|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
EXPOSITORY WRITING SKILLS IN ELEMENTARY SCHOOL CHILDREN FROM
THIRD THROUGH SIXTH GRADES AND CONTRIBUTIONS OF SHORT-TERM
AND WORKING MEMORY
CYNTHIA S. PURANIK
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
Cynthia S. Puranik
This dissertation is dedicated to my family, especially my sisters Sandra Coelho
and Sylvia deSouza, who never stopped believing in me and to my dearest friends H. S.
Uday Kumar and Rajat Mittal who decided it was their responsibility to keep me
emotionally and intellectually together-this might not have been possible without them.
I thank my committee members, in particular Dr. Linda J. Lombardino and Dr. Lori
J. P. Altmann, who have mentored me through this entire process.
Finally, I would like to thank The School Board of Hillsborough County, Florida,
for granting me permission to conduct this investigation and to the students, parents, and
teachers who so graciously participated in this research study.
TABLE OF CONTENTS
A C K N O W L E D G M E N T S ......... ................................................................................... iii
LIST OF TABLES ......................... ................................. ........ vi
L IST O F F IG U R E S .... ......................................................... .. .......... .............. vii
A B STR A C T ... .................... ............................................ ... ....... ....... viii
1 INTRODUCTION AND REVIEW OF LITERATURE .............................................1
T h e P ro cess o f W ritin g ...................................................................... .....................
A n aly ses of W ritin g ................................................................. .......... ..............5
Genre and Method for Collecting Writing Samples ..........................................7
M em o ry ..................... ....................................................... ................ 9
Defining Short-term Memory and Working Memory ............... ... ..............10
Tasks that Measure STM and WM .................................................. 12
Relationship between Memory and Writing.......................................................13
Experimental Questions, Hypotheses and Predictions ............................................15
2 M E T H O D S ........................................................................................................... 2 0
S e ttin g ............................................................................................................2 0
P a rtic ip a n ts ........................................................................................................... 2 1
D ata C o lle ctio n ..................................................................................................... 2 1
P h a s e I ........................................................................................................... 2 1
P h a se II ............................................................................................ ............ 2 2
Dependent Variables for W writing ............................................................ ........... 24
D isc o u rse L ev el ............................................................................................. 2 5
T -u n it L ev e l ................................................................................................... 2 5
S en ten c e L ev el............................................................................................... 2 6
W ord L evel ........................................................................... 29
Transcription, Coding, and R liability ....................................................... 29
3 R E S U L T S ....................................................................................... 3 1
Discourse Level Measures..................... ...............32
T -u nit L ev el M easu res.............................................................. .....................32
Sentence Level M measures ............................................................................. 33
W ord L ev el M easu res............................................................... .....................34
Sum m ary of R esults......... ................................................................ ....... .... ... 4 1
4 DISCUSSION ....................................................... ........... .............. 43
Changes in Writing across Grades and Factors underlying Dimensions of Writing..43
M measures of Productivity.......................................................... ............... 44
Measures of Complexity ............... ............... .................... 44
A accuracy M measures ......................... .. .................. .. .. ...... ........... 46
M echanics of W writing ................................... ........................................... ...46
Summary of Changes in Writing across Grades and Dimensionality of Writing.......47
Relationship between Memory and Writing.............................. ....... ............ 48
C clinical Im plication s........ ................................................................ ... .... .. .... 50
Limitations and Future Directions .. .. ......................... .....................51
General Conclusions ......... .. ..................... ........ ... ............................. 53
A STUDIES EXAMINING WRITING ........................................ ..................... 54
B IR B C O N SE N T F O R M S ......................................... .............................................60
C TESTIN G PR O TO COL ............... .................................................... ............... 67
D PROTOCOL FOR LINGUISTIC ANALYSIS .................................. ...............74
R E F E R E N C E S ........................................ ........................................................... .. 8 0
B IO G R A PH IC A L SK E TCH ..................................................................... ..................89
LIST OF TABLES
1 W written language variables by level....................................................................... 30
2 Means and standard deviations for writing measures at the discourse, T-unit,
sentence, and w ord level. ............................................... .............................. 35
3 Four-factor solution of expository written microstructure.......................................37
4 Standardized regression coefficient factor loadings for the four-factor solution.....37
5 Means and standard deviations for independent measures. ....................................39
6 Correlations between memory, vocabulary measures and writing variables..........39
7 Forward stepwise regression with memory and vocabulary scores as predictor
v a riab le s .......................................................................... 4 1
LIST OF FIGURES
1 A simplified and modified schematic of the writing process originally based on
the H ayes and Flow er (1980, 1987) m odel. ........................................ ...................3
2 The three-component model of working memory proposed by Baddeley and
H itch (1974). ...................................................... ................. 12
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
EXPOSITORY WRITING SKILLS IN ELEMENTARY SCHOOL CHILDREN FROM
THIRD THROUGH SIXTH GRADES AND CONTRIBUTIONS OF SHORT-TERM
AND WORKING MEMORY
Cynthia S. Puranik
Chair: Linda J. Lombardino
Major Department: Communication Sciences and Disorders
The primary goal of this study was to evaluate multiple dimensions of written
language in children at 4 grade levels and to explore the relationship between written
language variables and short-term and working memory. Written language samples were
collected from a hundred and twenty children from grades 3 through 6 using an
expository text-retelling paradigm. The written language samples were analyzed for 13
microstructural elements at the discourse, T-unit, sentence and word levels.
This study was designed to address the following three experimental questions: (1)
How does writing develop between grades 3 and 6 at different levels of language?, (2) Do
the thirteen language variables cluster into factors that categorize the dimensions of
written language?, and (3) What is the relationship between short-term memory (STM),
working memory (WM) and writing?
The results of this study show that measures of text length, lexical diversity, and
syntactic complexity improve steadily with age. Students in grade 3 showed poorer
performance than students in grades 4, 5, and 6; students in grade 4 showed poorer
performance than students in grades 5 and 6. It appears that children reach a plateau in
their writing at about grade 5 similar to one described in studies examining oral language.
An exploratory factor analysis confirmed that writing is a complex process and
can be categorized into four factors comprising productivity, complexity, accuracy, and
mechanics. Consistent with previous research, WM showed a higher correlation with text
generation processes and STM showed a higher correlation with transcription processes.
The results of this study provide preliminary data on changes in writing of normally
developing children from grades three through six and add to the existing evidence
regarding the differential role of WM and STM on writing. The findings of this study
have important clinical implications for both the assessment and treatment of written
INTRODUCTION AND REVIEW OF LITERATURE
Research on writing has lagged behind investigations of reading acquisition and
cognitive processes associated with reading. As noted by Treiman and Kessler (2005),
"Literacy research has concentrated on reading, but without the ability to write, a person
could scarcely be called literate" (p. 120). This study investigated changes in the writing
of children from grades 3 through 6 and examined the dimensionality of writing.
Additionally, this study examined the contribution of short-term memory and working
memory on writing because research suggests that memory plays an important role in
written language skills.
The Process of Writing
Writing requires the activation and coordination of orthographic, graphomotor and
several linguistic skills including, but not limited to, semantics, syntax, spelling, and
writing conventions (Scott, 1999; Singer & Bashir, 2004). Writing involves three primary
processes, (1)planning- the conception or thinking up of ideas a writer wishes to convey;
(2) translating- encoding thoughts and ideas into meaningful words, phrases, clauses, and
sentences; and finally (3) reviewing- reconsidering and revising what is written to the
writer's satisfaction (Hayes & Flower, 1980, 1987). These processes are recursive rather
than linear, meaning that they are not necessarily followed in any particular order and are
continuously interacting. Writing is a skill that develops after the foundations of reading
have been laid. Since writing requires the management and coordination of multiple
cognitive-linguistic processes simultaneously, it requires explicit instruction for its
Figure 1 contains a modified and simplified schematic of the Hayes and Flower
(1980, 1987) model of writing depicting the three processes or stages of writing. Their
model also specifies sub-processes for the planning and reviewing phases but not for the
translating phase. Sub-processes for the planning phase include generating ideas to write
about, setting goals for writing and organizing ideas such that writing goals are achieved.
Sub-processes involved in reviewing include reading what has been written, editing and
revising to ensure that the writer's goals have been met. Beminger and her colleagues
(Berninger, 1999; Beminger & Hooper, 1993; Berninger et al., 1992) suggested
modifications to the original Hayes and Flower (1980, 1987) model based on their cross-
sectional studies of nearly 900 children from grades 1 through 9 showing that for
children, the translation process involves two separate sub-components: text-generation
and transcription. Text-generation refers to the process by which the writer translates
his/her planned ideas into meaningful chunks of sentences, phrases, and words; while
transcription refers to the actual mechanics of converting those sentences, phrases, and
words into written symbols and includes spelling, handwriting, and punctuation. The
Hayes and Flower (1980, 1987) model was based on mature or expert writing and
assumed that writers had mastered the transcription processes of writing. While that holds
true for mature or expert writers, research by Berninger and her colleagues (Berninger,
1999; Berninger & Hooper, 1993; Berninger et al., 1992) has shown that children in
elementary grades have not mastered transcription skills thus affecting their writing skills
and output. Therefore, any model of writing in children must necessarily include the two
sub-components of the translation stage of writing.
Figure 1. A simplified and modified schematic of the writing process originally based on
the Hayes and Flower (1980, 1987) model.
Writing is a challenging activity for most children. The National Assessment of
Educational Progress (NAEP) has been conducting ongoing national assessments in
major academic subjects including writing. The NAEP has identified three achievement
levels for writing, (a) the Basic level implies partial mastery of the prerequisite
knowledge and skills that are fundamental for work at each grade; (b) the Proficient level
represents solid academic performance for each grade assessed; and (c) the Advanced
level signifies superior performance (U.S. Department of Education, 2003). For the year
2002, only 28% of 4th graders, 31% of 8th graders, and 24% of 12th graders performed at
or above the Proficient level of writing (U.S. Department of Education, 2003). These data
show that a large majority of school children acquire only the very basic writing skills.
Children with and without learning disabilities (LD) often experience writing difficulties
and exhibit poor writing skills (Baker, Gersten, & Graham, 2003).
