Modeling the rules of spelling
Simon
order to write an application which will be able to help people with
spelling difficulties understand which rules of spelling that do not
know in order to learn those rules and improve their spelling.
I am looking for people who are interested or who have or are doing
research in this area more specifically the functional decomposition
of the rules of spelling in to a more complex model delineating the
rules in to rules, meta-rules and exceptions to rules, in order to
code them.
I have come across much research in to rule based spelling models
however none that I have found reflect the rules that we use to learn
to spell.
I would be grateful for any correspondence on this or related
research.
Lester Zick
Surely, any of the rules applied by spell checkers on computers would
be of some help.
Regards - Lester
remove DEL in address for email
Simon
only if the rules applied in the spelling checker are used by people
to learn to spell.
Lester Zick
I'm not sure if people do use such rules to learn to spell. But the
rules spell checkers actually employ to decipher word spelling are
fascinating because they expose a fairly strict set of analytical
considerations that go into structure and spelling. So, I think you
could do a lot worse to begin there with a view towards determining
whether such considerations play any definite role in people's
approach to spelling.
Wolf Kirchmeir
>Surely, any of the rules applied by spell checkers on computers would
>be of some help.
>
>
>
>Regards - Lester
Spell checkers on computers don't apply rules. They just look up the words on
the fly. So-called hash-tables and binary search trees enable very fast
searching. The search algorithms use the same general principles as the party
game known as 20 Questions or Animal-Vegetable-Mineral - the next step is
always an exclusive either-or step. If you have 20 questions, you can find
any one of 2^20 items - which is an awful lot. There are about 2^10 words in
English, so at most 10 steps are needed to find a word. That takes
microseconds with a modern processor.)
Actual spelling rules used by humans are not easy to establish. IMO, it's an
empirical question, but in all my years as a teacher, there was precious
little done. Pretty well all research on "spelling" that I've seen is
partisan -- it tries to support one side or the other in the phonics vs
whole-word war. I've not cluttered my memory or notebooks with references to
it. :-)
The primers used in schools have limited value as teaching/learning tools;
their value is heavily teacher-dependent in my experience. They don't IMO
encode the rules people actually develop and use as they become skilled
writers (always excepting typos, which is a different issue neurologically.)
Bottom line: Simon is pretty well on his own, and IMO will have to devise and
do some research himself.
HTH
Best Wishes,
Wolf Kirchmeir
Blind River, Ontario
..................................................................
You can observe a lot by watching
(Yogi Berra, Phil. Em.)
..................................................................
Lester Zick
<wwol...@sympatico.ca> wrote:
>On Mon, 06 Jan 2003 15:56:49 GMT, Lester Zick wrote:
>
>>Surely, any of the rules applied by spell checkers on computers would
>>be of some help.
>>
>>
>>
>>Regards - Lester
>
>Spell checkers on computers don't apply rules. They just look up the words on
>the fly. So-called hash-tables and binary search trees enable very fast
>searching. The search algorithms use the same general principles as the party
>game known as 20 Questions or Animal-Vegetable-Mineral - the next step is
>always an exclusive either-or step. If you have 20 questions, you can find
>any one of 2^20 items - which is an awful lot. There are about 2^10 words in
>English, so at most 10 steps are needed to find a word. That takes
>microseconds with a modern processor.)
I was wondering when you were going to check in on this subject since
you obviously have the most practical experience. I did get to
wondering about the accuracy of my reply since my experience dates
from the '70's when machines had a much more limited capacity.
There are much more than 2^10 words in English because that's only
1024, and I believe the average vocabulary runs at least 2-3000 words.
And when you include grammatical inflections, possessives,
contractions, alternate spellings, etc, the number goes much higher.
But even so the number is problably not more than 20-30 thousand at
most. And I would think a PC could easily handle a binary search of
that magnitude.
>
>Actual spelling rules used by humans are not easy to establish. IMO, it's an
>empirical question, but in all my years as a teacher, there was precious
>little done. Pretty well all research on "spelling" that I've seen is
>partisan -- it tries to support one side or the other in the phonics vs
>whole-word war. I've not cluttered my memory or notebooks with references to
>it. :-)
IBM did do some work on the subject in the '70's at least. I can
remember seeing references to it for their word processors, and I can
distinctly remember being impressed with their ability to make some
sense of it all. What I can remember seemed pretty ingenious. You
might check out the IBM research journal from the period. You might
also post the query to the newsgroup comp.lang.asm370 because they
have a lot of posters with some long experience who might remember.
