 74

Saturday 2 April 1994

Contents

o Re: Spell checking
J. Taggart Gorman
Pete Mellor
Castor Fu
Les Earnest
o Re: Language ability is not entirely learned
Paul Colley
o Re: Spelling, punctuation, poor language technology
Bill Stewart
o Re: English spelling design
Christopher Upward
o Info on RISKS (comp.risks)

Re: Spell checking (RISKS-15.71)

Jonathan Haruni <jharuni@london.micrognosis.com>
31 Mar 1994 13:57:32 GMT
The only risk I see in spelling checkers is that people may trust them too
much, or even expect too much of them.  This is a widepsread risk of every
technology.

 1) You suggest that "it is easy to go unconscious in front of the mouse
    and press 'replace' one too many times".  Surely a program which replaces
    words of your document without any knowledge of their meaning or intention
    is not one with which you should go "unconscious", especially when
    applying it to a document which, as you suggested, could lose you your job.

 2) Your use of "business" and "profane" are a bit misleading.  What you
    really want is a checker which does not suggest words which are
    inappropriate to the purpose of your document (making money from
    customers).  How is that possible ?  Whatever a document says, there are
    valid words which express the opposite ideas.  You cannot omit from the
    dictionary all the words which might cause offence to anyone doing
    "business".  Even if you did, add or omit the word "not" in a suitable
    place in a business proposal, and you could lose a customer.  Can a
    business document not contain profanities if it suits the purposes of the
    document ?

 3) Can you really blame the SPELLING checker for suggesting a common English
    word in place of a proper name ?  I have never seen a spell checker which
    did not allow you to augment the dictionary.  The first thing a company
    should do is add all the names and addresses of every organization it
    deals with.  Have you done that ?

I suggest that you have bought a useful tool (spelling checker), failed to
make the effort of tailoring it to your needs (adding your correspondents to
the dictionary), and expected too much of it (to provide you with a document
which fulfills your intentions, rather than merely one without spelling
errors).  Then you fed it a document with so many errors that you became
complacent about its power (not paying conscious attention to its prompts for
confirmation).  In spite of all this, it corrected your spelling mistakes and
you DID notice the "goddamn" suggestion when prompted, so you ended up with a
better document.

Where are the risks ?

Jonathan Haruni.


Re: RISKS of RISKS on spelling-checkers

J. Taggart Gorman <taggart@scopus.com>
Fri, 1 Apr 94 17:09:51 PST
  It seems that some RISKS contributors are engaged in an activity that should
not be discussed on a computer-related list such as this: spellcasting.
  However, it is obvious that they are using their computers to help with
their witchery, for they keep on mentioning using "spell-checkers".  At least
we know that modern technology is helpful to all, including spellcasters.
  On the side, I've never seen a "spell-checker" for sale in a computer store.
Are they commonly available in occult stores?  (Available on the Microsoft
Occulta CD?)

  :) (Not April's Fools, but with a light content for the day.)

Taggart Gorman  taggart@scopus.com


Re: Risks of spelling checkers (Girard, RISKS-15.71)

Pete Mellor <pm@csr.city.ac.uk>
Sat, 2 Apr 94 21:02:14 BST
> with the suggestion that the word "Goldman" (as in a large company we all
> know) should be replaced with "goddamn". The word processor involved was MS

      ------From the Daily Mail, Friday April 1st 1994, p21------------

      Don't you dare be sexist says the PC PC, by Suzanne O'Shea

The new computer program promised to help users write better English.

But buyers have ended up with more than they bargained for. As well as a
guide to glitch-free grammar and scintillating syntax, they get a lesson
in political correctness every time they switch on.

The use of words such as `wife', `policeman' and `housewife' meets with
a sharp rebuke from the software, which flashes up a message that they are
`gender-specific' then provides `gender-neutral' options such as `spouse',
`police officer' and `homemaker'.

Anyone foolish enough to test the PC personal computer with words such as
`little woman' or `girlie' is sternly informed that they are `sexist
expressions'. No alternative is offered here, only the ominous message:
`Avoid using this word.'