In an effort to estimate the prevalence of writing problems among middle school
students, Hooper and Montgomery (1993) examined the writing of a large cohort of
middle school students located in three different regions of the United States. Using the
Spontaneous Writing portion of the Test of Written Language-2 (TOWL-2; Hammill &
Larsen, 1988), Hooper and Montgomery (1993) found a high incidence of writing
problems in these children who were not diagnosed with learning difficulties or in special
education classes. Only about 6% of those students were receiving some special
education services and none were in a full time special education classroom. If children
developing normally have difficulty with writing, this difficulty is expected to be marked
among children with learning disabilities. In fact, the extensive research by Graham and
colleagues (Graham & Harris, 1997; Graham, Harris, & Troia, 2000) with learning
disabled children has shown that most of these children struggle with some or all aspects
Several studies have examined written language samples of elementary and middle
school children with LD (Houck & Billingsley, 1989; Newcomer, Barenbaum, & Nodine,
1988; Nodine, Barenbaum, & Newcomer, 1985; Taylor, 1980). A compilation of these
studies is shown in appendix A. Conclusions drawn from these studies underscore that
learning disabled children exhibit less lexical diversity (Morris & Crump, 1982), produce
fewer sentences (e.g., Houck & Billingsley, 1989) and shorter T-units (e.g., Scott &
Windsor, 2000), include fewer components in their stories (Laughton & Morris, 1989;
Newcomer et al., 1988), produce less coherent stories (Graves, Montague, & Wong,
1990; Montague, Maddux, & Dereshiwsky, 1990) and demonstrate poorer performance
in writing conventions (Houck & Billingsley, 1989; Poteet, 1979) when compared to
normally developing peers. Investigations into the writing of children with a history of
spoken language impairment have shown that they use fewer words, make more syntax
and spelling errors, and produce writing that is grammatically less complex compared to
their age matched peers (Fey, Catts, Proctor-Williams, Tomblin, & Zhang, 2004; Gillam
& Johnston, 1992; Mackie & Dockrell, 2004). Examination of the writing of dyslexic
subjects has shown that they exhibit inferior performance when compared to age-matched
peers on measures of spelling and construction of grammatical sentences (Puranik,
Lombardino, & Altmann, in press).
Analyses of Writing
For many decades, speech-language pathologists (SLPs) have used analyses of
spontaneous oral language samples as part of their diagnostic methods for assessing the
spoken language capabilities of children with language impairments (Hunt, 1965; Loban,
1976; Paul, 2001). Comparing language-impaired children's abilities with their normally
developing peers aids in (a) the identification of language impairments, (b) an
understanding of the nature of these language difficulties, and (c) the formulation of
intervention goals (Nelson, 1998).
Various methods are used to elicit oral language samples and range from
spontaneous or casual conversations, spontaneous story generation (Nelson & Van Meter,
2002) to story retelling after listening to a story (e.g., Gazella & Stockman, 2003), and
story generating while viewing pictures (Vallecorsa & Garriss, 1990) or watching video
productions (Berman & Verhoevan, 2002; Scott & Windsor, 2000). Language sampling
has been used mainly with preschool and early elementary school children and a
spontaneous story telling format is the most commonly used method for eliciting oral
The use of systematic language sampling procedures for identifying and
documenting writing impairments has lagged behind the use of similar procedures in
testing the spoken language of children. One reason that the analysis of writing has
received little attention may be because the assessment of writing can be a time-
consuming and tedious task. Moreover, standardized assessments for writing are scarce
compared to assessment instruments for oral language and reading processes. Typically,
SLPs do not assess writing and therefore generally do not address this skill when
formulating intervention goals (Silliman, Jimerson, & Wilkinson, 2000). A primary
reason could be that traditionally, the role of SLPs has been confined to assessing and
treating disorders of oral language. Other major reasons are likely to include lack of age-
appropriate sampling procedures for older children and data for determining the quality
of the sample. Hux, Morris-Friehe, and Sanger (1993) suggest that the decreased use of
oral language sampling with older children may be due to lack of age-appropriate
guidelines. This lack of age-appropriate guidelines appears to be even more pronounced
in the case of writing. With the American Speech Language and Hearing Association's
(ASHA, 2001) recommendations for additional roles and responsibilities for SLPs,
greater attention must be given to empirically based methods for assessing and treating
written language disorders and underscores the need for establishing normative databases
for evaluating written language skills.
Genre and Method for Collecting Writing Samples
It is widely accepted that type of genre (i.e., narrative, expository) and nature of the
method for collecting a language sample may affect performance (Gazella & Stockman,
2003; Scott, 1989, 1999). One remarkable finding from the literature is that the
progression of written language skills across different tasks and the effect of situational
variables on various language measures in writing have not been documented (Hadley,
1998; Scott, 1994). On the whole, data on writing development in children with language
learning disabilities is meager (Silliman et al., 2000). Most studies have investigated only
a small number of variables such as mean length of T-units, total number of words,
number of different words. Studies on expository texts are scant; most studies have
analyzed written language samples using spontaneously generated narratives. As is often
the case, writing studies employ a range of methods for collecting language samples,
hence, comparisons of the performance of children across studies for clinical or
educational purposes lack precision and are difficult to interpret (Scott & Windsor, 2000).
Given the importance of expository writing for success in school, this lack of age-
appropriate norms for writing skills underscores an educational and clinical need for data
of this nature. This study was motivated by the need for expository data in school age
children and was designed to examine changes in the expository writing of children from
grades 3 through 6. A future goal is to standardize a text-retelling task that can be used by
SLPs as an efficient and scientifically validated assessment procedure for identifying
children who have deficient writing skills.
Previous investigations of written language have yielded retell data following an
audio-only condition, an audio-visual condition (a video was shown at the same time that
the story was heard), or following the reading of a paragraph (Gambrell, Koskinen, &
Kapinus, 1991; Gambrell, Pfeiffer, & Wilson, 1985; Morrow, 1985). Retelling has been
used successfully as an instructional strategy for improving writing (Gambrell et al.,
1991; Gambrell et al., 1985; Morrow, 1985). Retellings are more commonly used
following reading of a paragraph or short story to assess comprehension and provide an
efficient method for assessing and evaluating students' memories, reactions, and reading
comprehension (Gambrell et al., 1991; Gambrell et al., 1985; Harp, 2000).
Gazella and Stockman (2003) provided several supporting arguments for using
story-retelling task as a screener for assessing syntax in children's oral language. They
assert that story-retelling formats allow clinicians to control the stimulus input and
provide optimal potential to standardize procedures. Further, when clinicians are familiar
with the stimulus input they can more reliably assess the degree of information recalled
and the accuracy of propositions and inferences made. Similarly, Merrit and Liles (1989)
underscored the advantages of a story retelling over a story generation task for oral
language sampling. They found that retold stories are easier to transcribe and are often
longer, allowing for a greater number of story grammar components to be expressed.
Findings from research using retelling for oral language sampling indicate that it can be
an efficient tool for amassing data on syntactic complexity and story grammar elements.
In summary, retellings could provide much needed information on how students
handle text level content and hold compelling potential for assessing older children
because, in school settings, children are required to listen to new information through
listening to lectures and then summarize this new information in written form. The format
for this study closely resembles a "typical" academic exercise in which students are
required to comprehend and demonstrate learning of material presented orally by an
Findings from various research studies suggest that memory capacity is closely
linked to several academic and language skills. Memory capacity has been shown to play
a vital role in reading comprehension (Swanson & Beminger, 1996; Turner & Engle,
1989), vocabulary (e.g., Dixon, LeFevre, & Twilley, 1988) and scholastic development
and attainment (Gathercole, Pickering, Knight, & Stegmann, 2004; Towse, Hitch, &
Hutton, 1998). Specifically, the results of several investigations have shown that memory
is related to writing (Bourke & Adams, 2003; Hoskyn & Swanson, 2003; Swanson &
Beminger, 1996). Evidence of the relationship between memory and writing comes not
only from studies with normally developing children and adults but also from studies
with children and adults who have language, reading, and learning disabilities.
Writing is a complex task, requiring the intricate activation and coordination of
several cognitive-linguistic skills (e.g., Scott, 1999; Singer & Bashir, 2004). These
cognitive-linguistic skills include linguistic and meta-linguistic knowledge, prior content
knowledge, general thinking ability, processing speed, and short-term and working
memory (see Figure 1); variables that are believed to constrain or facilitate the writing
process (Hayes & Flower, 1980, 1987; Singer & Bashir, 2004). Although several
cognitive skills have been identified in models of writing, in this study, the focus will be
on the contribution of short-term memory (STM) and working memory (WM) to writing,
and important area of inquiry because children with spoken and/or written language
impairments have been shown to have deficiencies in STM and WM (de Jong, 1998; Ellis
Weismer, 1996; Ellis Weismer, Evans, & Hesketh, 1999; Gathercole & Baddeley, 1990,
1993a; Gillam, Cowan, & Day, 1995; Kibby, Marks, Morgan, & Long, 2004;
Montgomery, 1995; Torgesen, 1985).
Defining Short-term Memory and Working Memory
STM is that part of memory that is capable of holding little information for a
limited amount of time. It refers to the ability to hold information in immediate
awareness and then use it in a few seconds. It is generally thought of as a static buffer of
seven plus or minus two storage units (Miller, 1956). STM requires the storage and
retrieval of information and is critical for just about every human activity. For example, it
is impossible to engage in a conversation without keeping in mind what one's partner has
just said in order to respond appropriately. An often cited example of a STM task in
everyday life includes looking up a number in the telephone book and dialing the number
from memory. Similarly, while writing, STM is needed to hold a thought in mind while
the writer is planning to execute thoughts through print. Finally, another function
attributed to STM is that it is a quick route or a temporary store for information retrieved
from long-term memory (Carpenter, Miyake, & Just, 1994).
On the other hand, WM, a term first coined by Baddeley and Hitch (1974), refers to
the ability to hold information in immediate awareness while performing a mental
operation on the information. A transformation or inference is required on WM tasks
(Baddeley, 1986, 2003; Baddeley & Hitch, 1974; Daneman & Carpenter, 1980; Just &
Carpenter, 1992; Swanson, 1993). As shown in Figure 2, Baddeley and his colleagues
(Baddeley, 1986; Baddeley & Hitch, 1974) identified three important components of
WM; the central executive, the phonological loop, and the visuo-spatial sketch pad. The
central executive component is the most important, being responsible for regulating the
flow of information within working memory, coordinating and retrieving information
from other memory systems such as long-term memory, and storing and processing
information (Gathercole & Baddeley, 1993b). It is supported by two other components or
"slave systems" which are responsible for the processing and temporary maintenance of
material within a particular domain. Verbal information is held in a temporary, passive,
phonological storage buffer called the phonological loop. The phonological loop is
responsible for verbally coded information. In contrast, visuo-spatial sketchpad, the other
'slave system' is involved in the storage of visual or spatial elements. All three
components of WM operating together are important for a variety of mental tasks
including a wide range of linguistic skills.
