On Learning to Write about Ideas

Author(s): Sandra Stotsky

Source: College Composition and Communication, Vol. 37, No. 3 (Oct., 1986), pp. 276-293
Published by: National Council of Teachers of English
Stable URL: http://www.jstor.org/stable/358045
Accessed: 21/09/2008 01:50

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=ncte.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.

National Council of Teachers of English is collaborating with JSTOR to digitize, preserve and extend access to
College Composition and Communication.

http://www.jstor.org
On Learning to Write About Ideas

Sandra Stotsky

Introduction

As students improve as writers, what changes in their writing can we detect?

This question has long stimulated research on ways to measure growth in
writing ability. As part of the effort to develop and refine useful measures,
many studies have investigated the value of syntactic units of analysis, such as
sentence length and clause length. Although these measures have often proven
to be reliable indicators of syntactic differences between writers at different
grade levels and between good and poor writers at one grade level, they do not
always shed light on the kind of growth these writers achieve over time. 1 Un-
fortunately, few studies have explored the use of lexical units of analysis to
measuregrowth in writing ability. Those studies that have done so have tend-
ed to look only at the number of different words a writer has used in a specific
text, or at their frequency of occurrence in the English language, in order to
determine how varied or how easy or difficult a vocabulary the writer has
used.2 These studies have found that good writing displays greater diversity in
vocabularythan does poor writing. Apparently, the use of a larger number of
different words is one sign of growth in the ability to express complex mean-
ing in writing. But it is not clear from these studies exactly what the use of
larger numbers of different words reveals about growth in writing ability or
how it contributes specifically to the quality of writing. Moreover, the mea-
sures used in these studies do not provide information on how words are used
to create meaning in written texts; they tell us, instead, only what words are
used to create meaning. As a result, we have a limited understanding of how
students grow in their ability to construct meaning with written language.
Recently, two different approaches have been proposed for examining how
words are used to create meaning in written discourse. One approachexamines
the subject of each clause in a text in order to explore how meaning is con-

SandraStotskyis a ResearchAssociateat the HarvardGraduateSchoolof Education.She has

publishednumerousessayson languageand reading, in Research in the Teachingof Englishand
LanguageArts as well as in CCC. She is now completinga book on civic writing, which she
beganwith the aid of a Mina ShaughnessyScholarshipfrom the Fundfor the Improvementof
PostsecondaryEducationduring 1984-1985.

276 College Composition and Communication, Vol. 37, No. 3, October 1986
On Learning to Write About Ideas 277

structed in clauses or sentences. The other analyzes the number and kinds of
cohesive ties a writer creates across sentences in order to explore how meaning
is constructed over the complete text. In this paper, I present and discuss the
results of a study that used both these approaches in order to illuminate the
differences among a group of essays written by twelve developing writers. My
purpose in this paper is to show that we may well expand our understanding
of what precisely constitutes growth in learning to write about ideas and how
that growth contributes to the quality of writing by analyzing both the kinds
of subjects writers choose for their sentences as well as the network of semantic
relationships they create among these sentences.
In the first part of this essay, I describe how the writing samples were se-
lected, why these particular approaches for examining the construction of
meaning were chosen, and how the samples were scored. In the second part, I
present the results of the study and offer an interpretation of the findings. The
essay concludes with a discussion of the implications of these findings for the-
oretical models of how the ability to use written language develops and for a
working hypothesis about what constitutes progress in learning to write about
ideas.

Selection of the Writing Samples

The essays used in this study were written for a holistic evaluation of writing
by Grade 10 students in a small town high school at the end of the school
year. All students had been asked to take and defend a position on whether
the granting of a high school diploma should be contingent upon a student's
attaining competence in all the basic skills. (See the Appendix for the exact
wording of the topic.) The 11 papers judged lowest in composition quality
(those receiving a rating of 2) and the 8 papers judged highest in composition
quality (those receiving a rating of 8) were selected for examination.
There were two reasons for using these essays. First, the differences between
what is judged good and poor at one grade level by experienced teachers of
composition often provide more useful information than do differences be-
tween older and younger writers. The differences between these two groups of
essays suggest the direction that the writers of the low-rated essays probably
need to move in to improve their writing. Teachers might be able to help
poor writers to express and develop complex ideas more effectively than these
writers now do if the teachers better understood the ways in which language is
used to create and organize complex ideas in writing. Second, it seemed that
results from a small exploratory study might be more generalizable if one ex-
amined the writing of students from a small, relatively homogeneous commu-
nity with a stable school population rather than from a large, heterogeneous
community with a transient school population. Most of the students in this
town have attended both one of the town's two elementary schools and its
278 and Communication
CollegeComposition

junior high school and, thus, have had the same teachers. Differences in their
writing more likely reflect normal variation in a native English-speaking pop-
ulation rather than differences in community norms or school curricula.

The Analysis of the Grammatical Subject of a Clause

My examination of the grammatical subjects of clauses was motivated by sev-

eral sources. One source was an article by Lee Odell proposing that changes in
intellectual processes are one dimension of growth in writing ability.3 In this
article, Odell claims that we can learn a great deal about the way a writer per-
ceives and thinks by examining the words used as the grammatical subject, or
focus, of each clause. He further suggests that an examination of the changes
in the grammatical focus of each clause can also help us to determine what in-
tellectual processes students are using in their writing. Andrea Lunsford fol-
lowed up this suggestion in a study of the essays written for a college place-
ment examination by 320 basic and skilled writers in their senior year of high
school. She analyzed in detail the grammatical focus of each clause in an essay
written by one writer in each group, and found marked differences between
the types of words chosen by each writer, particularly in the use of personal
pronouns and abstract concepts. Finally, the work of Bruno Snell, a scholar of
the classics, also suggests the intellectual significance, for the development of
academic language, of using abstract concepts or ideas as the subjects of sen-
tences.4 I therefore expected that an analysis of the types of words used as the
subjects of clauses in my samples would also be highly informative.

