Glowing Review or Tough Love: Effects of Perceived Audience and Feedback Type on Self-Editing in Writing PDF Free Download
1 / 55/55
100%
Glowing Review or Tough Love:
Effects of Perceived Audience and Feedback Type on Self-Editing in Writing
An Honors Thesis Submitted
in Partial Fulfillment of the Elon University Honors Program
By Alyson J. Hignight
April, 2012
Approved by:
____________________________________
L. Kimberly Epting, Thesis Mentor
____________________________________
Alan Scott, Department of Psychology
____________________________________
Paula Rosinski, Department of English
Abstract
Self-editing allows us to monitor our communication with others to ensure that it is
appropriate and effective. Skinner (1957) proposed this judgment is based on past
experience with an audience’s reaction. A positive reaction advances the conversation
and reinforces the speaker’s behavior; negative reaction alerts the speaker’s monitor to a
problem. The message is re-phrased until a positive reaction is achieved. This process is
applicable to writing. In addition to feedback, a perceived audience of an expert versus a
peer evaluator may influence a student’s writing and editing behavior. This remains
largely uninvestigated among studies concerning self-editing; thus, this study investigates
whether feedback from an expert occasions differences in writing compared to feedback
from a peer, and if positive and negative feedback have different effects on writing. To
test these hypotheses, participants wrote two essays using a computer program that
measures writing behavior. They were told a professor, peer, or non-specified reviewer
would read and provide feedback on their essays; however, the feedback was randomly
assigned and delivered by the researchers. They received either positive or negative
feedback on their first essay before writing the second essay. Results showeda significant
effect of feedback type on the amount of time spent viewing the prompt as well as a
significant effect of feedback on the ratio of typed keystrokes to submitted characters.
Exploratory analyses also suggested interactions between audience and feedback type on
the number of keystrokes typed and the number of characters submitted. These findings
suggest that the intended audience and type of feedback affect how a writer self-edits,
offering implications for effective methods of education and potential future directions
for investigation.
Table of Contents
Background of Literature………………………………………………………………….1
A General Overview of Language……………………………………………………..1
The Behavioral Approach to Language..………………..……………………..………….2
Self- Editing......………………………………………………………………………..4
Self-Editing in Writing….…………………………………………………………...........7
Effects of Audience Type on Self-Editing.…………………………………….………….9
Effects of Feedback on Self-Editing….………………………………………………….15
The Present Study…………………………………………………………………….18
Method….………………………………………………………………………………..19
Participants……………………………………………………………………………19
Materials……………………………………………………………………………...19
Transcription Task……………………………………………………………….….…...19
Writing Task…..………………………………………………………..…………..……20
Procedure………………………………………………………………………………...23
Results……………………………………………………………………………………24
Discussion………………………………………………………………………………..31
Effects of Feedback…………………………………………………………………...31
Interaction Effects…………………………………………………………………….34
Limitations……………………………………………………………………………37
Future Implications……………………………………………………………....…...41
References………………………………………………………………………………..45
Appendices……………………………………………………………………………….51
Acknowledgements
I could not be more grateful for my thesis mentor, Dr. Kimberly Epting, for being
a constant source of guidance throughout my college career and the completion of this
thesis. My education and growth have been strongly influenced by her sharp intelligence,
stimulating questions, and unfailing support. I have learned a great deal from her and I
am proud to have completed this project under her mentorship.
I am also grateful for my fellow SCOPE laboratory members, Brittany Bowers,
Jennifer Cox, Hayley D’Antuono, and John Hollander for their assistance in data
collection, their presence and support at several presentations and conferences, and their
valuable insight and encouragement throughout the production and completion of this
thesis.
I would like to thank Duke Hutchings for writing the software program used to collect
data for this study. Thanks is also due to Professors Alan Scott and Paula Rosinski for
being a part of the thesis defense committee and process; the time they volunteered to
evaluate this thesis is very much appreciated.
The Elon Honors Program and Undergraduate Research Program made this thesis
possible; the resources and assistance offered by both programs have greatly facilitated
the process and aided in several presentations.
Self-Editing in Writing 1
Background of Literature
A General Overview of Language
Language is a phenomenon that has garnered the interest of scientists, scholars,
and the average person for centuries. It is a tool that allows us to transmit ideas, unite in
community and culture, and benefit from the knowledge of others. Through language, we
are able to identify and understand what we encounter in the world, allowing us to make
sense of our surroundings and to predict what may occur in our environment; moving
beyond its practical functions, it also evokes emotion whether through poetry, metaphors,
or humor. Language is one of the most crucial features that sets humans apart from any
other creature on Earth (Deacon, 1997; Ratner, Berko-Gleason, & Narasimhan, 1998).
Yet for all of its complexity and importance, it is a process that even toddlers are able to
manipulate. As Steven Pinker states, “the ability comes so naturally that we are apt to
forget what a miracle it is” (p. 1, 1994).
Language exists in countless forms and captures our interest for its dual nature of
being both universal and idiosyncratic. Not only does it exist in every culture and in
every region of the planet, but it also forms in linguistically isolated communities.
Studies have shown that children, even without formal instruction, are capable of
constructing their own language complete with vocabulary, grammar, and slang
(Coppola, 2002). Humans possess an innate ability to create a structured and fully
functional mode of communication on their own. It is not necessary to teach a child to
communicate in order for him to have the ability to do so; however, that is not to say that
Self-Editing in Writing 2
language develops without a significant amount of feedback and influence from one’s
environment.
The Behavioral Approach to Language. As with all human behavior, language is
subject to factors presented by a certain situation, including one’s culture, past
experiences, future expectations, intent, reactions of others, and so on. Cognitive
psychologists, who define language as a mental process that forms naturally, often
dismiss behaviorism’s approach to language; however, the study of language can benefit
from both perspectives (Epting & Critchfield, 2006). While a cognitive approach
introduces the important roles of attention, perception, and memory in language
production, it is essential to consider the impact of environmental factors and their
influence on our language development and production (Ludvig, 2003; Schlinger, 2002).
Skinner discusses how language is shaped behaviorally in Verbal Behavior (1957),
stating that spoken language is met with positive or negative reactions from the
environment. The environment includes one’s audience, and as a child, this is frequently
a parent. For example, when a child exclaims, “Look at the gooses!” the parent may
respond with, “Yes, look at the geese,” thus correcting the child’s grammar and
reinforcing appropriate verbal behavior. Through numerous exposures to the shaping
effects of reinforcement and punishment, children develop verbal behavior appropriate to
their society, their native language, and for various social situations (Epting &
Critchfield, 2006; Fishbach, Eyal, & Finkelstein, 2006). The individual continues to add
to his or her linguistic repertoire in novel situations, such as when a child begins school
and forms a peer group. Given the variety of situations that one encounters both on a
daily basis and throughout a lifetime, approaching language from a behaviorist standpoint
Self-Editing in Writing 3
has many benefits. We make observations that reveal what influences language, what
shapes it, and how we may manipulate it. The ability to observe how different
environments influence verbal behavior enables us to measure behavior and adjust
conditions of the environment to gain further insight. From these observations we may
infer mental processes that underlie language.
Through such observations, one conclusion is made clear: errors often occur when
we speak. For most people, language comes so naturally that at times it eludes our
“mental censors,” necessitating an active and refined self-monitor (Pinker, 1994).
Spontaneous verbal language is rife with disfluencies (the use of “um,” “uh,” etc.),
pauses, stutters, and other components that detract from the intended fluency of our
speech. Fox Tree (2001) suggests that disfluencies play a role in speech production and
comprehension by giving the speaker more time to plan what he or she is going to say
and to alert the listener that the speaker is not finished communicating; nevertheless,
prolonged or frequent disfluencies tend to detract from the intended message and are thus
viewed as errors. Other issues stem from the listener not comprehending what is said.
Issues of ambiguity, vagueness, or speaking incoherently result in breakdowns in
communication (Pinker, 1994). Regarding these speech errors, we as speakers are alert to
their occurrence (Harley & MacAndrew, 2001) so that we may communicate effectively.
The reaction of listeners indicates whether our message has been effective; based on this
feedback, we adjust our behavior until we reach the intended effect. This helps develop
the process of self-editing, a crucial component of verbal and written behavior that is
affected by variables such as audience type, audience reaction, and feedback.
Self-Editing in Writing 4
Self-Editing
The process of self-editing allows us to monitor our speech production to ensure
that it is appropriate and effective. Speech production is commonly separated into four
stages: conceptualizing, formulating, executing, and monitoring (Epting & Critchfield,
2006). Conceptualizing is the act of developing an idea or message that is relevant for
communication. This may be the initial message in a conversation or during a
conversation in response to another’s verbalization. The next step, formulation, involves
structuring the most effective way to deliver the intended message. This step considers
syntactic components and pragmatics. Execution is the actual delivery of the message
(speaking). Monitoring, which includes the crucial component of self-editing, judges
whether or not the message had the intended effect on an audience.
