A comparison of the effects of self-evaluation and peer evaluation on the revising behaviors and overall composing proce...

MISSING IMAGE

Material Information

Title:
A comparison of the effects of self-evaluation and peer evaluation on the revising behaviors and overall composing processes of college freshmen
Physical Description:
xii, 218 leaves : ; 28 cm.
Language:
English
Creator:
Olendzenski, Michael Felix, 1948-
Publication Date:

Subjects

Subjects / Keywords:
English language -- Composition and exercises -- Evaluation   ( lcsh )
Editing   ( lcsh )
College freshmen -- Florida -- Gainesville   ( lcsh )
Curriculum and Instruction thesis Ph. D
Dissertations, Academic -- Curriculum and instruction -- UF
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1991.
Bibliography:
Includes bibliographical references (leaves 210-217).
Statement of Responsibility:
by Michael Felix Olendzenski.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001678747
oclc - 24915710
notis - AHZ0659
System ID:
AA00002104:00001

Full Text














A COMPARISON OF THE EFFECTS OF
EVALUATION ON THE REVISING
COMPOSING PROCESSES OF


SELF-EVALUATION AND PEER
BEHAVIORS AND OVERALL
COLLEGE FRESHMEN


MICHAEL


FELIX


OLENDZENSKI


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY


























Copyright


1991


Michael


Felix


Olendzenski



















This


dissertation


dedicated


Felix


Irene


. Olendzenski


Olendzenski


gave


a lifetime


life

of love.
















ACKNOWLEDGMENTS


author


wishes


to acknowledge


debt


owes


to the


members


of his


doctoral


committee,


helped


him


complete


this


project.


First,


thanks


go to Dr. Margaret


Early


infinite


patience


steadying


influence.


Next,


. Paul


Jensen


offered


invaluable


advice


and


counseling


from


beginning


addition,


to the


Rodman


of the


Webb


author


helped


s doctoral


author


program.


understand


nature


of qualitative


research


provided


sympathetic


listening


when


was


most


needed.


Also,


Dr. Linda


Crocker


provided


essential


interpretation


instruction


of quantitative


implementation


research


methods.


and


Next,


Robert


Wright


shared


with


author


indispensable


insights


on the


nature


of composition


and


problem


of teaching


he also


gave


author


gift


of laughter


many


times


when


it was

former


sorely


needed.


committee


To these


members


names m

Theodore


rust


be added


Hipple


those

Dr.


Ruthellen


Crews.


They


each


provided


author


with


valuable


guidance


during


their


tenure


on his


committee.


Finally,


people


were


involved


in the


generating


of the


data

















TABLE


OF CONTENTS


ACKNOWLEDGMENTS . . . ... ....................


LIST


LIST


OF FIGURES................... ................... ....


ABSTRACT................................................

CHAPTERS


INTRODUCTION


TO THE


STUDY .....................


Hypot
Signi
Limit
Defin
Organ


ses T
cancer
ions
ions
action


ed i
the
the
Term
the


the
Study
tudy.


Study.



rtatio0


S...... ... .
n .... 0.. 0....0...


REVIEW


OF RELATED


Developing


Revisi
St
Audien
Other
Self-E
Co
Peer E


on
ud
ce
Re
va
mp
va


in
ents
Awa
visi
luat
osit
luat


LITERATURE


Model


of


Writing


Revis
of Co


. .. .. .. . 9


.sition




of

of


Revision

Writing
S ....0.
Writing


Composit


Student


THREE


METHODOLOGY


S S S S S S S S S S S 5 *** *~ S S S S S S S S S


election
ssignmen
assroom
say and
script


of Instructor


t of
Pro
Rev
on o
nf


I|I J[ II I i I n I


S.. ... .
to Treatment


de Data....
Treatment s
l*t f l 4 t -


.000.. 0. .
...... .. .


S 5
*. S ..S .S .* .S
.... .........55


Pace


OF TABLES .


"









Treatment
Revision
Interview
Reportin


of Qualitative D
Guides.........
w Data..........
g of Qualitative


ata. .


Data


.* 5 t t .
* .. .


FOUR


QUANTITATIVE


FINDINGS . .. ................


Analyses of Ho
Inter-Rater
Descriptive
Three-Factor
Analyses of Re
Descriptive
ANOVA Tests


listic S
Reliabil
Means an
ANOVA R
visions.
Means an
of Signi


cores


ity .......
d Standard
esults....

d Standard
ficance...


Deviation


* S S
. S


Deviation


FIVE


QUALITATIVE


FINDINGS ...... ................ ..


Specific


Revisions


Responses.....
Instructor One..
Instructor Two..
Student Responses
Students' Self-R
Student Percepti
Student Percepti
Improvement a
Summary.........


and

* ..


Revi


sion


........
.0. 00....


Writ


erv
of
Re
Th
ers


Guide


.....0...
*. ..... .* ..


iew Questions.....
Revising Behavior
vision Guides.....
eir Own


a


* SS S*S~t S*S 5 164


CONCLUSIONS


AND


RECOMMENDATIONS.....


Related
Related


Quantitative
Qualitative


Data
Data.


Teaching
Research


APPENDICES


HOLISTIC


QUALITY


FIRST-TO-FINAL


DRAFT


CHANGE S


IN QUALITY


SCORE..


REVISIONS


MADE


IN FIRST


DRAFTS ........


REVISION GUIDES
THE STUDY....


USED
S*....


IN CONJUNCTION


WITH


REFERENCES CES ..


SCORES....

















LIST


OF TABLES


Table


Response
Study..


Student


Distribution

Inter-Rater


to Request


of Males


Reliability


Product-Moment


to Participate
.... .. ...... 59


Females ............


Coefficients


Correlations)


(Pearson


Raters


Reading


Inter-Rater


Essays


Instructor


Reliability


Product-Moment


Coefficients


Correlation


s Students.


(Pearson


Raters


Reading


Essays


Instructor


Two'


Students.


Mean


Holi


stic


(Instructor


cores


One),


of Sample
Scale: 1-1


Essays
Point


s ........


Mean


Holi


stic


(Instructor


Scores


Two),


of Sample


Scale


: 1-18


Essays
Point


.... ....


Three-Factor
Comparing


ANOVA
Holisti


Test


of Significance


Scores


of Self-


Evaluating


(Instructor


and


Peer


One) .


-Evaluating


Students


5 05 55 555 5C SS SC S* *587


Three-Factor
Comparing
Evaluating


(Instructor


ANOVA


Holistic


and


Two)


Test


cores


Significance
of Self-


Peer-Evaluating


Students


S SS SSS SC SSS SS SSS SS S589


Means


per 100
Levels


Standard


Words


Deviations


of First-Draft


of Textual


Complexity


Revi


Text


sions


at Six


(Instructor


One)


..... .. ...... ... ....... ............ 91


Means


per 100
Level s


Standard


Words


nf Texty1 l


Deviations


of First-Draft


I' amrnn 1. 4~ 4...CI


Revi


Text


sions


at Six


I11 *I- a-4.--.a4


Race


I 7rr nC m.r IIC A Y









ANOVA


Tests


Textual


of Significance


Complexity


at Six


(Instructor


One) ....


S*.... 97


ANOVA


Tests


of Significance


at Si


Two)....


Textual


Complexity


structor


... a99


Student
You D


Responses
o When You


(Instructor


to the
Revise


Question,


Your


First


"What


Draft?"


One)


Student


Respon


Do When


(Instructor


ses


to the
Revise


Question,


Your


First


Two)


"What


Draft?"


Student
Effect
on the


Responses
Did the
Way that


(Instructor


to the
Revision


Wro


Question,
Guide Ac
te Your I


One).


:tivity
east Es


Have
ay?"


Student
Effect


on the


Responses


Way


(Instructor


to the
Revision


that


Two).


Question,
Guide Ac


Wrote


Your


:tivity


Last


ssay?"


.. a...... .....a.a s .....Q


.. 144


Student R
Anythin
Essays


Have


esponses
g Changed
as a Resu


Experienced


(Instructor


to the


Question,
Way that


of the


"Has


Activities


1101


This


Write
You


Year?"


One).


Student Responses
Anything Changed


Essay


to the


as a Result


Question,
Way that


"Has


Write


Activities


Have


Experienced


(Instructor


1101


This


Two)


Year


First


Draft


Scores


--Instructor


(Scale


-18)


Final


Draft


cores


--Instructor


(Scale


-18)


First


Draft


cores--


Instructor


Scale


-18)


Final


Draft


Scale


cores


--Instructor


-18)


Levels


Level


"Uhat


"What


Have










First-to-Final


Draft


Changes


Quality


Score,


Instructor


Revisions


One,


Revisions


Made


Instructor


Made


First
One..


In First


Drafts


Drafts


--Sample


--Sample


Essay


Essay


Two,


Instructor


One..


Revi


sions


Made


First


Drafts


--Sample


Essay


Three,


Instructor


*One


Revi


sions


Made


First


Drafts


--Sample


Essay


Four,


Instructor


One.


Revision


One,


Revisions


Two,


Revisions


Three,


Made


Instructor


Made


Instructor


Made


First
Two..


First
Two..


First


Drafts


Drafts


Draft


--Sample


--Sample


--Sample


Essay


Essay


Essay


Instructor


Revisions


Four,


Made


Instructor


First
Two.


Drafts


--Sample


Essay

















LIST


OF FIGURES


Figure


Pagea


Procedure


followed


students


as they


completed


their


ass


signed


essay


* .... 63


Assignments


essays


collected


as data


S.... 64


Quantitative


Revision


research


coding


design.


system......


* S S * S S 50555

* S S S S *s*~5**~* a.


Interview


questions..........


. ..... 79















Abstract


of Dissertation


Presented


to the


Graduate


School


of the


University


of Florida


Requirements


Degree


in Partial F
ee of Doctor


Fulfillment of the
of Philosophy


A COMPARISON


EVALUATION


OF THE


ON THE


EFFECTS


REVISING


OF SELF-EVALUATION


BEHAVIORS


PEER


OVERALL


COMPOSING


PROCESSES


OF COLLEGE


FRESHMEN


Michael


Felix


Olendzenski


May


1991


Chairman


Margaret


Early


Major


Department


: Instruction


and


Curriculum


Two

classes


instructors


University


students


of Florida


freshman


participated


composition


this


study.


engaged


each


in self


instructor,


-evaluation


randomly


of first-draft


selected


essays,


students


responding


to written


Another


rev


group


vision

of 10


guides c

students


containing prompts

exchanged essays


revising.

responded


revision


guides


each


other


s essays.


Holistic


scorers


rated


quality


of all


first


and


final


drafts.


revi


sons


were


identified,


coded,


counted,


and


tabulated


researcher.


Sixteen


of the


participating


students


were


interviewed


regarding


their


perceptions


of the


course.


Findings


included


a significant


difference


between


first









revising.


No statistical


difference


mean


holistic


scores


between


treatment


group


was


found.


Also,


no significant


difference


was


found


effect


of topic


on changes


holistic


scores


from


first


to final


drafts.


Interviews


suggested


an inability


on the


part


of students


to reconstruct


accurately

pleasing i


composing


nst ructors


behavior


than


and


learning


a greater


how


concern


to write


with


well.

















CHAPTER


INTRODUCTION


TO THE


STUDY


students


progress


through


school,


they


too


seldom


become


aware


of the


benefits


of revision


in compositional


writing,


partly


because


fewer


than


one


quarter


of public


school


students


are


asked


to revise


papers


they


hand


a regular


part


of writing


instruction


(National


Assessment


of Educational


Progress,


1986) .


many


students,


writing


remains


a secretive


process


which


they


do their


writing


private


evaluates


and


also


submit


private.


teacher


This


traditional


reads


pattern


teaching


kind


of writing


of audience


does


awareness


little


that


to help


professional


students


writers


gain


use


turn


unsucces


sful


first


drafts


into


successful


final


drafts.


Composition


instructors


have


responsibility


helping


students


to become


sensitive


to the


need


revision


to learn


to revise


effectively.


In their


efforts


meet


this


responsibility,


writing


teachers


may


see


k guidance


in research

scholarly


literature.


literature


As shown


related


Chapter


to writing


Two,


instruction


contains










sources


of this


feedback


are


teacher,


students


peers,


students


themselves.


study


described


in this


dissertation


focused


attention

evaluation


on self-provided


as sources


evaluation


feedback


peer-provided


first-draft


essays.


These


sources


of feedback


have


been


isolated


so that


their


effectiveness


prompting


needed


revi


sions


may


compared.


practical


question


addressed


in this


study


was,


students


evaluate


each


other


s first


drafts


as a


regular


feature


of instruction


engage


different


revision


behaviors


produce


better


final


drafts


than


students


evaluate


their


own


first


drafts?"


answer


this


question,


researcher


created


a series


"revi


sion


guides,


" based


on models


(Beach,


1979;


Flanigan


Menende z,


1980)


described


research


literature.


These


revi


sion


guides


were


introduced


to students


as a normal


component


of classroom


instruction.


Half


of the


students


parti


cipating


evaluate


their


in th

own


study


first


used


these


drafts


revision


essays


guides


written


freshman


composition


to evaluate


be described


first


course.


drafts


Chapters


Four


other


of their


and


half


peers


Five,


used


essays.


essays


guides


As will


of self-


evaluating


peer-evaluating


groups


were


compared


using


bohnth


anant i t1-t I v


rra n i t iFMUx


I-~~p *~ I -*


1-n ring-a rrn4 na


"il I I I


T ij l










Hypotheses


Tested


in the


Study


compare


results


of the


revision


based


on self-


evaluation


revis


based


on peer


evaluation,


following


primary


null


hypotheses


were


tested:


There
score


will


between


averaged


no difference


first


over


drafts


mean


and


treatment


second
group


holisti


drafts


four


topi


There


will


treatment


change
final


no difference


groups
holisti


draft;


treatment


terms


score
there


between


of the


from
will


interaction.


degree


first


draft


no draft-by-


There


will


holistic
different


no difference


scores
topics


achieved


addr


essays.


ess


among


on each


ed in


mean


of the


four


students'


For
will


each
be


groups i
students


of six


levels


no difference


terms


of the


' essays.


texual


between
amount


complexity,


there


treatment


of revision


In addition


to these


hypotheses,


following


research


questions


were


investigated:


will


specific re
first-draft


What


nature


visions
essays?


will


specific
students


written
to the


made


nature of
responses
revision g


study?


of a sample


these


a sample
made by


uides


students


of
these


employed


their


will


their


responses


a sample


own


to interview


of these


revising


students


questions,
character


how


behavior


CS.


What










In t
will


their


heir


respon


a sample


own


course


ses


to interview


of these


improvement
the study?


students


as writers


questions,


how


characterize


over


study


was


replicated


using


instructor


thus,


each


hypotheses


questions


was


investigated


independent


samples


of students.