To summarize, STM refers to the temporary storage or memory for information
that lasts only for a few seconds after one hears or reads something or receives a
stimulus, while WM refers to short-term memory when it is used to process information,
solve a problem or perform a task with the information (Cowan, 1996). To illustrate, in a
sentence recall task, a participant is given a short sentence and asked to repeat it. In a
STM task, the subject is required merely to hold the sentence very briefly in the
phonological loop before repeating it quickly. In a WM task, when the subject is asked to
correct an ungrammatical sentence, he or she is required to hold that information in the
phonological store (STM) while performing a mental operation (done by the central
executive) to correct the syntax of the sentences.
Figure 2. The three-component model of working memory proposed by Baddeley and
Tasks that Measure STM and WM
Since STM requires the storage and retrieval of information, it is typically
measured on memory span tasks, i.e., reproduction of information in an untransformed
fashion. Some common tasks used in research to measure STM are digit span, word span,
and nonword repetition. For example, in a typical digit span task, subjects are presented
with a string of numbers, which they are asked to repeat. The tasks always starts out easy,
generally with two stimuli, and gets progressively harder as the number of syllables,
words, or digits is increased.
WM refers to the combined storage and processing of information held in memory.
Tasks commonly used to measure WM are complex span tasks and include numbers
reversed, digit ordering, sentence/listening span, and reading span. In a numbers reversed
task, a subject hears a string of numbers, starting with three numbers, and is required to
sketch pad loop
repeat the numbers in the reverse order. What differentiates this WM task from the above
STM tasks is that the subject has to hold what he or she has heard in memory and then
complete some mental processing or manipulation, in this case remember them in a
reverse order. Digit ordering is similar to numbers reversed except the subject has to
repeat the numbers back from the smallest to the largest number. Sentence/listening span
tasks require that the subject listen to groups of sentences, provide answers to simple
comprehension questions, then recall the last word of the sentence. Again, as the test
progresses, the number of sentences in a group increases, increasing load on WM.
Finally, reading span is similar in format to sentence/ listening span, except that the
subject reads sentences rather than listens to them.
Relationship between Memory and Writing
Studies that have examined the contribution of WM and STM to writing tasks
suggest that that these two memory systems operate independently of each other (Cantor,
Engle, & Hamilton, 1991; Swanson, 1993). The findings of these few studies suggest that
WM is correlated with higher order aspects of writing (e.g., text generation) while STM
is correlated to lower-order skills (e.g., transcription). Text generation is considered a
higher order process, because it involves language processing at the word, sentence, and
text level, and requires translation of ideas in memory into linguistic representations.
Transcription is considered a lower order process because it primarily draws on
orthographic and phonological recoding processes in STM (retrieval from LTM) to
translate these linguistic representations into written symbols. To produce written text, a
writer has to juggle with several cognitively demanding processes- generating ideas,
organizing thoughts, translating those ideas into temporally sequenced discourse, and
revising his or her writing. All of these processes need to be prioritized by an executive
system prior to written output and hence, coordinating these operations is hypothesized to
be supported by a WM system (Berninger, 1999). Research suggests that the storage and
processing function of WM is limited in its capacity (Hitch, Towse, & Hutton, 2001; Just
& Carpenter, 1992). Thus in children whose transcription processes (spelling,
handwriting, and punctuation) are not yet fully automatic, greater resources consumed by
STM could mean fewer resources available for the text generation processes controlled
by WM. If transcription processes are fluent and automatic when writing, it may help to
reduce WM load allowing for improved or better writing (McCutchen, Covill, Hoyne, &
Even though specific writing variables analyzed by studies cited above vary,
findings suggest that STM and WM correlate with different processes of writing. WM
has been associated with complex components of text generation (Berninger, 1999;
Hoskyn & Swanson, 2003; McCutchen, 2000; Swanson & Berninger, 1996) and text
quality (Bourke & Adams, 2003). Swanson and Berninger (1996) found that WM was
significantly correlated with higher-order text generation processes (i.e., number of
words, number of clauses, and compositional quality) while STM best predicted lower-
order transcription processes (i.e., spelling, handwriting, and punctuation). In another
study, Hoskyn and Swanson (2003) examined the relationship between working memory
and writing performance across three age groups (mean ages 15, 30, & 77) and found that
age-related changes in text generation were moderated by limitations in WM.
A larger, replicated, and validated body of empirical work is needed to more fully
understand the relationship between memory and writing and how memory impacts
writing. In the majority of studies discussed above, writing samples were obtained
through spontaneous narrative generation. In this study, the objective was to investigate if
the relationship between STM and WM and writing continues to hold true when genre
and method for collecting the writing samples vary. Writing samples for this study were
collected using a text-retell paradigm and the stimulus paragraph was expository.
To summarize, this study was designed to 1) investigate changes in expository
writing using a text-retell format in the writing of children from grades 3 through 6, 2)
explore the dimensionality of writing, and 3) examine the role of memory to writing. A
long-term goal of this research is to provide SLPs and educators a database for assessing
the writing of LD children and in developing intervention procedures for improving
children's expository writing skills. Additionally, this study also examined the role of
short-term and working memory to writing. This information should be instructive in
providing children with disabilities additional support with writing tasks.
Experimental Questions, Hypotheses and Predictions
The following experimental questions were addressed to examine changes in
expository writing of children from grades 3 through 6, to explore the dimensionalities of
writing, and to evaluate the contribution of individual differences in STM and WM to
children's written language samples.
1. How do writing skills change between grades 3 and 6 at the discourse, T-unit,
sentence, and word level?
a. No differences were expected on discourse, T-unit, sentence, and word level
variables for children at adjacent grades (e.g., grade 3 and 4).
b. Differences were expected for discourse, T-unit, sentence, and word level
variables in children at lower grades (3-4) when compared to children at
higher grades (5 6).
c. Spelling and punctuation errors were expected to decrease significantly
between children in lower grades (3 4) when compared to children in higher
grades (5 6).
Written language samples were collected using a retelling paradigm. This
procedure allows us to focus on the translating process of writing. The written language
variables analyzed at the various levels of text were restricted to examination of the
microstructure of writing or the product of writing. Details on the variables analyzed are
provided in the next chapter.
In a cross-linguistic study of seven languages, Berman and Verhoeven (2002)
found a significant effect for age between students in 4th, 7t, 11th grade and university
graduate students on text length, suggesting that the amount of text produced increases
with age. It is expected that total number of words will be sensitive to grade differences,
although, it is anticipated that the differences in writing at the discourse level will not be
significant for all adjacent grades. Significant differences were expected between lower
and higher grades.
Nearly all studies that examined variables at the T-unit and sentence level have
involved a comparison of the writing capabilities of learning disabled children with
normally developing controls. Nelson and van Meter (2003) found no differences in
number of T-units and mean length of T-units for adjacent grades. However, Berman and
Verhoeven (2002) found a significant effect across grades for number of words per clause
and Blair and Crump (1984) reported that syntactic density increased from grades 6, 7
and 8. In this study, it was hypothesized that sentence level variables would improve with
age, while differences between adjacent grades might be too small to be considered
Similarly, significant differences at the word level were not expected between all
adjacent grades, but significant differences were expected between lower and higher
grades. While, Nelson and van Meter (2002) found that number of different words was
sensitive to grade level, other studies of grade school children have not found significant
differences in adjacent grades. Berman and Verhoeven (2002) also reported a significant
increase in lexical diversity between junior high and high school students. For oral
language, number of different words has been shown to increase steadily between the age
of 3 and 13 (Miller, 1991). Given the rapid increase in vocabulary for this age group,
significant differences are expected across grades. As expected, other studies found
significant differences in spelling accuracy across grades (Moran, 1981; Nelson & Van
Meter, 2002). During the elementary school years, children's phonics skills and
orthographic knowledge improves rapidly, change that is expected to be reflected in their
2. Do the thirteen language variables cluster into factors that categorize the
dimensions of written language?
a. It was expected that three factors would emerge to capture the dimensions of
written language: 1) Productivity, comprised of total number of words,
number of ideas, T-units, sentences, and NDW; 2) Complexity comprised of
number of clauses, percentage of grammatically correct sentences, sentence
complexity, errors per T-unit, clause density, and mean length of T-unit; and
3) Mechanics comprised of spelling and punctuation.
Recently, Justice et al. (2006) examined the dimensionality of the micro-structural
elements of oral narratives. Their exploratory factor analysis produced two clear factors
accounting for 76% of the variance in narrative micro-structural elements. The first factor
was Productivity and included the total number of words, number of different words,
total number of T-units, and total number of complex T-units and to a lesser extent, the
total number of coordinating and subordinating conjunctions. The second factor,
Complexity consisted of the mean length of T-unit and proportion of complex T-units
and subordinating conjunctions to a lesser extent. Using their factor analysis as a
guideline, written language variables were assigned to either the dimension of
productivity or complexity. Since written language was analyzed in this study, a third
factor, namely mechanics was hypothesized to emerge as a factor.
3. Do STM and WM correlate with writing measures at the discourse, T-unit,
sentence and word level?
a. It was expected that WM memory would be correlated with text generation
processes including: (1) total number of words and ideas at the discourse
level, (2) number of T-units and clauses at the T-unit level, (3) sentence
complexity at the sentence level, and (4) NDW at the word level.
b. It was expected that STM would be correlated with transcription processes at
the word level including: (1) spelling, and (2) punctuation.
Differences in sentence generation and lexical retrieval are related to individual
differences in writing (McCutchen et al., 1994). Results of a study by Swanson &
Beminger (1996) indicate that WM was significantly correlated with text generation,
which included number of words and number of clauses while STM was more closely
related to transcription processes, which included spelling, handwriting and punctuation.
In a study exploring the relationship between memory and oral language in young and
older adults, Kemper and Sumner (2001) reported that grammatical complexity was
correlated to measures of WM in young and older adults.
4. Do STM and WM contribute unique variance to writing after controlling for age
a. It was expected that WM would contribute unique variance to the text
generation processes of writing after controlling for the effects of age and
b. It was expected that STM would contribute unique variance to transcription
processes of writing after controlling for the effects of age and vocabulary.