According to Skinner (1957), self-editing is the speaker’s examination and
alteration of verbal behavior. This examination, which is conducted by the monitoring
process (the proposed fourth step of speech production listed above), is initially based on
the audience’s reaction (Vaughn & Michael, 1982). A positive reaction (e.g., a smile or a
nod) allows the conversation to move forward and reinforces the speaker’s behavior for
future situations. A negative reaction (e.g., a frown, silence) alerts the speaker that there
is a problem. When this happens,the message is sent back through formulation and
execution, where it is re-phrased until a positive reaction is achieved. By reacting to both
positive and negative past reactions, the speaker refines the monitoring process and
builds this self-editing skill through an ever-growing collection of interactions with
various audiences. For example, an individual speaking too loudly in a place meant for
quiet, such as a library or a church, may be admonished by others and encouraged to
Self-Editing in Writing 5
speak more quietly in the future (Epting & Critchfield, 2006). Other examples that often
lead to punishment include swearing, lying, rude or crass comments, semantic errors, and
grammatical errors (as illustrated in the previous example of gooses being corrected to
geese). With each new environment, more opportunities are available to reinforce and
punish verbal behavior so that self-editing is refined and, eventually, automatic.
Over time, self-editing evolves from relying solely on the audience’s reaction to
relying on the speaker’s internal monitor to determine if a message is effective; it is
gradually learned as a result of the individual’s history of punishment and reinforcement.
It is a function of automatic reinforcement, which is a result of one’s own behavior and
does not include another person’s feedback (Vaughn & Michael, 1982). Rommetveit
(1970) viewed self-editing as “anticipatory decoding” to emphasize how the speaker
learns to predict, or anticipate, how the external listener will decode a message. With this
skill, a message may be suppressed mid-sentence or rejected by the speaker before it is
released rather than adjusting communication that has already been produced (Epting &
Critchfield, 2006). Self-editing, then, encompasses the process of becoming a functional
and savvy communicator in the world. It allows us to learn, through operant conditioning,
which verbal behaviors are most effective given specific circumstances so that we are
prepared in the future. Without self-editing, every new social encounter would lack the
aid of past experiences and our communication would remain ambiguous and awkward.
It is through self-editing that we become our own best audience (Vaughn & Michael,
1982). Hyten & Chase’s definition (1991), in line with Rommetveit’s concept of
anticipatory decoding, emphasizes how self-editing enables a person to be a self-listener
by anticipating an audience’s reaction and adjusting behavior accordingly: “the process
Self-Editing in Writing 6
of modifying one’s own verbal responses, or the products thereof, to make them more
appropriate to the social context or more effective in controlling the behavior of the
listening audience” (p. 67).
A recent study by MacDonald, Johnson, Forsythe, Plante, & Munhall (2011),
illustrates the development of self-editing and advances the notion that it emerges with
environmental input. In this study, adults, four-year-olds, and two-year-olds said “bed”
repeatedly while hearing themselves through headphones. The researchers modified the
playback so that participants heard themselves saying “bad” instead of the intended
“bed.” Upon hearing this error, adults and four-year-olds immediately adjusted their
verbal behavior until they heard the intended message, “bed,” in the playback. Two-year-
olds did not adjust their verbal behavior; instead, they ignored their erroneous speech and
could not detect the mistake on their own. If self-editing were a purely innate mechanism
then two-year-olds should be just as adept at monitoring their speech as the older
participants. The absence of a functional self-monitor in two-year-olds suggests that it is
a mechanism that must be formed and refined through experience. Although the disparity
between the two-year-olds and older participants may be partially attributed to brain
development, self-monitoring as a process is subject to the environment’s influences and
can be observed in everyday interactions between children and adults, such as in the
earlier example of “gooses” being corrected to “geese.” Over time, the toddler does not
need to rely on external feedback to determine if her speech is correct, and develops her
own internal self-monitoring mechanism. As this study indicates, children have
developed a self-monitoring mechanism by approximately age four. Self-monitoring
Self-Editing in Writing 7
continues to develop with more experience, and is refined according to the intended
message and various environmental factors such as audience type.
Self-Editing in Writing. The process of self-editing in verbal behavior is highly
applicable to writing and improving one’s writing ability. This is very important to
studying self-editing, since more of writing can be made an observable, overt process.
Verbal behavior is more difficult to measure due to our inability to observe how one
mentally edits a statement.By studying self-editing in writing (revision), we can
investigate similar processes in verbal behavior. Here it should be clarified that although
the terms editing, revision, reviewing, and planning have been distinguished from each
other in composition research, the behaviorist view may approach these processes as one
response class because they share the function of improving the effect of written material
on the reader. (Epting & Critchfield, 2006; Epting, Gallena, Hicks, Palmer, & Weisberg,
under review). Thus, here we use “self-editing” to refer collectively to the range of these
factors.
Commentary on a written piece, usually from an expert (most often a teacher in
academic contexts) improves a written piece and advances skills, such as self-editing, that
will help a student evaluate his or her writing in the future (MacArthur, 2007). As with
verbal behavior, external feedback is delivered until a self-monitoring process is
developed and the writer is able to write skillfully and effectively; just like we become
self-listeners regarding verbal behavior, we become self-readers when writing. Skinner
(1969) suggests that as with verbal behavior, writing offers automatic reinforcement since
we are able to actually see ourselves progressing through a piece, how the sentences flow
from one to the other, and the physical final product; we participate as readers during the
Self-Editing in Writing 8
writing process. Willingham (1990) states that the goal of feedback, through
reinforcement and punishment delivered in commentary, is to improve future writing. He
contends that the crucial point in providing feedback is to encourage the student to be his
or her own editor, or a self-reader. In other words, the best way to improve a student’s
writing for the long-term is to encourage the self-editing process. If a writer can read a
piece of written work with the intended audience in mind, it will increase the
effectiveness of the work and the author's writing skills. This is supported by Flower,
Hayes, Carey, Schriver, & Stratman (1986) in their detailed review of revision and its
underlying processes. They point out that simulating anaudience’s response to a written
work helps a writer diagnose problems; eventually, this ability to diagnose issues
becomes automatic. Diagnosis is more involved than merely detecting that a problem
exists (e.g., realizing that a particular sentence seems “off”). It pinpoints what exactly is
wrong with a part of the text and what measures should be taken to improve it (realizing
that a sentence is too wordy and would be more effective if it were broken down into two
separate sentences). The ability of a writer to do this internally is an accurate predictor of
how well his or her ideas will be communicated. This link between self-editing in spoken
behavior and in revision is clear: adjusting behavior in response to an audience’s reaction
is crucial to developing a reliable, internal self-monitor over time.
Flower et al.(1986), in the same review, call revision a “powerful, generative
process” (p. 16). It is a mechanism that is constantly evolving as it interacts with one’s
environment and personal experiences; it is informed by both sources of knowledge to
effectively evaluate writing, ensuring that the intended message is accurately represented.
Revision, then, is affected by a writer’s knowledge of a topic, his or her mental
Self-Editing in Writing 9
representation of the topic, and personal standards of writing quality. Sometimes a
writer’s mental representation of an idea and the actual text used to explain or describe it
is incongruent. When this occurs, issues of ambiguity arise, as the writer has failed to
convey the mental representation in a way that makes it as clear to the audience as it is to
the writer. Hence the goal of revision is to make the written piece offered to the audience
more congruent with the intention of the writer (Della-Piana, 1978). This skill is
developed, refined, and adjusted in response to varying audiences and environments. The
aspects of a paper that are approved through positive feedback will be reinforced for the
author, while aspects of a paper that are punished through negative feedback will be
avoided in the future. The writer will tailor his writing behavior in such a way that it will
be rewarded and avoid punishment. Revision is an adaptive process, a useful term as this
aligns perfectly with the already-introduced behavioral principles that have been applied
to speech.
Effects of Audience Type on Self-Editing. While feedback on written assignments
is highly beneficial to students, it is not always an ideal exercise for the evaluating expert.
Reading an essay thoroughly enough to provide comprehensive advice is time-consuming
for a teacher, especially with many student papers to grade. This can be a problem when
planning lessons and budgeting time, and also puts the quality of the feedback itself at
risk. As a result of being pressed for time, teachers may have little choice but to leave
brief, sometimes vague comments on papers instead of in-depth responses.While this
may improve the paper at hand, it does little to advance students' writing ability for the
long-term. Often, instructors reduce or stop assigning written tasks entirely. To deal with
this issue, peer review and feedback is often used as an alternative or supplement (Cho &
Self-Editing in Writing 10
MacArthur, 2010). This raises questions of whether peer review has a different influence
on writing and editing processes, given it is from a different source.
Indeed, studies have shown that peer review is qualitatively different from expert
review. This is largely due to different bases of knowledge. Experts, with knowledge
specific to their discipline, recognize errors faster (Chiesi, Spilich, & Voss, 1979), can
consult their own memory instead of external sources when reviewing a paper (Besnard
& Bastien-Toniazzo, 1999), are better at solving problems, and are more likely to
experiment with the text rather than trying just one solution to an issue (Hull, 1983).
Their knowledge is more organized and they generally have more of a strategy when
responding to a paper and giving feedback (Chi, Feltovich, & Glaser, 1981; Anderson &
Lebiere, 1998). Additionally, experts often employ heuristics and technical jargon in their
feedback due to their extensive experience. By contrast, peer review tends to include
more colloquial, age-appropriate language and lacks the technicality of expert feedback,
which often includes esoteric terms and concepts that may not prove as useful to non-
experts seeking assistance.