Si anificance


of the


Study


Teachers


of writing


make


a major


decision


when


they


include


part


or exclude


of instructional


self-evaluation


self-evaluation


methodology.


been


recognized


or peer


Although


several


evaluation


need


researchers


(Beach,


1989;


Beaven,


1977;


Flanigan


Menendez,


1980),


there


is disagreement


about


merits


peer


evaluation.


group


of researchers


(Bruffee,


1984;


Elbow,


1981;


Lagana,


1972;


Moffett,


1968;


Nystrand,


1986,


1989;


Nystrand


Brant,


1989)


claim


that


peer


evaluation


an appropriate


salutary


According


method


to these


student-writer


of promoting


researchers,


an audience


with


learning


peer


other


about


writing.


evaluation


than


provides


instructor.


Moreover,


peer


evaluation


allows


writer


see


or her


work


from


another


person


s perspective


before


handing


a paper


in to the


instructor.


Furthermore,


peer


evaluation


allows


writer


see


how


other


students


write,


thereby


providing










Ziv,


1983)


question


value


peer


evaluation.


They


point


that


peers


know


enough


about


writing


to offer


real


help


to their


classmates.


Peer


evaluation,


they


say,


a cas

with


of the


strong


blind


leading


personalities


blind.


talk


fello


Secondly,

w students


students

into


taking


bad


advice


or rej


good


advice


offered


less


assertive


students.


Furthermore,


these


researchers


have


found


that


students


are


reluctant


to criticize


each


other


writing,


because


doing


so might


interpreted


insulting


have


negative


interpersonal


consequences.


Finally,


peer-


evaluation


students


activities


to waste


have


been


time


characterized


engaging


invitations


in irrelevant


Soc


chatter.


This


study


was


an attempt


to document


effects


self


-evaluation


and


peer


evaluation


under


controlled


conditions.


evaluation


isolating


as treatment


self-evaluation


variables


and


comparing


peer


effects


they


have


on revising


behavior


and


ess


quality,


researcher


hoped


to develop


evidence


which


help


clarify


some


of the


points


of disagreement


regarding


these


instructional


techniques.


Limitations


of the


St udv


I ** ~ O trar~1 t1 n Lt'i -n.- t


Qtiit'it


lanaQ


*1 mtril-n


1


I I I -


1 n


W-T T










typical


college


freshman


writing


than


would


have


been


case


example,


students


enrolled


in freshman


composition


University


of Florida


a given


semester


been


studied


detail.


Secondly,


preserve


naturalness


made


of the


no attempt


four


classes


environment


to directly


being


observe


involved


studied,

or record


study.


Data


researcher


dynamics


directly


bearing


on the


findings


this


study


may


have


been


missed


because


of this


limitation.


Finally,


fact


that


control


groups


were


included


design


of this


study


limited


range


conc


lusions


which


could


be drawn


from


data.


design


of the


study


allowed


only


conclusions


comparing


each


of the


treatments


against


other.


Definition


of Terms


following


terms


used


this


study


should


understood


Revision


light



guides


of the



.are r


following


definitions


esearcher-prepared


written


questions


and


directions


designed


to guide


students


they


cons


idered


their


first


drafts


before


making


needed


revisions.


A different


revi


sion


guide


was


designed


each


essay


written


students.


Sel f-Pva 1 it- -i nn


-I nrn/r'orli ra -n th -l 4 nj n-t.--.-


rC(r~nnC rr


ok: nk


HAC~


Y rl


1 n


r'










Peer


evaluation


a procedure


which


pairs


students


face-to-face


during


class


time


to exchange


written


papers


responses


to the


to discuss


each


revision


other


guides


s first


each


drafts.


other


These


discussions


were


based


on but


limited


to the


students'


written


revision


guide


responses.


Revision s


are


differences


between


first


final


drafts


essays


discernible


drafts


through


, occurring


a line-by-line


at six


levels


comparison


of textual


complexity,


following


Bridwell


1980)


surface,


word,


phrase,


clause,


sentence,


multisentence.


A first


draft


first


typed


draft


submitted


class,


after


which


students


responded


to revi


sion


guides.


These


drafts


or may


have


been


preceded


preliminary


drafts


written


students


before


first


draft


was


due.


A final


draft


draft


an essay


submitted


a final


grade.


some


cases,


students


wrote


additional


drafts


between


first


final


drafts


submitted


class.


Organization


of the


Dissertation


Chapter


this


dissertation


contains


introduction


to the


subject


of the


study


includes


t in (n+1 00 a a a -


a. a C2 a a a


ptll~tl


raaaar nk


Frr rhnC: rr nrr


v










study


presented


Chapter


Three.


Quantitative


findings


are


given


Chapter


Four.


Qualitative


findings


are


given


Chapter


Five.


A general


discussion


of findings


with


conclusions


recommendations


presented


in Chapter


Six.

















CHAPTER


REVIEW


OF RELATED


LITERATURE


This


nature


chapter


a review


revision,


of scholarly


self-evaluation,


and


writing


peer


related


evaluation


composing


process.


In addition,


scholarly


writing


which


reveals


effects


that


revi


sion,


self-evaluation,


peer


evaluation


have


on the


quality


of student


writing


reviewed


this


chapter.


first


section


of this


chapter


contains


of revision


section


a description


reflected

a survey o


of the


in the


f scholarly


developing


research


writing


theoretical


literature.


regarding


models


second

nature


of revision


writing


of composition


students.


third


section


a discussion


of self-evaluation


effects


section


about


on the


of thi


peer


quality


chapter


of student


a review


evaluation


effects


writing.


fourth


of scholarly


on the


writing


quality


student


writing.


Developing


Model 1


of Revision


To be meaningful,


discussion


of how


concept


I *


I


t I r


I t


I t










writing


process.


Fitzgerald


(1987


offered


following


synopsis


of the


dynamic


forces


which


combined


to propel


writing


research


forward:


1970s,


dramatically
(Scardamalia
educational


about
the s


writ
tudy


rapidly
point o
1970s,
field.


several
affect t


Bereiter,


community


ing skills
of psycho-


expanding.
occurred in
experiment


researchers


(Humes,
as case


factors


study
1986)


were


converged
of writing


. The


public


increasingly


cognitive


linguistic


Also a m
writing re
1 research


quantitative


to the


1983) .
studies


ever-growing
engendered a
of writing.


study


Marrying


interest


conc


erned


psychology


processes


ethodological


search. P
dominated


emphasis


of written


other


naturalistic i
t in cognition


allowed


more


rese


rio
th


tended


were
turning
r to the
e writing
to bind


products


methodologies s


inquiry) w
probably


arch


on the


ith


such
the


both
process


. 482)


State-of-the-art


thinking


about


writing


before


emergence


of the


kind


of research


activity


Fitzgerald


was


referring


was


represented


Rohman


Wlecke


1964)


three-part


stage


model


composing


process.


This


model


defined


writing


as a linear


process


including


prewriting,


writing,


rewriting.


beginning


of the


assault


on this


view


of the


composing


process


was


Emig


s The


CompDosing


Processes


Twelfth


Graders


(1971).


Emig


found


that


discrete-stage


model


of the


composing


process


was


incongruent


with


data


collected


from


professional


writers.


In Emig


s words,


Clearly, fo
of writincr


these


are


AA- L


authors,


fixed


I 11


so-called


an i nPYnrashl


"stages"


In the










From


observation


of 8


high


school


students


as they


composed


aloud,


Emig


was


able


to construct


following


six-


part


model


of the


writing


process


stimulus,


prewriting


planning,


moment


of starting,


reformulation


(her


term


revision),


moment


of stopping,


contemplation


product.


Although


Emig


distinctly


introduced


concept


of recursion


when


describing


what


had


learned


from


study


of professional


writers,


characterized


only


element


of planning


as recurs


in her


model


of the


writing


process


Emig


divided


revis


into


three


activities--


correcting,


revi


sing,


rewriting--but


discuss


revision

research


may

into


operate

writing


recursively

proceeded,


in the


writing


concept


process.


of recursiveness


writing


process


was


more


fully


articulated.


Using


Emig


s composing-aloud


methodology,


Flower


Hayes


(1980)


took


what


Emig


characterize


as essentially


an external,


making


recurs


mostly


inferences


ive,


linear,


about


unobservable


observable


nature


process


of the


cognitive


and


internal,


processes


began


fully


enable


which


human


beings


to produce


written


texts.


Borrowing


from


information-processing


theory,


Flower


Hayes


their


associates


developed


a convin


cing


model


of how


brain


operate


discourse


goes


The main


about


work


coanitive


of creating


n rnrsss s p


i i il


written


i


Fr 1 w M


L


g..










process


characterized


essentially


these


linear


subprocesses


earlier


as recursive,


models


of Emig


1971)


or Rohman


and


Wlecke


(1964)


Flower


and


Hayes


accounted


recursiveness


of the


writing


process


introducing


what


they


called


"monitor"


function.


This


function


thought


as a source


of executive


control


ove r


several


whole


writing


writing


process.


researchers


have


As will


explored


be shown


this


below,


function


found


to be


an indispensable


feature


writing


process


of successful


writers.


Revi


sion,


according


to the


Flower


Hayes


model


composing,


a result


of a mismatch,


perceived


reviewing


process


posited


in the


mode l,


between


existing


text


produced


writer


goals


established


writer


text.


mismatch,


editing


When


function


writer


may


perceives


activated,


such


following


which


text


changed


to bring


into


line


with


established


goals,


or the


goals


are


changed


to bring


them


into


line


with


existing


text.


Nold


(1982)


developed


a model


of composing


that


complemented


Flower


Hayes


model


delineating


kinds


review


concerns

ed and e


monitor


evaluated.


function


Nold


divide


attends

d the "


to when


complex


text

task"


of writincr


into


subtasks


motor.


aranhical


L..


.


L V


P


YI rL~









forming


of letters


with


a pen,


typewriter,


word


processor,


some


other


writing


implement.


graphical


subtask


involved


attention


usage


written


paid


subtask


language


of complete


lexical


individual


to such


concerns


included


as opposed


sentences


subtask


words


as spelling


attending


to spoken


express


corresponded


express


to the


thought


to the


thought


punctuation.


requirements


language,


in written


writer


use


language.


s choice


syntactic


subtask


was


construction


of sentences,


including


cons iderat ions


of how


when


to embed


subsentence


structures


within


sentences.


Daiute


(1981)


observed


that


this


subtask


sometimes


made


more


difficult


natural


tendency


of the


mind


Nold


to translate


s formal


syntactic


subtask


structures


required


into


writer


semantic


to make


ones.


choices


regarding


paragraphing,


sectioning,


or arrangement


of the


text


on paper,


including


choice


of appropriate


transitional


expression


cohe


sive


devi


ces


(Halliday


Hasan,


1976).


generic


subtask


was


related


to the


application


of knowledge


about


requirements


of particular


forms

text


of writing


being


(e.g


., newspaper


written.


articles,


rhetorical


lim


subtask


ericks) as new

included


considerations


of the


needs


of the


audience.


Flower


(1979)


pointed


out


that


writers


may


fail


carry


this


subtask


successfully


because


their


rhetorical


stands


. liii










inclusion


or exclusion


of information


text;


in other


words,


decisions


vis-a-vis


about


emerging


relevance


text.


of certain


Finally,


purpose


information


subtask


required


writer


to formulate


a reason


initiating


of writing


in the


first


place.


Nold


s model


of writing


important


because


adds


entire


dimension


represented,


which


in the


only


Flower


implied,


Hayes


explicitly


model.


Although


Flower


and


Hayes


model


a great


improvement


over


those


which


preceded


one


still


to believe


Flower


Hayes


model


that


writing


a simpler


process


than


really


In the


Flower


Hayes


model,


events


happen


linear


sequence.


Planning


followed


translating,


which


followed


reviewing,


although


each


be interrupted


either


other


two.


Nold


s model


nicely


complements


Flower


and


Hayes


model 1


one


superimposes


Nold


model


on the


Flower


Hayes


model,


then


writing


becomes


sequential,


linear


cognitive


process


which


one


thing


happens a

gridlocks


t a time,


log-jams


a more


are


precarious


constant


sources


process


in which


of potential


breakdowns.


hierarchical


subtasks


simultaneously


compete


writer


s limited


attention


every


turn


from


planning,


through


translating,


and


on to reviewing.


This


more


complex


reDresentation


of the


ritin


nrones. s


1irni no


n


%


||


W


11










Elbow


(1987)


questioned


Flower


s categorical


admonitions


against


egocentric,


"writer-based"


prose,


and


contention


that


prose.


writers


Elbow


should


argued


strive


that


to produce


much


only


attention


"reader-based"


to audience


needs


can


stymie


a writer


s efforts


to create


new


meanings.


In place


of Flower


s conception


"writer-based"


"reader-


based"


writing


as necessarily


"bad"


and


"good"


writing,


respectively,


Elbow


posited


need


writers


to dismiss


cons


iderations


of audience


during


earlier


stages


writing,


order


more


attention


to their


own


thinking


about


their


subjects.


According


to Elbow,


"reader-based"


prose


call


ed for


Flower


best


attained


only


after


"writer-based"


prose


brought


through


revision,


at which


time


writer


s attention


shifts


to audience


needs.


Nold


(1982)


asserted


that


there


are


"two


kinds


revising:


revising


intentions"


18) .


to fit


conventions


Convention-based


revising


revi


sing


to fit


often


referred


as proofreading.


Murray


1982)


called


this


type


of revising


"external";


"surface


Faigley


Intention-ba


Witte


revi


(1981)


sing


referred

more


complex,


since


writ


ers


must


evaluate


their


own


writing


against


accepted


rules


norms


against


their


own


sense


of what


they


are


trying


to accomplish


with


their


writing.


Murray


(198


rnf r o nr n4- n.


cnal Pd


this


t ttnrs


II : nt am ~ 1 11 I


1


L .










Flower


and


Hayes


produces


an improved


representation


this


aspect


of writing.


Flower


Hayes


have


constructed


plausible


working


model


of how


mind


operates


as text


being


produced,


Nold


contributed


a convincing


taxonomy


of the


numerous


demands


that


act


of writing


makes


mind.


Other


researchers


constructed


models


of the


writing


process


which


share


important


features


common


with


Flower-Hayes


and


Nold


models.


Britton


et al.


1975)


gave


names


"conception,


incubation,


and


production"


what


they


considered


to be


most


important


subproces


ses


writing.


Such


a conception


emphasized


what


Flower


Hayes


referred


as planning


adequately


account


revision


as part


of the


writing


process.


Murray


(1978)


gave


names


processes


"previ


sion,


Flower


vision,


Hayes


revision"


called


planning,


to the


translating,


reviewing.


Scardamalia,


Bereiter,


Goelman


(198


used


verbs


"compare,


diagnose,


operate"


. 67)


to label


subprocesses


they


inferred


to be at work


within


process


Flower


name

has


and


the

calle


Hayes


planning

d for co


called


process


ncerted


reviewing.


of Flower

research i


Witte


and


nto


(1987)


Hayes

this i


chose


"pretext"


important


elusive


component


writing


process.