Computer writer Mark Smithson, 51, of Bedford, risked the wrath of the
{pounds} 250 Microsoft Word 6 package when he typed in the word `freeman'.
The computer promptly spat back `citizen'.

`I couldn't believe it,' he said yesterday. `Then I started going through
lots of other sexist and "gender-specific" words and, sure enough, the
same thing happened.

`It's like Big Brother. Manipulating what people write is a form of
censorship. I am the last person to be deliberately sexist but this is
downright frightening.'

In the politically correct world of Word 6 - produced by an American firm -
users are advised to replace `mankind' with `humankind' or `humanity' -
although `womankind' passes through without a hitch - and to replace `fireman'
with `stoker'.

Its scope is limited when it sees words which it has not been told are sexist.
While `little lady' may result in the reprimand `sexist expression, avoid
using this phrase', followed by the explanation that `this term is considered
by many to be inappropriate and belittling when used to refer to women', the
word `floozie' is freely allowed.

No mention of the programme's political correctness was mentioned
[sic: Perhaps Ms. O'Shea should use a style checker! - PM :-) ]
in publicity material when Word 6 was launched in Britain recently. Neither
is the feature listed in the 830-page manual.

A Microsoft spokesman - sorry, spokeswoman, we mean spokesperson - defended
the program yesterday.

`It does not force users to change what they write,' she said. `It simply
highlights words that might be regarded as sexist and suggests alternatives.

`Microsoft is trying to bring its programmes in line with real life and how
people actually work. This type of thing is a sign of the times, as people
do say chairperson instead of chairman nowadays.'

  [Disclaimer: I don't *think* this is an April fool joke (if only because, if
  it were not true, Bill Gates would sue the Mail), but if it is, I didn't
  make it up! :-)  Peter Mellor, Centre for Software Reliability, City Univ.,
  Northampton Sq., London EC1V 0HB  +44 (71) 477-8422  p.mellor@csr.city.ac.uk


More spelling checker stories

Castor Fu <castor@drizzle.Stanford.EDU>
Fri, 1 Apr 1994 18:01:42 -0800 (PST)
When cleaning up one day we found a portable spelling checker.  To test the
size of its vocabulary, we tried out some proper names.  We were dismayed to
find it suggesting "a**hole" [censored by PGN] as a correction for "Achille",
my housemate's name. This was particularly unimpressive, as "Achilles", the
more common spelling, was actually in its dictionary, but was not among any of
the alternatives, which included a number of other unflattering possibilities.

-Castor Fu   castor@drizzle.stanford.edu


Risks of spelling checkers (RISKS 15.72)

Les Earnest <les@sail.stanford.edu>
2 Apr 1994 03:26:54 GMT
The earliest spelling checker was evidently one that was part of a pen-based
computer system for cursive writing recognition that I developed at MIT
Lincoln Lab in the 1959-61 time period.  It was set up to recognize the 10,000
most common English words.

Sometime in 1961 a film crew from BBC came to the lab and asked to photograph
the handwriting recognizer as part of a television program on advanced
technology, to which I agreed.  After setting up, they asked if the system
could recognize the word "television."  I agreed to give it a try but pointed
out that it sometimes listed more than one word if it wasn't sure.  After I
wrote the word on the CRT with a light pen, the system paused only a second or
two before responding:
  TEDIOUS
  TELEVISION

The film crew loved it and zoomed in for a close-up!  I've often wished
that I had asked for a copy of their film.

Les Earnest (Les@cs.Stanford.edu)               Phone:  415 941-3984
Computer Science Dept.; Stanford, CA 94305    Fax:  415 941-3934


Language ability is not entirely learned (Ranum, RISKS-15.69)

Paul Colley <colley@qucis.queensu.ca>
Sat, 26 Mar 1994 23:41:54 GMT
>Spelling mistakes are a result of inattention to detail, ignorance, or apathy.

Which makes poor spelling sound like a deliberate decision.  I assure you I am
neither inattentive to detail nor apathetic about my poor spelling abilities.
I like to think I'm not ignorant...