Although peer review differs from expert review, this does not mean it cannot be
as effective. Willingham (1990) even reminds educators to think about the student’s
perspective when providing feedback; otherwise, the comments may not be as useful.
Peer review naturally includes the perspective of the student, and may result in
commentary better suited to the writer’s level of understanding. Evidence for this idea
has been provided by Cho, Chung, King, and Schunn (2008) in which students rated peer
feedback on questions related to technology as more helpful than expert feedback
because the novice reviewers were easier to understand. Falchikov and Goldfinch (2000)
Self-Editing in Writing 11
conducted a meta-analysis of 48 peer assessment studies and found that correlations
between peer ratings and expert ratings were strongest for assessments of the overall
quality of a paper as opposed to assessments of the individual dimensions of a paper.
Additionally, correlations were strongest for papers concerning academic processes
(involving topics such as classes, school, education, etc.) as opposed to professional
papers (involving more specific topics). The authors suggest that this may be due to
students’ familiarity with academia, supportingfindings by Chiesi et al. (1979) that
differences in peer and expert knowledge bases affect how they evaluate papers. On the
other hand, experts provide new information to writers, pushing them to learn new
concepts and improve their skills more than the peer reviewer may be able to, again due
to different knowledge bases. Thus, while the peer’s feedback may be more relatable to
the student, it may not be challenging them enough. Zhu (1995), however, found that
students may be taught to provide specific and relevant feedback to their peers as well as
teachers can, given formal instruction on how to best read and respond to papers.
Beach (1979) also investigated self-editing in writing, emphasizing that
experienced writers, compared to novice writers, possess a greater understanding of the
importance of self-editing in communicating effectively. In Beach’s study, high school
students composed essays over the course of three weekly writing sessions. After each
session, the draft was evaluated: a teacher evaluated the draft, no one evaluated the draft
(the control group), or the draft was self-evaluated by the student. All measures of
evaluation were conducted using the same rubric. Results showed the drafts evaluated by
the teacher had a significantly higher degree of change from the initial draft to the final
version than the other two conditions. Beach explains this finding by discussing the lack
Self-Editing in Writing 12
of self-editing (which he terms “self-evaluation”), stating that “The teacher’s evaluation
gives students another reader’s perspective on whether or how well the intended meaning
has been communicated. Without that external perspective it may be difficult for students
on their own to recognize whether their intended meaning has been communicated” (p.
117). For both verbal behavior and writing behavior, reinforcement and punishment
provided by an audience encourages effective communication, with the development of
the speaker/writer becoming a self-editor as the ultimate goal. Time and experience are
necessary for individuals to fully develop and refine their internal monitor that allows
self-editing to function.
Cho and MacArthur (2010) emphasize that although studies show that peer
review is an adequate means to give feedback, its ultimate effect in improving the writing
process remains largely uninvestigated. Skinner (1957) also argues that the type of
audience can affect the degree to which people edit. A salient example is illustrated when
comparing a typical conversation with an employer and a conversation with a close
friend. It is not likely that one edits his or her speech as much when talking to the friend;
this tends to be a less formal interaction. The friend is likely to be more tolerant of speech
errors than the employer, a probability learned over time by the speaker. The same holds
true for writing, such as when one writes to a boss versus a friend. If the intended
audience has an impact on the speaker’s or writer’s behavior, then a perceived audience
of an expert versus a peer evaluator may affect how a student writes and edits both before
and after feedback. That is, the writing process and the quality of a written piece may be
influenced by the author’s knowledge of the person giving the feedback, and not just the
feedback itself. In Cho and MacArthur’s (2010) study, a class of undergraduates
Self-Editing in Writing 13
participated in a double-blind study in which they submitted essay drafts which were
reviewed by a single expert, a single peer, or multiple peers. The researchers were testing
whether or not peers could provide commentary that was of the same quality as the
expert’s, and if the feedback from the different sources led to different types of revisions.
They found that feedback from multiple peers led writers to improve their essays the
most due to the numerous perspectives and comments. However, the authors point out
that if the writers had known that an expert was providing the feedback, their writing
behavior may have been impacted by this knowledge to some degree; in fact, they state
that it is likely that the perception of an authority figure would have affected their writing
behavior. In line with behavioral approaches, the data suggest that the perceived audience
has an impact on one’s writing behaviors regardless of the quality of feedback. This
phenomenon may also have influenced Beach’s (1979) results such that the increased
revision in teacher-evaluated, in contrast to self-evaluated, essays may be due in part to
the perception of the teacher as a credible authority figure and therefore someone to
impress. Placing more emphasis on who is providing feedback is an important avenue to
investigate, as it inspires still-unanswered questions of the influence of audience type on
learning and behavior.
Other studies outside of writing have suggested that feedback from a person of
authority will be revered more than if it is a peer providing advice. Nobes and Pawson
(2003) asked children to evaluate rules mandated by adults and by other children. They
heard stories in which a moral, a cultural convention, an adult-invented convention, or a
child-invented convention was broken by either an adult agent or a child agent. The
children had to determine and explain which rules should always be followed and which
Self-Editing in Writing 14
rules could be observed more leniently. Conventions created by an adult, the typical
authority figure, were considered to be less open to alteration or transgression than child-
invented conventions. Additionally, children were deemed to have as much authority as
adults to break and alter child-invented rules. When it was their peers inventing the rules
– not a figure of authority such as the adult – children had more freedom to transgress
from or alter the rules. These findings suggest that we tend to feel more comfortable
breaking rules made by those who share the same level of authority as we do. This can be
explained by considering histories established by punishment and reinforcement.
Breaking a rule mandated by an authority figure generally has more severe consequences
than breaking a rule mandated by a person who does not have authority, where there may
not be any consequences at all. Therefore, ignoring or transgressing from an authority’s
rule is punished more effectively, and this behavior occurs less often. Applied to the issue
of peer versus expert review, students may be more likely to ignore or deviate from
feedback given by other students than feedback from the authority figure; expert review,
as suggested in Nobes and Pawson’s study, may be considered more seriously since
following an authority’s advice has been reinforced.A study by Laupa and Turiel (1986)
provides further insight into how rules mandated by different sources are perceived. In
interviews, they presented children with hypothetical conflicts between two children. In
these scenarios, a peer authority, an adult authority, or an adult non-authority determined
a resolution to the conflict. The participants were asked to judge which resolution should
be applied to the situation. The results showed that children were more likely to follow
rules given by an adult authority rather than a peer authority, but were more likely to
follow rules given by a peer authority than an adult non-authority. Ultimately, authority
Self-Editing in Writing 15
rather than age was the crucial factor in adhering to rules. In a college setting, the person
with higher position is the professor rather than fellow students. Based on these studies, it
is likely that feedback on writing would be placed in higher regard and have more – or at
least a different – influence on behavior when it is from a professor compared to when it
is from a student.
Effects of Feedback on Self-Editing. If the perception of feedback and its effect
on writing behavior is influenced by who is providing it, perhaps the type of feedback
itself also plays an important role. We have seen how the designation of an authority
figure versus a peer may bring about different results; perhaps positive feedback versus
negative feedback will also have different effects on writing behavior. Willingham (1990)
advises that reviewers, when commenting on a student’s paper, should be sure to include
positive comments, suggesting that positive feedback has more benefits on editing and
writing. Hyten and Chase (1991) offer empirical evidence for feedback’s effect on self-
editing. In this study,students composed a series of eight short essays on computers in
response to questions provided by the experimenters. What the student typed was also
displayed on the researchers’ computer in another room, allowing the researchers to see
not just a final written product but observe how that product was systematically
composed; hence, they were able to observe self-editing. The participants did not know
that their writing was being observed, ensuring that editing remained as natural as
possible. After submitting their answers, participants received randomly assigned
feedback stating that they were either correct (the comments from the researcher were
positive, and they received points for good answers) or incorrect (the comments were
negative and they lost points). Their editing behavior was measured on subsequent
Self-Editing in Writing 16
essays, with the result that self-editing increased in response to negative feedback and
decreased with positive feedback. Participants tended to type answers more slowly and
waited longer to submit a final product in response to negative feedback, suggesting they
monitored their essays more meticulously. They also wrote more characters overall when
their writing behavior resulted in negative feedback, and they were more likely to delete
already-composed material in order to write an entirely different answer upon receiving
negative feedback. Hyten and Chase’s study thus provides strong evidence that negative
feedback increases self-editing overall.
Outside of writing and revision, Fishbach, Eyal, and Finkelstein (2010) reviewed
studies that explored how positive and negative feedback motivate goal pursuit, finding
that the most effective type of feedback largely depends on the individual under review.