Addinar


t-n thp>.s


mnri1 q


nf hnto


SI,*,,j


Y~n~n "F


iinr- flflO c


rrlir4 nfl










based


on psycholinguistic


theory,


especially


work


Kahneman


(1973,


cited


Reed,


1982)


on the


allocation


limited


capacity


cognitive


capacity.


of short-term


memory


According


places


to Daiute,


considerable


limited


constraints


on a writer


as he


or she


generates


ideas,


formulates


propositions,


accesses


lexical


items,


plans


clauses


sentences,


translates


from


semantic


to orthographic


representations,


anticipates


forthcoming


units


of text


(Daiute,


1981,


. 9).


Daiute


showed


that


many


types


errors


written


texts


are


explainable


terms


overextended


overflow


short-term


of information.


memory


attempting


Similarly,


to juggle


de Beaugrande


(1984)


characterized


this


predicament


of a limited


short-term


memory


trying


cope


with


multifarious


demands


text


production

elements o


as a case


of simultaneous


f discourse-creation


being


hierarchical


forced


through


inadequate


funnel


into


linear


and


time-ordered


orthographic


representations.


Flower


taboo,


against


(1988)


called


established


considering


for

New


a writer' s


a breaking

Critics in


intentions


of the


literary


when


.ongstanding

studies,


analyzing


a text.


argued,


"Writing,


as a rhetorical


act,


carried


research


within


into


a web


cognitive


of purpose"


processes


537).


of wri tars


ongoing


rPv a 1 sei


[


w --


c- I~










Writers


at work


textbooks


purposes
They may
negative
for what
out how
content


decide


advise;


They set
respond to
evaluation,


they
to do
and l


provisional,


achieve,
ditch or
("Do I i


have
it.


eap


they


goals,


one


which


to do
They


back


tentative


working


even
rant .


a workable
loggerheads
they don't
dimensional


radiates
unknown.


plan
with
want


on "their


create


toss
those


can


only


think


into


goa


idea


hypotheses


forget.
S ?"),


They


that
worr


confusion


or cherished


other
to thro


network


in all


dire


goals c
w away.


goal


a web


purpose"
of


up possibilities.


ideas


lead


with


new


they


through pc
1 setting,


what t
they
y over
s, and


could


criteria


figure


)ssible
generating


hey
may


want
soon


questions
conflicts


seems


ir a piece


They


of information--a


actions,


anchored


create
web


text
e a multi-
that


to points


. 531)


chief


compose-aloud


differences


protocols


between


of expert


networks


and


apparent


inexperienced


writers


collected


Flower


were


that


those


developed


expert


writers


were


more


elaborate


flexible,


allowing


reorganization


of the


whole


purpose


of attaining


a new


goal


established


to replace


unworkable


or preliminary


goals.


Inexperienced


writers,


contrast,


rarely


abandoned


initial


goals,


worked


from


more


skeletal


networks,


often


resulting


Murray


writing


1978)


which


suggest


seemed

d that


forced

with s


or di


sjointed.


efficient


practice


writer


can


automatize


many


of the


subtasks


involved


writing


after


which


"some


pieces


of writing


come


easily"


92) .


To help


students


achieve


control


over


their


own


writing


processes,


to experience


automaticity


with


which










what


task


of writing


entails


which


emerged


over


last


years


will


help


writing


instructors


meet


that


challenge.


next


section


this


literature


review


contains


a discussion


of scholarly


writing


regarding


how


revision


operates


practice.


Revision


in the Writ I nP flnmnn~ 11- no S?1-iitinnl- a
- -~ .~ .~ p~ L.. a


In a 1959


study


cited


Braddock,


Lloyd-Jones,


and


Schoer


(1963),


Buxton


compared


effects


of three


experimental


freshmen


treatment


conditions


University


on the


of Alberta.


writing

In the


ability


control


condition,


students


merely


completed


courses


comprising


teacher-preparation


program


lasting


months.


first


treatment


condition,


instructional


program


students

wrote w


enrolled


weekly


in the


500-word


same


essays


addition

minimal


to their

guidance


normal c

regarding


oursework.


topic


These


selection


students

or how to


received


develop


their


essays


and


received


only


positive,


encouraging


feedback


from


instructors.


No revision


of their


essays


was


required.


second


treatment


condition,


students


were


expected


write


on assigned


topics


"and


to include


some


critical


thinking,


a central


idea,


material


that


was


organized


developed"


Braddock,


Lloyd-Jones,


Schoer,


61).


e!S S a S


nf the


sasrnnd


t" ra+ m nont-


ft WCV11 V


tot r0


^ar h A a A a


n j~' n n I


in the


Wr~tinn


Rf rnmnn9 i t i nn


St rr rl Cr lit C!










students,


were


then


given


one


50-minute


class


period


"correct


Using


errors


indicated


an evaluation


form


their


which


papers"


assigned


61).


points


quality


of features,


such


as title,


diction,


logical


sequencing,


raters


effectiveness


assigned


scores


of the


to pretest


conclu


posttest


ion,


essays


two

written


subjects.


These


raters


achieved


and


reliability


coefficients


on their


pretest


and


posttest


scoring,


respectively.


significantly


Buxton


outperformed


both


found

the


that


revision


writing


group


control


groups


terms


of overall


essay


quality


several


of the


subcategories


scoring


instrument.


basis


Buxton


s results,


Braddock,


Lloyd


-Jones,


Schoer


(1963)


drew


following


conclusion:


College


thoroughly


their
their
writing
grades
their


relative
factors.


freshmen


whose


marked


papers
writing


in light


more


receives


inten


papers.


tha


a few


sive
(It


influence


Unfortunately,


as the


writing i
criticized


of these
n college
general


marking


clear,
each of


graded


revi


matters


can


freshmen w
suggestions
who do not


however,


these


parenthetical


improve


hose


and


revise


what


three


statement


points


out,


Buxton


did


control


variables


sufficiently


to support


inferences


about


effects


of revision


alone


on the


quality


of writing.


However,


since


revi


sion


was


1 tnnnrI-nni- ~%flfl~lflfl 0n4- 4-h a t. nfl *4 P4 nn- it nC t1 a. -


above


; rnn ~ rt s n t


r~mnhnant~


nF Ck


CI*CA~ II 1 Cl


mllmA


n










regarded


as providing


experimental


support


including


revision


as part


writing


instruction.


A follow-up


to the


Buxton


study


was


carried


Effros


1973)


using


intact


sections


freshman


composition.


Each


of five


instructors


taught


one


experimental


section,


which


included


guided


revi


sions


students


' essays


with


grades


delayed


until


revi


sion


was


completed,


incidental


one


revi


control


sion


section,


immediate


which


grades


included


after


only


first


drafts


were


completed.


Effros


used


two


instruments


measure


differences


between


treatments.


first


was


Engli


sh Expression


Test


(Educational


Testing


Service, 1960


other


was


essay-rating


scale


developed


Buxton.


Buxton


scale


was


used


raters


to evaluate


pretest


and


posttest


essays


written


in class


students


participating


study.


Effros'


English


results


Expression


Test,


not

the


support

control


Buxton


sections


On the


significantly


outperformed


experimental


sections.


pre-post


essay


scoring


produced


no significant


differences


between


treatments.


However,


these


results


were


called


into


question


Effros


on the


basis


of strong


interactive


effects


in her


statistical


analysis,


especially


between


instructor


t- rC'a+-t no -


I~ffrr\a


la7~


FCFrho


*Inn


ylnn A ~IC A ~


CClclL


n










teaching


or different


predispositions


of the


classes


difficult


to surmise


from


results"


20) .


Moreover,


Effros


noted


large


different


ces


pretest


scores


from


section


to section,


a condition


which


also


confounded


findings.


Finally,


Effros


pointed


another


failing


in her


research


design


Judging
themes


effects


that


were


of revising


written


at home


nd rewriting
without the


press


sure


of time,


and


revi


under


same


conditions,
class under


also be


ques


results


cne pressure
tionable. (


an essay


of test


written


conditions,


may


. 21)


main


scathing


uninspired


thrust


still


manner


of Emig


quite


which


1971)


accurate


writing


study


indictment


instruction


amounted


of the


carried


our


little


schools.

or no time


Emig


observed


to plan


what


that

they


her


wrote;


8 subjects


in other


took

words,


they


engaged


in almost


none


of the


pre-text


revi


sion


that


Witte


(1987)


claimed


an ess


ential


feature


of composing.


reason


Emig


subjects


took


little


or no time


to plan


what


they


wrote


that


little


or no time


was


built


into


class

tended

might


writing


assignments


compose


be programme


algorithmic

d to write,


their school.

essays, the

devoid of an


The

kind


students

a computer


commitment


to the


exploration


ideas


or to taking


risks


name


learning


to write


well


Moreover


none


of Emig










products


showed


little


or no satisfaction


or excitement


about


their


work


87) .


They


did


stop


to contemplate


what


they


wrote


were


indifferent


to what


their


writing


might


about


they


were


or the


effects


that


their


writing


might


have


on an audience.


Emig


s 1971


finding


that


students


do not


typically


revise


what


they


write


consistent


with


other


result


(Applebee,


1984;


Applebee


et al.,


1981;


Hoetker


Brossell,


1979;


National


sessment


of Educational


Progress,


1977,


1986;


Pianko,


1977)


which


show


that


since


Emig


s study


little


changed


in the


that


writing


taught.


Writing


years


after


Emig,


Gere


(1986)


pointed


out


that


composition


instruction


resulting


still


lacked


in what


a coherent


called


"the


philosophy


dominance


to guide


of mechanical


features


(syntax,


spelling,


modes,


style,


and


punctuation)


today


instruction"


It is

literature


important


to keep


on composing,


that


mind,


studies


as one

such a


reviews


s Buxton


which


tested


methods


teaching


composition


including


a fully


articulated


revision


component,


amount


to research


into


what


might


happen


happening


schools


there.


This


into


what


well-documented


typically


between


theory


practice


explain


why,


as Hillocks


(1986)


shown,


1 *4-la


r flC ao.~ EN -~ a a a 4-. a 4- 1~ .21 --A-- -- w


34)


aaaa rnk


* n h A A A


Ltl


r


r%










environments


available


composition


where


research


research.


revision


settings


Lacking


going


been


natural


on.


lack


a serious


settings


obstacle


where


students


looked


regularly


revise


differences


written


between


more


work,


researchers


successful


and


have


less


successful


writers


an attempt


to tease


distinctive


features


characterizing


these


groups.


was


that revision

Unlike Buxton,


until


Sommers


itself became

Sommers was


the

not


(1978)


focus


wrote


dissertation


of a major


interested


study


comparing


effects


of revision


against


no revi


sion


in the


composition


classroom

revising


analyzed

freshmen


Instead,


practices


an extensive


examined


competent


corpus


experienced


differences


novice


of revisions

professional


between


writers.


made


the

Sommers


college


writers


interviewed


them


about


their


writing.


found


many


differences


between


revising


practices


of the


groups.


example,


Soinmers


found


that


inexperienced


writers


tended


their


to equate


revi


texts


sion


and


with


did


single-word


think


replacements


of revision


as a


process


which


complete


texts


may


be reshaped


Moreover,


inexperienced


writers


governed


their


writing


questionable


"rules


revi


sion"


such


"never


end


Q flfl On C.O = ~nrnarnr C' 4 i- 4 nI I -0C.


ai~


~ an t an ~~


bt n?~


I










restructuring


putting


text,


together


of finding


again


argument,


132-134).


of taking


They


apart


looked


overall


framework


or design


emerge


from


their


rough


drafts.


In addition,


unlike


shmen


writers


study,


Sommers


their


writing


' experienced


(pp.


writers


138-140),


were


this


aware


an audience


awareness


affected


their


writers


revising


saw


efforts.


revision


Most


important,


as a never-ending


experienced


process


operating


levels


of textual


organization


. 140-142).


Perl


(1978,


1979)


used


a case-study


approach


to reveal


composing


processes


of five


unskilled


writers


composition


course


was


teaching


at a junior


college


York


City


Perl


reported


successful


testing-out


of a research


instrument


she devised,


composing


style


sheets.


These


style


sheets


recorded


observable


behavior


subjects


engaged


in during


act


composing


written


texts


cluded


notations


identifying


episodes


of such


subprocesses


as planning,


assess


ing,


repeating,


reading,


editing,


several


others


(1978,


54-64).


Perl


found


that


subjects


began


writing


after


only


minutes


subjects


of prewriting


displayed


activity


consistent


311).


composing


five


behavior


of her


that


they


engaged


some


form


of prewriting,


following


comnosina.


interspersed


with


i..ti ncr


Psrl


I L -


fnnln r


~lfll









movement


from


a consideration


of what


been


written


to the


production


more


writing


to a reconsideration


of what


been


written,


and


so on


315).


This


finding


consistent


with


model


of the


writing


process


constructed


Flower


Hayes.


Regarding


revision,


Perl


reported


that


subjects


often


attempted


to resolve


complex


problems


syntax,


usage,


organic


zation


using


a set


of editing


"rules"


which


were


quite


inadequate


achieving


success


in improving


their


texts.


Subjects


miscorrected


about


as many


errors


they


were


able


to correct


appropriately


323).


Furthermore,


Perl


reported


that


subject s


typically


wrote


from


decidedly


egocentric


point


of view.


In her


words,


"They


see


making


necessity


connections


of making


among


their


their


referents


ideas


explicit,


apparent,


carefully


explicitly


relating


one


phenomenon


with


another,


or of placing


S. generalizations


within


orienting,


conceptual


framework"


. 326).


concluded


from

too


this

much


observation

cognitive ca


that


pacity


perhaps


unskilled


to editing


the


writers

y are c


commit


omposing,


thus


closing


flow


of idea


and


obscuring


relationships


among


ideas.


To make


matters


worse,


Perl


found


that


subjects


confined


their


editing


to the


most


erf i


cial


levels of


text.


much


attention


was


baid


.










were


more


concerned


with


spelling,


pronoun


use,


and


subject-


verb


agreement.


Clearly,


Perl


s findings


and


conclusions


are


close


accord


with


those


Sommers.


A study


Faigley


Witte


1981)


measured


differences


revis


behaviors


of six


expert


adult


writers,


advanced


composition


students,


inexperienced


designed


students


students


cruited


deficient


from


a writing


writing


laboratory


skills"


406).


Faigley


and


Witte


found


that


of their


inexperienced


writers


' revisions


were


surface


changes


which


affect


meaning


their


texts.


advanced


composition


students


expert


adult


writers,


taken


as groups,


each


made


times


as many


meaning


-changing


revisions


as the


inexperienced


affected


writers,


global


times


structure


more


of their


revi


texts.


sions


These


which


results


are


consistent


with


findings


of S


ommers


and


Perl,


discussed


above,


with


those


Bridwell


(1979,


1980),


be discussed


next.