The Analysis of Lexical Cohesion

In Cohesionin English, Michael Halliday and Ruqaiya Hasan propose a way to

analyze how words are used to construct meaning in whole pieces of dis-
course.5 Their work describes various types of semantic relationships that link
together sentences and paragraphs, units of discourse that are structurally in-
dependent of each other. Some types of relationships are created by gram-
matical structures, others by vocabulary items. Halliday and Hasan use the
term Lexical Cohesion to refer to the semantic relationships created by spe-
cific lexical items. According to their theoretical perspective, these relations in
meaning create the cohesive quality that helps to produce in the reader a sense
of connected discourse. Thus, the number and types of lexical cohesive ties
used in an essay are clearly important to examine. Moreover, according to a
study of popular scientific discourse by Robert Hopkins and a study by Ste-
phen Witte and Lester Faigley of essays written by college freshmen, lexical
cohesive ties appear to constitute the majority of cohesive ties in essay-
writing. However, my study did not use the scheme for analyzing lexical co-
On Learning to Write About Ideas 279

hesion proposed by Halliday and Hasan in Cohesionin English, since their

scheme is based on an analysis of only conversational and literary discourse. In
a recent essay, I demonstrated how their scheme did not account clearly and
completely for all types of lexical cohesion in expository essays, and I proposed
a modification and reorganization of their scheme for use with exposition.7
This new scheme consisted of two categories, as did the original, but, in the
first, words were related only through systematic semantic relationships, and,
in the second, they were related only through collocation. This scheme was
used for the present study and is outlined and exemplified below.
I. Semantically related words: a type of cohesion in which a lexical ele-
ment (or a group of lexical elements) is systematically related to a pre-
vious element (or a group of elements) through:
1. repetition: e.g., test/test;high schooldiploma/highschooldiploma
2. synonymy or near-synonymy: e.g., difficultlhard
3. opposition or contrast: e.g., simpleldifficult
4. inclusion as a superordinate, subordinate, or coordinate member in
an ordered or unordered set of general or specific terms: e.g., the
days of the week constitute coordinate members of an ordered set;
the names of occupations, such as teacher/janitor/cook, constitute co-
ordinate members of an unordered set.
5. derivation or repetition of a derivational element: e.g., benefit/bene-
ficial; employer/lworker
II. Collocationally related words: a type of cohesion in which one lexical
element is related to another only through frequent co-occurrence in
similar contexts: e.g., worker/skills/job
Part of the rationale for this new scheme was to distinguish systematic asso-
ciations in the language from contextual ones. It seemed theoretically more
defensible to differentiate cohesive ties in which words are related to each
other within the language system itself, and thus have a stable relationship re-
gardless of context, from those ties in which words are related only through
frequent co-occurrence in similar contexts. Further, it seemed that a more
suitable scheme than the one proposed by Halliday and Hasan might facilitate
informative distinctions. In Marion Crowhurst'sstudy of cohesion in the argu-
mentative essays of students in Grades 6, 10, and 12, all lexical ties, although
originally classified according to Halliday and Hasan's taxonomy, were col-
lapsed for analysis into just two categories, "same lexical item" and "other lex-
ical item," because all types other than repetition contained too small a
number of examples for meaningful analysis.8 Use of a scheme for studying
cohesion that was more suitable for essay writing than Halliday and Hasan's
taxonomy might have enabled Crowhurst to make further distinctions among
lexical ties across grade levels than her study showed. Only "other lexical
items" distinguished among students' work at all three grade levels in her
study.
280 and Communication
CollegeComposition

Scoring Procedures

From the original group of high- and low-rated essays, six in each group were
selected for analysis. The excluded low-rated essays were so incoherent or non-
fluent that data, where obtainable, from these essays would have distorted the
mean scores obtained from the other essays and made the differences between
the two groups even greater than they turned out to be. To counterbalancethe
exclusion of these essays, and to create an even number, the two most fluent
high-rated essays were excluded.
A number of guidelines were formulated and used to analyze and tally lex-
ical cohesive ties. First, in counting the number of words in each essay, an idi-
omatic expression, such as high school,pay attention,or do away with, as well as
an individual word that functioned semantically or grammatically as an inde-
pendent and indivisible unit, was counted as one word. (Almost all examples
in this section may be found in the two sample essays in the Appendix.)
Second, in counting the number of different words in each essay, only the
appearanceof an identical word or of a word varying in an inflectional or com-
parative ending, such as givinglgiven,fairfairer, was counted as an instance of
repetition; following the definition offered in Hans Marchand'stext on word-
formation, words related as derivatives, such as benefitlbeneficial, were consid-
ered to be different words and classified as creating a separatetype of cohesion,
as outlined above.9
Third, in counting the number of ties in each essay, a repeated phrase (e.g.,
givescredit),as well as a repeated word or idiomatic expression, was counted as
only one instance of repetition, no matter how many words were included in
the phrase.
Fourth, both intra-sentence ties and inter-sentence ties were counted, but
each type was coded separately.10
Fifth, phrases and clauses as well as individual words were counted as single
lexical items whenever they appeared to enter into cohesive relationships with
other words, phrases, or clauses. I explicitly extended the concept of a "lexical
item" beyond the level of a single word or idiom to include phrases or clauses
as lexical elements, not only to be consistent with the way in which Halliday
and Hasan coded the sample passages they offer at the end of their text but
also to explore the significance of the use of such lexical items. 11
Sixth, both systematic associations specific to a particular text as well as
those inherent in the language system itself were recorded as systematic se-
mantic ties. 12 (See the previous section of the essay for an explanation of "sys-
tematic associations.") For example, in the first sentence of the high-rated es-
say in the Appendix, only thosestudentswho have donethe requiredwork functions
as a subordinate concept to all students.Even though these phrases might not
be related systematically to each other in any other text, they are so related in
this text; their relationship was therefore coded as a systematic semantic tie.
Text-specific systematic semantic ties occurred only in the high-rated essays in
OnLearningto WriteAboutIdeas 281