Positive feedback is most effective for novices (which are often students) as they tend to
seek encouragement while pursuing a goal. Novices must evaluate their commitment to a
goal before adhering to it; positive feedback tends to boost this commitment. Negative
feedback is most effective for experts since they seek to maintain the skills they have
already achieved. They have already committed to the goal and wish to track their
progress, balancing their current knowledge with skills refined through criticism. Positive
feedback, then, boosts commitment to a goal while negative feedback boosts progress
(Fishbach, Dhar, & Zhang, 2006). In an academic setting, novices may be first-year
students and experts may be students in their final year; the difference in how each group
responds to feedback has yet to be extensively studied. Van Duijvenvoorde, Zanolie,
Rombouts, Raijmakers, & Crone (2008) developed a task in which participants responded
to either the color or shape of a stimulus. Participants were grouped by age: 8- to 11-year-
Self-Editing in Writing 17
olds, 12- to 13-year-olds, and 18- to 25-year-olds; they received either positive or
negative feedback after each attempt, which was delivered randomly, regardless of their
actual answers. All participants performed more accurately and faster on trials after
positive feedback, with the youngest age cohort having the most difficulty with learning
after negative feedback. This suggests that positive feedback may be more effective in
learning, and that the impact of feedback is influenced by the age and development of the
person receiving the feedback. The findings also support Hyten and Chase’s result that
negative feedback slows performance.
Valence of the feedback may also have an impact on a behavior; that is, different
degrees of criticism or praise may result in different levels of editing. Kleivin (1973)
measured how much participants remembered from observing a speaker. There were
three conditions of the speaker’s opinions and the valence of his opinion: a low-level
condition, in which the speaker used a monotone voice, few adjectives, and paused often;
a medium-level condition, in which the opinions were delivered in a straightforward,
normal tone of voice and without excessive emotion, and a high-level condition, which
included more opinions that were delivered forcefully and loudly. Participants
remembered more content and details in the medium-level condition, suggesting that the
way information is delivered has an impact on what we retain. Boring or harsh exchanges
will be remembered in less detail than a straightforward exchange. This may mean that
the most effective feedback, whether it is negative or positive, should be informative
enough to remain interesting but should be neither overly laudatory nor excessively
harsh. Based on these studies, it is clear that the nature of feedback and its effect on
various tasks, including self-editing, remains inconclusive. Finding more evidence for the
Self-Editing in Writing 18
effect of feedback on editing processes could hold important implications for how written
and verbal skills are developed from a psychological and educational standpoint.
The Present Study
The purpose of this study was to investigate whether feedback ostensibly from an
expert occasioned differences in writing and editing processes as compared to feedback
ostensibly from a peer. Further, this study investigated if positive feedback had a
different effect on editing than negative feedback. It was hypothesized that the type of
feedback would influence self-editing behaviors, and that a professor’s feedback would
affect self-editing differently than a peer’s feedback. It was also hypothesized that the
specific effects of feedback would depend on the reviewer. The present study intended to
further our understanding of the effects of praise and criticism on writing and editing, as
well as the degree to which authority influences these processes. As stated previously, the
influence of different sources of reinforcement on writing quality and the editing process
remains largely uninvestigated. Attention has been given to the self-editing process but
not to how it is influenced by a perceived audience or feedback (with the exception of
Hyten & Chase, 1991, who investigated the effect of feedback), making this a unique
study. Additionally, this study examines the components of individualized writing and
real-time editing (e.g., pauses, substitutions, deletions, etc.), which have not been
extensively examined previously in this way.
Self-Editing in Writing 19
Method
Participants
This study included 193 college undergraduate volunteers at Elon University; 41
were male and 152 were female. There were 98 freshmen, 66 sophomores, 17 juniors,
and 12 seniors. Some may have received credit for a class, and all were offered a gift card
worth five dollars. All were native English speakers.
Materials
Participants used a computer running Windows 7 to complete a transcription task
and then a writing task to write two separate essays.
Transcription Task. Before writing their first essay, participants completed a
transcription task to measure their individual typing style. This provided a baseline
measure of normal typing behavior and established an individualized criterion for what
was counted as a pause in the subsequent writing task. Six passages of one to two
sentences were presented one at a time to the participant. For each trial, the entire passage
appeared on the screen. Participants read the sentence(s) and then reproduced the
sentence(s) in a text box. They were able to edit their work as needed so that their
sentences exactly matched the presented sentences. They submitted each passage and
then clicked NEXT to initiate the next trial. The passages were presented in random order
across participants.
As a participant typed, the program recorded every keystroke, inter-keystroke-
interval, and any typographical error corrections. Because the participants merely copied
a sentence in this task, inter-keystroke-intervals reflected their natural typing speed
Self-Editing in Writing 20
without interference from additional thought processes (e.g., thinking of an answer to a
question or making up a sentence of their own). The 99th percentile inter-keystroke-
interval designated pause criterion for the subsequent writing task. This meant that all
hesitations except those exceeding this percentile were considered part of the writer’s
normal typing style; any hesitation that exceeding the 99th percentile was measured as a
pause.
Writing Task. Participants were asked to write two separate essays, each at least
two substantial paragraphs long. They were informed that the essays would be in
response to an open-ended prompt, and that their essays would be evaluated by a
reviewer in another room. They were randomly assigned to an audience condition and
told that a student, a professor, or a non-specified reviewer would give feedback based on
a numeric rubric with accompanying comments:
1 – Poor: “This answer is vague and does not completely answer the question; it requires
improvement.”
2 – Unsatisfactory: “This answer contains some relevant information, but lacks clarity
and detail; it could be improved.”
3 – Satisfactory: “This is a good attempt at answering the prompt; it is a bit basic but
contains relevant points.”
4 – Excellent: “This is a very good answer. It is clear, well-developed, concise, and
answers the prompt very well.”
Self-Editing in Writing 21
Although participants were aware of scores from 1-4, they were randomly assigned to
receive either a 1 - Poor score or a 4 - Excellent score in order to compare the effect of
the most negative feedback and the most positive feedback. The writing prompts were:
a) What does the term "responsibility" mean to you? Please write about a personal
experience that illustrates this term in at least two complete paragraphs.
b) What does it mean to you to be pushed outside of your comfort zone? Please write
about a time when you experienced this in at least two complete paragraphs.
These prompts were chosen because they are broad enough for participants to
write a substantial amount; a narrow prompt may have made it more difficult for them to
answer. They were concepts familiar to the average undergraduate student, so it was
unlikely that a participant would not have a response. Additionally, with such broad
topics it is easier to give arbitrary feedback regardless of the answer. Because these
questions have no absolute correct or incorrect response, it was easier to maintain the
deception and make the feedback seem credible. Hyten & Chase (1991) also tasked their
participants with defining and giving examples of abstract concepts, setting a precedent
for the selection of prompts for this study. The order of the prompts was randomized.
When participants began writing, the prompt disappeared; however, it was viewable any
time by clicking and holding down “SHOW PROMPT.”
As participants wrote their essays, the writing program recorded keystrokes, inter-
keystroke-intervals, and error corrections. The dependent variables constituted
components of self-editing, including:
Self-Editing in Writing 22
a) Reviews/planning: number of times the prompt was viewed and how long
participants viewed the prompt (in milliseconds).
b) Pauses: number, average length, and elapsed time in milliseconds between the last
keystroke and final submission of the essay.
c) Edits: deletions (pressing the “delete” key or “backspace” key), substitutions
(highlighting the text and typing new material over it), insertions (clicking within
the text and typing), and total number of edits. The edits were further separated
into two categories: edits that immediately followed a pause and edits that did not
immediately follow a pause. They were converted into ratios of edits out of total
characters to obtain an accurate representation of how much participants edited in
relation to the length of the essays.
d) Overall content: total number of characters submitted, total number of keystrokes
typed, ratio of keystrokes to characters (how much the writer “fiddled” with the
text throughout the writing process), and total time spent writing.
After participants received the feedback for the first essay, they began the second
essay; again, the keystrokes and inter-keystroke-intervals were recorded. By comparing
the editing measures during the first essay to the editing measures during the second
essay through difference scores (post-feedback essay minus pre-feedback essay), the
effect of the perceived audience (professor vs. peer vs. unknown reviewer) and the effect
of feedback (positive vs. negative) on self-editing were analyzed.
Self-Editing in Writing 23
Procedure
Participants were seated in a room with a table, a chair, and a computer that ran
the program. They completed the transcription task and writing task privately.
Participants completed the transcription task before moving on to the writing task. In the
writing task, 65 participants were randomly assigned to the "professor" condition, in
which they were told that a professor would provide feedback on their first essay. Sixty-
four participants were randomly assigned to the "peer" condition, in which they were told
another student volunteer would review their essay. An additional 64participants were
randomly assigned to the "unknown" condition; for these participants, the person
reviewing the essay was identified only as “the reviewer.” Once participants completed
their first essay, they clicked SUBMIT and notified the researcher. They were told that
the reviewer was busy reading their essay, and then completed a short demographic
survey as a filler task. The survey inquired about their age, year in college, their major,
and if they had heard of Elon’s Campus Writing Center; if they had, a follow-up question
asked if they have ever taken work to the Writing Center. After completing the survey,
the researcher told the participant that they were going to retrieve the essay from the
reviewer. The researcher returned after approximately three minutes with the rubric in an
unsealed envelope. Based on the participant’s feedback condition, the 1 - Poor score or a
4 - Excellent score was circled on the rubric. Ninety-eight participants (33 in the
professor condition, 31 in the peer condition, and 34 in the non-specified reviewer
condition) received positive feedback while 95 participants (32 in the professor condition,
33 in the peer condition, and 30 in the non-specified reviewer condition) received
negative feedback. The second essay prompt was then given to the participant, and
Self-Editing in Writing 24
keystrokes and inter-keystroke-intervals were recorded once again as they composed their
essay. Difference scores were calculated to measure the change in number and type of
edits from the pre-feedback essay to the post-feedback essay.