An experimental


study


focus


on the


question


of what


student


do when


they


revise


their


essays


was


carried


Bridwell


(1979,


1980) .


Bridwell


analyzed


revi


sions


twelfth


processes


graders


during


three


writing


stages


of first


their


drafts,


composing


between


dr~ ft c! n A


AIir nn


r~~~~ ~ 1 .A t d-at.. f- E.


7-rb-4 4-~ 5.. n


~,, EL,


n~-? t 'I


T--nn


r










improvement


as measured


Diederich


(1974)


Analytic


Scale.


Next,


Bridwell


found


that


there


was


a strong


correlation


between


scores


revi


given


sions


extending


to those


essays,


length


implying


essays


that


quality


opportunity


students


had


to revise


resulted


in idea-elaboration,


which


turn


improved


quality


of the


writing


second


drafts


their


essays.


Consistent


with


findings


reports


Sommers


(1980),


Perl


(1979),


Pianko


(1977


Ramig


(1982),


majority


revisions


in Bridwell


s sample


occurred


at the


word,


phrase,


surface


mechanics


levels,


indicating


a lack


of attention


to whole


discourse


concerns


revi


sions


these


students


made.


When


Bridwell


essays


rated


analyzed


lowest


essays


terms


rated


of quality


highest


(1979,


128-134)


found


that


lowest-rated


essays


had


received


only


4% of the


total


revi


sions


performed


on them


at the


between-draft


stage.


Ninety-six


percent


of the


revi


sions


made


on these


poorest


essays,


therefore,


were


performed


writer


was


producing


new


text.


inference


which


can


be drawn


from


this


finding


that


these


students


read


contemplate


their


first


drafts


as much


as more


successful


students


did


before


beginning


their


second


drafts.


iA 1.4 un~a ..i..&. a --- -


~_.-11


1


~ *Irrr


'I' *


I


*










what


they


had


written,


contemplate


need


revision,


make


improvements


their


sting


texts


before


moving


on to


completion


of their


second


drafts.


Audience


Awareness


Revision


Other

awareness


researchers

of audience


have


explored


revising


relationship


practices


between


of student


writers.


Monahan


(1984),


using


compose-aloud


procedures,


studied


differences


between


four


basic


and


four


competent


writers


they


wrote


essays


audiences


their


teachers


writers


their


"were


peers.


generally


He found


capable


that


of the


both


same


types


types


revision,


same


purpose


. 299).


However,


Monahan


"basic


writers


[made]


most


of their


revisions


teacher


audience


competent


writers


[made]


most


their


revisions


peer


audience"


. 300).


addition,


Monahan


found


that


competent


writers


made


more


revi


sions,


a wider


range


of revisions,


longer


revi


sions


than


basic


writers.


Roen


between

including


Willey


audience

a revisi


(1988)


also


awareness


on.


explored


students


Specifically,


these<


relationship


' composing p

e researchers


processes ,

studied


effects


of giving


one


group


of freshmen


composition


0*1 rlnon4- a


0~ ~ h a 4- nC' a 1%r 4---A. 5 -- 4 -- -_--


~ baC


,,, ~


LL


,.,,,, L


1










questions


after


they


had


completed


first


draft


essay.


Twenty


students


were


randomly


assigned


to each


these


treatment


conditions


A third


group


of 20


students


received


no questions


as they


wrote


and


revised


their


essays.


Following


holistic


scoring


of the


essays


trained


raters


using


a 6-point


scale,


Roen


and


Willey


found


that


students
audience,
revised,


were


either
produced


significantly


revised
asked t


Furthermore,


quality


asked


as they
revised


higher


versions


o focus


they


of revised


to focus


attention


drafted or as
versions that


in overall


produced


attention


found


ess


quality


students


on audience


no significant


written


they


were
than


who
(P.


judged


were


difference


students


between


in the


questions-before-drafting


questions-after-drafting


conditions.


Other


Revision


Studies


Bernhardt


(1988)


investigated


effects


time-delay


on revis


writers,


ing.


when


He found


given


that


3 days


of 117


to do


college


so at home,


freshmen


were


basic


able


successfully


Bernhardt


revise


drew


their


writing


conclusion


that


at all


text


there


levels.


"need


to be


cautious


skills


general


when


zing


conditions


about


. students


of testing


severely


' revising


constrain


performance










correlation


between


either


kinds


.g.,


mechanical,


lexical,


syntactic,


stylistic,


figurative,


rhetorical)


or the


amount


of revi


sing


done


students


resulting


scores


on a 4-point


holistic


scale


assigned


essays.


Similarly,


Hillocks


(1986)


found,


a meta-analysis


selected


studies


published


between


1963


and


1982,


statistically


measurable


connection


between


revis


ion,


taken


itself,


improvement


of student


writing


219).


However,


Larson


(1987)


and


Schwegler


1988)


both


strongly


questioned


validity


of Hillocks


' findings


regarding


revi


sion.


Schwegler


asserted,


"The


literature


Hillocks


reviews


on the


effects


of revision


. is


skimpy


clearly


focused"


452).


Onore


(1983)


questioned


wisdom


of expecting


improvement


in texts


as a necessary


and


immediate


result


bringing


writing-process-based


pedagogy


to bear


composition


classroom.


Onore


and


trained


experienced


readers


working


as a committee


applied


facilitativee"


comments


Onore


to the


s study


first-draft


revealed


essays


that


of three


students


college


' revisions


freshmen.


produced


as much


diminution


of quality


subsequent


drafts


they


improvement.


Onore


concluded


that


students


were


some


points


disrupted


("exposed


to chaos")


their


composing


n ~ ~ ~ ~ ~ ~ ~ k 'w a % .--ii.n aaa-- .- 1 -- _- -


_r


L1.


E


L


r"r vlA A A A CI A A k


A


4.










writing


seemed


important


development


within


students


of the


tacit


understanding


of what


means


compose


effective


written


texts.


Part


of this


understanding,


according


to Onore,


involves


acceptance


of the


fact


that


chaos


an integral


part


of the


writing


process.


Allowing


texts


to become


chaotic,


and


therefore


less


successful,


Onore


claimed,


may


a necessary


step


some


students


on the


road


to writing


skill.


These


students


need


to learn


how


to call


forth


from


that


chaos


truer,


more


satisfying


meanings


which


will


contribute


toward


improving


later


versions


of their


texts.


Elaborating


this


argument,


Onore


(1989)


called


sharing


judging


power


between


quality


including


quality


teachers


of writing


of their


students


students


revisions.


do for


relative


school,


In her


words,


traditional


textbook


classroom

precepts


settings,

prevented


authority


student


writers


teachers


from


asserting


ownership


over


their


writing"


231).


What


needed,


according


to Onore,


are


classrooms


which


students


and


teachers


negotiate


with


each


other


concerning


quality


student


texts


such


a way


"that


neither


pure


idiosyncracy


nor


tyranny


results"


. 232).


Self-Evaluation


Writing


fla in n, 44n 'S


c,,~A*IL~


1 I h~


T 2










1978)


"a mismatch"


(Flower


Hayes,


1980)


between


their


communicative


intentions


and


execution


of those


intentions


through


use


of language.


This


between


intentions


execution


may


be perceived


writer


before


(Witte,


1987),


during,


or after


writing


and


correspond


one


or more


of Nold


s (1982)


levels


writing


subtasks.


Unless


students


are


aware,


through


some


form


of evaluation,


of discrepancies


between


their


communicative


intentions


degree


success


with


which


they


them


have


satisfied


to revise


what


those


they


intentions


write


language,


virtually


requiring


an empty


exercise,


a guessing


game


in which


goal


"give


teacher


what


or she


wants"


(Pianko,


1977,


. 258)


and


not,


should


achieving


research


move


writer


examined


a piece


s goals.


of writing


A considerable


self-evaluation


closer


amount


writing


process,


following


review


will


show.


Emig


1971)


revealed


that


subject,


Lynn,


engaged


self-evaluation


as she


composed


aloud


that


this


self-


evaluation


process


was


strongly


influenced


a particular


teacher


Lynn


s dictum:


attempted


abstractness


into


clear,


translate


sets


concise


[these]


of behavior


memorable"


directives


[could]


60).


of high


enact"


- a a Ir-- --- ..-. .SS


73\ rt,,


I I


1 I










aloud


sessions,


Emig


counted


self-accepting,


self-


congratulatory,


neutral,


and


self-critical


comments


In addition,


Emig


found


evidence


that


part


of Lynn


self-evaluation


source


awareness


and


concern


audience


64) .


Lynn


also


displayed


an ability


identify


irrelevant


digressions


writing


. 66).


important


to note


that


Lynn


s application


of self-


evaluation


occurred


exclusively


during


creation


new


text.


Emig


pointed


(pp.


67-68)


that


research


design


encourage


subjects


to revise


after


completion


first


draft.


However,


Emig


gathered


from


observations


conversations


with


Lynn


that


this


student


"[had]


a view


of the


writing


process


as a no-nonsense,


no-dawdle


task


which


one


devotes


a given


amount


of time,


and


no more"


g placed


blame


this


compartmentalization


Lynn


s self-evaluation


process


squarely


on the


kind


writing


instruction


Lynn


experienced


school.


According


Emig,


Lynn


s writing


process


"matches


views


of her


teachers


provide


school


time


earlier


[i.e


., planning


later


[i.e.,


revi


ing]


portions"


Lynn,


revising


a piece


of writing


after


completion


of a draft


was


"punishment


work"


68),


something


which


was


required


only


a piece


of writing


oii FfP ora


f ram


Ur1'n flmn 7 ar


-r-..


~rrhrO


n










hand,


their


awareness


of a need


major


revision


in their


writing


. 86-87)


and,


on the


other


hand,


fact


that


they


"did


no reformulating"


(i.e


revising)


of pieces


produced


Emig


s inquiry


. 87)


No revi


sing


logically


implied


that


either


no self-evaluation


took


place


on the


part


these


students


or that


their


self-evaluation


yielded


complete


sati


sfaction


with


writing


they


had


produced


basis


of Emig


s comment


that


subjects


were


"dutiful


enough


to want


to please--minimally;


no more,


one


conclude


that


no self-evaluation


took


place


once


these


students


completed


their


first


drafts.


Moving


beyond


Emig'


exploratory


research,


Stallard


(1974)


collected


analyzed


data


related


to the


composing


processes


of "good"


and


"average"


student


writers


purpose


of systematically


comparing


these


two


groups.


Stallard


observed


subjects,


seniors


at a Virginia


high


school,


as they


composed


essays


about


recent


news


events


208).


After


writing


these


essays,


students


were


interviewed


concerning


their


recollections


of the


experience


of writing


essays.


Stallard


found


that


both


groups


demonstrated

mechanics.


over


a concern

Moreover,


issues


for

both


of audience


correct

groups


needs


spelling

evidenced


global


good


a lack


essay


concern


structure.