this study and were tallied together with other systematic semantic ties ac-
cording to type.
Seventh, multiple ties were noted when a group of words functioning as a
lexical element in one type of cohesive tie contained individual words that en-
tered into other types of cohesive relationships with previous lexical elements.
For example, three different ties were noted for only thosestudentswho havedone
the requiredwork. Not only does this lexical element cohere with all students,
but it also contains a repetition of the word studentsas well as a phrase required
workthat collocates with school.
Finally, only one type of tie was recorded in those cases where a lexical item
seemed to enter into different types of cohesive ties simultaneously with two
or more previous elements. A systematic semantic tie was recorded in prefer-
ence to a purely collocational one (e.g., fails to pass in sentence 1 in the low-
rated essay in the Appendix was listed both as a contrast to pass as well as con-
taining a repetition of pass, rather than as a collocation of 12 years of school);an
inter-sentence tie was recorded in preference to an intra-sentence tie (e.g.,
flunk in Sentence 2 in the low-rated essay was listed as a synonym of fails to
pass in Sentence 1 rather than as a collocation of test in Sentence 2).
Also tallied were the total number of words in each essay, the total number
of different words used in each essay, the total number of lexical ties in each
essay, the average number of words in each tie, and the total number of words
entering into at least one lexical tie in each essay. The reason for the last two
counts is as follows: the average number of words in a tie tells us whether the
lexical ties in an essay are created mainly by single words (2.0 words being the
minimum number for a tie) or also by longer semantic units (phrases or claus-
es); the total number of words in at least one lexical tie in an essay, as a
proportion of the total number of words in the essay, tells us what percentage
of the words contribute to the cohesiveness of the essay. 13

Findings

An examination of the grammatical subjects of each clause revealed large dif-

ferences between the low-rated and the high-rated essays. The data from this
analysis appear in Table 1. Of the total number of 71 clauses in the low-rated
essays, 54 subjects (or 76%) were pronominal. Of these 54 pronominal sub-
jects, 50 were personal or indefinite pronouns. Of these 50 personal or indefi-
nite pronouns, you was the subject of 9 clauses, I the subject of 10 clauses, and
we the subject of 9 clauses. Altogether 39% of the subjects of all clauses in the
low-rated essays were you, we, or I. Eleven subjects (or 15%) were personal
nouns, such as students,teacher,kids, and only 6 subjects (or 8%) were non-per-
sonal in nature. The words used as non-personal subjects in the low-rated es-
says are listed in Table 1.
In contrast, of the total number of 98 subjects of clauses in the high-rated
essays, only 50 (or 51%) were pronominal. Of these 50 pronominal subjects,
282 CollegeCompositionand Communication

Table 1

Types of Grammatical Subjects of Clauses in Six High-Rated and

Six Low-Rated Essays

Low-RatedEssays High-RatedEssays
Total (% of Total Total (% of Total
Number Number) Number Number)
GrammaticalSubjects 71 98
PronominalSubjects 54 (76%) 50 (51%)
Personalor Indefinite
PronominalSubjects 50 (70%) 36 (37%)
You 9 (13%) 0 (0%)
1 10 (14%) 5 (5%)
We 9 (13%) 0 (0%)
PersonalSubjects 11 (15%0) 21 (20%)
Non-PersonalSubjects 6 (8%) 27 (28%)
WordsUsed as Non-Personal
Subjects reportcard, reason, depart- purpose, jobs, type, state,
ment, test, material test, Massachusetts,system,
testing, majority, idea, fu-
ture, work, grade, schools,
diploma

only 36 were personal or indefinite pronouns. Of these 36 personal or indefi-

nite pronouns, you or we was never the subject of a clause and I was the subject
only 5 times. Altogether 21 subjects (or 20%) were personal nouns, and 27
(or 28%) were non-personal. The words used as non-personal subjects in the
high-rated essays are also listed in Table 1.
The analysis of ways of achieving lexical cohesion in these essays also
showed differences (see Table 2). In the low-rated essays, the total number of
all systematic semantic ties constituted 65% of all lexical ties, whereas they
constituted 79% of all lexical ties in the high-rated essays. While almost all
essays contained many examples of repetition, the low-rated ones contained
few examples of other types of systematic semantic ties, e.g., the essay labeled
A-3 contains no other systematic semantic ties, while the one labeled C-14
contains only one other such tie. The high-rated essays tended to contain at
least several examples of these other types. Only one multiple tie was found in
the low-rated essays; 15 multiple ties were found in the high-rated essays
(e.g., in the high-rated essay in the Appendix, othersystemin Sentence 8 is re-
lated through opposition to presentsystemin Sentence 6 and contains a repeated
word as well). The low-rated essays contained on the average 82 words and 54
 Table 2

Number of Each Type of Lexical Tie, Total Number of Lexical Ties, Average Number of
of Words Entering into Lexical Ties, Total Number of Words, and Total Number of Diff
High-Rated Essays

Type of LexicalTie
Total
Number Avera
k -e ^n of Numb
0 0
C CC
Lexical of
o
.*~? C
0
000?~~t
?r
aa Ct
!r:
.. Ties Word
?** V
4*
.2. .S
. (as a % of in
o 04
0, 3.1u 0t S -c5 Number Lexic
A4
CA
c:
0 Q p^ Q v-
o