In my defense, I'll note that there is some strong evidence that language is
based, at least in part, on genetics.  Thus some portion of language skill is
beyond the control of the individual.

Quoting from Jay Ingram's book, "Talk Talk Talk", pp.133-141, there is...

    "...a gene that makes it possible for most of us to be able to
    add an `s' to a word to make it plural, or choose `he' instead
    of `they' when it's appropriate, or add `ed' to a word when it
    happened in the past!  Apparently if you inherit a faulty
    version of this gene you will never be able to do any of those
    automatically.

    [...]

    These people aren't aware that they have a problem making
    plurals or past tenses, [...]

    [...]

    This discovery [...] makes it much more difficult to argue that
    language is simply a byproduct of learning, [...]

The defect occurs in non-English speakers also.  The gene seems to only
affect language, and only the ability to make plurals and past tenses.
If there's a gene for plurals, there are probably genes for other
components of language.

Reference:  Myrna Gopnik, linguist at McGill University, "Linguistic
Properties of Genetic Language Impairment," address to the American
Association for the Advancement of Science, February 10, 1992,
Chicago.

- Paul Colley      colley@qucis.queensu.ca    +1 613 545 3807


Re: Spelling, punctuation, poor language technology

<wcs@anchor.ho.att.com>
Sun, 27 Mar 94 03:46:51 EST
Aside from all the flames about whether spelling and punctuation
errors come from poor language design (:-) or poor user education or
differences in values, there *are* some new technology-related problems.
Many maga- zine articles, especially in the com- puter industry,
are suffering from leftover hyphen- ations, which come from re-for- matting
word-processed text and not checking whether -'s at the ends of lines are
intentional dashes or are hyphens put in to accommodate line-breaks before
including the - and space in the new text.

"Wired" is one of the worst offenders, probably because most of its
authors use a variety of computer systems to write on.

        Bill Stewart

    [RISKS readers will notice that I try to REMOVE hyphenations whenever
    I spot them.
      Other comments on this subject were received from
      brewer@cs.wmich.edu (Steven D. Brewer) and
      albaugh@agames.com (Mike Albaugh).  PGN]


English spelling design

<c.upward@aston.ac.uk>
Tue, 29 Mar 1994 18:38:29 +0000
I just picked up the Don Norman/Mark Jackson/Alayne McGregor exchange on
'its', 'it's', and English spelling design generally.

Don is right about bad spelling design being the cause of endless problems
of written English. But Halle & Chomsky were wrong about underlying deep
consistency in English spelling. For one thing, their analysis ignored such
fundamental inconsistencies as <ea/ee> in 'speak/speech'. For another
thing, they ignored the whole historical dimension, which Don Norman
rightly alludes to.

The truth is that for 1,000 years no one has been able to ensure consistency,
deep or otherwise, in English spelling, ie since the Norman Conquest of
England in 1066, English spelling, unlike that of most languages, has not been
"designed with the user in mind", as Don Norman very sensibly puts it.
Webster's contribution was a small step in the direction of greater
consistency, which the British have still largely failed to follow. Various
people have tried using extra symbols (Benjamin Franklin was one), but they
have always run up against the problem of needing to teach all th millions
(billions?) of potential readers what these new symbols stand for.

As for the apostrophe,the deep INconsistency of English rears its head there
too. Mostly the possessive apostrophe precedes final <s> with singular nouns:
'the dog's kennel', but follows it in the plural: 'the dogs' kennels'. But
sometimes we find the reverse: 'men's' is plural, but 'Achilles'' is singular.

A different set of inconsistencies affects the possessive pronouns mentioned
by Mark Jackson. As he rightly says, most don't use apostrophes, so that we
write 'hers', 'ours', 'yours', 'theirs', and of course 'its', and not 'her's',
'our's' etc.  But 'one's' is an exception: for some reason we DO write that
with an apostrophe.  However, the craziest inconsistency is 'whose', where we
add an <e> at the end!

If Alayne McGregor implying that all languages are written as inconsistently
as English, he is mistaken. English is unique - as are its problems of
illiteracy.  Both the USA and Britain have recently published major reports on
its appalling extent.