Language measures including reading and writing are sensitive to age and receptive
vocabulary is highly correlated to oral language measures and memory tasks (Dixon,
LeFevre, & Twilley, 1988; Hoskyn & Swanson, 2003; Kemper & Sumner, 2001). Hence,
it is reasonable to hypothesize that children who have larger receptive vocabularies would
reflect this knowledge in greater productivity and lexical diversity in their written
language. In support of this hypothesis, Hoskyn and Swanson (2003) found that WM
continued to exert an influence on writing across age even after the effects of reading
comprehension and vocabulary were partialed from the analysis.
This study was designed to: (1) examine changes in expository writing of children
from grades 3 through 6 using a text retell paradigm, (2) determine if writing could be
categorized to reflect the dimensions of writing, and (3) investigate how WM and STM
are related to writing. The purpose of this chapter is to describe the procedures for subject
recruitment and participation, describe data collection procedures and the testing
protocol, and define the dependent writing variables.
Subjects were recruited from 7 schools in Tampa, Florida. The University of
Florida Institutional Review Board (IRB-02) approved the procedures and consent forms
for this study (see Appendix B). The consent forms described the goals of the study, tests
that would be administered and potential contributions of the findings. Permission was
obtained from Hillsborough County Schools to conduct research. Consent forms were
distributed to children through their classroom teachers. The schools were chosen to
represent children from varying socioeconomic backgrounds. Schools chosen for
participation included 3 elementary schools, one with less than 25% children on free
lunch, one with 50% children on free lunch, and one title one school with 75% children
on free lunch. Four middle schools were chosen including 2 title one schools with 75%
children on free lunch, one with less than 25% children on free lunch, and one with 50%
children on free lunch.
Approximately, three hundred children returned consent forms signed by their
parents giving permission to participate in the study. From this group, thirty subjects each
from grades 3 through 6 were randomly selected if they were monolingual English-
speaking children with no history of language or reading deficits, sensory or neurological
deficits, cognitive impairments or overt behavioral deficits as determined by parent and
Participants included an equal number of males and females at each grade. The
mean chronological ages for the participants were as follows: 8.7 years (SD = 0.6) for 3rd
graders, 9.7 years (SD = 0.5) for 4th graders, 10.8 years (SD = 0.7) for 5th graders, and
11.7 years (SD = 0.5) for 6th graders. The sample was ethnically diverse: 3% Asian-
American, 15% African-American, 66% Caucasian, 10% Hispanic, and 3% other.
Mother's level of education was used as an index of socioeconomic status and included a
range: 43% college or college plus, 23% high school plus, 29% high school, and 4% less
than high school.
All data were collected in the spring of the school year. Data were collected in 2
phases. In phase 1, a vocabulary test was administered and a written language sample
collected. Testing in phase 1 was administered to participants in groups of 3-5 children.
All children were administered a modified version of the Peabody Picture Vocabulary
Test-3 (Dunn & Dunn, 1997). The PPVT-3 is a test that is administered individually. To
enable group administration, two versions were created, one for grades 3 and 4 and
another for grades 5 and 6. The original starting point for children in 3rd and 4th grades
who are usually between 8 and 9 years of age is item number 73. The vocabulary gets
progressively difficult as a child advances through the items in the original test. Since it
was group administered, one needed to ensure student success and keep children who had
poorer vocabulary motivated and on task. Hence the test was modified to begin with an
earlier set, i.e., at item 61 and from there every second item was chosen until 50 items
were obtained. Then the chosen items were randomly distributed such that easy and more
difficult vocabulary words were interspersed. The same protocol was followed for the 5th
and 6th grades, except that the starting point was item number 73. The modified PPVT-3
forms are shown in Appendix C. To allow for comparison of vocabulary scores between
the grades, scores were recalculated based on 29 items common to all four grades.
Next, writing samples were collected from the participants. Participants were read
a modified expository passage (see Appendix C) about the 'Where people live' from the
Qualitative Reading Inventory (Leslie & Caldwell, 2001). The examiner read the passage
to the students who were told to listen carefully because they would have to write what
they remembered from memory on a piece of lined paper. Then the examiner told the
student that she or he would hear the passage one more time just in case they had missed
any important information during the first reading. This procedure was used to minimize
memory constraints and allow for more time to commit the text details to memory. The
children were given as much time as they needed to complete their written retelling of the
text. Most children finished writing in 10 minutes. All testing was conducted in a quiet
room at the children's school.
In phase II, children in grades 3 and 5 were given additional memory tests. The
memory for digits, a commonly used task to measure STM, was taken from the
Comprehensive Test of Phonological Processing (Wagner, Torgesen, & Rashotte, 2001).
Working memory was evaluated using the Competing Language Processing Task (CLPT;
Gaulin & Campbell, 1994) and a Digit Ordering task. All memory tests were
administered individually and presented on audiotape.
Memory for Digits: This subtest contains 4 practice items and 21 test items. On
this test, the child hears a series of numbers and is required to say the numbers back in
the same order as they heard it. The number of digits presented increases systematically,
starting out with 2 numbers and increasing to 8 numbers. No repetition of test items is
permitted. Testing is discontinued after the participant misses 3 items in a row (see
Competing Language Processing Task: The CLPT is an adaptation of a sentence
span task originally designed by Daneman and Carpenter (1980) to evaluate verbal
memory span in adults. It was modified by Gaulin and Campbell (1994) to assess verbal
working memory in school-age children. In this task, children are presented with groups
of 1 to 6 short sentences. Children are first required to respond yes or no to each sentence.
The questions are included to ascertain comprehension and ensure that children are not
focusing exclusively on the word-recall task. Sentences are divided into two groups with
six different levels of difficulty. The number of sentences increases by one as the level
increases, reaching a total of 6. For example, level 1 contains only one sentence per group
and requires comprehension of one sentence and recall of one word. Level 6 contains six
sentences per group and requires answers to 6 comprehension questions and recall of 6
words. After all the sentences in a group are presented to the participants, they are
required to recall the last word of each sentence. All sentences contain three words, are
controlled for length and difficulty with each sentence having one of these structures-
subject-verb-object, subject-verb-modifier, or subject-auxiliary-main verb. The protocol
consists of 4 practice items and 42 test sentences (see Appendix C). Two types of scores
are obtained, one for comprehension and one for memory.
Digit Ordering: Digit Ordering is a task commonly used to assess working
memory (e.g., Altmann, Kempler, & Andersen, 2001; MacDonald, Almor, Henderson,
Kempler, & Andersen, 2001). The test contains 2 practice items and 24 trial items. Each
level has four items. Levels increase in difficulty starting with two digits at level 1 and
increasing to 7 by level 6. Testing is discontinued when all items at a given level have
been administered and participant makes two or more errors at that level (see Appendix
Dependent Variables for Writing
All written samples were transcribed into a computer database according to
Systematic Analysis of Language Transcript conventions (SALT, Miller & Chapman,
2001). T-units were used as the unit of segmentation, as suggested by Scott and Stokes
(1995) and Scott and Windsor (2002). In this study, assessment was restricted to the
microstructural elements of writing. Microstructural elements are concerned with the
writing product as opposed to the macrostructural elements, which are concerned with the
writing process (e.g., planning, organizing, revising, and editing). Written samples were
then analyzed at the discourse, T-unit, sentence, and word levels using a modified
protocol developed by Nelson, Bahr, and Van Meter (2004). See Appendix D for the
At the discourse level, two variables were analyzed to measure written fluency:
(1) total number of words written, and (2) total number of ideas expressed. These
variables have been widely used by researchers when measuring productivity in both
spoken and written language (Berman & Verhoeven, 2002; Houck & Billingsley, 1989;
Mackie & Dockrell, 2004; Nelson, Bahr, & Van Meter, 2004; Nelson & Van Meter,
2002; Scott & Windsor, 2000). "Text length is a consistently good predictor of holistic
quality ratings of writing" (Scott, 1998, p. 249). Total number of words was the number
of words produced in writing by the subject and calculated automatically by SALT.
Words or phrases that did not pertain to the original paragraph such as "The end" or
"that's all I remember" were deleted when calculating total number of words. Total
number of ideas was calculated so as not to penalize participants who wrote concisely
while reproducing ideas/propositions from the original sample. For example the original
sample contained sentences such as, "People lived in different places" and "Farmers sell
their crops". The first sentence was counted as one idea and the second sentence was
counted as another idea. The stimulus paragraph contained a total of 25 ideas.
Information produced by the participants that was not present in the original sample was
not counted as an idea (see appendix D).
At the T-unit level, five variables were examined: (1) number of T-units, (2) mean
length of T-unit, (3) number of clauses, and (4) clause density and (5) errors/T-unit.
Number of T-units and mean length of T-units (MLT-unit), frequently used measures of
syntactic complexity, were calculated using the formula proposed by Hunt (1965). A T-
unit, which is the most common method for investigating competence in writing (Scott,
1998), is one main clause with all subordinate clauses embedded in it. Clauses that begin
with the coordinating conjunctions and, but, or or are considered to comprise a new T-
unit. Written samples were entered into SALT in T-units, hence number of T-units was
calculated automatically by the program. MLT-unit was determined by dividing the total
number of words in the writing sample by the total number of T-units in the sample.
A clause was defined as a group of related words containing a subject and a verb.
The number of clauses in each sentence was coded in SALT. The clauses produced were
summed to provide the total number of clauses produced by the subject. Clause density,
as suggested by Scott and Stokes (1995), is another index of syntactic complexity that
can be used with children and adolescents. They define clause density as "a ratio of the
total number of clauses (main and subordinate) summed across T-units, and divided by
the number of T-units in the sample" (p. 310). Finally, errors/T-unit was the ratio of the
total errors divided by the total number of T-units. Descriptions of what was counted as
an error along with examples are provided in the next section. The complete scoring
protocol along with a written language transcript with all the codes marked is shown in
Sentence level analyses included three variables, (1) number of sentences (2)
percentage of grammatically correct sentences, and (3) sentence complexity. Just as
length of T-units and clause density provide a general index of syntactic complexity,
examination of sentence types provides information on subjects' sentence level skills
(Nelson et al., 2002, 2004). Findings from various studies have been mixed and
inconclusive regarding the effectiveness of T-units to capture syntactic complexity
adequately (Scott, 1989, 1999 for a review). Therefore, writing at the sentence level was
examined as well. Unlike a T-unit, which is based on a preset definition, a sentence was
counted based on child's own punctuation (i.e., the presence of a period). If a period was
omitted but a new sentence was begun with a capital letter, or a change of topic in spite
of lack of appropriate punctuation, it was counted as sentence. It should be noted that
more than one T-unit could make up a sentence. For example, "Some people live in the
city/ and some people live in the country", is one sentence but divided into 2 T-units (/
indicates division of T-unit).
