@part(intro,root "thesis.mss")



@chapter(Introduction)
@set(page=1)
@pageheading(center <@value[page]>,immediate)
@label(background)

@section(Children with learning difficulties in spelling: population
and problem)
@label(pop-problem)

This thesis is concerned with children who, despite being intelligent,
fail to learn how to spell.

Characteristically, they have spelling ages (as measured by standardized tests)
which are two years or more behind their chronological ages.
They are often not able to spell even common words correctly.
In addition, they may have particular difficulty in specific areas:
in perceiving patterns in words; in identifying rhymes; and in the sequencing 
of letters and numbers.
The written work produced by the children is usually poorer than
would be expected for their general level of ability.
To demonstrate the extent of these difficulties, 
three examples of the writing produced by these children 
are given in figure @ref(pop-example).
@begin(figure)
@begin(quotation)
to day i am waching the tahgfring i ask dad to waching the talhfi
it  is  n on the talhfi at 7,15.  am dad ask turnofr i askto am
dad it is started.  the hammer house horror the hammer house 
horror then i turnofr to waching the news started at 10 , oclock.
(S.S. aged 10)
@end(quotation)
@begin(quotation)
I live in a mansion it was bilt in 1717 many people were kild.  I
get  nightmars  this  is  what  the nightmars is all about. It is
winter and it it verey cold I alwas leave the  window  open.  The
butler opens the door and the man come in and gave the butler got
some monye and shut the door and the man comeing up the stare she
got  to  the  top of the floor so I ran to my room. I was washing
the mag from my room the man was the doctor I was supris  a  gasp
come  out  of  me.  He pulld out a big sharp knif he went into my
mum`s room a squaek come from the floor bord`s and then a  skraem
come  from  the  room  thiar  was a big howl of wind I turned the
crtane whigld and fell on the floor. I was going to scraem but  i
never.  So  i went to my mum`s room i saw blode but not my mum. I
went to my room and i got my case and packed and i ran away and i
never cam back again
(J.M. aged 11)
@end(quotation)
@begin(quotation)
90 years ago in a lonly qorie in  Midleton  as  work  bigan  some
workers  at  the front talkin the then some one said I wonder how
the old full got on with his gohst hunting  they  got  to  bottom
look overthere thier was a body lay cut up one of us said hes 
warin a brass skurt hay hes a roman they did out what abuot my  dad
look  heres  a note join the legon at midnight then the body came
together it said hulo its me dad remember to be here to night
(L.B. aged 11)
@end(quotation)
@caption(Samples of children's writing)
@tag(pop-example)
@end(figure)
Commonly, these children have experienced earlier difficulties in reading,
although some may show normal progress. Their reading difficulties are
overcome more easily than their spelling difficulties.

In the literature they are referred to as children with learning
difficulties in spelling, or with specific spelling disabilities, or more
generally (particularly in cases where there is also a problem with
reading) as 'dyslexic' @cite(ellis81).

The major causes of the child's difficulty are often unspecified,
or unknown.
In some cases earlier language problems could be a contributory
factor to learning difficulties, though causes such as deafness will 
have been ruled out.
Emotional or personality problems have been experienced by some of 
the children: it is difficult to ascertain in these cases whether 
the emotional problems have caused the learning difficulty or vice-versa. 
Behaviour problems may also be evident. 
Part of the child's current lack of ability might be
attributable to 'learnt failure': the child's teacher
initially has a problem trying to convince him that he@foot(
The children with learning difficulties in spelling are predominantly male:
their remedial teachers are predominantly female. In general, the
children will be referred to as 'he' and the teachers as 'she'.)
may now succeed where he has always previously failed.

Traditional teaching has often failed to help these children.
Even specialized remedial teaching has not always been successful.
To help each child it is necessary to ascertain his specific
areas of difficulty.
Observation of the child at work, and collection and analysis of his errors,
can provide the information needed to do this. 
However, acquiring this information is tedious and time-consuming 
for the teacher,
given currently available schemes for classification of spelling errors.
Additionally, this approach assumes regularity in the child's errors.
If these regularities can be shown to exist this information can be used to 
aid the child, both in the planning of remedial teaching and as a basis for
supportive tools.
These tools might also be used to motivate the child. They might take
advantage of the child's abilities and compensate for his disabilities.
There are few such tools designed to be used with this population:
traditional tools such as dictionaries are of little use.
Computer tools are advocated here as a means
of enabling the children to communicate more effectively in writing.
The development of a tool to be used by children with learning
difficulties in spelling is described in this thesis.

@section(Outline of the research)
@label(outline)

Children with spelling disabilities were observed by the author, and their
errors were recorded. 
Based on analysis of these errors, a scheme of error classification
was devised. 
It was hypothesized that there were regularities in the errors.
Two computer programs were developed,
both to test this hypothesis and
as potential components of a larger program to be used as a compensatory
tool.

Initial testing and refinement of the programs were carried out.
The two programs developed and implemented were both spelling correction
programs. The first, the EDITCOST program, offered corrections for
a misspelling by comparing it with potential corrections from a 
dictionary: those words that the misspelling could 
be altered to match most cheaply were offered.
The cost of this matching was determined by the analysis of errors
made by the children. 
Regularities noted in misspellings were used to reconstruct the
intended word.
The second correction program, the PHONCODE program, used
information about the children's phonetic confusions to match
the error to 'phonetically equivalent' corrections.
Again, information about the regularities in the errors (in the
phoneme-grapheme correspondences) was used to reconstruct the error.