There was no time limit on writing. After the participants completed the second,
post-feedback essay, they were debriefed, given their $5 gift card compensation, and
released from the study.
Results
Means and standard deviations for each of the dependent variables are shown in
Table 1. Each edit type is reported as proportions of the number of submitted characters,
and all measures of time are reported in seconds. The difference scores (post-
feedback/essay 2 minus pre-feedback/essay 1) were analyzed using a 3x2 between-factors
MANOVA. The multivariate tests resulted in a significant omnibus effectof feedback,
F(17, 171) = 1.683, p = .05, ηp2 =.143. There was no significant omnibus effect of
audience, F(34, 342) = 1.098, p = .330, ηp2 = .098, and no significant omnibus interaction
between audience and feedback, F(34, 342) = 1.302, p = .126, ηp2 = .115.
Univariate tests for all dependent variables were run to investigate the significant
main effect of feedback; the results are reported inTable 2.Only significant findings will
be discussed in depth. The effect of feedback on the amount of time spent viewing the
prompt was significant, F(1, 187) = 9.174, p = .003, ηp2 = .047. As shown in Figure 1, on
average, participants increased the amount of time spent (in seconds) viewing the prompt
during their second essay if they received negative feedback on their first essay (M=
4.309, SD = 2.039) and decreased the amount of time spent viewing the prompt if they
Self-Editing in Writing 25
received positive feedback on their first essay (M = -4.358, SD = 2.008). The effect of
feedback on the ratio of typed keystrokes to number of submitted characters was also
significant, F(1, 187) = 4.461, p = .036, ηp2 = .023, see Figure 2. Specifically, the ratio of
keystrokes to characters decreased after feedback, but decreased significantly more
following positive feedback (M = -.137, SD = .034) compared to negative feedback (M =
-.035, SD = .035). Finally, the effect of feedback on the number of substitutions
immediately following a pause approached significance, F(1, 187) = 3.569, p = .06, ηp2
=.019. There was an overall decrease in the proportion of pause-associated substitutions
on essay 2, an effect that was more pronounced when negative feedback was given (M = -
.033, SD = .012) compared to positive feedback (M = -.001, SD = .02).
Because this is one of the few empirical studies of self-editing with this
framework and methodology and it examines a wide range of variables, exploratory
univariate tests were conducted to investigate possible isolated effects of audience and
the audience-feedback interactions in order to find potential recommendations for future
research. Table 2 shows the F-values for these tests. There were no significant main
effects of audience, but there were several significant interaction effects. The interaction
between audience and feedback on the amount of time spent viewing the prompt
approached significance, F(2, 187) = 2.931, p = .056, ηp2 = .03; see Figure 1. Simple
main effects tests showed that the effect of feedback was time viewing the prompt after
receiving negative feedback (M = 3.87, SD = 11.39). A similar effect was found when
writing for a peer, F(1, 64) = 7.16, p = .01, ηp2 = .103; participants viewed the prompt for
their second essay less after positive feedback, (M = -6.39, SD = 18.76), and more after
Self-Editing in Writing 26
Table 1
Means and Standard Deviations for All Dependent Measures
Professor
Peer
Non-specified
Positive Negative
Positive Negative
Positive Negative
Measure
M SD M SD
M SD M SD
M SD M SD
Prompt Reviews
Number of reviews
-0.88 2.18 -0.31 2.31
-1.10 3.02 0.45 2.82
-1.15 3.76 -0.83 2.85
Total review time (s)
-2.42 10.3 3.87 11.39
-6.39 18.76 11.72 33.01
-4.26 19.81 -2.66 16.55
Pause Measurements
Total # pauses
-21.88 40.64 -12.72 30.63
-9.26 33.86 -22.39 54.45
-3.24 23.67 -5.97 28.97
Ave pause length (s)
0.11 1.75 -0.15 1.82
-0.19 1.30 0.24 1.45
0.28 1.59 0.09 1.50
Post-typing pause (s)
-3.02 11.43 4.52 1.56
-0.63 18.54 -6.76 25.04
2.86 18.79 2.03 6.45
Pause Associated Edits
Deletions
0.02 0.14 0.01 0.06
-0.02 0.04 0.02 0.11
-0.02 0.06 0.01 0.08
Substitutions
0.02 0.11 -0.01 0.07
-0.01 0.12 -0.05 0.16
-0.02 0.08 -0.04 0.01
Insertions
0.01 0.02 0.01 0.01
-0.01 0.02 -0.01 0.02
0.02 0.01 -0.02 0.01
Total PAEs
0.04 0.18 -0.02 0.10
-0.03 0.15 -0.03 0.23
-0.03 0.18 -0.03 0.15
Non-Pause Edits
Deletions
-0.65 1.15 -0.68 1.17
-1.01 1.96 -0.67 2.34
-0.46 1.27 -0.82 1.43
Substitutions
0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00
Insertions
-0.11 0.35 -0.11 0.31
-0.11 0.35 -0.11 0.31
0.07 0.38 -0.06 0.38
Total NPEs
-0.16 0.37 -0.14 0.33
-0.09 0.37 -0.08 0.52
0.05 0.40 -0.07 0.40
Overall Content
Number of Keystrokes
-403.64 854.72 -192.12 728.53
-41.10 761.15 -483.18 1021.51
-229.32 775.71 20.03 698.71
Number of Characters
-114.15 418.41 -35.75 453.93
122.90 418.67 -270.21 585.88
-13.71 378.13 96.73 91.99
K/C Ratio
-0.10 0.39 -0.08 0.22
-0.20 0.33 0.01 0.45
-0.1 0.27 -0.03 0.31
Total Writing Time (s)
-148.93 316.93 -94.44 263.94
-76.14 344.64 -195.63 501.29
-76.88 242.59 -50.00 14.11
Total Edits
-0.1 0.45 -0.16 0.34
-0.12 0.47 -0.11 0.67
0.02 0.45 0.10 0.47
Note 1: Measures of edits are reported in proportion of edits to number of submitted characters
Note 2: All measurements are difference scores (post-feedback/essay 2 minus pre-feedback/essay 1)
Self-Editing in Writing 27
Table 2
F-Values for Univariate Tests for Each Dependent Variable
Measure
Audience
Feedback
Audience*Feedback
Prompt Reviews
Number of reviews
0.874
3.827
0.827
Total review time (s)
1.587
9.174**
2.931˚
Pause Measurements
Total # pauses
2.275
0.177
1.477
Ave pause length (s)
0.064
0.104
1.299
Post-typing pause (s)
1.616
0.005
1.928
Pause Associated Edits
Deletions
0.335
1.402
2.163
Substitutions
1.852
3.569˚
0.042
Insertions
0.471
1.580
0.392
Total PAEs
1.624
0.647
0.764
Non-Pause Edits
Deletions
0.297
0.007
0.733
Substitutions
0.889
0.892
0.889
Insertions
1.433
1.254
0.526
Total NPEs
1.931
0.292
0.590
Overall Content
Number of Keystrokes
1.017
0.003
3.361*
Number of Characters
1.430
1.116
6.362*
K/C Ratio
0.075
4.461*
1.161
Total Writing Time (s)
0.808
0.067
1.204
Total Edits
0.694
0.497
0.293
Note 1: * indicates p < .05
Note 2: ** indicates p < .01
Note 3: ˚ indicates marginal significance, p ≤ .06
Self-Editing in Writing 28
Figure 1: Average (± 1 SE) change in time viewing the prompt after positive feedback
(solid bars) and negative feedback (striped bars) for each audience condition.
Figure 2: Average change (± 1 SE) in the ratio of keystrokes to characters after positive
feedback versus negative feedback.
Self-Editing in Writing 29
negative feedback (M = 11.72, SD = 33.01). When writing for a non-specified reviewer,
participants decreased the time viewing the prompt after positive feedback (M = -4.27,
SD = 19.81) and after negative feedback (M = -2.65, SD = 16.55); the amount decreased,
however, was not significantly different, F(1, 64) = .122, p = .72, ηp2 = .002.
There was also a significant interaction between audience and feedback on the
total number of typed keystrokes, F(2, 187) = 3.631, p = .028, ηp2= .37; see Figure 3.In
absolute terms, participants decreased their number of keystrokes in the professor and
peer conditions after receiving either positive or negative feedback, as well as in the non-
specified reviewer condition if positive feedback had been given. Simple main effect tests
showed that the change in number of keystrokes was marginally significantly different
after positive feedback (M = -41.1, SD = 761.15) compared to negative feedback (M = -
483.18, SD = 1021.5) only in thepeer condition, F(1, 64) = 3.815, p = .055, ηp2= .058.
The number of keystrokes also decreased in the professor condition following positive
feedback (M = -403.64, SD = 854.72) and negative feedback (M = -192.12, SD = 728.53),
although the differences was not significant,F(1, 65) = 1.15, p = .28, ηp2= .079.When
writing for a non-specified reviewer, there was no significant difference in the change in
number of keystrokes following positive feedback (M = -299.32, SD = 775.71) compared
to negative feedback (M = 20.03, SD = 698.71), F(1, 64) = 1.806, p = .184, ηp2= .028.