We do need to get to grips with this question of spelling design. Let me now
attach a recent paper put out by the Simplified Spelling Society on the
subect.

Simplified Spelling Society World HQ c/o Bob Brown, 133 John Trundle Court,
Barbican, London, EC2Y 8DJ, tel. 071-628 5876.
US HQ c/o Ken Ives, 401 E 32, Apt 1002, Chicago IL 60616.


CUT SPELLING

A Streamlined Writing System for English

a proposal for modernizing English spelling by removing redundant letters
Enquiries to Chris Upward
Chairman of the Society's Cut Spelling Working Group
61 Valentine Road, Birmingham, B14 7AJ, England
Tel. 021-444 2837, Fax. 021-359 6153.


THE BACKGROUND

Why reform English spelling?

English spelling is notoriously hard to master. It is a centuries-old writing
system whose contradictions and eccentricities were never designed for a fully
literate society. We all suffer from its clumsiness and inconsistency: it
takes far longer to learn than more regular systems; it limits people's
ability to express themselves; it causes mispronunciation, especially by
foreign learners; most people acquire at best an erratic command of it (even
skilled writers are prone to uncertainty and error); and many millions are
condemned to functional illiteracy. It is therefore small wonder there is such
concern about standards of literacy in English-speaking countries today. Yet
many of those countries have in recent decades seen the benefit of modernizing
equally antiquated systems of currency and weights & measures. Similar
modernization of English spelling is badly needed.

Is reform possible?

Spelling reform is an unfamiliar idea to the English-speaking world, but other
languages show it is feasible and indeed a normal way of preserving a writing
system from obsolescence. The letters of the alphabet were designed to stand
for the sounds of speech, but pronunciation evolves in the course of time, and
confusion sets in when letters and sounds cease to match: the way we speak
words now no longer tells us how to write them, and the way they are written
no longer tells us how to speak them. That is the central problem of English
spelling. In the past century many languages have modernized their spelling to
improve this match between letters and sounds, and so aid literacy. To ensure
continuity, only small changes are usually made, and while schoolchildren
learn some new, improved spellings, most adults continue to write as before.
It may therefore take a lifetime before everyone uses the new forms. Ideally,
spelling reform needs to be an imperceptibly slow, but carefully planned and
continuous process.

Problems of regularizing

Many schemes have been devised for respelling English as it is pronounced, but
apart from some small improvements in America none has been adopted for
general use. Several fully regularized systems have however been tried in the
past 150 years in teaching beginners, with dramatic success in helping them
acquire basic literacy skills, the best known recently being the i.t.a.
(initial teaching alphabet). However, all these schemes have required learners
to transfer to the traditional irregular spelling as soon as they can read and
write fluently, and much of the advantage is then lost.
        Ideal though total regularization may ultimately be, the effect such
schemes have on written English is so drastic as to be a major deterrent to
their adoption. The following sentence, in the Simplified Spelling Society's
New Spelling (1948), perhaps the best thought-out and most influential of
these fully regularized orthographies, demonstrates the effect:"Dhe langgwej
wood be impruuvd bie dhe adopshon of nue speling for wurdz". Less radical
proposals have therefore been made since then, so as to avoid such visual
disruption, suggesting for instance that at first only the spelling of one
sound, like the first vowel in any, should be regularized; or a single
irregularity, like <gh>, should be removed.  However, the immediate benefit of
such a reform would be slight.
        A new approach is called for if today's readers are not to be
alienated, yet learners are to benefit significantly.

STREAMLINING

Cutting redundant letters

In the 1970s the Australian psychologist Valerie Yule found that many
irregular spellings arise from redundant letters. These are letters which
mislead because they are not needed to represent the sound of a word.  Writers
then cannot tell from a word's pronunciation which letters its written form
requires, nor where to insert them, while readers are likely to mispronounce
unfamiliar words containing them. A group within the Simplified Spelling
Society therefore decided to explore which letters are redundant in English,
and the effect their removal has on the appearance of the resulting 'cut'
text. This Cut Spelling (CS) is now demonstrated.