C of Words) Ties
Low-Rated AU vAW AU
Essays A WS A W A WV~~
AW A W A W A W
H-22 8 5 1 1 1 1 00 0 00 0 1)
(1 10 10 2.9
H-4 7 2 00 00 00 00 2 3 (0) 9 5 3.0
A-3 8 1 00 0' 00 00 00 2 3 (0) 10 4 2.2
C-14 5 0 01 0 0 00 00 3 1 (0) 8 2 2.0
A-14 1 0 00 01 00 00 30 (0) 4 1 2.4
H-9 3 000 1 0 00 10 2 3 (0) 7 3 2.0
Total 32 8 1 1 2 2 00 1 0 12 13 (1) 48 25(15%)
High-Rated
Essays
F-21 18 2 2 0 10 10 00 1 4 (0) 22 6 3.5
C-8 15 1 5 1 3 0 1 0 1 1 2 4 (2) 27 7 3.6
B-15 9 2 3 0 0 0 06 04 2 2 (4) 14 14 3.3
G-19 11 2 1 0 1 0 0 3 0 0 5 1 (2) 18 6 3.9
H-68 1 0 0 0 1 1 0 1 1 1 6 (1) 11 9 3.3
D-19 21 5 4 4
40 0 1 0 0 3 6 (6) 33 12 3.7
Total 82 13 15 1 9 1 4 10 2 6 14 23 (15) 125 54(21%)

A = acrosssentences
W = within sentences
284 CollegeCompositionand Communication

different words; the high-rated essays contained on the average 145 words and
84 different words. Inter-sentence ties constituted 66% of the total number of
lexical ties in the low-rated papers and 71% in the high-rated papers, a small
difference suggesting that the overall ratio of inter-sentence to intra-sentence
ties may not have developmental significance.
The average number of words in lexical ties in the low-rated papers was
2.4; in the high-rated essays, it was 3.6. In the low-rated essays, the total
number of lexical ties constituted 15% of the total number of words, i.e., one
tie every 6.6 words, while the total number of words that entered into at least
one lexical tie constituted 22% of the total number of words. On the other
hand, in the high-rated essays, the total number of lexical ties constituted
21% of the total number of words, i.e., one lexical tie every 4.9 words, but
the total number of words that entered into at least one lexical tie in these es-
says constituted 44% of the total number of words. In other words, not only
were the writers of the high-rated essays creating more lexical cohesive rela-
tionships relatively and absolutely than the writers of the low-rated essays (a
finding that has been corroboratedin other research), but, more importantly,
they were also using longer semantic units to create these relationships, with
almost half of their words contributing to the cohesiveness of their essays.

Discussion

Many of the ideas expressed in the high-rated and low-rated essays were sim-
ilar; all these students have fundamentally the same understanding of the
world in which they live. Yet, the writers of these two groups of essays were
clearly using words to create meaning in different ways. What were the writ-
ers of the high-rated essays doing? First, they were using more different words
than were the writers of the low-rated essays, and they were using more of
them as the grammatical subjects of clauses. Moreover, they were making as-
sertions as often about nouns or noun phrases as they were about pronouns,
and more often about concepts or objects than about people. The writers of
the high-rated essays were also creating proportionately more lexical ties of all
kinds within and across sentences, frequently using phrases and clauses as well
as individual words to create these ties, with individual words often entering
simultaneously into different types of ties in semantic units of varying
lengths.
What these writers were doing to construct meaning was reflected in their
essays in a number of ways. The high-rated essays generally contained a clear
introduction that referred to the topic, an explicit statement of the writer's
position, a development of the reason(s) for this position that often referred
back to the topic and to the writer's position, and an explicit concluding
statement of the argument. The writers of the high-rated essays also expanded
and clarified their ideas with examples or details. Their arguments tended to
center on the positive relationship of competency testing to the occupational,
On Learning to Write About Ideas 285

political, and social needs of students in later life or on the question of fairness
for students currently in school, i.e., the wisdom of rewarding students for
working harderor for taking harder courses.
On the other hand, not only did the writers of the low-rated essays use a
smaller number of different words than the writers of the high-rated essays,
they also used very common words and an extremely high proportion of pro-
nouns of various kinds, with few lexical ties other than repetition connecting
their ideas. Few low-rated essays had explicit introductory statements address-
ing the topic or had explicit conclusions to an argument; in several, the con-
clusion was simply a reiteration of the writer's original position. In a few es-
says, one or two reasons supporting the position taken by the writer were
simply stated and not developed at all. In several others, the writer developed
his/her argument with only a few brief statements. It should be noted, howev-
er, that the coherence and organization of some low-rated essays were not so
inferior to those of some high-rated essays as the disparity in their ratings
would suggest. Poor control of sentence structure and of conventions of writ-
ten language probably contributed as much to the rating a low-rated paper re-
ceived as did its relative lack of structure and coherence.
In the low-rated essays, common nouns that referred to people and pro-
nominal references to the writer or to a partner in a dialogue were most often
used as subjects of clauses. Few subjects were nouns that referredto objects or
concepts. In other words, the writers of the low-rated essays tended to make
assertions about themselves, their audience, or people, not about non-personal
concepts. Their heavy use of first- and second-person pronouns suggests that
they viewed essay writing more as an engagement in a dialogue with another
speaker than as a transaction with a reader. Their arguments tended to center
on the question of fairness for poor students, i.e., whether competency testing
would penalize students who had received poor teaching or who had poor test-
taking ability. No argument in a low-rated paper went beyond the student's
experience; a few writers commented directly on the effect of competency test-
ing upon themselves.
As an explanation of some of their findings, Witte and Faigley suggest that
the writers of the low-rated papers in their study do not have "working vocab-
ularies capable of extending, in ways prerequisite for good writing, the con-
cepts and ideas they introduce in their essays" (p. 198). However, lack of a
working vocabulary may not necessarily be the whole explanation for the
failure of the writers of the low-rated papers to elaborate their ideas. If we ex-
amine the words used as the grammatical subjects of clauses in the high-rated
essays in this study, it seems probable that all these words are within the read-
ing vocabulary of all the writers of the low-rated essays in this study; more-
over, many of them were used in these essays in other than subject position.
Yet, the writers of these low-rated essays rarely used them as the focus for
their sentences. Instead, they or their audience tended to become the focus for
their statements through the use of personal pronouns-the typical subjects of
conversational utterances. Overuse of the structures of conversational utter-
286 CollegeCompositionand Communication