There was a significant interaction between audience and feedback on the total
number of submitted characters, F(2, 187) = 6.362, p = .002, ηp2 = .64; see Figure 4
Self-Editing in Writing 30
Figure 3: Average change in the number of typed keystrokes (± 1 SE) following positive
feedback (solid bars) and negative feedback (striped bars) for each audience condition.
Figure 4: Average change in the number of submitted characters (± 1 SE) following
positive feedback (solid bars) and negative feedback (striped bars) for each audience
condition.
Self-Editing in Writing 31
In absolute terms, participants decreased the number of submitted characters when they
received positive or negative feedback from a professor, when they received negative
feedback from a peer, and when they received positive feedback from a non-specified
reviewer. They increased the amount they submitted after receiving positive feedback
from a peer and negative feedback from a non-specified reviewer. Simple main effect
tests again revealed a significant feedback effect only for the peer condition, F(1, 64) =
9.429, p = .003, ηp2 = .132, with participants increasing the amount of submitted
characters following positive feedback (M = 122.9, SD = 418.87) and decreasing the
number of characters after negative feedback (M = -270.21, SD = 585.88). In the
professor condition there was no significant difference in change of number of characters,
F(1, 65) = .525, p = .47, ηp2 = .008, between positive feedback, (M = -114.15, SD =
418.39) and negative feedback (M = -35.75, SD = 453.93).There was no significant
difference in change of number of characters in the non-specified reviewer condition,
F(1, 64) = 1.314, p = .25, ηp2 = .021, between positivefeedback (M = -13.71, SD =
378.13) and negative feedback (M = 96.73, SD = 391.99).
Discussion
The present study found that students write and edit differently, in some respects,
depending on the type of feedback given on a written task and the perception of who is
giving the feedback.
Effects of Feedback
Participants who received negative feedback on the first essay spent significantly
more time viewing the prompt on their second essay; conversely, they spent less time
Self-Editing in Writing 32
viewing the prompt while writing their second essay if they received positive feedback on
the first essay.For this study, negative feedback was accompanied by the comment that
the essay “did not completely answer the question.” Participants receiving this evaluation
may have taken this into consideration for the next essay, spending more time reading
over the prompt to remedy the issue. If viewing the prompt indicates that the participant
is reflecting on the topic and planning what to write, these datasuggest that negative
feedback motivates students to plan and reflect on atopic more than positive feedback
does. After their first essay was met with criticism, perhaps they approached the second
essay with greater care, ensuring that they fully understood the prompt by reading it more
thoroughly. Those who received an excellent score on the first essay may not have felt
the need to read the prompt as carefully as before, given their approach with the first
essay was apparently successful.
Evidence from Cho and MacArthur (2010) suggesting that negative feedback
occurs less often than positive feedback mayalso help explain this finding. In their study,
which investigated the effects of peer review on students’ writing, negative feedback
occurred so rarely that it was not included in the final analysis. If negative feedback is a
less common occurrence than positive feedback, it is possible thatits noveltyoccasions a
stronger response from the writer. Students, more accustomed to praise than criticism,
may take the criticism more seriously and try to write more carefully on the next task.
Feedback type also affected the ratio of keystrokes to characters such that the
ratio decreased after feedback, especially if positive feedback was given. This
measurement reveals the proportion of editing compared to the actual volume of the
essay that the participant submitted; it may be thought of as how much the writer fiddled
Self-Editing in Writing 33
with the text throughout the writing process before sending it to the reviewer.A higher
ratio indicates more fiddling with the text, which could mean the writer tried various
approaches, strategies, and styles as they completed the assignment.This decrease in the
ratio may be due to fatigue, as each session was approximately 45 minutes long and some
writers may have tired by the second essay. While this may have had some effect on
participants’ behavior, the finding that there was significantly less editing, by this
measure, following positive feedback as compared to negative feedback suggests that
explanations other than fatigue are possible. It may be due to the same reason that may
explain the effect of feedback on time spent viewing the prompt: perhaps the writers felt
that getting positive feedback on the first essay meant that they did not have to put as
much effort into editing the second essay as they wrote.
The two variables – time viewing the prompt and the ratio of keystrokes to
characters – reflect the planning and editing occurring throughout the writing process and
brings to mind Flower et al.’s (1986) discussion of effective writing and editing. A skilled
writer diagnoses issues as she writes; that is, a problem is recognized and solved by the
writer herself during the writing process and not by an external editor. This development
of an internal self-editor is the primary goal of self-editing (Beach, 1979; Skinner, 1957;
Willingham, 1990), and thus, it is highly relevant for this study that participants
demonstrated a change in some aspect of self-editing in response to the type of feedback.
These findings indicate that students do react differently in response to praise or
criticism. This could have implications for teaching students to be effective, skilled
writers. Perhaps providing constructive criticism rather than praise may lead students to
consider the writing topic more while they write. In addition, as suggested by feedback’s
Self-Editing in Writing 34
effect on the ratio of keystrokes to characters, they may try different approaches and
writing styles as they complete an assignment.
Interaction Effects
Two measurements of writing volume were influenced by feedback differentially
depending on audience: the number of typed keystrokes and the number of submitted
characters. As explained previously, these variables were analyzed to investigate isolated
interactions for exploratory purposes in order to add to the general knowledge of self-
editing and to provide potential future directions for research; because of this, the
inferences that are drawn are speculative. While the number of typed keystrokes
measured every time a key was struck (so material that was deleted at a later point in the
writing process still contributed to the keystroke measurement), the number of submitted
characters measured each letter or symbol submitted in the final product. These two
measurements set up an interesting comparison: the keystroke measurement allows us to
track the editing process as the participant types, while the character measurement shows
us what the participant considers to be the complete, fully revised piece.
Tentative results showed that when writing for a peer, participants tended to type
significantlyfewer keystrokes on their second essay if they had received negative
feedback on their first essay compared to the change in keystrokes following positive
feedback. The reason the number of keystrokes significantly decreased after being
criticized by a peer is still unclear; however, it may be inferred, given that the effect of
feedback depended on the perceived audience, that student participants viewed a peer
reviewer as less credible than an expert such as a professor. Upon receiving a negative
Self-Editing in Writing 35
evaluation from a peer, participants seemed to stop investing as much in the essay and
greatly decreased how much they typed. Again, we see how a history of interaction
between peers and figures of authority comes into play: disregarding a peer’s evaluation
has, in past scenarios, likely resulted in less punishment than disregarding an expert’s
evaluation (Skinner, 1957). This echoes Nobes & Pawson’s(2003) study, which found
that children judged rules mandated by authority figures as less open to transgression
than rules mandated by peers. The same judgment of an expert’s evaluation versus a
peer’s seems to be occurring in participants’ reactions to feedback. A student being
criticized by a peer responds with something of a “What do they know?” effect and stops
writing as much. In fact, students may have a history that has taught them specifically to
disregard peer review in an academic setting because it is often not done well. In this
case, the participants lacked any information about the credentials of and criteria used by
the peer reviewer (beyond the grading rubric). Thus, it is not just the type of feedback
that impacts self-editing, but who is giving it. Furthermore, in Hyten and Chase’s (1991)
study, the number of characters and overall self-editing increased after negative
feedback, opposing the findings of the current study. Hyten and Chase’s study, however,
did not include the perceived audience as an independent variable. The finding that
negative feedback from a peer elicited the opposite effect on self-editing compared to
negative feedback from Hyten and Chase’s non-specified reviewer is of high interest and
perpetuates the notion that writers are more likely to dismiss a critical review from a peer.
A similar effect was also seen regarding the number of submitted characters.
When writing for a peer, participants submitted significantly much less after receiving
negative feedback compared to positive feedback. Again, this suggests that criticism from
Self-Editing in Writing 36
a peer is viewed with less credibility. These results supplement Cho & MacArthur’s
(2010) study of peer review. In their research, some of the comments given to student
writers were actually from experts, but this was unknown to the students. The authors
raise the question of whether students would change their writing behavior if they had
known that an expert was evaluating their work. Based on the present study’s findings, it
would seem that the expectation of who is giving the feedback does have an impact on
writing and self-editing.
Participants’ second essays were also shorter than the first essay when writing for
a professor. Similar to the measurement of typed keystrokes, this was seen in both
feedback conditions.The finding that the second essay was generally shorter when writing
for a professor may be explained by the students’ perspective of the professor as an
expert. Chiesi et al.(1979) and Willingham (1990) discuss how experts sometimes
become set in their evaluation and grading styles; it is usually not the teacher who adjusts
his grading method, but the student who must change her behavior to meet the standards
of the professor. In this study, participants may have seen no need to drastically change
their behavior, and so wrote less for the professor whether they received positive
feedback or negative feedback; if their writing style earned a “Poor” grade the first time,
it is possible they felt there was little they could have changed to get a better grade on the
second essay, and so they did not change their writing behavior very much. If they
received positive feedback, perhaps they felt that what they wrote for the first essay
sufficed, and so there was no need to change.