- I 'I ----2 U 1- -- 1- --


~~~?L


nL,'I 'I


II----111


L


1~ I


*










higher


levels


of syntactic


complexity.


They


more


often


read


thought


about


what


they


were


writing


as they


composed,


they


more


often


expressed


a concern


having


a clear


purpose


their


writing


. 216)


In other


words,


Stallard


evaluation


"good"


of their


writers


engaged


writing


sus


more


tainted


wide-ranging


that


self-


engagement


longer


than


randomly


selected


("average"


group.


These


diff


erences


between


more


and


less


competent


writers


are


consi


stent


with


findings


of Sommers


1978)


and


Perl


(1979)


discussed


above.


As Sommers


' inexperienced


writers


revised


on the


basis


of self-evaluation,


they


made


mostly


single-word


or short-phrase


substitutions


. 155).


These


revisions


were


process"


"dictated


154) .


In sharp


their


a-theoretical


contrast,


Sommers


[composing]


' experienced


adult


writers


were


capable


of self-evaluation


at all


levels


of textual


organization


were


sensitive


to revi


sion


cues


which


college


freshmen


writers


were


largely


unaware.


Similarly,


unable


Perl


(1979)


to evaluate


their


found


own


that


writing


subjects


beyond


were


word


phrase


levels


and,


like


Sommers


' inexperienced


writers,


were


heavily


influenced


their


self-evaluation


attempts


apply


an unsystematic


of "rules


revi


sion"


to their


writing.


Perl


s subjects


were


consistently


unable


aim 1 iitn t-ho4r ,an ti4-l-.aea1 a..n a. a -] i- a --ar n4


tar4 4- flf


~tr~ 1 ir 3 i" a


tha~r


katrhn ~


Ckhnn


" vlA rll~ nC I


nwn


r










revise


effectively


can


seen


as resulting


from


their


dysfunctional


self-evaluation


processes.


Beach


(1976)


conducted


an informal


study


involving


undergraduates


in a writing


methods


course.


He had


them


write


first

first


drafts


papers


draft


then


draft.


they


which


they


"free


tape-recorded


students


shed


were


were


wrote"


their


allowed


asked


(Elbow,


evaluations


to write


1981)


of that


as many


to tape-record


their


evaluations


of each


succeeding


draft.


Beach


found


that


students


fell


neatly


into


distinct


groups


extensive


revisers


nonrevisers.


extensive


revisers


demonstrated


kind


of self-evaluation


process


which


characterized

experienced a


extensive


Stallard


dult


s "good"


writers.


reviser


group


writers


In Beach


tended


s words


to conceive


Sommers'

: "Students


of revising


a process


of making


major


alterations


content


substance


contrary,

problems


their


of their


"mimicked


their


problems


drafts"


formulaic


drafts.


in a checklist


161) .


textbook


. They


manner,


nonrevisers,


language


tended


noting


to conc


each


on the


describing

eive of


problem


a separate


category"


. 161)


In short,


Beach


s nonrevisers


displayed


characteristics


similar


to Stallard


, Sommers


Perl


ess-able


writers.


They


revised,


when


they


H1~! AA


^ all^1


-~ ~~~ A.. '-. .'- S


A





t


LL L r










concerning


although


one


relative

tempted


quality

to infer


of the

that


two groups'

the extensive


writing,

revisers


wrote


better


papers.


interesting


to note


that


although


Beach


been


taken


to task


assuming


that


"extensive


revisers


necessarily


produce


better


quality


paper"


(Faigley,


Cherry,


Joliffe,


Skinner,


1985,


. 57),


make


this


claim


anywhere


in his


1976


article,


focusing


instead


differences


between


self-evaluation


processes


extensive


revisers


nonrevisers.


Beach


s expressed


assumption,


own


words,


was


"the


fact


that


students


often


revise


their


drafts


reflects


their


inability


effectively


evaluate


their


own


writing"


(Beach,


1976,


p.160).


A later


study


Beach


(1979)


compared


effects


teacher


evaluation,


student


self-evaluation,


evaluation


of first


drafts


on the


revi


sions


10th-


and


11th-


grade


students


made


on those


first


drafts.


Beach


controlled


teacher


bias


having


a trained


professional


other


than


classroom


teacher


involved


study


fill


a four-


part


essa


y-evaluation


form


teacher-evaluation


treatment


group.


Self-evaluators


filled


same


form


their


own


essays.


no-evaluation


group


was


simply


asked


to revise


their


first


drafts


before


handing


them


grades.


measure


effects


of these


three


treatments,


1 I S


'1~ ~~~~ aw r '. A 4 L--a- aa --a.-S..-


C1an


~Mn 1 hr rA ~


c 1,,,,


L


~I


rl


r r


r\


r










maintained


inter-rater


reliability


above


categories


of the


instrument.


Beach


found


that


teacher-


evaluation


treatment


group


"showed


significantly


higher


degree-of


-change


scores,


fluency


scores,


final


-draft


support


ratings


"supporting


an argument


with


examples


logical


reasoning")


than


either


self-evaluation


or no-


evaluation


subjects"


. 117)


However,


since


Beach


did


include


any


peer-evaluation


treatment


experimental


design,


results


are


limited


in their


scope,


indicating


only

more


that


teachers


effective


thar


' evaluation

n students'


of students'


evaluations


writing


of their


may


own


writing;


results


cannot


indicate


how


peer


evaluation


might


fare


comparison


with


other


treatments.


Miller


professional


(1982),


student


an informal


writers,


study


comparing


discovered


major


differences


these


groups


evaluated


their


own


writing.


First


students


"thought


of good


writing


as writing


teacher


liked"


. 179) .


Professional


writers,


sharp


contrast,


well


"experienced


finished


quality


product


of a piece


fulfilled


relation


writer


intention"


179) .


Furthermore,


Miller


found


that


profe


ssional


writers


of their


I'ouh


best


writing


one


ece


in which


they


had


learned


or mastered


a particular


I
I-. a -


-~~~ ~~ -1 J -


L


,,,,,, -t


II 1


r t ~ r.


.J


*










sake


of discovery.


Miller


s findings


add


an important


dimension


to this


discussion


of self-evaluation


Expert


writers


apparently


have


internalized


criteria


against


which


they


judge


their


own


writing,


while


students


more


typically


view


those


criteria


as external


to themselves.


Another


study


related


to the


issue


of how


well


students


assess


quality


Matsuhashi


Gordon


their


own


(1985).


writing


They


was


found


conducted


that


students


were


directed


to add


information


to their


texts


when


those


texts


were


fresh


memory


unavailable


rereading


produced


significantly


more


revisions


whole


-text


level


than


with


students


their


were


texts


similarly


available


directed


rereading


to add


information


or student


were


directed


simply


to revise


they


shed


with


their


texts


available


rereading.


suhashi


and


Gordon


experimental


subjects


(i.e.,


"blind"


ones)


were


forced


their


classroom


situation


to evaluate


their


texts


as wholes,


since


they


could


focus


on the


bits


and


pieces


of their


texts


without


rereading.


According


to Matsuhashi


and


Gordon,


kinds


of environmental


factors


presented


to students


their


instructors


makes


a great


deal


of difference


terms


of learning


outcomes.


Matsuhashi


Gordon


drew


a distinction


between


I nt-aa


a a- 2 2 -


_-1


-


-. UU


a. -


4 r r f4 ^


-Y *


I


*


-


*










revising


which


elicit


"the


writer


s ability


to orchestrate


myriad


processes


involved


in writing


revising.


This


kind


of knowledge


S. represents


a knowing


a tacit


ability


to traverse


full


range


of revisions"


231).


results


of thi


study


indicate


that


students


have


more


gain


from


writing


instruction


which


stimulates


building


of tacit


knowledge


about


revi


sion


than


they


do from


instruction


Mat suhashi


which


Gordon


aimed


only


refer


at building


as knowing


what


that


knowledge.


This


same


distinction


underlying


revision


process


ses


of the


more


successful


success


writers


studied


Stallard,


Sommers,


Beach,


Perl,


and


Miller,


reviewed


above


Beach


(1989)


described


some


of the


difficulties


students


encounter


as they


attempt


assess


effectiveness


their


own


writing.


First


among


these


difficulties,


according


to Beach,


assessment


tendency


as a result


of students


of being


to terminate


"overwhelmed


self-


or frustrated


certain


problems


in their


writing"


129) .


Next,


Beach


cited


fact


that


"students


often


have


an unclear


sense


their


own


rhetorical


role


or persona


writing


within


academic


contexts"


130)


. A third


problem


inhibiting


students


' ability


engage


fruitful


self-a


assessment


takes


- -


C -n*


,E i, a


aI


- -J


a a


a -- -- a - -


t. '--1 -


hP~ra,










Sommers


(1980).


Another


reason


students


have


difficulty


with


self-assessment,


essential


according


step


to Beach,


self-assessment


that


they


process.


skip


They


do not


describe


themselves


what


they


are


doing


in their


writing,


leap


features


unless


ahead


of the


students


instead


text


can


to make


. 134,


describe


prec


135) .


their


ipitous


Beach


writing


judgments


pointed


accurately,


about


that


they


will


have


great


difficulty


judging


how


can


be improved.


Perhaps

to be a


owing


to these


n unhelpful


factors,


technique


students f

prompting


mnd


self-assessment


effective


revision


(Freedman,


1987).


Peer


Evaluation


Comnos


ition


the Writing
Students


Pierson

evaluation o


Pierson


method


classes


only


(1967)


compared


f ninth-grade


isolated


of evaluating


of similar


difference


teacher


essays


variable


first


students


ove r


was


drafts


taught


between


evaluation and

a period of 8


interested


peer

months.


in--the


of essays--by


same


groups


pairing


instructor.


taught


each


three


instructors


was


that


one


group


received


conventional


teacher


correction


of first


drafts,


while


other


broke


into


groups


of four


or five


students


engaged


peer


discussion


first


drafts.










groups


assigned


to each


of his


three


instructors,


leading


to recommend


peer


discussion


method


as preferable


on the


grounds


that


relieves


instructors


many


hours


of work


grading


first


drafts.


validity


of the


STEP


test


employed


Pierson


been


called


into


question


Buros


' Fifth


Mental


Measurements


Yearbook


(1959),


because


requires


no writing


asks


instead


that


test


-takers


evaluate


what


others


have


written.


In spite


this


flaw,


Pierson


s study


is still


good


example


of how


to set


an experimental


study


writing


in a live


school


setting.


Pierson


findings


cannot


be regarded


as part


as strong


of composition


evidence


instruction,


favor


peer


but,


interaction


same


time,


they


offer


no disconfirmation


of that


method


either.


Another


experimental


study


comparing


effects


of two


ways


of teaching


writing


was


basis


Lagana


(1972)


dissertation.


Lagana


use


d two


10th-grade


classes


was


ass


signed


to teach


experimental


environment


study.


compared


self-designed


approach


composition


instruction,


including


use


of individualized


writing


with


folders


traditional


peer-group


method


discussion


of teaching


of student


writing


essays,


Penn


Hills

whole


, Pennsylvania,


- cxronnl


at that


i nest rir+- i nn


time,


a method


organized


around


1-14~~~~~mr i--o Lynn, eha4a4.a-ra


tt n~~


C n r ~L CI H










Lagana,


following


example


of Pierson,


used


STEP


Writing


Test


measure


effects


of her


treatments.


reported


individualized


that


experimental


folders


peer


group,


discussion,


using


significantly


outperformed


traditional


group.


Also,


Lagana


claimed


that

with


experimental


a more


positive


subjects

attitude


emerge

toward


from


writing


that

and


treatment

presented


long


list


of testimonial


statements


from


students


support


this


claim.


Moreover,


reported


that


experimental


subjects


were


able


to arrive


informal


"grades"


each


other


s work


which


closely


matched


official


grades


arrived


Lagana,


agreement


with


Pierson,


found


that


paper-correcting


burden


was


greatly


reduced


experimental


peer-interactive


method


treatment


over


class,


traditional


an argument


approach.


Unfortunately,


design.


classroom


about


Test


somewhat


Lagana


fact


teacher


objectivity.


s study


that


an obvious


Also,


compromised


open


to criticism


experimenter


weakness,


use


validity


was


raising


STEP


of her


concerning


also


questions


Writing


findings.


Benson


(1980)


conducted


a study


involving


seventh-


eighth-grade


experimental


students


treatments


measure


on a variety


effects


of three


of dependent


variables.


t reatmsnt.


first


of these


1 i n'r 'rtm .an


Ir~llP~


W;I 4


* -


,


,










each


other.


second


treatment


was


"reinforcement


feedback"


and


involved


peer


interaction


using


a loosely


structured


evaluative


form


eliciting


such


responses


"favorite


sentence"


and


"favorite


word.


This


treatment


tended


to evoke


more


itive,


complimentary


responses


than


information-feedback


form.


third


treatment


was


control


group


receiving


no peer


interaction.


Compared


to the


results


of the


control


group,


both


experimental


treatments


"achieved


statistical


significance


or beyond


five


dependent


variables


quality


writing,


paragraph


revision,


sentence


revision,


word


revi


sion,


total


revision"


iv).


effects


of the


experimental


treatments


were


statistically


significant


comparison


with


each


other.


basis


of th


ese


findings,


Benson


recommended


use


peer


interaction


as part


of composition


instruction.


Gere


Abbott


(1985)


conducted


a study


to determine


what


kinds


of conversations


take


place


when


peer


groups


meet


in composition


classrooms.


These


researchers


recorded


conversations


of 5th-


8th-


11th-grade


students


were


part


cipating


writing


discussion


groups.


Nine


such


groups


were


minimum


after


involved


of four


which


study,


times.


students


each


recordings


' conversations


group


were


were


was


recorded


trans


cribed,


divided


into


4 Ann


rn; 4 -


'Fnll 1,%tt, nfl


rkn'-'4


I',flOfl\


.....I a .1 AJa.*.. a -


* -


ru rA


A


mLn










directing,


or eliciting.


researchers


were


interested


identifying


focus


of attention


each


idea


unit.


They


counted

writing


number

those


which

which


were

were


directed

directed


toward the

toward other


students

issues,


such


as how


relations


group


between


was


group


functioning


members.


or interpersonal


Finally,


researchers


tallied


number


of idea


units


which


focused


on writing


processes,


form,


content,


context,


and


reference,


respectively.


Gere


Abbott


found


that


majority


of the


students'


idea


units


focused


on the


student


' writing


and


frivolous


or irrelevant


subjects


of conversation


This


finding


countered


objections


which


may


be raised


instructors


opposed


to the


use


peer


groups


as part


composition


instruction


on the


grounds


that


such


peer


interaction


presents


much


temptation


students


waste


time


in social


chatter.


Gere


and


Stevens


(1985),


working


Gere


with


a different


Abbott


(1985),


portion


supported


data


discussed


reinforced


general


finding


that


students


expected


to remain


on task


when


they


are


asked


to meet


peer


groups


to discuss


each


other


writing


that


responses


they


give


each


other


in these


groups


frequently


more


personally


relevant


directly


ral matoA


4-r 4-ho


4-o~r-c r n~r nnfl 7~ 4-1ar~ f An ar a. a~ 4 .


I


tavt a


rh~l~nn rl


Ck31~


1" n nrt n ~3 ttn










Additional


evidence


supporting


use


peer


editing


groups


as part


of composition


instruction


provided


Karegianes,


Pascarella,


Pflaum


(1980)


Working


with


low-


achieving


10th-grade


students


(the


subjects


mean


grade


equivalent


score


was


on the


Lata s MacGin it ie


Reading


res


organized


searchers


writing


presented


instruction


students


including


with


well


assignment


one


essay


week.


Twenty-six


students


participated