ances may have far-reaching consequences. Roger Cayer and Renee Sacks, in a
study of basic writers at the college level, noted that a major difference be-
tween oral and written language is the greater expansion in writing of the sub-
ject portion of the utterance. 14John Mellon, in a discussion of research on
sentence-combining, also noted that growth in the length of dominant NPs
(the subjects or objects of verbs in main clauses or the objects within preposi-
tional phrases modifying verbs in main clauses) is a characteristicof growth in
writing ability, and, according to Mellon, reflects growth in the conception
and complexity of ideas. 15 The use of personal and indefinite pronouns as the
subjects of clauses almost automatically precludes the expansion of the subject
portion of an utterance and, thus, may preclude some of that growth. It may
also preclude many possibilities for lexical cohesion as well.
Too much reliance on conversational structures by developing writers of es-
says may retard growth in writing ability in another way. Lunsford found in
her researchthat basic writers used a vocabulary characterized"by a high per-
centage of personal pronouns, especially those relating to first person, by a rel-
atively low degree of nominalization, and by the use of concrete diction and
simple concepts" (p. 287). These writers showed a "tendency to egocentricity,
to focus in their writing most often on themselves" (p. 285). Although Witte
and Faigley did not note the types of words used by the writers of the low-
rated essays in their research, they noted that these writers "seem to lack in
part the ability to perceive and articulate abstract concepts with reference to
particular instances, to perceive relationships among ideas, and to reach be-
yond the worlds of their immediate experience" (p. 199). It is possible that
the use of an abstraction or generalization to govern the development of one's
ideas and the mental playing with abstract verbal concepts that is charac-
teristic of mature stages of essay writing may have its intellectual roots in the
simple act of focusing on non-personal objects or concepts as the subjects of
predicates. Writers who appear to rely on models derived primarily from di-
alogue when they attempt to structure expository statements may be limiting
their ability to use concepts and objects as the initiators of actions in their
writing. As a consequence, they may also be limiting their ability to concep-
tualize ideas or objects as manipulating, modifying, or otherwise influencing
other ideas or objects in their writing. The failure of poor writers to develop
and elaborate ideas in their writing may be caused in part by a meager vocabu-
lary. It may also be caused in part by the ways in which they are accustomed
to use the words they know. 16
The writers of the low-rated essays in this study, and others similar to
them, seem to need more familiarity with the patterns of formal written Eng-
lish. Where are the models for them? The social context for the speech act is
not apt to provide them; conversation is, in fact, not intended to serve the
same purposes as written texts and, therefore, does not require its semantic
and structural features. As Jean Simon suggests in his five-year study of chil-
dren's writing development, models of the semantic and structural features of
written language can be found primarily in the written language the student
On Learning to Write About Ideas 287

is provided to read. 17 Weak writers clearly need more reading experiences

with well-written exposition. Although there is little research on this topic,
the little that has been done suggests that poor writers are apt to be poor read-
ers. 18 Weak writers also need to be guided away from the structures of conver-
sational utterances by pedagogical strategies that can help them frame ideas in
more intellectually beneficial ways. 19 Probably weak writers need both in-
creased exposure to well-written formal English and carefully-structuredwrit-
ing assignments to help them make the shift from self- or person-centered dis-
course to idea-centered writing. In fact, ease in using concepts and objects as
the subjects of predicates may be necessary before students can move beyond
an egocentric perspective to decentered writing and thinking.20

Speculations about the Nature of Growth in Ability at Essay Writing

There appear to be two basic theories about the development of the ability to
understand and use written language. In the first theory, written language is
considered either as spoken language written down or as an alternate but par-
allel form of spoken language; in either case, spoken and written language are
considered identical in system and substance.21 According to this theory, ex-
perience with spoken language always determines meaning in both reading
and writing; what skilled readers/writersunderstand or write is never seman-
tically and syntactically more complex than what they have absorbed from
their experience with spoken language.
In the second theory, written language-more specifically, the language of
formal schooling-is seen as qualitatively different from oral language.22 Ac-
cording to this theory, although experience with the spoken language of oth-
ers determines meaning in beginning reading and writing, the written lan-
guage of others becomes a far more powerful source of influence on the
development of meaning as the student proceeds to higher levels of literacy;
the language that developing readers/writersunderstand or use, although ini-
tially dependent upon their experiences with spoken language, becomes in-
creasingly less dependent on spoken language and more directly influenced by
their experiences in reading written language.
The results of this study also suggest that good essay writing is not an alter-
nate but parallel form of spoken language; rather, it is a systematically differ-
ent way of creating meaning. Further, they seem to suggest a way of conceiv-
ing growth in writing ability that recognizes and incorporates the essential
features that distinguish the act of composing a text from the act of generating
an utterance-the writer's planning an overall semantic structure for a text
and his/her putting together and shaping ideas in language for others to read
within that semantic structure. In contrast to the writers of the low-rated pa-
pers in this study, the writers of the high-rated essays were creating longer se-
mantic units, placing a larger number of these units in cohesive relationships,
and establishing a greater variety of cohesive relationships among these units
spanning all portions of the text. In particular, their papers were just begin-
288 and Communication
CollegeComposition