Self-Editing in Writing 37
Limitations
Although some significant results were found, audience and feedback did not
show significant effects on pauses (neither number nor length) and types of edits
(substitutions, deletions, and insertions). There are several limitations to this study that, if
amended in future research, could reveal why these variables were unaffected and direct
researchers to important avenues to study self-editing further.Beach (1979) points out that
personality and demographic factors influence revision in writing. The sample here was
fairly homogenous; all were undergraduate students enrolled in the same private
university, and most (78.8%) were female. The sample also included mostly
underclassmen (84.9% were first-years or sophomores). Such a skewed proportion may
have affected the results, given that upperclassmen should be generally better writers and
have a more developed self-editing mechanism as a result of up to three more years of
schooling than underclassmen. Perhaps juniors and seniors have different editing styles
than freshmen and sophomores; a study examining this idea would yield very interesting
results relating to changes in self-editingduring college.
Additionally, many of the participants were enrolled in an introductory or other
psychology course, possibly leading them to expect some degree of deception. This may
have led them to doubt there was an actual reviewer. During debriefing, some
participants did reveal that they had learned of psychological studies and deceptions in
their classes, leading them to suspect that deception was present in this study and that
there may not be an actual reviewer; these participants were excluded from analyses.
Students studying areas other than psychology may have been deceived more easily since
they are generally more removed from the subject. A more diverse sample, with a greater
Self-Editing in Writing 38
variety of characteristics and backgrounds among participants, may reveal more about
self-editing and what shapes it.
The anonymity of the essays and the lack of any enduring consequence may have
introduced limitations as well. Participants bore no lasting penalties for sub-par writing in
this study, since participation credit was not contingent on performance. In a university
setting, where one puts his or her name on nearly every article of submitted work, a
writing task which requires no name may have discouraged participants from putting as
much effort into the essays as they would normally, when their name is on it. If there had
been some positive consequence for quality work or a more meaningful penalty for poor
work, participants may have edited more overall, possibly resulting in a greater effect of
audience and feedback on self-editing. The lack of a lasting consequence could explain
why more significant effects of audience and feedback on the dependent variables were
not found. Specifically, pauses and editing variables were not significantly affected by
the audience or feedback (see Tables 1 and 2), possibly because the task did not engender
a meaningful reward or consequence. Hyten and Chase (1991) included a system of
rewards and punishments beyond positive or negative feedback in their study of self-
editing, which may have resulted in a stronger effect of feedback. Upon receiving
positive feedback for an essay, participants were given points; negative feedback
deducted points. At the end of the session, these points were converted to money. Thus,
these participants had an additional motive to write well and adjust their self-editing in
response to the feedback, even though, like the present study, the feedback was randomly
assigned regardless of the quality of the essay. Future studies examining specific
Self-Editing in Writing 39
influences of anonymity and incentives on how individuals self-edit would clarify such a
connection.
A more naturalistic setting would also be beneficial since many students were
psychology majors or at least registered in a basic psychology class, giving them
knowledge of the experimental process and potentially compromising the deception in
the study. The use of deception itself raises more questions – if students knew that the
feedback was arbitrary and/or there was no real reviewer, this may have affected the
results. Specifically, this could explain why there was not a stronger effect of the
independent variables on the self-editing measures; if participants knew that the feedback
they received was not contingent on their writing behavior, they may not have edited as
much as they would if they believed that the feedback was actually based on their essay.
Carpenter, Sachs, Martin, Schmidt, & Looft (2012) suggest that students are more likely
to pay attention to and correct plausible feedback; that is, feedback that is more positive
or more negative than what is usually earned is not reflected upon as much as feedback
that aligns with previous experience. Thus, students who had histories of receiving
primarily high praise for their writing and received a poor score may not have believed
the comments and not put as much effort into the second essay, limiting the results. In
fact, some participants did reveal during debriefing that they did not believe their score.
Some skilled writers thought that something was wrong with the reviewer’s grading or
there was no reviewer at all if their essay received the worst score, while some writers
who usually struggle with essays were suspicious if they got positive feedback, since the
highest possible score constituted positive feedback. This could also help explain the
significant decrease in the number of keystrokes and characters following negative
Self-Editing in Writing 40
feedback from a perceived peer reviewer; if a peer reviewer is considered as a less
reliable source of feedback than the expert reviewer, then feedback that does not match
previous experiences may be viewed as an error on the part of the evaluator and not the
writer, resulting in the participant decreasing overall writing behavior. The writer may be
responding to the surprising feedback in addition to the perception of the peer as less
credible, and so in response, they shorten the volume of the second essay. Moreover,
Rommetveit (1970) discusses the importance of face-to-face interaction. Histories and
scripts of authority are generally learnedand carried out in person (Schlinger, 2002;
Skinner, 1957; 1969; Vaughn & Michael, 1982) , and it is this interaction, rather than a
score on a rubric, that brings about a stronger change in behavior (Rommetveit, 1970).
The method of the study could be adjusted to investigate the variables further and
find more significant effects. It is possible that participants did not have time to write
enough material for analysis. This could explain why no significant results were found
for pauses or edits; perhaps there was not enough of these variables present in the data to
investigate, and so there were no distinctive findings. A future study may ask participants
to write for longer or write more to have more material to analyze and, hopefully, more
results to explore. Hyten and Chase (1991) incorporated this notion in their study by
having participants write a series of eight brief responses with feedback delivered after
each one. This resulted in more content available for analysis and greater variance for
each writer than in present study, which only asked for two essays (although the essays in
the present study were longer). More essays would also mean different types of prompts
could be posed to participants, such as open-ended prompts (like the prompts used in this
Self-Editing in Writing 41
study) in addition to fact-based prompts (e.g. asking them to reflect on a current issue),
giving different writers more flexibility to write in a way most natural to them.
Future Implications
This study found that individuals adjust aspects of self-editing in response to
specific audiences and feedback types, holding implications for writing behavior, verbal
behavior, and education more generally. It also builds on previous research; Hyten and
Chase’s study, for example, measured self-editing only in terms of backspacing and
deleting the entire response at once. The present study measured self-editing across
numerous dimensions and found significant effects for variables that had not been
previously investigated (prompt review time, ratio of keystrokes to characters, and total
typed keystrokes) using such a methodology; non-significant findings were still important
to direct future research since they have not been investigated in such a way before.
Additionally, Hyten and Chase (1991) found that the number of submitted characters
increased after negative feedback, while the present study found that when writing for a
peer, characters in fact decreased after negative feedback compared to when positive
feedback was given, introducing an important new dimension to an earlier result.
This relationship between a perceived peer reviewer and negative feedback also
holds implications for education. Specifically, educators may reconsider the use of peer
evaluation since it seems to result in different writing and self-editing behaviors than
expert evaluation. This is especially relevant when teaching writing and editing skills
since criticism from a peer seems to result in less overall self-editing. This finding is
consistent with suggestions that expert review holds more benefits than peer review
Self-Editing in Writing 42
(Anderson & Lebiere, 1998; Beach, 1979; Besnard & Bastien-Toniazzo, 1999; Chi et al.,
1981; Chiesi et al., 1979; Hull, 1983) and may help inform what will most benefit student
writers as they learn to self-edit and revise. Zhu (1995), however, suggests that students
may be trained to give feedback that is just as effective as feedback from experts. Given
enough time and training, they can learn to recognize, diagnose, and provide solutions to
problems in a piece of work as well as experts can. Peer review may be improved, but
students’ perception of a peer reviewer as less knowledgeable or less credible may still
influence how they react to feedback. An interesting avenue for future research may
replicate the conditions of this study but inform the participant that the peer reviewer has
been trained to read assignments and give helpful feedback. This perception of the peer
reviewer as someone who is skilled at evaluating written assignments would clarify the
effect that the perception of an audience has on self-editing.
The lack of a significant effect of the independent variables on pauses and editing
variables (substitutions, insertions, and deletions) may be due to the fact that
undergraduate students do not revise as much as is expected. Other studies (Epting et al.,
under review; Rau & Sebrechts, 1996) have also tasked participants with completing
written assignments as a means to study editing variables, but found that they did not
revise extensively throughout the writing process. This may also be due in part to the
potential limitations discussed previously, such as fatigue, lack of an enduring
consequence, or the prompts not encouraging sufficient writing for analysis. However,
with increasing converging evidence from various studies finding fewer edits than
expected, it must be considered that students do not edit as much as is anticipated in the
first place, regardless of the independent variables that are introduced. Furthermore, the
Self-Editing in Writing 43
level of revision was not addressed in this study. Cho and MacArthur (2010) detailed the
varying classes of revisions, including surface level changes (simple repairs), micro-level
meaning changes (complex repairs) and macro-level meaning change (new content and
organization). The present study explored simple repairs more so than changes in the
content, including micro-level and macro-level changes; these were outside the scope of
this study and were not analyzed. A future study may investigate how audience and
feedback affect written content in terms of meaning and organization rather than simple
edits exclusively.
Although the demographic survey distributed to participants in this study was
primarily used as a distractor task while the researcher retrieved the feedback from the
perceived reviewer, students’ responses were in some ways potentially useful. In the
responses to the demographic survey question asking participants if they had visited
Elon’s Writing Center, most underclassmen reported that they had been to the Writing
Center as part of an assignment in the required College Writing course. This may be a
recent curriculum-wide assignment that was not mandatory when the upperclassmen were
younger, since most of the upperclassmen reported that they had never been to the
Writing Center. Many of the comments added that the Writing Center had been very
helpful and would be visited again; future visits would likely continue to refine students’
self-editing skills, which could alter the findings of this study such that overall self-
editing would increase. Perhaps students who are encouraged (or in current College
Writing courses, required) to take work to the Writing Center gain a better understanding
of revision, detection of problems, and as Flower et al. (1986) emphasize, diagnosis of
problems. A study tracking the self-editing of students who regularly attend the Writing
Self-Editing in Writing 44
Center and those who do not may inform us of the importance of such resources on a
college campus and how their use influences self-editing.