What I can't tell is whether any machine rules devised have any
relevance to the way people either do or could learn to spell. But at
least they seemed pretty systematic and were certainly impartial.
>
>The primers used in schools have limited value as teaching/learning tools;
>their value is heavily teacher-dependent in my experience. They don't IMO
>encode the rules people actually develop and use as they become skilled
>writers (always excepting typos, which is a different issue neurologically.)
>
>Bottom line: Simon is pretty well on his own, and IMO will have to devise and
>do some research himself.
>
>HTH
>
>
>Best Wishes,
>
>Wolf Kirchmeir
>Blind River, Ontario
>
>..................................................................
>You can observe a lot by watching
>(Yogi Berra, Phil. Em.)
>..................................................................
>
>
Regards - Lester
Wolf Kirchmeir
>There are much more than 2^10 words in English because that's only
>1024, and I believe the average vocabulary runs at least 2-3000 words.
Sorry, my arithmetic mistake. There are roughly 2^18 (500K) words in English,
not counting inflected forms. A search program can find a word in 18 binary
steps or less. Spell check dictionaries generally have around 100K words, or
2^16 entries. Anyhow, the program search progressively. Eg, "able" is found
in four binary steps:
1) a vs not-a
2) ab, vs not-ab
3) abl vs not-abl
4) able vs not-able...
In actual practice, it's more complicated than that, since a step may entail
further search, but it's pretty quick all the same - hence the on-the-fly
spell check functions available with current word-processors. (Which IMO is
not as useful as it must have seemed when implemented, since I find the
constant highlighting of partial words irritating.)
"Vocabulary" is a slippery concept. We recognise many more words than we use,
and recognise more words in text than in speech. In fact we can all
understand words we encounter in text but can't (or don't) pronounce
properly. We write more words than we speak, and we never write some words
that we speak, and vice-versa. The vocabulary-size test I used with my senior
high school classes indicated that most of them had a 4-6,000 word
recognition vocabularies, with some in the 8-10,000 word range.
BTW, "big words" are generally spelt far more regularly that the "little"
everyday words, and homonyms (especially homophones) cause no end of
trouble.
It's text-to-speech programs that use "spelling rules". They would be a good
place for Simon to start his investigations.
Lester Zick
This seems like a good idea. Your numbers and observations seem quite
reasonable. Once I noted that those who typically misspell learn words
from speech whereas those who typically mispronounce words learn them
from reading.
>Best Wishes,
>
>Wolf Kirchmeir
>Blind River, Ontario
>
>..................................................................
>You can observe a lot by watching
>(Yogi Berra, Phil. Em.)
>..................................................................
>
>
Regards - Lester
Simon
Would you say then that there is any truth in the hypothesis that:
"People who can not spell generally do not understand the rules of
spelling or are not using the rules of spelling correctly".
I was told this recently when I was Tested for dyslexia by an
educational psychologist, hence my interest in this subject(and my
poor command of writen english). I have found some quotations that
agree with this but not much empirical evidence
Wolf Kirchmeir
The problem with this diagnosis is that it's too vague. Also, it's woefully
incomplete.
"Not understanding the rules" is unclear. Does it mean or include simple
ignorance? Some misspellings are certainly the result of ignorance. However,
carelessness, inattention, and spelling context are major factors. EG, I
found that students who misspelled words in their compositions nevertheless
scored perfect or near-perfect scores on spelling tests. And we all know (or
should know) that we notice spelling mistakes and typos in other people's
writing but not in our own. Spelling errors happen even with people who
"understand" the rules very well.
Spelling is a largely unconscious skill, and I suspect it is a skill learned
largely unconsciously -- it is a conditioned skill, IOW. Explicit rules don't
help much, if only because you have to also explicitly state the exceptions,
which makes the rules cumbersome. In my experience, voracious readers tend to
be good spellers (and good writers generally), and people who read little
tend to be bad spellers - regardless of their overall intellectual abilities.
Eg, the notoriously bad spelling of engineers and computer gurus.
IOW, spelling is far more complicated than "understanding the rules."