ning to suggest a network of synonymous, contrasting, and inclusive rela-

tionships connecting all parts of their essays-relationships that would have
been noted only as repetitions, if noted at all, if analysis had been confined to
the word level alone or to language-based ties only, rather than to both lan-
guage- and text-based ties. For example, in the high-rated essay in the Appen-
dix, we see an opening section with inclusive ties (relationships between su-
perordinate, subordinate, and coordinate members of a set) created by the
writer's decision to establish a general category of students-all students-and
two subcategories-students who do the required work and students who
come to school only because they have to be there-as the focus of her argu-
ment. The opening section is followed by a series of contrasting phrases which
serve to link statements about the effects of a proposed system and the current
system for obtaining a high school diploma either on all students or on mem-
bers of either subcategory. These contrasting phrases extend from this systemin
Sentence 3 to presentsystemin Sentence 6 and to othersystemin Sentence 8, and
collaborate with a series of synonymous phrases referring to the writer's major
concern-the first subcategory. These synonymous phrases extend from only
thosestudentswho have donethe requiredwork in Sentence 1 to thosewho deserveit
in Sentence 5 and to studentswho try harderthan othersin Sentence 6 and to one
who tries harderin Sentence 9. The writer's concern for the first subcategory
culminates in a final sentence that is linked in several ways to the opening
sentence and to the topic itself, thus completing the structure of the essay. In
the final sentence, tests, giving, and Massachusettsare repetitions of tests, give,
and Massachusettsin Sentence 1, while otherstates functions contrastively with
thestateof Massachusetts in Sentence 1.
By linking a large number of phrases and clauses as well as words in various
ways, the lexical ties in the high-rated essay create a texture whose richness
stands in sharp contrast to the sparsenessof the texture in the low-rated essay.
All the high-rated essays in this study differed from the low-rated essays in the
number and variety of interconnections among the semantic units they con-
tained, and, hence, in the richness of the texture they created. This richness
may be gauged by noting both the number of different ways in which each
sentence in an essay is linked to other sentences in the essay and the number of
other sentences to which it is related. For example, from the protocols in the
Appendix, we find that each sentence in the high-rated essay is directly relat-
ed, on the average, to three other sentences and by more than one type of sys-
tematic semantic relationship; in contrast, each new sentence in the low-rated
essay tends to be related to only the first and/or the previous sentence, pri-
marily through repetition.
Overall, the differences in the ways in which the developing writers in this
study used words to construct meaning in their essays suggest that growth in
the ability to write expository discourse is growth in the ability to present in-
formation and ideas with an increasingly larger number of increasingly inter-
connected semantic units of increasing length. This general hypothesis may
explain why the use of a larger or more diverse vocabulary is an indication of
On Learning to Write About Ideas 289

growth or quality in writing. Poor writers may write on the average so much
less than good writers, and younger writers may write on the average so much
less than older writers, not only because they have smaller vocabularies but
also because they cannot yet readily make assertions about non-personal ob-
jects or concepts and use different kinds of lexical cohesive relationships with
words, phrases, and clauses to control extended stretches of expository dis-
course. It seems likely that (1) the more concepts a writer can form to make
assertionsabout, the more interconnections among these concepts there should
be; (2) the larger the number of different words used in an essay, the more lex-
ical cohesive ties there should be; and (3) the more this network of cohesive
ties develops, both within and across sentences, with increasingly longer se-
mantic units entering into many different types of ties spanning all portions of
a text, the better organized and developed the essay should be. These specific
hypotheses could easily be investigated in longitudinal case studies of develop-
ing writers. The increasing use of non-personal objects or concepts as the sub-
jects of their sentences and of different types of systematic semantic rela-
tionships among these sentences may well be the best indications of these
writers' progress in learning to write about ideas.

Notes

1. For example, see the conclusions reached by John Brereton, Sondra Perl, and Richard
Sterling in the Final Report of the Writing DevelopmentProject, Fund for the Improvement of
Postsecondary Education, Washington, DC, May, 1981.
2. For example, frequency counts were used by Henry Rinsland in A Basic Vocabularyof Ele-
mentarySchoolChildren(New York: Macmillan, 1945) and by Robert Hillerich in A Writing Vo-
cabularyof ElementaryChildren (Springfield, IL: Charles C Thomas, 1978), the two most recent
compilations of children's writing vocabulary; no similar compilations exist at the secondary or
college level. Among the measures used by Cary Grobe in "Syntactic Maturity, Mechanics, and
Vocabulary as Predictors of Quality Ratings," Researchin the Teachingof English, 15 (February,
1981), 75-86, were indices of diversity in vocabulary and type-token ratios.
3. Lee Odell, "Measuring Changes in Intellectual Processes as One Dimension of Growth in
Writing," in Evaluating Writing. Describing,Measuring,Judging, ed. Charles R. Cooper and Lee
Odell (Urbana, IL: National Council of Teachers of English, 1977).
4. See Andrea Lunsford, "The Content of Basic Writers' Essays," CCC, 31 (October, 1980),
and Bruno Snell, "The Forging of a Language for Science in Ancient Greece," ClassicalJournal,
56 (November, 1960), 50-60.
5. Michael A. K. Halliday and Ruqaiya Hasan, Cohesion in English (London: Longman
Group Ltd., 1976).
6. Robert Morris Hopkins, "Popular Scientific Discourse: A Rhetorical Model for Teaching
Writing and Reading," Diss. University of Missouri-Columbia, 1979, and Stephen Witte
and Lester Faigley, "Coherence, Cohesion, and Writing Quality," CCC, 32 (May, 1981),
189-204.
7. Sandra Stotsky, "Types of Lexical Cohesion in Expository Writing: Implications for De-
veloping the Vocabulary of Academic Discourse," CCC, 34 (December, 1983), 430-446.
8. Marion Crowhurst, "Cohesion in Argumentative Prose Written by Sixth-, Tenth-, and
Twelfth-Graders," paper presented at the Annual Meeting of the American Educational Re-
search Association, Los Angeles, California, April, 1981.
and Typesof Present-Day
9. Hans Marchand,TheCategories 2nd ed.
EnglishWord-Formation,
(Munich, Germany: C. H. Beck, 1969).
10. In the conclusion to her dissertation, Mary Ann Eiler recommended that future research
290 CollegeCompositionand Communication