Sentence codes were added after the child's written language sample was divided
into separate T-units. Sentences produced by participants were divided into
grammatically correct or incorrect sentences. A sentence was coded as incorrect if it
contained grammatical or lexical errors. Error types analyzed included errors of verb or
pronoun tense, agreement, or case, omitted or incorrect inflection, substitution of
grammatical elements, and violation of word order, lack of referent, and wrong word
choice. Percentage of grammatically correct sentences was calculated by dividing the
number of correct sentences by the total number of sentences produced by the subject.
A sentence was coded as simple if it contained only one clause and as complex if it
contained two or more clauses. They were coded as simple correct or simple incorrect,
complex correct or complex incorrect as shown in the examples below:
* Simple correct: Most people in the country are farmers.
* Simple incorrect: The farmers have chicken, ducks, and cow.
* Complex correct: For fun you can walk in the woods or you can find a river and go
* Complex incorrect: There are many thing to do in the country.
Finally, sentences were coded from levels 1 through 7 based on available literature
of the development of complex sentences in children (Miller, 1981; Rosenberg, 1974)
and the work of Rosenberg and Abbeduto (1987) with adults. The various levels are
shown below (Rosenberg & Abbeduto, 1987, p. 26).
Level 1: Embedded infinitival complement with subject identical to that of the matrix
clause (e.g., Try to brush her hair. I am going to meet John).
Level 2: wh-infinitive clause (e.g., Remember where it is?); sentence conjoined with a
coordinating conjunction (e.g., I brought candy and Peter cleaned up);
compound sentence (e.g., John and Mary left early).
Level 3: Object noun phrase relative clause (e.g., The man scolded the boy who stole
the bicycle); object noun phrase complement (e.g., John knew that Mary was
Level 4: Gerundive complement (e.g., I felt like turning it); comparative (e.g., John is
older than Mary).
Level 5: Subject noun phrase relative clause (e.g., The man who cleans the rooms left
early today); subject noun phrase complement (e.g., For John to have left
Mary was surprising); nominalization (e.g., John's refusal of the drink
Level 6: Two sentences conjoined with a subordinating conjunction such as if
because, before, so (e.g., They will play today, if it does not rain).
Level 7: More than one use of sentence combining in a given sentence (e.g., John
decided to leave Mary when he heard that she was seeing Mark).
Sentence complexity was calculated by obtaining the weighted average of the
sentences depending on the complexity level. To simplify, each sentence was multiplied
by the number that was assigned to it based on its complexity level and divided by the
total number of sentences produced by the subject. For example if the subject produced
two sentences at level 1, one sentence at level 3 and one at level 4, his/her sentence
complexity would be (2xl+3xl+4xl)/4 = 2.25.
At the word level, three variables were examined (1) number of different words
(NDW), (2) percentage of spelling errors, and (3) writing conventions. NDW is a widely
used measure of lexical diversity (e.g., Berman & Verhoeven, 2002; Houck &
Billingsley, 1989; Moran, 1981; Nelson & Van Meter, 2002; Nelson et al., 2004; Scott &
Windsor, 2000). Number of spelling errors, extensively used in studies examining writing
(Mackie & Dockrell, 2004; Moran, 1981; Nelson et al., 2004; Nelson & Van Meter,
2002; Poplin, Gray, Larsen, Banikowski, & Mehring, 1980), was the number of words
spelled incorrectly. A word was counted as a spelling error only once if the child used the
same (incorrect) spelling. If however, a word was spelled incorrectly but differently, each
incorrect spelling was counted as an error. The percentage of spelling errors was
calculated by dividing the number of spelling errors by the total number of words
produced by the subject. With a few exceptions, punctuation is a matter of choice
(Simone, 1996). Most children resort to the very basic punctuation such as periods at the
end of a sentence or use of capital letters to begin a sentence (Scott, 1998). For writing
conventions, use of end periods and initial capital letters was examined.
Transcription, Coding, and Reliability
All written samples were transcribed into SALT by a research assistant. Every
transcript was checked by the author to ensure that all writing samples were correctly
transcribed. After practice and establishing coding guidelines, the author and a trained
research assistant coded 20% of the written samples independently to obtain reliability.
Interrater reliability ranged from 80% to 100% for coded items across transcripts. Scoring
differences between the author and the research assistant were settled by consensus
In summary, thirteen written language variables were examined and organized into
four levels. The levels and the variables examined are shown in the table below:
Table 1. Written language variables by level.
Language Level Dependent Measures
DISCOURSE Total Words
T-UNIT No. of T-units
Mean Length of T-unit
No. of Clauses
Errors per T-unit
SENTENCE No. of Sentences
% Grammatical Sentences
WORD Number of different words
% Spelling Errors
This primary purpose of this study was to examine changes in expository writing of
children from grades 3 through 6 using a text retell paradigm. Thirteen written language
variables were analyzed at the discourse, T-unit, sentence, and word level. The second
aim of this study was to determine if the writing variables analyzed could be categorized
into factors to reflect the dimensions of writing. Finally, this study investigated how WM
and STM are related to writing.
Question 1: How does writing develop between grades 3 and 6 at the discourse, T-unit,
sentence, and word level?
Data for the dependent writing measures were analyzed using a multivariate
analysis of variance (MANOVA) at each level of language with grade as the between-
subjects factor. Table 2 contains the means and standard deviations for all writing
variables across grades. Pairwise comparisons, adjusted for multiple comparisons, were
used to compare the performance of the 4 grades, corrected for Type I error using a
Bonferroni correction. Measures of effect size have been reported using partial eta
squared (q ). Effect sizes measure the degree of association between the effect and the
dependent variable and have values ranging from 0 to 1.0. Effect sizes less than .10 are
considered small, effect sizes between .10 and .25 are considered moderate and effect
sizes greater than .25 are considered large.
Discourse Level Measures
Two variables were analyzed across the groups at the discourse level, total words,
and total ideas. The MANOVA for the dependent variable of total words showed a
significant main effect for grade F(3,116) = 17.33, p < .0001, partial f = .31. There was
a clear increase in overall productivity reflected in the total number of words used at
every grade. Pairwise comparisons showed that children in grade 3 differed significantly
from children in grades 4, 5, and 6 (p <.001, for all three grades) and children in grade 4
differed significantly from children in grade 6 (p < .05). The mean total words for
children in each grade level was higher than the mean at the preceding grade levels
showing an increase in overall productivity in the total number of words used from 3rd to
Similarly, for total ideas, a significant main effect for grade was found, F(3,116) =
13.7, p < .0001, partial f = .26 indicating that the number of propositions children are
able to recall is sensitive to grade. 3rd graders produced significantly fewer ideas than 4th
(p < .05), 5th (p < .001), and 6th graders (p < .001) and 4th graders produced significantly
fewer ideas than 6th graders (p < .03). Again, the mean number of ideas at each grade was
higher than mean at the preceding grade showing an increase in the total number of ideas
from 3rd to 6th grades.
T-unit Level Measures
Five variables were analyzed across the groups at the T-unit level, number ofT-
units, mean length of T-unit, number of clauses, clause density, and errors T-unit. The
analysis of the number of T-units produced showed a significant main effect for grade,
F(3,116) = 15.25, p< .0001, partial 2 = .28. Performance of 3rd graders differed
significantly from 4th graders (p < .002), 5th(p < .0001) and 6th graders (p < .0001) and 4th
graders produced significantly fewer T-units than 6th graders (p < .007). Similar to
discourse level variables, the number of T-units produced from 3rd to 6th grade showed a
steady increase with grade.
For the variable, mean length of T-units, no statistically significant difference
between grades was noted, F(3,116) = 1.51, p = .22. Similarly, there were no statistically
significant differences between grades for clause density, F(3,116) = 0.14, p = .94.
A significant main effect for grade was observed for the number of clauses,
F(3,116) = 14.5, p < .0001, partial r = .27. Children in grade 3 produced significantly
fewer clauses than children in grades 4 (p < .02), and children in grades 5 and 6 (p <
.0001 for both). Performance of children in grade 4 differed significantly from children
in grade 6 (p < .001). Once again a steady increase was noted from grade 3 to 6 in the
number of clauses produced. Finally, for the number of errors/T-unit, no significant effect
was obtained for grade, F(3,116) = .88, p= .46.
Sentence Level Measures
At the sentence level, number of sentences, percentage of grammatically correct
sentences, and sentence complexity were examined. Number of sentences showed a main
effect for grade F(3,116) = 13.6, p < .0001, partial r2 = .26. 3rd graders were significantly
different from 4th (p < .004), 5th (p < .0001), and 6th graders (p < .0001). 4th graders were
significantly different from 6th graders (p < .001) and 5th graders were significantly
different from 6th graders (p < .02). Similar to the trend noted on several other variables,
the number of sentences produced by the participants also increased with grade. This was
the only measure on which the difference between grade 5 and 6 was significant.
There was no significant main effect for grade for percentage of grammatically
correct sentences, F(3,116) = 1.43, p = .24. However, a significant main effect for grade,
F(3,116) = 4.34, p < .006, partial rI = .10 was noted for sentence complexity. Significant
differences were found between 3rd and 4th graders (p < .05), 3rd and 5th graders (p < .05),
and 3rd and 6th graders (p < .001). The means for the 4th and 5th graders was nearly
Word Level Measures
At the word level, three variables were examined, (1) NDW, (2) percentage of
spelling errors, and (3) writing conventions. For the dependent variable NDW, a
significant main effect was found for grade, F(3,116) = 18.87, p < .0001, partial rI = .33.
Not only does overall productivity in writing increase with grade level, but lexical
diversity also increases with grade. 3rd graders produced significantly fewer different
words than 4th graders (p < .04), 5th graders (p < .0001), and 6th graders (p < .0001) and
4th graders produced significantly fewer different words than both 5th (p < .05) and 6th
graders (p < .0001). Mean scores showed steady increase in lexical diversity from 3rd to
Percentage of spelling errors also showed a main effect for grade, F(3,116) = 6.11,
p <.001 partial 2 = .14. 3rd graders made significantly more spelling errors than 5th (p <
.02) and 6th graders (p < .0001) and 4th graders made significantly more spelling errors
than 6th graders (p < .01). 3rd and 4th graders and 5th and 6th graders showed the same
percentage of spelling errors. Finally, no differences were observed between grades for
writing conventions, F(3,116) = .80, p = .49. All four grades showed between 82% and
90% accuracy for punctuation.