As for "not using the rules correctly:" That's also more complicated. "Using
rules" implies the ability to distinguish between correct and incorrect forms
-- dyslectics can't do that, to varying degrees, and for a variety of
reasons. The most common form of dyslexia appears to be an inability to
separate phonemes. It is possible to train a dyslectic to improve this skill,
but it's a very difficult task - which implies that there's a neurological
glitch.
An observation that might be helpful: Many English speakers have trouble
spelling non-English names -- and also have trouble pronouncing them.
Misspelling often takes the form of reversing letters <ks> instead of <sk>,
for example. This suggests that among other things we associate
sound-clusters with letter groups, not just individual sounds with individual
letters (or letter groups).
HTH
Lester Zick
<wwol...@sympatico.ca> wrote:
>On 9 Jan 2003 01:12:41 -0800, Simon wrote:
>
>>Thank you for you interaction.
>>
>>Would you say then that there is any truth in the hypothesis that:
>>
>>"People who can not spell generally do not understand the rules of
>>spelling or are not using the rules of spelling correctly".
>>
>>I was told this recently when I was Tested for dyslexia by an
>>educational psychologist, hence my interest in this subject(and my
>>poor command of writen english). I have found some quotations that
>>agree with this but not much empirical evidence
>
>The problem with this diagnosis is that it's too vague. Also, it's woefully
>incomplete.
>
>"Not understanding the rules" is unclear. Does it mean or include simple
>ignorance? Some misspellings are certainly the result of ignorance. However,
>carelessness, inattention, and spelling context are major factors. EG, I
>found that students who misspelled words in their compositions nevertheless
>scored perfect or near-perfect scores on spelling tests. And we all know (or
>should know) that we notice spelling mistakes and typos in other people's
>writing but not in our own. Spelling errors happen even with people who
>"understand" the rules very well.
You know, I've noticed this last myself. I wonder if spelling problems
could have a contextual origin. We would naturally be less familiar
with other people's language structures, and spelling errors might
therefore jump out at us in unfamiliar verbal contexts. (On the other
hand we might just be paying closer attention to words in unfamiliar
contexts.)
Conversely, we might learn to spell isolated words correctly and then
be unable to apply the spellings accurately in the unfamiliar written
context of writing out our thoughts. If so, it would be a very
difficult problem to correct. This might also be why those who read
and write a lot tend to spell well, because they are exposed to a wide
variety of different verbal contexts and modes of expression.
Regards - Lester
Eto Demerzel
news:3e1ca3fe...@netnews.att.net...
> This seems like a good idea. Your numbers and observations seem quite
> reasonable. Once I noted that those who typically misspell learn words
> from speech whereas those who typically mispronounce words learn them
> from reading.
>
Don't people learn words from both speech and reading?
Also - is it not possible to mis-pronounce words after hearing them?
I have heard people refer to their 'crotch' as their 'crutch', and people
saying they were being 'pacific' instead of 'specific'. Perhaps there is a
tendency to hear a word and assume it's a different meaning of a similar
word we already know, rather than a completely different word altogether.
Me../
(No professional background in this. Just a bundle of mostly unorganised
thoughts.)
Lester Zick
<eto.demerzel@../../^wfijivillage.com> wrote:
>"Lester Zick" <lester...@worldnet.att.net> wrote in message
>news:3e1ca3fe...@netnews.att.net...
>
>> This seems like a good idea. Your numbers and observations seem quite
>> reasonable. Once I noted that those who typically misspell learn words
>> from speech whereas those who typically mispronounce words learn them
>> from reading.
>>
>
>Don't people learn words from both speech and reading?
Of course. Therein lies the difference between mispronunciation and
misspelling, I suspect.
>
>Also - is it not possible to mis-pronounce words after hearing them?
It would depend on the length of time and other factors. One might
deliberately mispronounce a word as a dialectical inflection or for
effect, but I don't see any special reason one would do it for other
reasons if the pronunciation is correctly remembered.
>
>I have heard people refer to their 'crotch' as their 'crutch', and people
>saying they were being 'pacific' instead of 'specific'. Perhaps there is a
>tendency to hear a word and assume it's a different meaning of a similar
>word we already know, rather than a completely different word altogether.
Things like crotch and crutch, pacific and specific aren't really
different pronunciations: they're really mistakes in word meaning and
selection. I am assuming that the original word known would have been
known correctly. It's always possible to misconstrue what you hear -
we do it all the time.