Esy readng for continuity

One first notices that one can imediatly read CS quite esily without even
noing th rules of th systm. Since most words ar unchanjed and few letrs
substituted, one has th impression of norml ritn english with a lot of od
slips, rathr than of a totaly new riting systm. Th esential cor of words, th
letrs that identify them, is rarely afectd, so that ther is a hy levl of
compatbility between th old and new spelngs. This is esential for th gradul
introduction of any spelng reform, as ther must be no risk of a brekdown of
ritn comunication between th jenrations educated in th old and th new systms.
CS represents not a radicl upheval, but rather a streamlining, a trimng away
of many of those featurs of traditionl english spelng wich dislocate th smooth
opration of th alfabetic principl of regulr sound-symbl corespondnce.

FURTHR ADVANTAJS

Savings

Th secnd thing one notices is that CS is som 10% shortr than traditionl
spelng. This has sevrl importnt advantajs. To begin with, it saves time and
trubl for evryone involvd in producing ritn text, from scoolchildren to
publishrs, from novlists to advrtisers, from secretris to grafic desynrs.  CS
wud enable them al to create text that much fastr, because ther wud be fewr
letrs to rite and they wud hesitate less over dificlt spelngs.  Scoolchildren
cud then devote th time saved in th act of riting (as wel as that saved in
aquiring litracy skils) to othr lernng activitis. Simlr time-saving wud be
experienced by adults in handriting, typng, word-procesng, typ-setng, or any
othr form of text production. Th reduced space requiremnt has typograficl
benefits: public syns and notices cud be smalr, or ritn larjr; mor text cud be
fitd on video or computer screens; fewr abreviations wud be needd; and fewr
words wud hav to be split with hyfns at th ends of lines. Ther wud also be
material savings: with around one paje in ten no longr needd, books and
newspapers wud require less paper (alternativly, mor text cud be carrid in th
same space as befor), and demands on both storaj and transport wud be less.
And th environmnt wud gain from th loer consumtion of raw materials and enrjy
in manufacturng and from th reduction in th amount of waste needng to be
disposed of.

Targetng spelng problms

Less imediatly obvius is th fact that CS removes many of th most trublsm
spelng problms that hav bedevld riting in english for centuris. Ther ar thre
main categris: ther ar silent letrs, such as <s> in isle or <i> in business,
wich ar so ofn mispelt eithr as ilse, buisness, or as ile, busness; th latr ar
th CS forms. Anothr categry is that of variant unstresd vowls, as befor th
final <r> in burglar, teacher, doctor, glamour, murmur, injure, martyr, wich
CS neatly alyns as burglr, teachr, doctr, glamr, murmr, injr, martr. Thirdly
ther ar th dubld consnnts, so ofn mispelt singl today, as found in such words
as accommodate, committee, parallel(l)ed; CS simplifys these to acomodate,
comitee, paraleld.


RULES OF CUT SPELLING

Cutting rules

These three problem areas of traditional spelling correspond to the three
main rules of Cut Spelling (CS).

Rule 1  Letters irrelevant to pronunciation

About 20 of the 26 letters of the alphabet are sometimes used with no bearing
on pronunciation at all. Some, like <e> in love, <gh> in though and <w> in
answer, were once sounded, but fell silent centuries ago. Others were taken
from foreign languages, like <ch> in yacht (Dutch), <h> in honest (French),
and <p> in psyche (Greek), but are always silent in English. Yet others were
inserted by analogy (<gh> in haughty to match naughty, <l> in could to match
would) or to show a dubious or imagined derivation (<b> in doubt, <c> in
scythe). Two vowel letters are often written when the pronunciation only needs
one; thus <a> in measure, <e> in hearth, <i> in friend, <o> in people, <u> in
build are all redundant. CS removes letters such as these from hundreds of
often common words; most strikingly, CS eliminates that most grotesque of all
English spelling patterns, the <gh>.

Rule 2a Unstressed vowels before <l,m,n,r>

Thousands of English words contain 

                    
    

Please report problems with the web pages to the maintainer

Top