Assumptions about the general methods of use of the programs were tested.
The performance of the both correction programs was assessed:
they were tested on data generated by two groups of children
with spelling difficulties.
Additionally they were tested on a corpus of errors
generated by a sample of children of 'normal' spelling ability.

The two correction programs both used on-line dictionaries to
provide the correction: this provided the means of retrieval
of the word and compensated for the child's inability to do so.
The child's problem became a recognition problem rather than
a retrieval one.
Other ways in which the program could be developed to help the
child and his teacher were also considered.

@section(Questions addressed in this thesis)
@label(intro-questions)

@begin(verbatim)
@b(Theoretical Questions:)
@end(verbatim)
The following questions, relating to the children's 
difficulties with spelling, are addressed.
@begin(itemize)
Do the errors made by the children show regularity, or are they random?

If a classification scheme is developed, based on the errors,
can it provide adequate information to enable a computer program
to 'debug' the children's errors and to reconstruct the intended
words?

Are the children able to recognize intended words (correctly spelt)
even if they cannot produce them?
@end(itemize)

@begin(verbatim)
@b(Practical Questions:)
@end(verbatim)
Questions relating to the practicality of a computer program as a tool
are also considered.
@begin(itemize)
If a computer-based tool is provided to help the child:
@begin(itemize)
is he able to use it?

is he willing to use it?
@end(itemize)

How well does the tool succeed in correcting his errors?

Do the answers to these two questions vary according to the 
individual using the tool?
@end(itemize)

@begin(verbatim)
@b(Additional Questions:)
@end(verbatim)
Additional questions, not addressed directly, are also considered 
in this thesis:
@begin(itemize)
Could such a tool provide us with:
@begin(itemize)
information about the phoneme-grapheme correspondences used by
individual children?

other information that might be of use to the remedial teacher?
@end(itemize)

Could the tool be used by other groups of children?
@end(itemize)

@section(Plan of the thesis)
@label(intro-plan)

Chapter @ref(spellit): provides a review of the literature on spelling,
including discussion of the theoretical psychological literature relevant
to this thesis. 

Chapter @ref(complit): provides a review of computer programs
designed to detect and correct spelling errors.

Chapter @ref(theory): describes the classification scheme 
developed in this thesis; the way in which the program will be
used in teaching; the requirements of the computer program
that will incorporate the error classification and be used
by the children.

Chapter @ref(design): presents the overall design of a
computer program, incorporating the spelling correction programs,
and gives examples of the way in which it should be
used.

Chapter @ref(assumptions): in this chapter two pilot studies
are described in which various assumptions about the use of
the program are tested. These studies are also used to collect data
about errors.

Chapter @ref(detail): details of the EDITCOST and 
the PHONCODE programs are described  in this chapter.

Chapter @ref(perform): the performances of the EDITCOST
and PHONCODE programs are evaluated.
The questions posed above are discussed in this chapter.

Chapter @ref(conclude): conclusions drawn from the work in this thesis
are summarized. Further work is proposed.

Conventions for representation of phonemes throughout the thesis
text are given in figures @ref(intro-vowels) and @ref(intro-cons).
@begin(figure)
@begin(verbatim)
                  International Phonetic  Notation
       Examples        Alphabet(IPA)    used in text
      late, day             eI               @b(eI)
      air, care                              @b(eE)
      bat, add                               @b(ae)
      car, aunt                              @b(a:)
      about, silent                           @b(E)
      beat, keep             i                @b(i)
      here, ear             I                @b(IE)
      end, let                                @b(e)
      maker, urn                             @b(e:)
      ice, high             aI               @b(aI)
      ill, bit               I                @b(I)
      boat, know            o                @b(EU)
      port, saw                              @b(o:)
      pot, soft                               @b(O)
      food, rude             u                @b(u)
      foot, book                              @b(U)
      cube, unite           ju               @b(ju)
      up, son                ^                @b(^)
      oil, boy               I               @b(oI)
      out, cow              a                @b(aU)
@end(verbatim)
@caption(Notation for vowel phonemes)
@tag(intro-vowels)
@end(figure)
@begin(figure)
@begin(verbatim)
                  International Phonetic   Notation 
       Examples        Alphabet(IPA)     used in text
      bad, rub               b                @b(b)
      bad, day               d                @b(d)
      fat, rough             f                @b(f)
      go, big                g                @b(g)
      hit, behind            h                @b(h)
      gin, joke             dz               @b(dz)
      keep, cock             k                @b(k)
      loud, kill             l                @b(l)
      mad, jam               m                @b(m)
      man, no                n                @b(n)
      pit, top               p                @b(p)
      run, bread             r                @b(r)
      sit, loss              s                @b(s)
      trap, step             t                @b(t)
      very, love             v                @b(v)
      wash, when             w                @b(w)
      yellow, yet            j                @b(y)
      zoo, beds              z                @b(z)
      chair, lunch          t                @b(ch)
      ethics, accent         -               @b(ks)
      quick, aqua           kw               @b(kw)
      sing, along                            @b(ng)
      sugar, bush                            @b(sh)
      theatre, thank                         @b(th)
      that, with                             @b(tv)
      garage,pleasure                        @b(zh)
@end(verbatim)
@caption(Notation for consonant phonemes)
@tag(intro-cons)
@end(figure)