Table 2. Means and standard deviations for writing measures at the discourse, T-unit, sentence, and word level.
Grades 3 4 5 6
M SD M SD M SD M SD F-Ratio
Discourse Level Measures
Total Words 61.0a 20.8 89.2ab 30.1 97.7a 24.0 109.8ab 32.7 17.34***
Total Ideas 6.8a 3.3 9.4ab 3.5 10.6a 3.3 12.9ab 4.9 13.7***
T-unit Level Measures
No. ofT-units 6.5a 2.0 8.5ab 2.4 9.5 2.7 10.6a,b 2.8 15.25***
MLT-unit 9.6 2.3 10.5 1.8 10.5 1.7 10.3 1.7 1.51
Clause Density 1.78 0.5 1.77 0.4 1.83 0.4 1.82 0.4 0.14
No. of Clauses 11.2a 3.6 15.0ab 5.1 17.1" 5.0 19.2ab 5.7 14.45***
Errors/T-unit 0.30 0.27 0.21 0.21 0.28 0.24 0.27 0.20 0.87
Sentence Level Measures
No. of Sentences 5.9a 2.1 7.9ab 2.4 8.4ac 2.4 10.0a,b,c 3.0 13.59***
% Grammatical 0.73 0.24 0.81 0.2 0.73 0.2 0.73 0.2 1.43
Sentence 8.8a 6.3 14.7a 9.4 14.4a 11.3 17.1a 9.4 4.34**
Word Level Measures
NDW 33.8a 9.7 41.7ab 12.2 49.3ab 9.9 53.7ab 12.0 18.87***
% Spelling Errors 0.072 0.06 0.055b 0.03 0.047a 0.04 0.027ab 0.03 6.12**
Writing 88.6 16.4 90.1 16.1 86.3 24.2 82.3 24.1 0.80
Note: p < 0.5; **p < .01; *** p < .0001. Values sharing the same superscript significantly different from each other at p< .05
Question 2. Do the thirteen language variables cluster into factors that categorize the
dimensions of written language?
To measure the dimensionality of the thirteen written language variables analyzed
in this study, an exploratory factor analysis (EFA) was done in SPSS with a principal
component analysis factor extraction method, component correlation matrix estimates,
and Promax rotation (power = 4). The EFA produced four clear factors with eigenvalues
greater than average in the correlation matrix with estimated communalities on the
diagonal and a scree plot according to the Kaiser-Guttman rule (eigenvalues of
correlation matrix > 1). The four factors accounted for 86% of the variance in the
microstructure of expository written language, with correlations between factors ranging
from .04 to .29. Eigenvalues and standardized loadings for the writing variables and the
four factors are presented in Tables 3 and 4 respectively. The first factor, Productivity,
resulted in loadings of total number of words, ideas, T-units, sentences, clauses, and
number of different words. The second factor, Complexity, resulted in loadings on MLT-
unit, clause density, and sentence complexity; the third factor, Accuracy, resulted in
loadings on errors/T-unit and % of grammatically correct sentences; and the fourth factor,
Mechanics, resulted in loadings on spelling errors and punctuation. Clearly, these factor
loadings did not completely match the predicted factors. It was predicted that number of
clauses would load on the factor, Complexity instead it loaded on Productivity. Accuracy
was not predicted to be a separate factor; instead errors/t-unit and percentage of
grammatically correct sentences were hypothesized to load on the Complexity rather the
Accuracy factor. The EFA indicates that when assessing written language, four aspects
are crucial to examine- overall productivity in writing, complexity of writing, accuracy in
writing and the mechanics of writing.
Table 3. Four-factor solution of expository written microstructure.
Factor Initial Eigenvalue % of Variance Rotations Sums of
Table 4. Standardized regression coefficient factor loadings for the four-factor solution.
Writing Variables Productivity Complexity Accuracy Mechanics
1. Total words .93 .17 .03 .003
2. Total ideas .88 .03 .03 .003
3. No. of T-units .97 -.23 .01 .06
4. MLT-unit .08 .86 .06 -.06
5. Clause Density -.02 .94 -.04 -.05
6. No. of clauses .88 .28 -.03 .01
7. Errors/T-unit -.02 .01 -.97 -.01
8. No. of sentences .93 -.34 -.00 -.07
9. % Grammatical Sentences -.02 .02 .98 -.00
10. Sentence Complexity -.07 .87 -.01 .11
11. NDW .92 .12 -.05 .00
12. % Spelling Errors .02 -.03 .03 .86
13. Writing conventions .01 -.03 .03 -.83
Question 3. Do STM and WM correlate with writing measures at the discourse, T-unit,
sentence and word level?
Table 5 contains the means and standard deviations for the independent measures.
As expected, PPVT-3 vocabulary scores showed a significant main effect for grade,
F(3,116) = 12.75, p < .0001. 3rd graders had significantly lower vocabulary scores than
both 5th (p < .0001) and 6th graders (p < .0001) and 4th graders had significantly lower
scores than 5th graders and 6th graders (p < .004) On the CLPT WM subtest, 3rd graders
did not differ from 5th graders, however, 3rd graders performed significantly more poorly
than 5th graders on the WM subtest of Digit Ordering, F(1,58) = 8.61, p < .005. Finally,
no significant effects were found between 3rd and 5th graders on the STM-Memory for
Digits subtest (p = .76).
Correlations between vocabulary, memory measures and writing variables are
shown in Table 6. Significant correlations were found between vocabulary and working
memory tasks and several writing variables. At the discourse level, total words was
significantly correlated to WM tasks and vocabulary. Similarly, total ideas was
significantly correlated to WM measures and vocabulary (p < .01) in addition to STM (p
At the T-unit level, significant correlations were found for number of T-units,
number of clauses, WM and vocabulary (p < .01). No significant correlations were noted
between MLT-unit, clause density, and errors per T-unit and STM, WM measures or
At the sentence level, significant correlations were found only between number of
sentences and vocabulary (p < .01). Finally, at the word level, NDW was significantly
correlated to WM and vocabulary (p < .01) while conversely, spelling was significantly
correlated only to STM measure but not to WM or vocabulary. Finally punctuations did
not show any correlation to either memory measure or vocabulary.
Table 5. Means and standard deviations for independent measures.
Grades 3 4 5 6
M SD M SD M SD M SD
PPVT-3 Modified 18.1 3.2 18.9 5.5 22.4 3.4 23.2 2.7
Memory for Digits 10.2 2.6 10.4 2.4
CLPT Memory 27 4.3 28.9 3.6
Digit Ordering 13.7 2.5 15.4 2.1
Note: For vocabulary, n = 120; for STM and WM, n = 60.
Table 6. Correlations between memory, vocabulary measures and writing variables.
STM WM WM Vocabulary
Memory for CLPT Digit PPVT-3
Digits Memory Ordering Modified
1. Total Words .23 .38** .32* .47**
2. Total Ideas .27* .39** .37** .45**
3. No. of T-units .20 .29* .26* .46**
4. MLT-unit -.004 .16 .04 .05
5. No. of Clauses .24 .28* .29* .43**
6. Clause Density 0.5 -.02 -.007 -.02
7. Errors/T-unit -.09 -.14 .008 -.08
8. No. of Sentences .16 .20 .25 .44**
9. % Grammatical .05 .12 -.02 .09
10. Sentence Complexity .02 .06 .04 -.07
11. NDW .18 .35** .30* .46**
12. % Spelling Errors -.31* -.18 -.13 -.15
13. Writing Conventions .20 .04 .16 -.02
Note. p* <.05; p** <.01
Question 4. Do STM and WM contribute unique variance to writing after controlling for
age and vocabulary?
To explore the relationship between memory and writing measures, the correlations
were examined more closely using a stepwise regression analyses. To simplify the
analyses, only correlations that were found to be significant were examined. Of particular
interest was the relationship between writing variables and memory measures while
controlling for age and vocabulary. Therefore variables were entered into the regression
equation with age first followed by WM and STM measures and vocabulary. A
composite WM score was calculated by averaging scores from the CLPT-memory and
Digit Ordering tests. Table 7 shows the results of the stepwise regression analyses.
As shown in Table 7, when the effects of age were partialed out, WM accounted for
5% of the unique variance in total number of words and 9% of the unique variance in
number of ideas. No other variable entered significantly into the equation. As also shown,
STM contributed 9% of the variance in spelling. No other variable entered significantly
into the equation. After controlling for the effects of age, WM, STM, and vocabulary did
not enter significantly into the regression equation for other writing variables examined,
namely, number of T-units, number of clauses, number of sentences and number of
Table 7. Forward stepwise regression with memory and vocabulary scores as predictor
Proportion of Increment in R2 F ratio
Number of Total Words
1. Age .44 46.58**
2. WM (composite) .49 .05 27.64**
Number of ideas
1. Age .28 23.05**
2. WM (composite) .37 .09 17.05**
Number of T-units
1. Age .56 26.32**
Number of clauses
1. Age .34 29.51*
Number of sentences
1. Age .26 20.19**
1. Age .42 42.36**
1. STM .09 6.3*
Note. p* <.01, p** <.001
Summary of Results
Written language samples collected from one hundred and twenty children using a
text-retell paradigm were analyzed at the discourse, T-unit, sentence, and word level.
Four experimental questions were addressed: 1) How does writing develop between
grades 3 and 6 at the discourse, T-unit, sentence, and word level?; 2) Do the thirteen
language variables cluster into factors that categorize the dimensions of written
language?; 3) Do STM and WM correlate with writing measures at the discourse, T-unit,
sentence and word level?; and 4) Do STM and WM contribute unique variance to writing
after controlling for age and vocabulary?
Measures of productivity and lexical diversity, which include total number of
words and produced, number of T-units, clauses, and sentences, and number of different
words were sensitive to grade. On most of these measures, results were statistically
significant between grade 3 and grades 4, 5, and 6, and between grade 4 and grade 6. The
means for grade 6 were higher than grade 5 but the differences were not significant,
except for the number of sentences. The difference between grades was not statistically
significant for mean length of T-unit, clause density, errors per T-units, percentage of
grammatically correct sentences, and punctuation. Finally for spelling, 3rd graders
showed poorer performance compared to 5th and 6th graders and 4th graders showed
poorer performance compared to 6th graders.