peer


editing.


They


each


exchanged


essays


with


one


classmate


fill


ed out


an editing/rating


form


provided


researchers.


other


students


involved


in the


study


their


essays


using


same


edited


rated


editing/rating


classroom


form.


Except


instructor


this


variation,


fact


that


students


teacher-editing


treatment


spent


engaged


in class


peer-editing


scussion


students


to balance


interacting


with


time

each


other,


instruction


during


10 weeks


of study


was


kept


identical


as poss


ible.


Pretest


and


posttest


essays


study


were


scored


experienced


English


teachers


using


Diederich


analytic


scale (Diederich,

"flavor" category.


1974)

The


minus

raters


difficult-to-interpret


achieved


reliability


coefficients


pretest


posttest


roannnl- 4 iral ita oCa4-n-nJ a nC a- e a a


*naltrn~a


kA A


Test)


rElpnart: ttn 1 11


ac29alt~


,L cb,,,










scored


higher


than


teacher-edit


essays,


researchers


argued in favor

classroom on the


peer


basis


interaction

of increased


composition


time-on-task


associated


with


peer-edit


treatment.


They


described


this


effect


follows:


The practice
stimulate st


task


or peer
udents t


of composition.


teacher-edit
questions of
their papers,


were
class


intensively


smate


criteria


to the


task


s essa
or tha


was


editing


itself


an increased


(Recall


participants
the teacher


rticipants
involved
ys against
t lesson.)
voluntary


were


that


may .
attention


while


engaged


concerning


her


to the


the
n asking
editing


peer-edit


evaluating


group


their


instructional


Thus,


while


attention


students


teacher-edit


group,


peer-edit


function
focusing
thereby,
time.


was


group


as a useful


extending


providing
. 206)


ess


entially


Editing


itself


teaching technique
attention to task


an effective


use


mandatory
may, then,
for
and,
classroom


Karegianes,


Pascarella,


Pflaum


also


pointed


that


since


their


subjects,


inner-city,


low-achieving,


Latino


students,


experienced


a significantly


better


outcome


composition


instruction


using


peer


interaction,


arguments


that


this


type


of instruction


only


applicable


appropriate


middle-class,


high-achieving


students


may


be valid.


Taking


another


tack,


Graner


(1987)


conducted


a study


peer


editing


teachers


are


based

not c


on the


comfortable


assumption


with


that


giving


many


composition


control










essays


with


effects


a treatment


called


Revision


Work

prep


shops.

school,


Working

Graner


with

had


two

one


English


class


classes


engage


at a military

peer-group


discussions


of student


essays,


using


evaluation


forms


which


asked


students


to comment


on issues


ranging


from


correctness


of spelling


to the


overall


structure


an essay.


These


evaluation


forms


were


tailored


to the


varying


types


writing


., description,


character


sketch,


process


analysis)


students


were


assigned.


In the


Revision


Workshops,


students


did


read


comment


on each


other


essays,


nor


they


meet


groups


discussion,


Instead,


they


used


same


evaluation


forms


as did


peer


editing


students


applied


them


to sample


essays


chosen


researchers.


important


These


strengths


sample


and


essays


weakness


were


ses


selected


writing


to show


to provide


basis


teacher-led,


whole-class


discu


ssioi-s


of the


sample


essays


Graner


found


that


Revision


Workshop


group


peer


editing


group


both


improved


significantly


writing


ability,


based


on the


differences


between


scores


earned


on pre-


posttest


essays


as measured


Diederich


analytic


scale


(Diederich,


1974)


Importantly,


there


was


significant


difference


between


groups


terms


4h4 -r


* nn -


- -- A A .


~e:~


L


I


r


.. I t










helping


students


improve


their


writing,


regardless


whether


came


form


face-to-face


peer


interaction


form


of generalized


discussions


of the


issues


pertinent


to the


assignments


these


students


were


working


on.


Thus,


Graner


concluded


that


Revision


Workshop


was


effective


alternative


peer


editing


recommended


instructors


wish


to remain


focal


point


of their


classrooms.


In a study


Graner


whose


s evidence,


findings


Nystrand


ran


1986)


somewhat


compared


counter


effects


intensive


peer


review


with


effects


of instruction


writing


University


of Wi


sconsin-Madison


that


did


employ


this


technique.


Nystrand


described


intensive


peer


review


follows:


Students


five,
over


and
the


sharing


. meet


sa


course


and


instructor
students n
discussion


prepare
for pre


meeting.
paper or
They are


which
they
and u


membe
work.
best
hult .


are
sage


regularly


me groups
of the te


critiquing


assigns few
o checklists


Rather,


ces


sentation


Students


meet


rm


each


group


three
the


other


topics


of four


times
purpose


writing.


and


a week
of


The
the


gives


monitoring


students


f exposition
to classmate


are


a substantial


instructed


author
to avoid


;


of their


th


require


revi


si


to consid


achieves
checking
ev are re


-S


group


Periodically


papers


I *r


d ns


from


each


ml 'l* 4 irol


with


keep
from t
s at e
ed to
on for
er the


or her


spelling,


quired


their


journals


hese
very


prepare
each c
extent


notebooks
class


a new


lass
to


purpose;
punctuation,


to provide


a photocopy


instructor


student fo
14 A -


each


of their


collects


evaluation,


4,a n*-


4 n -.. ncle.. 4


k










Nystrand


collected


essays


from


each


students


13 classes


and


compared


these


essays


using


a rating


scale


derived


from


Britton


et al.


(1975)


which


differentiated


writing


along


a continuum


from


reporting


to inferencing


generalizing.


students


classes


engaging


intensive


peer


feedback


(the


"studio"


classes)


showed


significantly


larger


increase


amount


inferences


generalizations


their


essays


compared


with


students


take


part


peer


feedback


activities


(the


"nonstudio"


classes).


Nystrand


also


reported


that


"nonstudio


students


editing


while


came


(i.e.,


studio


increasingly


attention


students


see


to matters


increasingly


revision


as a matter


of mechanics


treated


usage),


revision


matter


of reconceptualization"


. 184) .


Nystrand


s studio


students


also


"made


significantly


more


progress


development


their


writing


abilities


s" (p


. 184)


as measured


pre-


posttest


scores


on the


Univers


of Wisconsin


Placement


Test


College


Qualifying


Test/Verbal.


Unfortunately,


instruction


Nystrand


provided


discuss


nonstudio


nature


students.


of the


Such


oversight


makes


interpretation


of Nystrand


s study


difficult.


Nystrand


Brandt


(1989)


found


a follow-up


Nystrand


s 1986


study


that,


on the


basis


of blind


readings


,~~~~~~~~~~~~, Al a4-a 4-a. '.-a-- a t T


t-rn, noel


n-rn rra.1+- a


n rr rr ~ ~ rr


* n~wrn


1 %J I









statements


about


revision


made


studio


nonstudio


students


First, st
quality t
addition,


very


-udio
:han


student


those


studio


different


views


s' revisi
their co
nonstudio


of what


ons


were


unterpart


of higher
s. In


students express
required to


effectively


Studio


revise


students


counterpart s


more


clearly


were


clearly


specific


what


about
about
their


openly


their
had m
their
those


papers


Lore


revision
needs,


revisions


critical


particularized


were


of their


their


. (p
say than


. 213)
their


needs, (2)
3) explain


were
d more


to accomplish,


drafts,


revision


more


needs


rhetorical


revi


sion


manner


situation,


needs ir
(p. 215)


a more


coheren


scussed
t and i


their


integrated


generally


positive


picture


regarding


benefits


be expected


from


peer


feedback


activities


reflected


literature


surveyed


above


been


reinforced


writers


such


as Beaven


Menendez


(1977),


1980) ,


Bruffee


Gebhardt


(1984),


(1980),


Elbow


Lamberg


(1981),


(1980),


Flanigan


Macrorie


(1984),


this


Moffett


supportive


(1986),


view


Murray


peer


(1985).


feedback


In contrast


as an ingredient


composition


pedagogy,


Diana


George


1984)


found


that


only


some


peer


groups


attain


kind


of wished-for


outcomes


described


authors


cited


above.


other


kinds


peer


groups,


which


George


named


"leaderless"


and


"dysfunctional,


interaction


were


approach


troublesome

to composition


by-products


of the


instruction.


The


peer-

se types


peer


groups


characteristically


engage


in counterproductive










composition


classroom


Newkirk


(1984),


who


introduced


additional


factors


which


tend


to erode


effectiveness


peer


interaction


as an instructional


approach.


Newkirk


found,


a study


comparing


reactions


students


instructors


University


of New


Hampshire


accepting


to student


of each


essays,


other


that


s writing


students


than


were


their


more


structors


They


tended


to read


missing


information


into


their


peers


writing


and


to be impressed


writing


styles


which


were


rejected


instructors


inappropriate.


This


kind


ineffective


feedback


on the


part


of students,


Newkirk


pointed


out,


must


addressed


instructors


use


peer


interaction


with


their


classes.


Berkenkotter


(1984)


reported


that


personalities


students


interfere


with


well


peer


interaction


helps


them


improve


their


ability


to write.


example,


Berkenkotter


found


that


immature,


defensive


types


tended


reject


categorically


good


peer


advice,


while


passive


and


vulnerable


types


were


more


likely


to accept


upon


poor


peer


advice,


thereby


lowering


quality


resulting


writing.


A fourth


investigator,


Carter


(1982),


found


study


which


isolated


peer


interaction


as a variable


study


of college-bound


high


school


seniors


that


having


peers


rP~ari


RflI nl ne- ri4-ar a 4- 1n ,nA.naA A


Snmmntnn-


nn is ~ ~h


ntknr


hyl~~l~nn~


r tr A L L A u


*


'LFh










(1983)


found


that


some


college


freshmen


participated


study


peer


evaluation


of student


writing


resented


critical


comments


about


their


writing


coming


from


a source


other


than


their


instructor


ignored


what


have


been


good


advice


because


of that


resentment.


Freedman


(1987)


found


that


students


responding


national


survey


indicated


that


they


find


peer-


response


Freedman,


group


when


very


students


helpful"


meet


. 87).


According


peer-response


groups,


talk


often


guided


students'


desire


to help


each


other


"get


their


writing


'right


I I


161);


in other


words,


they


instead


to identify


of using


what


peer


their


group


teacher


engage


wants


each


writing,


other


discussions


of their


experien


ces


they


read


each


other


writing.


generally


positive


image


peer


feedback


which


been


built


much


of the


research


literature


must


considered


light


of these


ess


supportive


reports.


Finally,


a study


of 12th-graders


comparing


peer


response


self-evaluation,


which


res


earcher


was


also the


cooperating


structor,


DiMento


(1988)


found


slight


advantage


self-evaluation


in improving


essays


from


first


to final


drafts.


However,


neither


approach,


according


to DiMento,


was


superior


stimulating


improvement


essay


- I1 ) A A I


mr rr i: Crl


,,, i


hrrnrr










perceive


writing


writing


instruction


(Herrington,


1989),


resulting


researchers


vigorous


to learn


and


more


continuing


about


effort


details


on the


of the


part


writing


process


and


to test


usefulness


instructional


techniques


that


are


intended


to promote


better


results


our


schools

created


complexity


been


of the


acknowledged


process


and


which


preliminarily


writing


explored.


concept


of revision


has been


clarified


so that


longer c

process,


consideredd


discrete


a permeating


fact


stage

r that


within

may b


writing


e activated


on the


basis


many


cues,


nature


of which


are


still


open


investigation.


importance


of raising


students'


consciousness


beyond


most


superficial


aspects


of text


when


they


read


their


own


written


work


also


been


acknowledged,


researchers


have


begun


to explore


ways


reaching


this


instructional


goal.


present


study


was


designed


to contribute


to the


expansion


peer


of knowledge


evaluation


about


within


revision,


context


self-evaluation,


composition


instruction.


Each


studies


discussed


this


review


literature


contained


valuable


insights,


some


of them


were


weakened


measurement


problems


they


in their


employed.


design


or in


example,


methods


Pierson


s (1967)


T.nnsa n


a 1Q6'Vt


a -


a a %C 4.a FlL.'*3 f~1a .. t.C S -


IL


AC C~LA


1__


i


L


I~ rr A


man m,,L










Perl


(1978)


Onore


(1983)


provided


rich


detail


their


examination


questions


these


of their


about


studies


subjects,


whether


were


they


particular


representative


were


students


or atypical.


weakened


involved


Moreover,


since


researcher


was


also


classroom


instructor


case


of Lagana,


Pearl,


and


Onore,


questions


about


objectivity


of data


interpretation


cannot


be entirely


dismissed.


Bridwell' s


(1979)


study


contained


important


data


related


to the


revi


sing


behavior


high


school


students


was


weakened


fact


that


conclusions


were


based


on a


single


piece


of writing


composed


students


participated.


present


study


collected


data


from


students,


analyzed


writing


composed


over


a 4-month


period,


and


used


holi


stic


scoring


measure


effects


on its


dependent


variables.


from


In addition,


qualitative


present


quantitative


study


sources,


drawn


data


demonstrating


how


these

each


two

other


research

within


methodologies


a study


be used


of composing


to complement


processes.


While


attempting


overcome


some


of the


problems


in the


experimental


design


of existing


studies


of revi


sion,


self-


evaluation,


design


a study


peer

that


evaluation,


would


this


contribute


researcher


toward


tried


a better


A a r LL. .


,S L1


---. I F I


1

















CHAPTER


THREE


METHODOLOGY


This


study


was


based


on data


drawn


from


four


sections


1101


semester


(Freshman


of 1987


Composition)


University


conducted


during


of Florida.


fall


Each


of two


instructors


evaluating


taught


section


one


self-evaluating


of ENC


1101.


one


sections


peer-


in the


morning


three


50-minute


periods


per


week


14 weeks.


Selection


of Instructors


instructors


were


used


this


study


in order


reduce


threat


to validity


which


would


have


resulted


only


one


instructor


been


involved.


However,


no attempt


was


made


compare


data


from


instructors


statistic


cally,


because


doing


so would


have


introduced


many


uncontrollable


variables,


such


as differences


between


instructors


way


they


taught


their


sections.


Instead,


instructors


represented


separate


environments


within


which


hypotheses


underlying


study


could


tested.


This


replication


made


p055


ible


to identify


both










summer


of 1987,


after


gaining


permission


English


sought


Department

composition


to conduct

instructors


this

who


study,

would


researcher


teaching


sections


of freshman


composition


in the


fall.


instructors


agreed


to parti


cipate


this


study


were


both


experienced


teachers


writing.


Instructor


was


assoc


iate


professor


of English


with


over


years


' experience


University


Florida.


Instructor


was


a doctoral


student


English


Department


University


Florida.


a M.A.


held


from


a B.A.


Teachers


English


College,


from


Columbia


Columbi

. Her


University


teaching


experience


included


years


as an English


teacher


high


school


level


Rhode


Island,


Texas,


and


Florida.


Assignment


of Subjects


Treatments


Random


assignment


of subjects


to treatments


was


possible


computer


this


study.


on a first


Students


come,


first


were


served


assigned


basis,


to sections


following


normal


registration


process


University


of Florida.


Which


of each


instructor


sections


would


receive


self-


evaluation


or peer-evaluation


treatment


was


decided


a coin


toss.


On the


first


of classes,


students


were


told