examine both intra-sentence and inter-sentence lexical cohesive ties ("Meaning and Choice in
Writing about Literature: A Study of Cohesion in the Expository Texts of Ninth Graders,"
Diss. Illinois Institute of Technology, 1979).
11. I could find no working definition of the unit of analysis that has been used to analyze
and tally lexical cohesive ties in previous research on cohesion. Halliday and Hasan themselves
offer no definition of "lexical item," and, even though their text frequently uses "word" to refer
to an item in a lexical tie, one may find phrases and clauses, as well as individual words, as
either cohesive items or presupposed items in the sample protocols at the end of their text, with
no explanation in their text for these variations.
12. Eiler also tallied both text-specific and language-based systematic semantic ties in her
research.
13. The scoring and tallying of lexical cohesive ties seemed to require the resolution of a
large number of important methodological questions. Surely a consensus among researchers on
all these questions is needed before the results of studies that analyze the ways in which writers
achieve lexical cohesion can validly be compared.
14. Roger Cayer and Renee Sacks, "Oral and Written Discourse of Basic Writers: Sim-
ilarities and Differences," Researchin the Teachingof English, 13 (May, 1979), 121-128.
15. John Mellon, "Issues in the Theory and Practice of Sentence Combining: A Twenty-
Year Perspective," in SentenceCombiningand the Teachingof Writing, ed. Donald Daiker, Andrew
Kerek, and Max Morenberg (Conway, AR: University of Central Arkansas, 1979).
16. Recent work on thematic progressions by Barbara Glatt ("Defining Thematic Progres-
sions and Their Relationships to Reader Comprehension" in What Writers Know:The Language,
Process,and Structureof Written Discourse, ed. Martin Nystrand [New York: Academic Press,
19821) suggests the desirability of a flow of information through sentences alternating between
what is given and what is new. It will be difficult for developing writers to provide a flow of
coherent discourse for their readers if they are unable to use the object of a predicate verb in one
sentence as the subject of the predicate in a succeeding sentence.
17. Jean Simon, "Evolution Genetique de la Phrase Ecrite chez lEcolier," Diss. University
of Paris, 1970.
18. For a comprehensive review of the research literature on reading/writing relationships,
see Sandra Stotsky, "Research on Reading/Writing Relationships: A Synthesis and Suggested
Directions," Language Arts, 60 (May, 1983), 627-642. I was informed by a teacher familiar
with all the students whose writing was examined in this study that almost all of the writers of
the low-rated essays were of below average reading ability but were not in special education
classes.
19. See, for example, the essay by Judith Boyce ("The Case for Complete Sentence Answers
to Questions," The Leaflet, 82 [Spring, 1983), 13-19) for several techniques for helping stu-
dents to write clear and complete expository sentences, and the composition program developed
by Anne Obenchain (Links to ForcefulWriting. Part Oneand Part Two [Reston, VA: Validated
Writing Systems, 1977}) for ways in which the writing of expository sentences and paragraphs
may be guided and sequenced.
20. For an overview of the historical development of academic language and the shift from
egocentric to ideocentric discourse, see Sandra Stotsky, "From Egocentric to Ideocentric Dis-
course: The Development of Academic Language," in Issuesin Literacy. A ResearchPerspective,
Thirty-fourth Yearbook of the National Reading Conference, ed. Jerome Niles and Rosary Lalik
(The National Reading Conference, Inc., 1985).
21. For a detailed comparison of these two theories, see Sandra Stotsky, "A Comparison of
Two Theories of Written Language Development," in Comprehending Oral and Written Language,
ed. Rosalind Horowitz and S. J. Samuels (New York: Academic Press, forthcoming). The first
theory is drawn from the work of the following writers: James Moffett and Betty Jane Wagner,
Student-Centered LanguageArts and Reading, K-13, 2nd ed. (Boston: Houghton Mifflin, 1976);
Frank Smith, "The Relation between Spoken and Written Language," in Foundationsof Lan-
guage Development,Vol. 2, ed. Eric Lenneberg and Elizabeth Lenneberg (New York: Academic
Press, 1975); Kenneth Goodman, "Behind the Eye: What Happens in Reading," in Language
and Literacy:The SelectedWritings of KennethS. Goodman, Vol. 2, ed. F. V. Gollasch (London:
Routledge & Kegan Paul Ltd., 1982).
22. The second theory is drawn from the work of the following writers: Lev Vygotsky,
Thought and Language, trans. Eugenia Hanfmann and Gertrude Vakar (Cambridge, MA:
On Learning to Write About Ideas 291
M. I. T. Press, 1962) and Mindin Society:TheDevelopment ed.
Processes,
of HigherPsychological
MichaelCole, VeraJohn-Steiner,SylviaScribner,and Ellen Souberman(Cambridge,MA: Har-
vardUniversity Press, 1978); AlexanderLuria, "SpeechDevelopmentand the Formationof
MentalProcesses,"in A Handbook ed. MichaelCole and Irving
SovietPsychology,
of Contemporary
Maltzman(New York: Basic Books, 1969); JeromeBruner, Rose Olver, and PatriciaGreen-
field, Studiesin CognitiveGrowth(New York:John Wiley and Sons, 1966). For a somewhat
more detaileddescriptionof this second theory, see SandraStotsky, "A MeaningfulModel of
Written LanguageDevelopment for Teachers,"in fforum:Essayson Theoryand Practicein the
Teaching of Writing,ed. PatriciaStock(Montclair,NJ: BoyntonlCookPublishers,Inc., 1983).