When the dimensionality of writing was examined, the exploratory factor analysis
produced four factors namely, 1) Productivity, 2) Complexity, 3) Accuracy, and 4)
Mechanics. Correlation between memory measures and writing variables showed that
WM was more highly correlated with text-generation process such as total number of
words and ideas, number of T-units and clauses and number of different words whereas
STM was more highly correlated with spelling which is a transcription process. In the
stepwise regression analyses, when age and vocabulary were controlled, WM added
unique variance to total number of words and number of ideas while STM added unique
variance to spelling.
The aims of this study were threefold: (1) to examine changes in expository writing
of children from grades 3 through 6, (2) to determine if writing variables analyzed could
be categorized into factors to reflect the dimensions of writing, and (3) to investigate the
contribution of WM and STM to writing. Written language samples were collected from
thirty children each from grades 3 through 6 using a text-retell paradigm. Thirteen
variables were analyzed to examine the microstructure of expository writing at the
discourse, T-unit, sentence, and word levels.
Changes in Writing across Grades and Factors underlying Dimensions of Writing
The first aim of this study was to examine how students' writing skills change
across grades. It was hypothesized that the differences between two adjacent grade levels
would not be significant (i.e., grades 3 and 4 or 4 and 5) while significant differences
between lower and higher grades would be found (i.e., grades 3 and 5 or 4 and 6).
Predictions made were partly validated. Differences between higher and lower grades
were significantly different. However, significant differences were also found between
some adjacent grades. The performance of 3rd graders was lower than 4th graders and 4th
graders showed lower performance than 5th graders on total number of words, T-units,
clauses, sentences, and different words, total ideas, mean length of T-unit, and sentence
complexity. The general trend was for performance to improve steadily with grade.
Specifically, performance of grade 3 students was significantly lower than that of
students in grades 4, 5, and 6 and performance of grade 4 students was significantly
lower than that of students in grades 5 and 6 on most variables Although the means for
most variables for grade 5 were lower than the means for grade 6, differences were
generally not statistically significant.
The second aim of this study was to explore the dimensionality of written language
similar to dimensions identified for oral narratives (Justice et al., 2006). An exploratory
factor analysis confirmed that writing is a complex activity comprising multidimensional
constructs -productivity (total number of words, ideas, sentences, T-units, clauses, and
NDW), complexity (MLT-unit, clause density, sentence complexity), accuracy (errors/T-
unit, % of grammatical sentences), and mechanics (% spelling errors, writing
conventions) of writing. The factor analysis provides an empirical basis for organizing
the microstructure of writing into validated categories. In the sections below, each of
these dimensions of writing is discussed.
Measures of Productivity
Measures of productivity at the various levels of language, (1) total number of
words and number of ideas at the discourse level, (2) number of T-units and clauses at
the T-unit level, (3) number of sentences at the sentence level, and (4) number of
different words at the word level increased steadily with grade. Children in grades 5 and
6 used more words and different words and recalled more ideas than children in grades 3
and 4. These results are consistent with findings from previous studies of written
language showing that measures of productivity are sensitive to changes in age and grade
(Berman & Verhoeven; Nelson & Van Meter, 2003).
Measures of Complexity
The steady trend of improvement with grade was not as consistent for measures of
complexity as it was for measures of productivity. No significant improvement or change
was noted for mean length of T-unit or clause density, however, sentence complexity was
sensitive to grade. The results of this study are consistent with results of previous studies
using T-units to examine competence in syntactic complexity showing a steady but very
slow increase in the mean length of T-units as a function of age (Hunt, 1965, 1970;
Klecan-Aker & Hedrick, 1985; Loban, 1976; Morris & Crump, 1982; O'Donnell, Griffin,
& Norris, 1967; Scott, 1988). This rate of growth is often so slow that statistically
significant differences between adjacent grades are not typically found (Scott, 1988).
Clause-density has shown little change with age as well, with several periods of short
plateaus (Loban, 1976) before it asymptotes at grade 8 (Hunt, 1970; Nippold, Ward-
Lonergan, & Fanning, 2005). Consistent with the results of previous research, in this
study, 4th and 5th graders showed a marginal improvement in clause density over 3rd and
In contrast, sentence complexity showed a steady increase with grade, showing as
expected, that children's ability to formulate complex sentences increases throughout
elementary school. For example to convey two ideas, a 3rd grader produced one simple
and one lower level complex sentence 'Most people who live in the country are farmers.'
and 'They grow their crops to sell or to feed the animals on the farm.' while a 6th grader
combined the two ideas by producing a higher level complex sentence 'The people who
live in the country are farmers who grow crops to sell and feed the animals'. Level of
schooling seems to be crucial in affecting changes in measures of productivity including
over-all text length and lexical diversity and syntactic development and complexity
(Berman & Verhoeven, 2002).
Unlike productivity measures, accuracy measures, which included errors/T-unit and
% grammatically correct sentences, did not show a steady increase with grade. Large
variability was noted for errors/T-unit, which may explain the lack of significant findings.
Almost no increase was noted for % grammatically correct sentences. One likely
explanation is that children in higher grades were attempting to produce more complex
sentences (as seen from significant differences noted in sentence complexity), which
resulted in them producing more errors than might be expected. This aspect will need to
be examined more closely in future studies.
Mechanics of Writing
The performance of students were in line with the predictions of this study, with 3rd
graders making more spelling errors than 5th and 6th graders and 4th graders making
significantly more spelling errors than 6th graders. This expected pattern of improved
performance with age and grade has been noted in other studies (Berman & Verhoeven,
2000; Moran, 1981; Nelson & Van Meter, 2003). Considerable variability was observed
in spelling performance as noted from the large standard deviations obtained. This
variability could be an artifact of the language sampling procedure used. Written samples
were obtained using a text-retelling paradigm in which children might have felt
compelled to use the words they had heard. It is highly likely that the students attempted
to spell words they had heard in the stimulus paragraph though they may not have known
how to spell those words. The most common spelling errors were for the words 'country'
(e.g., country, country) and 'suburbs' (e.g., soberbs, suberbs).
Students at different grade level did not differ significantly on writing conventions.
The punctuations measures used in this study were very basic and included only initial
capital letters and end periods. It is plausible that simple writing conventions may not be
a very sensitive indicator of punctuation measures in typically developing children
beyond 3rd grade. There is little research on the developmental changes in punctuation.
Researchers have suggested that children at this age are generally poor at punctuation and
also confine use to the most basic writing conventions (Ferreiro & Kucchermaglio, 1996;
Simone, 1996) which may very well explain why no differences were observed across
Summary of Changes in Writing across Grades and Dimensionality of Writing
Several variables examined in this study showed sensitivity to grade while some
variables did not. This suggests that there is a set of variables which includes total
number of words, ideas, T-units, clauses, sentences produced, number of different words,
and spelling errors that better differentiates performance across grades and should be
used to develop standardized norm referenced measures for evaluating writing.
The performance of grade 3 students was consistently lower than students in grades
4, 5, and 6 and performance of grade 4 students was lower than students in grades 5 and 6
on several variables. A marked shift in literacy skills occurs at grade 3 with children
having mastered the basic reading skills and starting to read for meaning as opposed to
learning to read prior to grade 3 (Chall, 1983). In addition, vocabulary shows a steep
growth with children learning about 9,000 new words from grades 1 to 3 but 20,000 from
grades 3 to 5 (Anglin, 1993). This shift is reflected in the writing of children as well.
Data from this cross-sectional study suggest that children's writing skills plateau at
grade 5 similar to plateaus reported for oral narrative productions (Justice et al., 2006).
With the exception of the number of sentences produced, there were no statistically
significant differences between grades 5 and 6 on the other variables examined. It is also
plausible, however, that the absence of significant changes between grades does not
reflect a plateau but rather is a result of methodological constraints and limitations. The
same paragraph was used for students in 3rd through 6th grade and might, at least at the
higher grades, reflect a ceiling effect. While using different levels of paragraphs for grade
6 might help tease this issue apart, it would be difficult to control variables using a retell
This scheme of examining writing along dimensions and at various levels of
language could be very relevant for teachers and clinicians when evaluating and planning
intervention programs that take into account an individual students' strengths and
weaknesses because research has shown that children can struggle with different levels of
language (Berninger, Mizokawa, Bragg, Cartwright, & Yates, 1994). A long-term goal of
this research is to design a clinical instrument for SLPs and teachers to use for evaluating
multiple dimensions of children writing skills. To that end, the data gathered for this
study will allow clinician's and educators who wish to use a story retell paradigm for
assessing the performance of 3rd through 6th grade children who are struggling with
Relationship between Memory and Writing
The third aim of this study was to examine the relationships between memory and
writing. The results of this study add to the current body of evidence indicating that WM
and STM are differentially related to writing (Berninger et al., 1992, 1994; Swanson &
Berninger, 1996). As predicted, text generation processes showed a higher correlation
with WM than STM. Text-generation processes draw on WM since they require
translating the writer's thoughts into linguistic representations in the form of sentences,
phrases and words. The text generation variables in this study that were correlated with
WM were number of total words, T-units, and clauses, total ideas and number of different
words. Further, WM was found to contribute unique variance to writing variables such as
total number of words and ideas when entered into a regression equation, after controlling
for age. It must be noted that when the contributions of age were partialed out, the
contributions of WM and STM to writing diminish, implying that age contributes
significantly to writing. As mentioned earlier, the changes in mean length of T-unit and
clause density are minimal across grades, which might explain why significant
correlations weren't obtained between WM and these variables. The elementary and early
middle school children studied sometimes omitted periods at the end of a sentence,
leaving the examiner to determine what constituted a sentence. It is possible the
examiner's judgment contributed to the lack of correlation between the number of
sentences and WM.
Spelling, a transcription process, showed a higher correlation with STM, once
again consistent with results obtained by other studies (Beminger, et al., 1992, 1994;
Swanson & Berninger, 1996). Transcription relies directly on orthographic and
phonological processes in STM. In this study, spelling, only one of the two measures of
transcription examined was correlated to STM. It was expected that punctuation would
also show a high correlation to STM based on previous investigations because it is
considered a transcription process However, perhaps because punctuation is still a
developing skill and because the punctuation measured (end periods and initial capital
letters) do not cover the whole range of writing conventions, it could have contributed to
the lack of correlation observed. An alternate explanation could be that punctuation is a
convention that does not require memory as much as it requires linguistic knowledge of
syntactic rules or conventions. Rather unexpectedly, number of ideas showed a
significant correlation to STM. Unlike spontaneously generated writing, a retelling task
probably shows a higher reliance on both WM and STM memory, which is echoed in the
correlation between WM, STM and number of ideas generated.