that


rnh ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~r n an4a n n- nan --1l n-A.


r~0~~ rrrk


nrh C hnC


trnA~


I\~AS~A~~rl


,, ,L










sections


chosen


study


agreed


to participate.


Table


shows


number


from


each


section


volunteered.


Table


Student


Response


to Request


to Participate


in the


Students
Enrolled


Section


Students


Agreeing


to Participate


Instructor

Instructor

Instructor

Instructor


One,

One,

Two,

Two,


Self-Evaluating

Peer-Evaluating

Self-Evaluating

Peer-Evaluating


From

subjects.


each s

Their


section,


identities s


students


were


were


never


randomly


revealed


selected


to their


instructors


during


study.


Of these


from


each


section,


students


were


again


randomly


selected


participation


tape-recorded


students


interviews


chosen


researcher


outside


interviews


one-to-one


of class.


agreed


interviews


Each


to meet


spaced


of the


with


about


month


apart.


Prospective


interviewees


were


contacted


telephone;


their


names


were


not


revealed


to the


instructors.


Table


displays


distribution


males


and


females


Study










Table


Distribution


of Males


Females


Self-Evaluating
Section


Peer-Evaluating
Section


Males


Females


Males


Females


Instructor


One


Total


Volunteers


Participants

Interviewees


Instructor


Two


Total


Volunteers


Participants

Interviewees


A questionnaire


filled


of the


students


four


sections


revealed


that


these


classes


were


fairly


homogeneous


from


a demographic


point


of view.


For


example,


of the


graduated


summer


1987


from


Florida


high


schools.


of the


students


randomly


selected


study


were


between


years


old,


graduated


from


high


school


summer


1987.


Thirty-










Classroom


Procedures


There


were


many


differences


between


instructors


way


they


usually


taught


their


sections


of ENC


1101.


However,


each


instructor


taught


two


sections


assigned


this


study


as similarly


as possible.


example,


sections


same


each


assignments,


instructor


commented


used


same


on student


textbook,


work


same


way,


followed


same


syllabus.


first


essay


assignment,


Instructor


One


asked


a response


to the


statement,


"All


writers


sell


reader


short


Instructor


Two,


on the


other


hand,


asked


students


compose


a hypothetical


application


letter


resume.


instructors


Following


organized


first


their


writing


essay


assignment,


assignments


both


according


"modes


of discourse"


approach


agreed


with


each


other,


independent


input


from


researcher,


to follow


same


order


of assignments.


As a result,


students


sections


wrote


their


second


through


seventh


(final)


essays


conforming


to the


following


modes


of discourse


following


order:


personal


narrative,


process


analysis,


definition,


classification,


Students


according


sections


to a four-step


cause/effect,


completed


procedure


and


essay


established


argumentation.


assignments


to promote










Some


clarification


necessary


concerning


timing


steps


nature


of the


revision


guides


mentioned


Figure


Step


2 always


commenced


one


week


following


Step


Step

Step


3 always


always


commenced

commenced


class

week


periods

following


following


Step


Step


In this


way,


students


always


one


week


to complete


their


revised


unrevised,


they


so chose)


final


drafts.


Essay


and


Revision


Guide


Data


researcher


collected


xeroxed


copies


of first


drafts


written


revision


guide


res


ponses


second,


fourth,


sixth,


seventh


essays


written


40 students


participating


study.


These


essays


are


referred


Sample


Essays


final


One,


draft s


Two,


of each


Three,

essay


and

were


Four,


res


pectively.


handed


When


researcher


xeroxed


these


drafts


returned


them


to the


instructors


within


24 hours.


given


four


Listed


essays


Figure


examined


are


this


assignments


study.


essays


revision


guide


responses


were


collected


from


instructors


outside


of class


Students


involved


interviews


see


researcher


after


first


of class


when


they


were


asked


to participate


in the


study,


until


last


class,


when


they


were


thanked


their


nz rt4


('4 V~~n*4nn tuIna flu wrl% r ..t -aaa


rc~n3f hn


rr L~1


,,,,,, ,E.


L A


1













SELF-EVALUATORS


PEER


EVALUATORS


STEP


Assignment


given.


Same.


STEP


dra
cop
ins
ion
ted.
wri
nses


ss.


fts due.
ies handed
tructor.
guides dis
Students
ting their
to revisi
stions in


Fir
Xer
in.
dis
stu
pee
beg


st
ox
R
tri
den
r's
in
rev


tions


drafts
copies
evisio
buted.
t rece
paper
writing
ision
in cla


due.
handed
n guides
Each
ives a
Students
g responses
guide ques-
ss.


STEP


Students


xero
writ
revi
tion
Stud
class
pape


cop
en r
ion
to
nts
on


hand
ies o
espon
guide
instr
work
revis


f
ses to
ques-
uctor.
in
ing


Students
copies o
response
guide qu
student
revision
with the
paper th
received


hand
f wri
s to
estio
discu
guid
peer
at st


xerox


sses
e re
who
uden


sion
Each

sponses
se
t


in Step


STEP


Final


drafts


submitted.


Same.


Fithure
their


1i. Procedure f
assigned essays.


followedd


students


they


completed













Number


Instructor


Instructor


2
(SAMPLE


ESSAY


ONE)


Discuss
became
"Gator.


how


Write


a Florida


narrative


about


memorable


event


your


life


Write


an extended


Same


(SAMPLE


ESSAY


TWO)


definition


"power,


" "beauty,


"success.


(SAMPLE


ESSAY


THREE)


Write
which
causes


an essay
discusses
, effect,


Same


both


effect


causes


some


phenomena.


(SAMPLE


ESSAY


FOUR)


Write
tive
favor


an argumenta-


Same


essa


or in


opposi-


tion to affirmative
action program.


Figure 2


Assignments


essays


collected


as data.


Essay










study.


In addition,


when


essays


were


collected


from


instructors,


researcher


received


whole


essays


each


subjects


section


without


picked


revealing


those


to the


of the


instructors


research


identities


of the


research


subjects.


In this


way,


biases


that


might


have


affected


instructors


interacted


with


subjects


were


precluded.


important


to note


that


instructors


differed


they


commented


on their


students


essays


in the


grading


poli


cies


they


applied.


Instructor


One


commented


detail


on all


first


drafts


assigned


each


one


a grade


which


would


rise


one-third


of a letter


grade


long


student-


writer


forth


a reasonable


effort


to revi


se.


example,


Instructor


assigned


a grade


of B to a first


draft,


final-


draft


grade


would


rise


to B+ if


was


sati


sfied


that


student


had


responded


to her


first-draft


comments


evidenced


an honest


effort


to improve


writing.


Lacking


this


response


teacher


comments


and/or


honest


effort,


one-third


grade


first


boost


draft s


would


with


be withheld.


grades


Instructor


comments


One


prior


always


to Step


returned


shown


Figure


Instructor


Two,


contrast,


commented


only


briefly


first


drafts.


did


not


assign


grades


until


students


had


fi ni haH


t-hca4


ranTi, ,n


nii 4 r~onn*tt 4: a7 -4rck4*


ntr~


Cr rr km: tCn rl


Cln~l


rT


i










both


sections


they


taught.


Neither


instructor


s grades


or comments


were


available


to the


holistic


scorers


rated


essays


this


study.


essays


were


required


to be


typed,


thereby


eliminating


problem


of handwriting


interfering


with


holistic


scoring


these


essays


Students


were


would


required


given (F

to hand


reedman,


xerox


1981;

copie


Charney, 1

s of their


984).

first


draft s


and


written


revision


guide


responses


on the


days


these


items


came


due.


This


procedure


was


a check


against


ess
f&GQ


being


lost


students


a way


of making


sure


they


completed


procedural


steps


each


essay


at about


same


time.


simplified


task


of collecting


data,


since


these


copies


were


handed


over


to the


researcher


instructors.


Description


of the


Two


Treatments


Students


wrote


responses


to revision


guides


each


their


essays


from


second


through


seventh


final


assignment.


Students


in the


self


-evaluating


sections


used


guides


to reflect


their


own


first


drafts


and


make


needed


revisions


before


submitting


their


final


drafts.


Each


student


peer-evaluating


sections


exchanged


papers


with


a peer


wrote

dyads

A..F A


revision


were


guide


chosen
-A 4A~


responses


researcher


n t.- J-


- a


that

for


- 4- - 4-


peer

each


- U- -


s papers.

assignment
- a _


Peer

; these










researcher


given


to each


instructor


prior


to Step


(see


Figure


each


writing


assignment.


After


students


written


their


revision


guide


responses


each


essay,


peer-evaluating


students


pairs


during


one


50-minute


class


period


to discuss


each


other


s first


drafts,


using


their


written


revision


guide


responses


as a basis


that


discussion.


evaluating


To balance


students


time


engaged


spent


cussions,


class


both


peer-


instructors


scheduled


a corresponding


50-minute


period


self-evaluating


students


to work


class


on their


own


essays,


using


their


own


revi


sion


guide


responses


to stimulate


decisions


to make


needed


revisions.


summary,


difference


between


treatments


was


that


peer


evaluators


exchanged


papers,


wrote


revision


guide


responses


conversation


each


about


other


each


s first


other


drafts,


s writing,


and


while


engaged


self-evaluators


wrote


revi


sion


guide


responses


their


own


papers


spent


time


in class


considering


themselves


what


revi


sions


to make


their


own


writing.


Treatment


of Ouantitative


Data


Holistic


Oualitv


Scores


In order


measure


relative


quality


students'










University


of Florida


Florida


Department


Education,


both


of which


administer


scoring


sessions


essay-


writing


tests


given


three


times


a year.


Following


procedures


used


these


universe


ity-


and


state-mandated


scoring


sessions,


researcher


holistic


scoring


sess


ions


on two


consecutive


weekends


January


1988.


difference


between


University


State


of Florida


scoring


sessions


ones


related


to the


present


study


was


scale


used.


In the


former


scoring


sessions


a 1-through


point


scale


was


used,


with


representing


lowest


quality


representing


highest


quality.


present


study,


an 8-point


scale


was


used.


Scorers


were


told


to divide


range


quality


represented


each


point


on the


4-point


scale


into


upper


and


lower


category.


There fore,


on the


8-point


scale,


scores


of 1


were


equivalent


to the


score


of 1


on the


point


scale,


so on up


scores


of 7


, representing


lower


and


higher


halves


of the


score


of 4


on the


4-point


scale.


using


this


8-point


scale,


researcher


hoped


to achieve


scores


which


would


more


precisely


represent


differences


quality


between


drafts


than


was


possible


using


4-point


scale.


Prior


to the


scoring


sessions,


research r


prepared


"range-finder


essays"


and


mixed


essays


to be


scored


into


1%fl+~~~~~r flna C'-t 4- Ccn a. -I 1 p4a


V^ rr


k~ C nk h n


CwAm C1*


rr,!


SAl~ Y


,A1 1 AAe A~ C


Y










batches


of Sample


Essay


One,


four


of Sample


Essay


Two,


on.


Each


batch


contained


between


15 and


essays.


Rangefinder


essays


are


sample


essays


used


holistic


scoring

scorers.


typically


to establish


Before


read


scoring


score


maintain

sessions


rangefinder


maximum

begin,


agreement

holistic s


essays,


compare


among


scorers


scores


they


give,


reached


discuss


regarding


disagreements


score


of each


until


a consensus


rangefinder.


When


acceptable


level


agreement


reached


maintained


least


three


consecutive


range finders,


scorers


are


ready


begin


reading


essays


that


will


receive


official


scores.


rangefinders


this


study


were


collected


from


students


in the


four


treatment


sections


been


selected


subjects


had


agreed


to participate


study.


research r


rangefinders,


searching


privately


through


scored


a large


essays


batch


collected


this


purpose


possible


until


of the


he found


eight


scoring


level


at each


of quality


as many


represented


point


scale


to be used


this


scoring.


researcher


was


able


to find


essa


at a minimum


of six


different


levels


each


four


assignments


involved


study.


Next,


years


a University


of experience


of Florida


as a holistic


graduate


scorer


student


independently


also


scored


rR nnraf4 nrcaa annA n' r a Qan-lAl


a~t a~


rbnn~r nknw


| |It "


nrF-% T-


n










rangefinders


was


established


between


this


reader


researcher.


When


holistic


scoring


session


this


study


convened,


researcher,


acting


as chief


reader,


distributed


rangefinders


first


four


batches


essays


to be


scored,


representing


one


four


assignments.


essays


from


each


assignments


were


scored


together.


order


which


four


assignments


came


before


readers


was


determined


random


lot,


scorers


were


aware


of when


semester


each


of the


assignments


was


written.


Such


knowledge


might


have


biased


their


scoring;


therefore,


precaution


random


ordering


was


taken


to deny


scorers


that


knowledge.


In addition,


no ess


contained


information


identifying


writer


or the


section


from


which


essay


came


or the


date


when


was


written.


Essays


were


identifiable


only


a 5-digit


code


number


written


on each


one.


After


randomly


scorers


rangefinders


selected


been


assignment


discussed


distributed


to be scored,


nature


of that


first


researcher


assignment.


scorers


then


scored


and


discussed


rangefinders


until


were


satisfied


that


agreement


was


an acceptable


level.


Scorers


submitted


their


scores


each


essay


entering


that


essay


s code


number,


followed


a numerical


score


between


I anrt W.~


flfl R Qi Ina ~ a a a -~l ,Z ~n %s-- --


n3 na


nn ~ Pt~n


Lnl*lr~


,|, *r%


n


~ LL










a particular


disagreement.


essay,


When


researcher


scoring


session


made


a note


ended,


these


of the

e discrepant


scoring


were


were


within


discussed,


points


of each


scorers

other.


negotiated


Of the


until

first


all

and


scores

second


drafts


included


in the


scoring


session,


or 18%,


required


that


scores


be adjusted


after


discussion.


A series


of Pearson


product-moment


correlation


calculations


was


performed


to determine


interrater


reliability


values.


Then,


scores


were


submitted


to a three-factor


analysis


of variance


(ANOVA),


with


two


levels


of treatment,


a draft


factor


representing


first


and


final


drafts,


a factor


representing


four


topics


of writing


assignments


collected


this


study.


Treatment


was


a between-groups


factor;


draft


and


topic


were


within-groups


(repeated


measures)


factors.


Figure


displays


design


employed


quantitative


treatment


of data


this


study.


As Figure


shows,


quantitative


design


this


study


was


a split,


split-plot


factorial


type,


with


treatment


(labeled


figure)


between-subjects


prepost


factor


(labeled


occasions


within-subjects


(labeled


factors.


Analysis


of Revisions


Another


type


of data


developed


this


study


was


derived


f r fi


n 1 naih-.-. 1 4: no


nnwnTn ri C


nfl nr -I-f $ T .-F-


4=4 -o+-


c: nal


rl r~ CC n


--Tt/"














































= Treatments
= Topics
= First and


Final


Drafts


= Subj


ects


Figure


Quantitative


research


design.










essays.


researcher


coded


and


counted


each


revi


sion


after


identifying


as belonging


one


of six


level


textual


complexity


see


Figure


Changes


from


first


to final


draft


punctuation,


spelling,


or capitalization


were


coded


as surface-


eve


1 revi


sion


Similarly,


other


revis


ions


were


counted


word,


phrase,


following


clause,


example


demo


sentence,

nstrates


and

the


multisentence


distinction


levels


between


clause


sentence


such


with


was


regard


delighted"


to thi


were


coding


deleted


scheme


from


If a sentence


an essay,


change


would


counted


as a sentence-


level


revi


sion


However,


sentence


"Because


fight


was


over,


was


delighted"


were


changed


to "Because


fight


was