Of course, there are people who might mispronounce crotch as crutch
and creek and crick. But these are just dialectical pronunciations
rather than mispronunciations.
>
>Me../
>(No professional background in this. Just a bundle of mostly unorganised
>thoughts.)
Good. That means you're right with the rest of us in tackling a very
vague subject.
Wolf Kirchmeir
>I have heard people refer to their 'crotch' as their 'crutch', and people
>saying they were being 'pacific' instead of 'specific'. Perhaps there is a
>tendency to hear a word and assume it's a different meaning of a similar
>word we already know, rather than a completely different word altogether.
I think this is an acute observation.
Bill Trowbridge
general rules that require a lot of exceptions.
In my adult life, some of the most valuable rules I use are based on word
origin. The English language has borrowed and modified words from many
other languages in many different time periods. Often, I somehow (?? See
note 1.) make a determination of the language origin of a word, and that
information gives me cues to the spelling of the word. Short choppy words
are often Germanic in origin, long multisyllabic words are often latin or
greek in origin.
For example:
philosophy ==> latin ==> "f" sound is spelled "ph"
passage ==> french ==> "-age" spelling
dart ==> germanic ==> simple spelling rules
Good knowledge of prefixes and suffixes is also very useful in spelling,
because it helps to decompose the word into smaller spellable parts.
My point is that you may not find such rules in textbooks, but they're
useful, so a good model would not omit them.
Note 1: Determining how would make a good research project of it's own.
"Auditory cues used in determination language origin of British English
words" ... or some such thing.
---- ----- -----
Bill Trowbridge, Information Power Company, Inc.
JustJim
> largely unconsciously -- it is a conditioned skill, IOW. Explicit rules don't
> help much, if only because you have to also explicitly state the exceptions,
> which makes the rules cumbersome. In my experience, voracious readers tend to
> be good spellers (and good writers generally), and people who read little
> tend to be bad spellers - regardless of their overall intellectual abilities.
> Eg, the notoriously bad spelling of engineers and computer gurus.
I would say correct spelling is a habit. As a matter of fact, all of
one's knowledge is habit. A person habitually answers to his name,
spells it correctly and goes to the correct house after school or
work.
Some people can develop habits of one kind or another more easily than
others. Good spellers have the physiological – which includes the
brain – facility to learn the habit of spelling words easily. I think
someone who has dyslexia will always be a bad speller and poor reader
unless they spend an inordinate amount of time practicing both skills.
Bad spellers sometimes make good athletes, carpenter and musician. To
means the part of the brain that takes care of word memory and
spelling is "disjoined" from all the other parts that work well. And
yes
Wolf Kirchmeir
>I would say correct spelling is a habit.
That's what I said.
etrout
>> Spelling is a largely unconscious skill, and I suspect it is a skill
learned
>> largely unconsciously -- it is a conditioned skill, IOW. Explicit rules
don't
>> help much, if only because you have to also explicitly state the
exceptions,
>> which makes the rules cumbersome. In my experience, voracious readers tend
to
>> be good spellers (and good writers generally), and people who read little
>> tend to be bad spellers - regardless of their overall intellectual
abilities.
>> Eg, the notoriously bad spelling of engineers and computer gurus.
>
>I would say correct spelling is a habit. As a matter of fact, all of
>one's knowledge is habit. A person habitually answers to his name,
>spells it correctly and goes to the correct house after school or
>work.
>
>Some people can develop habits of one kind or another more easily than
>brain - facility to learn the habit of spelling words easily. I think
>someone who has dyslexia will always be a bad speller and poor reader
>unless they spend an inordinate amount of time practicing both skills.
>
>Bad spellers sometimes make good athletes, carpenter and musician. To
>means the part of the brain that takes care of word memory and
>spelling is "disjoined" from all the other parts that work well. And
>yes
>
>> IOW, spelling is far more complicated than "understanding the rules."
When I think of habit I think of the conditioning of a behvior. Rote
learning
could be considered habit but active learning uses more of your senses and
connections as well. This is not conditioning because the individual is
making active conscious connections of learning. Spelling doesn't have to
be
a habit, we and our society and school systems just choose to make it that
way.
Glen M. Sizemore
etrout <etr...@temple.edu> wrote in message