Appendix

TOPIC: Many states now require students to demonstrate a certain level of skill on
tests of reading, writing, and mathematics in order to receive their high
school diploma. Only those students who pass the tests at the minimum level
would receive a high school diploma. The states are giving a certificate of at-
tendance instead of a diploma to those students who failed the tests and do
not wish to continue in school any longer. The states are giving extra help to
students who want to try to pass the tests a second time.
The state of Massachusetts is now considering giving such tests. Some are in
favor of having these tests, and some are opposed. Take one side of this issue.
Write an essay in which you state your position and defend it.

High-Rated Essay*

11 believe the state of Massachusetts should give the competency tests to all students
because it will give all students an incentive to come to school and it gives credit to
only those students who have done the required work. 2Some students come to high
school only because they have to be there. 3Now, with this system, one has to pay
attention in school in order to pass the test given in his/her senior year. 4All want to
show that they have successfully completed their years in high school by graduating;
therefore, they must study. 5This test only gives credit to those who deserve it, which
only seems fair. 6As the present system works, it doesn't give enough credit to stu-
dents who try harder than others, as all graduate that have gained enough credits and
pass the needed subjects. 7Some aren't ready still after completing the required work
but are still handed the diploma! 8Although, if the other system takes effect, each
student must prove him/herself able to graduate. 9This seems fairer to one who tries
harder. '0Overall, since the tests are already in other states, Massachusetts should be-
gin giving them.

Cohesive Sentence Presupposed Sentence Numberof

Item Number Item Number Type** Wordsin Tie
students 1 competency tests 1 Cw 3
give 1 give 1 Rw 2
all students 1 all students 1 Rw 4
school 1 students 1 Cw 2
gives 1 give 1 Rw 2
only those 1 all students 1 I-R-Cw 11
students who
have done
the required
work
(continued)
292 CollegeCompositionand Communication

Cohesive Sentence Presupposed Sentence Numberof

Item Number Item Number Type** Wordsin Tie
some students 2 all students 1 I-R 4
come to 2 come to school 1 R 6
high school
school 3 high school 2 R 2
pass 3 tests 1 C 2
test given 3 give . . tests 1 R 4
senior year 3 school 3 Cw 3
years 4 year 3 R 2
high school 4 school 3 C 2
graduating 4 high school 4 Cw 2
study 4 test 3 C 2
test 5 test 3 R 2
gives credit 5 gives credit 1 R 4
those who 5 only those 1 S 13
deserve it students who
have done
the required
work
present system 6 this system 3 O-R 4
doesn't give 6 gives credit 5 O-R 6
enough credit
students who try 6 those who 5 S 10
harder than deserve it
others
graduate 6 graduating 4 R 2
enough credits 6 enough credit 6 RW 4
pass 6 pass 3 R 2
needed subjects 6 required work 1 S 4
completing 7 done 4 S 2
required work 7 required work 1 R 4
diploma 7 required work 7 Cw 3
other system 8 present system 6 O-R 4
each student 8 all students 1 R 4
graduate 8 graduate 6 R 2
seems fairer 9 seems fair 5 R 4
one who tries 9 students who 6 R 10
harder try harder
than others
tests 10 tests 1 R 2
other states 10 the state of 1 O-R 5
Massachusetts
Massachusetts 10 Massachusetts 1 R 2
giving 10 give 1 R 2
N.B. The analysisof cohesiveties does not pick up the inconsistencyin the writer'sargument.
The writerfavorsgiving competencytests because, in her view, they will give credit to only
those who do the requiredwork;yet, the writerfrownsupon the presentsystem for grantinga
diplomato those who do not have the necessaryskills to graduate,even though they have done
the requiredwork.
On Learning to Write About Ideas 293

Low-Rated Essay*

1I think if a person passes all 12 years of school and then fails to pass their test, that
person will not be allowed to receive one's diploma. 2It's wrong to have to take a test
because some people freeze up and flunk the test even though the material on the test
was known by that person. 3Also people do forget material that was taught in the ear-
lier years of school. 4The test for one's diploma should be done away with in the states
that use the test. 5Also the test should not be allowed in Massachusetts.

Cohesive Sentence Presupposed Sentence Numberof

Item Number Item Number Type** Wordsin Tie
12 years of school 1 passes 1 CW 5
fails to pass 1 passes 1 O-] 4
test 1 pass 1 CW 2
person 1 person 1 RW 2
diploma 1 test 1Cw 2
test 2 test 1 R 2
flunk 2 fails to pass 1 S 4
test 2 test 2 RW 2
test 2 test 2 RW 2
person 2 people 2 Sw 2
people 3 people 2 R 2
material 3 material 2 R 2
years of school 3 years of school 1 R 6
test 4 test 1 R 2
diploma 4 diploma 1 R 2
test 4 test 4 RW 2
test 5 test 4 R 2
be allowed 5 be allowed 1 R 4
Massachusetts 5 the states that 4 0 7
use the test

N.B. The analysisof cohesiveties in this essay also does not pick up the lack of coherencebe-
tweenthe writer'sfirst statementand the argumenthe develops.The writerintendedto state in
his opening sentencethat it is wrong for a personnot to be allowedto receivea diploma if he
passesall 12 yearsof school, even though he fails to pass the test.
*All essayshavebeen edited for capitalization,punctuation,and spelling.
**R = repetition
S = synonymy
0 opposition
I = inclusion
D = derivation
C = collocation
The superscriptw indicatesthat the tie is within a sentence.