over,


was


filled


with


joy,


" the


change


would


be coded


as a clause-


eve


1 revi


sion


case


of revisions


containing


more


than


one


level


same


time,


eac


h change


was


cons


idered


to be a single


revi


sion


was


coded


highest


level


involved


For


example,


oyed


several


sport


were


changed


to "he


enjoyed


fishing,


bowling,


and


hunting,


" the


change


would


coded


as one


phra


se-


level

not b


revision;


e coded


punctuation


as a separate


event


involved

In the


revision


case


of revi


would


sions


which


a unit


text


one


level


was


replaced


a unit


text


at another


level,


revis


ions


would


be coded,


one


deletion


other


new


unit


being


added


SFor


I u


a~ ~ ~~~~~~~~~~~~~/ ,, Att a~ l a a a.a- a. a0 na f .raAI


a v ~ rnn 1


hn~^~l


~I~L rl n hA~


r( ~h nwC n


In IIUn


II LIF~ Yn


A


1




















Meaning


Revisions


made


capitalization,


n punctuation,
or spelling.


Revisions


Revisions


Revisions


Revisions


Revisions


involving


involving


involving


involving


involving


single words.


single phrases.


single clauses.


single sentences.


multiple sentences.


Figure 4.


Revision


coding


system.


Code










added


clause


"whatever


sport s


were


fashion"


would


count


as one


clause-level


addition.


Finally,


multiple-sentence


substitutions


were


counted


revi


sions


, one


a deletion


other


an addition.


multisentence


purpose


level


was


of thi


to account


approach


instances


to coding


in which


relatively


short


passages


of text


were


replaced


relatively


longer


ones,


or vice


versa.


To check


consistency


and


accuracy


with


which


these


coding


principles


were


applied


to the


students


' essays,


researcher


consulted


with


a doctoral


student


in linguistics


University


of Florida,


reviewed


pairs


of first


final


drafts


coded


res


earcher


corroborated


correctness


consistency


with


which


revi


sions


were


assigned


to categories.


After


coding,


revisions


made


each


student


on each


four


sample


essays


were


totaled


and


recorded


on tally


sheets.


process


first


was


five


levels


straight-forward--each


textual


revi


sion


complexity,


counted


this


as an


individual


unit.


However,


case


of multisentence


revisions,


count


was


determined


on the


basis


"words


revision


example,


if 35


words


of multisentence


text


were


deleted


replaced


words


of multi


sentence


text,


change


would


be counted


as 85 words


of multisentence


revision


This


approach


to counting


was


adopted


to account


widely


-I .~ -


II I


I I


1










data


representing


revisions


were


then


submitted


to a 2


repeated-measures


ANOVA


test


to determine


whether


differences


existed


between


treatment


groups


at each


levels.


each


instructor,


a series


ANOVA


tests


were


run,


one


each


level


of textual


complexity.


account


different


lengths


of student


essay


revision


counts


were


converted


to "number


of revisions


per


words


first-draft


text,


" in


case


surface,


word,


phrase,


clause,


sentence-level


revisions.


multisentence


revisions,


counts


were


converted


to "number


of words


of revision


words


of first


draft


text


this


precaution


been


taken,


a 20


-word


multisentence


revision


a 200-word


multisentence


revi


sion


would


both


have


been


counted


"one


revis


ion,


" causing


obvious


distort ions


in the


data.


Since


these


series


of ANOVA


tests


were


being


applied


identical


data


each


instructor,


alpha


level


testing


significance


at each


level


textual


complexity


was


.05/6,


.0083.


Treatment


of Oualitative


Data


Revision


Guides


After


completion


their


first


drafts


students


were


required


to respond


in writing


to a set


of questions


and


-R q --


S


I r I


* \


,r I










audience


needs


and


overall


text


structure


they


decided


where


much


revising


was


called


for.


These


sets


questions


and


directions


were


called


"revision


guides


(See


Appendix


complete


revi


sion


guides


used


this


study


example,


revision


guide


corresponding


to the


seventh


final


writing


assignment,


which


called


an argumentative


essay


, included


following


items


Evaluate


controversy


writer


underlying


more) conflicting
clearly defined?


opinions
Explain


introdu


ces


essa


regarding


your


topic


answer.


effectively


support h
an explic
reasoning


is
itl


doe s


or her p
y stated


in the


effectiveness


cannot
extent
result.


find


writer


position?


use


Giv


application


essay.


of the


such


to which


logic


e an example
of logical


Evaluate


example


an example,
ou think th


you cho
briefly


essay


discuss
suffers


well d
sections


your


Does


oes


to his


answer.


writer


support
fault?


use


writer
or her


employ


or her


emotional


argument


counter


own


anticipated


arguments


emotional


argument?
effectively


language


Explain


your


language
not, is


does


to advance


answer


Explain


this


the writer


or her


Interview


Data


of September,


beginning


of November,


beginning


of December


1987,


researcher


interviewed


o4,,~~~~~ ~ ~ ~ L A.n-r *- --A.- -- -1-


as a


ose,


nCr~~nnC r,


r


*


*


|


r


k rl


1 ~urr


.









students


would


have


just


finished


writing


their


final


draft


Sample


same


Essay


One.


relationship


Essays


and


second


time


Three.


to the


students


third


students


had


interviews


' writing


received


bore


of Sample


their


final


grades


essays


they


discussed


interviews,


researcher


had


read


essays


prior


to the


interview.


researcher


followed


interview


structure


shown


Figure


interviews


completed


this


study.


Only


occas


ionally


did


he deviate


from


this


format


in the


interest


following


"flow"


of conversations


with


students.


target


questions--numbers


, 11,


and


13--were


ones


considered


most


relevant


to this


study


researcher.


responses


to these


questions


are


presented


Chapters


Four


Five.


remaining


to "set


target


questions


questions,


interviews


to lead


were


intended


interviewee


into


frame


of mind


which


or her


responses


to the


target


questions


would


emerge


as part


of a conversation,


without


being


highlighted.


repeating


target


questions


at monthly


intervals,


researcher


attempted


to elicit


respon


ses


which


might


indicate


changes


that


might


have


occurred


students'


conceptions


of revision


in their


reaction


to the


experimental


treatments.











First


Interview


On1y


What


have


been


main


influences


on you


as you


have


learned


how


to write?


Do you change
situations?


that


write


different


Interviews


about


writing


an essay?


finish


one


sitting?


(TARGET


first


QUESTION)


What


do when


revise


your


draft?


easier


to revise


as you


are


writing


or after


have


finished


a draft?


How


know


when


an essay


finished?


Beginning


from


time


were


given


ass


ignment


write
story


your


last


of how


essay
went


about


1101,
e task


ease


tell


of completing


assignment.


become


stuck


any


point


process


of writing


that


essay.


long


take


to write


that


essay


(TARGET
activity


QUESTION)


have


What


on the


effect


that


did


wrote


revi


that


guide


ess


What
made


grade


would


arrive


give


that


yourself


on that


essay?


What


grade


Second


Third


Interviews


(TARGET


write


STION)


as a result


anything


your


changed


experience


in the


in ENC


that


1101?










Reporting


of Oualitative


Data


researcher


examined


qualitative


data


originating


in the


specific


guide


revisions


responses,


made


and


in student


interview


essays,


responses


written


order


revision


answer


five r

to the


research

target


questions

questions


posed


Chapter


interviews


One.

were


Student


responses


paraphrased


tabulated.


responses


Then,


with


researcher


revising


compared


behavior


these


evident


target-question


final-draft


essa


and


written


revision


guide


responses.


Results


of the


qualitative


dat a


analysis


are


reported


in Chapter


Five.
















CHAPTER
QUANTITATIVE


FOUR
FINDINGS


This


chapter


contains


results


of quantitative


analyses


applied


to data


collected


from


both


instructors'


students.


findings


each


instructor


will


be presented


under


separate


headings.


each


instructor,


applications


of statistical


analyses


will


be discussed


first,


mathematical


operations


associated


with


a three-


factor


ANOVA


test


of significance


holistic


scores;


second,

tests c


mathematical


>f significance


operations

revisions


associated


at each


with


of six


ANOVA

levels


textual


complexity.


Analyses


of Holistic


Scores


Inter-Rater


Reliability


Instructor


Dne.


Pearson


product-moment


correlation


values


displayed


inter-rater


Table


reliability


represent


was


achieved


degree


ess


to which


written


Instructor


s students.


These


inter-rater


reliability


values


were


lower


than










Table 3


Inter-Rater Reliability Coefficients


Correlations)


for Raters Readina


(Pearson


Essays bv


Product-Moment


Instructor One


Students


Raters 1 and


Raters 1 and 3


Raters


and 3


1st Draft


2nd Draft


1st Draft


2nd Draft


1st Draft


2nd Draft


Sample
Essay
One

Sample
Essay


(p<.000)


(p<.000


(p<.004)


(p<.004)


(p<.017)


(p<.000)


(p<.054)


(p<.000)


(p<.008)


(p<.001)


(p<.009)


(p<. 000)


Two


Sample
Essay
Three

Sample
Essay


(p<.001)


(p<.001)


(p<.000)


(p<.000)


(p<.000)


(p<.000)


(p<.001)


(p<.000)


(p<.000)


(p<.000)


Four


(p<.001)


(p<. 000)


six coefficients


that


did not


exceed


.60 occurred for


Sample Essay One.


three


Furthermore,


raters were summed to


scores used in


because scores across the


create the holistic quality


subsequent analyses,


the discrepancies


scores across the


three


raters


was not


as critical as


different


raters had scored different


essays.


As Crocker and


Algina


(1986)


have pointed out,


as the number of


raters


a ~ ~ ~ ~ ~~~ m n r' ri n Ce 4- *I, aH C l ~ -.n


- LtalJ, .J -- .


0 ^^


r:nn ~kn










others


' scores,


assuming


that


their


scores


are


summed


averaged


to derive


a final


score.


In other


words,


infinite


essays


number


raters


written


this


could


have


study,


been


one


assembled


might


have


score


perfect


confidence


in the


summed


or averaged


scores


obtained


each


essay.


Because


only


three


raters


were


employed


in this


study,


sum


of the


three


raters


scores


may


have


been


highly


accurate


estimate


of a student


s true


score.


Instructor


Iwca


Table


displays


Pearson


product-


moment


correlation


values


comparing


three


raters


' scoring


essays


written


Instructor


s students.


Similar


to the


values


shown


in Table


these


results


were


lower


than


desired.


Again,


reliability


among


raters


was


lowest


Sample


Essay


One.


This


time,


however,


of the


coeffi


clients


were


less


than


.60.


Issues


related


inter-


rater


reliability


in this


study


will


be discussed


in Chapter


Six.


Descriptive


Means


and


Standard


Deviations


To avoid


a proliferation


of quantities


with


decimal


fractions,


quality


score


assigned


to each


essay


draft


was


sum


of the


three


scores


given


holistic


raters.


lowest


score


actually


earned


a draft


in the


sample


was


LA ~rat ra a 00 fl i nn nnlrlI f, -iJ 2-


a
A- -- -


nF 1


- -


~_1__1


Ct vrflr" W


.


rrl


-


, --


*rnkr"










Table


Inter-Rater Reliability Coefficients


Correlations)


for Raters Reading Essays by


(es PootMmn


Instructor Two's


Students


Raters 1 and


Raters 1 and 3


Raters


2 and 3


1st Draft


2nd Draft


1st Draft


2nd Draft


1st Draft


2nd Draft


Sample
Essay
One

Sample
Essay


Sample
Essay
Three


Sample
Essay


(p<.195)


(p<.016)


(p<.003)


(p<.003)


(p<.066)


(p<.008)


(p<.000)


(p<.010)


(p<.004)


(p<.031)


(p<.002)


(p<.026)


(p<.034)


(p<.000)


(p<.000)


(p<. 001)


Four


(p<.111)


(p<.000)


(p<.000)


(p<.000)


(p<.000)


(p<.009)


(p<.003)


.71
(p<.000)


avoid the


problems of


interpretation


inherent


in a


4-21


scale,


the researcher reduced all


combined


scores


points,


yielding the


1-18


scale


used in reporting results


this


study.


Such a


linear transformation of the data had no


effect


on analysis


results


which comparisons were made.


Instructor One


Table 5 displays the mean holistic


scores


and standard deviations


o a m t r Tar4 +-+ an


a'4-. A -n.-4-


for each draft


7 ~ 8.- --.J-- -


of each sample


I


(Pearson P ro duct-Mome nt











Table


Nean
1-18


Holistic


Scores


of Samnle


Essays


(Instructor


One).


Scale


Points


Sample Essay


Sample


Essay


Sample Essay


Sample


Essay


Three


Four


Draft


Draft


Draft


Draft


Draft


Draft


Draft


Draft


Self-
Evaluating
Students


Mean


11.0


(Std.


Dev.,


(2.4)


(2.8)


(3.5)


(3.0)


(3.7)


(4.1)


(4.2)


Peer-
Evaluating
Students


Mean


11.5


11.0


10.1


11.4


(Std.


Dev.


(1.8)


(1.3)


(3.7)


(2.9)


(3.3)


(3.2)


(3.7)


(3.2)


Instructor


Two.


Table


contains


means


standard


deviations


holistic


scores


earned


students


Instructor


Three-Factor


pro's sections.



ANOVA Results


I~~~~~~~~~~~~~~i rIL n n oa r. i ,-n .nf l -~--aL


Q Ir~r~ C!


h


rr


nC stl


~r~ CC o


n C n ~ mn 1 A


I


I1Y











Table


Mean
1-18


Holistic


Scores


of Sample


Essays


(Instructor


Two).


Scale


Points


Sample


Essay


Sample


Essay


Sample


Essay


Sample Essay


Three


Four


Draft


Draft


Draft


Draft


Draft


Draft


Draft


Draft


Self-
Evaluating
Students


Mean


10.6


10.6


10.8


11.2


(Std.


Dev.


(2.4)


(3.2)


(2.8)


(3.3)


(2.9)


(3.2)


(4.6)


(3.7)


Peer-
Evaluating
Students


Mean


10.5


10.5


11.1


11.4


11.8


13.7


(Std.


Dev.


(1.6)


(1.4)


(3.0)


(4.1)


(3.5)


(2.5)


(3.5)


self-evaluation


peer-evaluation


draft


.e.


first


final


drafts),


topic


.e.


four


essay


assignments


coll


ected


this


study).


In this


analysis,


group


was


considered


a between-subjects


factor;


draft


topic


were


within-subjects


factors.


rP~1%1 a


7 4- rr I n4- r\ r l aI C -r* %C a 4a',14 -o 4


One


n T-


rrnnC ~: n


raarllCF


nC Cka


Ckrnn,


nn d


I


n-











Tabi


Three-Factor


ANOVA


Test


of Significance


Comrarinan


Holistic


Scores


of Self-Evaluatincr


Peer-Evaluating


Students


Instructor


Pnft)


Sum of
Squares


Source


Degrees
Freedom


Mean


Square


F-Value


Probability


13122


Mean
Group
Error


.506


213.906
727.463


13122


.506


324.70


213.906
40.415


.0000
.0336*


Topic
Topic
Error


x Grp


97.319
48.519
656.788


32.440
16.173
12.163


.0568
.2743


Draft
Draft
Error


61.256


x Grp


.056


51.063


61.256


21.59


.056
.837


.0002*
.8896


Draft
Topic


11.769


Draft
Topic
Group


Error


.969


.923


.2718


.656


158.388


.8795


.933


*Significant


at alpha


significant


reflecting


a difference


This


result


may


quality


interpreted


scores


earned


self-evaluators


peer-evaluat


ors


averaged


over


drafts


topics.


However,


from


Table


apparent


that


a a


r ~I


r


1 .










result


does


indicate


a treatment


effect


favoring


peer-


evaluation.


main


effect


of the


Topic


variable,


that


of the


four


different


writing


assignments


represented


sample,


was


significant,


not,


taken


indicating


themselves,


that


have


writing


a significant


assignments


effect


holistic


scores.


main


effect


of the


Draft


variable


was


significant


.0002


, indicating


a distinct


general


improvement


quality


of the


average


final


draft


compared


with


quality


of the


average


first


draft.


This


result


was


apparent


from


inspection


of the


means


in Table


remaining


values


Table


represent


effects


interactions


between


among


variables.


this


case


none


of the


interactions


involving


these


main


variables


achieved


threshold


to determine


statistical


significance


.05).


There


no evidence


of a positive


impact


peer-evaluation


treatment


final-draft


performance


significant.


because


In concert


draft-by-group


with


interaction


significant


was


effects


Group


Draft,


only


appropriate


conclusion


to be drawn


is that

overall


the

and


two

that


treatment


this


groups


difference


differed

remained


in writing


fairly


ability


constant


C~~~~~~l n -'-


f,,,


r!,, L


-* f


I I .~


..