The findings suggest that feedback has its own effects, but those effects also
sometimes depend on the perception of who is giving it. This means that our relationship
to and expectations of a person, influenced by his or her position, authority, familiarity,
and other factors (Rommetveit, 1968; Skinner, 1957), impacts how we react to praise or
criticism delivered by that person. This connects to the behavior-environment
contingencies (reinforcement and punishment) we encounter daily. The way that
participants reacted to audience and feedback in this study, and the way we react to
different interactions and environments in daily life, has been shaped by previous
episodes of reinforcement and punishment through feedback delivered by various
audiences. It is reasonable, therefore, to apply the finding that feedback depends on the
audience providing the feedback to multiple other instances, such as receiving an
evaluation from an employer or delivering feedback to an employee. With this
knowledge, we may become more effective communicators by considering our own and
others’ self-editing mechanism and how we alter it when interacting with certain people.
Thus, we gain empirical evidence for a phenomenon often taken for granted – the
adjustment of verbal and written behavior in response to a certain person. Recognizing
that behavioral principles influence these processes gives us a lens with which to
understand ourselves and how we interact with others. The current findings provide
useful insights into what impacts self-editing and writing and future directions that may
be taken to learn more.
Self-Editing in Writing 45
References
Anderson, J. R., & Lebiere, C. (1998). The atomic components of thought. Mahwah, NJ:
Erlbaum.
Beach, R. (1979). The effects of between-draft teacher evaluation versus student self-
evaluation on high school students’ revising of rough drafts.Research in the
Teaching of English, 13, 111-119.
Besnard, D., & Bastien-Toniazzo, M. (1999). Expert error in trouble-shooting: an
exploratory study in electronics. International Journal of Human Computer
Studies, 50, 391-405.
Carpenter, S., Sachs, R., Martin, B., Schmidt, K., & Looft, R. (2012).Learning new
vocabulary in German: The effects of inferring word meanings, type of feedback,
and time of test.Psychonomic Bulletin & Review, 19(1), 81-86. doi:
10.3758/s13423-011-0185-7
Chi, M. T. H., Feltovich, P. J., & Glaser, R. (1981).Categorization and representation of
physics problems by experts and novices. Cognitive Science, 5, 121-152.
Chiesi, H. L., Spilich, G. J., & Voss, J. F. (1979). Acquisition of domain related
information in relation to high and low domain knowledge. Journal of Verbal
Learning and Verbal Behavior, 18, 257-273.
Cho, K., Chung, T. R., King, W. R., & Schunn, C. D. (2008). Peer-based computer-
supported knowledge refinement: an empirical investigation. Communications of
the ACM, 51(3), 83-88.
Self-Editing in Writing 46
Cho, K., & MacArthur, C. (2010).Student revision with peer and expert reviewing.
Learning and Instruction, 20, 328-338. doi:10.1016
Coppola, M. (2002). The emergence of grammatical categories in home sign: Evidence
from family-based gesture systems in Nicaragua. Ph.D. Dissertation, Dept. of
Brain and Cognitive Sciences, University of Rochester, Rochester, NY.
Deacon, T. W. (1997).The symbolic species: The coevolution of language and the brain.
New York: W. W. Norton.
Della-Piana, G. (1978). Research Strategies for the Study of Revision Processes in
Writing Poetry. In Charles R. Cooper and Lee Odell (Ed.), Research on
Composing: Points of Departure, Urbana, IL: National Council of Teachers of
English.
Epting, L.K., & Critchfield, T. S. (2006). Self-editing: On the relation between behavioral
and psycholinguistic approaches. The Behavior Analyst, 29(2), 211-234.
Retrieved from EBSCOhost
Epting, L.K., Gallena, E., Hicks, S., Palmer, E., Weisberg, T. (under review). Read and
think before you write: Prewriting time and level of print exposure as factors in
writing. Journal of Writing Research.
Falchikov, N., & Goldfinch, J. (2000). Student peer assessment in higher education: a
meta-analysis comparing peer and teacher marks. Review of Educational
Research, 70, 287-322.
Self-Editing in Writing 47
Fishbach, A., Dhar, R., & Zhang, Y. (2006).Subgoals as substitutes or complements: The
role of goal accessibility. Journal of Personality and Social Psychology, 91, 232–
242.
Fishbach, A., Eyal, T., & Finkelstein, S. (2010). How positive and negative feedback
motivate goal pursuit. Social and Personality Psychology Compass, 4(8), 517-
530. doi:10.1111/j.1751-9004.2010.00285.x.
Flower, L., Hayes, J., Carey, L., Schriver, K., & Stratman, J. (1986).Detection, diagnosis,
and the strategies of revision. College Composition and Communication, 37(1),
16-55.
Fox Tree, J. (2001). Listeners’ use of um and uh in speech comprehension. Memory and
Cognition, 29, 320-326.
Harley, T. A., & MacAndrew, S. G. (2001).Constraints upon word substitution speech
errors. Journal Of Psycholinguistic Research, 30(4), 395-418.
doi:10.1023/A:1010421724343
Hull, G. (1983).The editing process in writing A performance study of experts and
novices.Diss. University of Pittsburgh.
Hyten, C., & Chase, P. (1991). An analysis of self-editing: Method and preliminary
findings. In L. J. Hayes & P. N. Chase (Eds.), Dialogues on verbal behavior (pp.
67– 81). Reno, NV: Context Press.
Kleiven, J. (1973). Verbal communications and intensity of delivery.
ScandinavianJournal of Psychology, 14(1), 111-113.
Self-Editing in Writing 48
Laupa, M. & Turiel, E. (1986).Children’s conceptions of adult and peer authority. Child
Development, 57(2), pp. 405-412.
Ludvig, E. A. (2003). Why Pinker needs behaviorism: A critique of The Blank Slate.
Behavior and Philosophy, 31, 139-143.
MacArthur, C. (2007). Best practice in teaching evaluation and revision. In S. Graham, C.
MacArthur, & J. Fitzgerald (Eds.), Best practice in writing instruction (pp. 141-
162). New York: Guilford.
MacDonald, E., Johnson, E., Forsythe, J., Plante, P., & Munhall, K. (2011). Children’s
development of self-regulation in speech production. CurrentBiology. Published
online 22 Dec. 2011.
Nobes, G. & Pawson, C. (2003).Children’s understanding of social rules and social
status. Merrill-Palmer Quarterly: Journal of Developmental Psychology, 49(1),
77-99.
Pinker, S. (1994). The language instinct: How the mind creates language. New York: NY
US: William Morrow & Co.
Ratner, N. B., Berko-Gleason, J., & Narasimhan, B. (1998). An introduction to
psycholinguistics: What do language users know? In J. Berko-Gleason & N. B.
Ratner (Eds.), Psycholinguistics (2nd ed., pp. 1-49). Belmont, CA:
Wadsworth/Thomson.
Self-Editing in Writing 49
Rau, P. S., & Sebrechts, M. M. (1996). How initial plans mediate the expansion and
resolution of options in writing. The Quarterly Journal of Experimental
Psychology, 49A, 616-638.
Rommetveit, R. (1968). Words, meanings, and messages. Academic Press and
Universitetsforlaget, New York, London and Oslo.
Rommetveit, R. (1970). Verbal communication and social influence. Communication and
Drug Abuse, 68-78.
Schlinger, H. D. (2002). Not so fast, Mr. Pinker: A behaviorist looks at the Blank Slate.
A review of Steven Pinker's the Blank Slate: The Modern Denial of Human
Nature. Behavior And Social Issues, 12(1), 75-79.
Skinner, B. (1957). Verbal behavior. New York: Appleton-Century-Crofts.
Skinner, B. (1962). Contingencies of reinforcement: A theoretical analysis. Englewood
Cliffs, N.J.: Prentice Hall.
VanDuijvenvoorde, A., Zanolie, K., Rombouts, S., Raijmakers, M., & Crone, E. (2008).
Evaluating the negative or valuing the positive? Neural mechanisms supporting
feedback-based learning across development. The Journal of Neuroscience,
28(38), 9495-9503. doi:10.1523/JNEUROSCI.1485-08.2008.
Vaughn, M. E., & Michael, J. (1982). Automatic reinforcement: An important but
ignored concept. Behaviorism, 10, 217-227.
Self-Editing in Writing 50
Willingham, D.(1990). Effective feedback on written assignments. Teaching of
Psychology, 17(1), 10-13.
Zhu, W. (1995). Effects of training for peer response on students’ comments and
interaction. Written Communication, 12, 492-528.
Self-Editing in Writing 51
Appendix A
Demographic Survey
What is your age? _____ What is your gender? M F
What year are you? Freshman Sophomore Junior Senior Other_________
Do you know about Elon’s Campus Writing Center? YES NO
If yes, have you ever taken work to the Writing Center?