The results of this study add to the existing evidence regarding the differential role
of WM and STM in writing. Most of the work examining writing and the role of memory
has involved spontaneously generated narratives. These findings support the unique
contributions of memory in writing obtained using a text-retelling format. Children may
have difficulty with writing obtained via retellings due to memory and linguistic
knowledge. Furthermore, it is likely that memory and linguistic knowledge interact to
affect writing performance. The contribution of memory and linguistic knowledge will
need to be teased apart in future studies probably by comparing performance between
written language samples obtained through retellings and spontaneous generation to help
determine the contribution of these two skills on writing.
This study is the first step in attempting to build a normative database for
evaluating expository written text in school age children. To date we have few systematic
methods for these that are evidence based. Writing is a language-based activity and given
the training and knowledge that SLPs posses in the area of oral language, they can make
valuable contributions to improving the writing of children. Since writing is a challenging
activity for most children, even those not in special education classrooms, it is very likely
that a large number of children struggling with writing in regular classrooms would
benefit from intervention by SLPs. In addition to addressing writing during one-on-one
therapy, SLPs can collaborate with classroom teachers to improve writing of grade school
The retelling format is an efficient and standardized method of evaluating writing
that should serve the SLP well in beginning to identify writing impairments. Evaluating
expository writing using this method in conjunction with story grammar for narratives
should give us an adequate database for (1) determining skill level, and (2) formulating
intervention goals. The variables analyzed in this study are quick and simple to calculate
using SALT. Given that SLPs are always strapped for time and generally overburdened
with large caseloads, this aspect if favorable. Since the variables are analyzed at various
levels of the text, clinicians can use this paradigm to identify relative strengths and
weaknesses in children's writing, which in turn will assist in setting clinical goals.
Further, finding a relationship between memory and writing supports the
importance of taking into account STM and WM skills when evaluating children with
written language difficulties and underscores the need to provide adequate support to
students with memory difficulties. Finally, the findings of this study strongly support the
need to consider lower-order processes of transcription and higher-order text generation
processes in the evaluation and remediation of writing difficulties (Berninger, 1999).
Limitations and Future Directions
Method of written language sampling: The elicitation procedure used in this
study is only one method of collecting a written language sample, other methods such as
watching an educational video or spontaneous writing can also be used. It is difficult to
predict if similar findings would result from writing samples collected in other contexts
such as report writing. It is very likely that performance might change on other types of
expository tasks. Given the scarcity of data on writing, it is clear that more work is
needed in this area to build a comprehensive database that will enable SLPs to assess and
evaluate written language in school-age children in a more comprehensive manner. This
study is an attempt in the right direction.
Data collection: Although participants for this study were chosen from schools
ranging in their socioeconomic status they were all recruited from one school district.
Building a normative database that would allow for the development of a reliable or valid
screening instrument would require stratified sampling, a larger number of subjects, as
well as subjects from different states to be fully representative. Also, data were collected
for only four grades. Normative data on writing for students in all grades should be
extended at least through high school. More research will also be needed to determine if
writing difficulties can be reliably identified using this protocol. Some comparison
groups might include children with language impairments and bilingual children at risk
for reading failure.
The PPVT-3 was modified for group administration with one modification
administered to grades 3 and 4 and another for grades 5 and 6. This made comparison of
vocabulary across grades difficult. While an attempt was made to correct this error by
recalculating scores on items common to all four grades, it is an issue to be mindful of in
Components of writing examined: The variables examined in this study were
confined to examining the microstructure of writing. More detailed analyses including
types of complex sentences, and examination of types of spelling errors would continue
to improve our understanding of grades level development in children. Studies in the
future examining the process and hence the macro-structural aspects (e.g., cohesion and
structural organization, revision and editing process) of writing are also needed.
Evaluation of memory: Examination of the role of memory to expository text was
preliminary and exploratory. Memory data were only collected for two grades (3 and 5)
and a limited number of tests were used. One goal for future research would be to include
a larger range of STM and WM tasks and to examine the role of memory across more
grades. The role of WM and STM to writing might be different for higher grades when
lower order transcription processes are more automatized. SLPs frequently work with
children with language impairments and reading deficits, both of whom have memory
deficits. Studies exploring the use of various strategies to support these students and its
effect on writing could be the subject of future research as well.
In summary, this study examined changes in the expository writing of children
from grades 3 through 6. Analysis of writing included several variables at the discourse,
T-unit, sentence, and word levels. Results showed that variables measuring productivity
and complexity of written language are sensitive to grade. These results allow us to
identify changes in writing among children from grades 3 through 6 and will lead to
procedures for assessing and treating children with written language disabilities. In
addition, this study demonstrated the important role of memory to writing. Once again,
these findings provide preliminary but standardized data for developing a system to
evaluate the performance of students who are struggling with writing.
STUDIES EXAMINING WRITING
Genre/ Studies Task Participants/ Summary of Results
Narrative Mackie & PLST was used to assess writing. 11 children with SLI (mean age 11 yrs.), 11 CA matched, and 11 LA
Written Dockrell (2004) Children were asked to look at a matched (mean age 7.3 years). Measures: total number of words,
picture, which was visible words per minute, content, proportion of syntax errors, and spelling
throughout the task and then errors. SLI children wrote shorter stories and used fewer words than
write a story about it. Children CA group and produced more syntax errors than both CA and LA
were allowed 30 minutes. matched group.
Fey, Catts, Children were allowed to a 538 children, 262 with typical language, 111- SLI, 75-NLI, and 90-
Proctor-Williams, choose set of pictures, asked to LNIQ children in 2nd and 4th grades. Stories of normally developing
Tomblin, & Zhang tell story with examiner prompts children had more different words, more grammatical complexity,
(2004) and then allowed to choose fewer errors and better overall quality than other 3 groups. Oral
another set of pictures and asked stories were better than written stories but greater improvements
to write a story. were seen for written stories from grades 2 through 4.
Nelson & Van Children were asked to write a i3 ND; 76 special needs; 6 ELL Grades 1 through 5. MLT-unit
Meter (2003) story about a problem and what d not differ from lst-2nd or 4th-5th grade nor differentiate special
happens- Data collected at 3 -eds students till mid 5th grade. NDW and % of words spelled
points in the school year rrectly was sensitive to grade level and special needs
Participants were shown a video
depicting scenes of conflict
(moral, social, physical) between
people. They were asked to tell a
story about an incident or
situation in which they
experienced similar problems
20 each- 4th, 7t, 11th grade and university graduate ND students.
Study was conducted in seven languages. Measures: total number of
words, number of words per clause, lexical diversity, and textual
coherence. Effect of age for text length was significant. Significant
effect for genre- narratives longer than expository text. Lexical
diversity greater in expository text than narratives. Younger subjects
made more spelling errors.
Scott & Windsor
Children were shown a 19-
minute video about Yanis and his
dream and asked to write a story.
The children were allowed 5
minutes to write.
60 students total- 20 with LLD, mean age 11.5 yrs. 20 matched for
CA. 20 matched for LA, mean age 8.11 years. Productivity measures
included total T-units, total words, total time, T-units/minute,
words/minute, % of T-units with mazes. For lexical diversity, NDW
was examined. Grammatical complexity measures included
words/T-unit, and clauses/T-unit. Grammatical errors included
errors/T-unit. Effect of genre and modality were consistent across
groups. Productivity measures were higher for narratives than
expository text. LLD children's performance significantly lower
than CA peers on all measures. The only measure that distinguished
LLD children from both CA and LA matched peers was extent of
Gillam & Johnston Children were shown 3 pictures. 10 LI, 10 each RA and CA matched. LI children produced higher %
(1992) Choose 1 of 3 to write a story. of grammatically unacceptable T-units compared to matched peers.
No differences were found for other measures. Grammatically
unacceptable T-units were more dramatic for written samples.
Written samples more complex than oral samples for all groups. LI
& RA children had different speaking-writing relationships from CA
& LA groups (i.e., percentage of complex T-units were higher in
spoken than written sample for LI & RA).
Graves, Compared stories under 3 30 LD, grades 5 and 6. No differences in fluency were noted.
Montague, & conditions- Provided story Differences were noted in story elements, coherence and
Wong (1990) grammar & characterization organization.
Issacson & Compared stories under two 42 LD children; grades 5 and 6. No difference in the holistic aspects
Mattoon (1990) planning conditions (goal & of writing.
Vallecorsa & Children were shown 3 pictures 23 LD, 23 NLD; grades 6 and 7. Written compositions evaluated for
Garriss (1990) and asked to compose a story. quality, fluency, and cohesion. LD children less coherent and fluent
and wrote fewer internal responses.
Laughton & Compared stories under 3 96 LD, 96 NLD; grades 3,4,5, & 6. LD children included fewer
Morris (1989) conditions. Story prompts, components in their stories.
production of fables with hidden
Houck & Subjects were asked to write 48 LD, 48 NLD (16 LD & 16 NLD in each grade 4, 8, and 11).
Billingsley (1989) about a trip and given a few Compared to NLD, LD children wrote fewer words and sentences,
directions. Allowed 20 mns. wrote more words per sentence, produced fewer words with 7 letters
or more, higher capitalization and spelling errors. No difference for
number of T-units & number of morphemes per T-unit.
Montague, Compared retold and written 36 LD, 36 NLD children; grades 4-5, 7-8 and 10. LD children less
Maddux, & stories for quality, fluency, and fluent and less coherent. No task or developmental effect.
Dereshiwsky coherence for grades.
MacArthur & Compared dictated, handwritten 11 LD children, grades 5 and 6. Dictated compositions were longer
Graham (1987) and word processed and of better quality.
Newcomer, Subjects shown a picture of a LD group- 16 third, 16 fifth graders and 15 seventh graders. Low
Barenbaum, & girl on a bike for oral retell; pig achieving group- 15 third, 17 fifth, and 14 seventh graders. NA
Nodine (1988) in the kitchen for written group- 19 third, 19 fifth, and 17 seventh graders. Modality (oral vs.
composition. written) did not affect story production. Overall LD children wrote
fewer stories and were less fluent at all three grade levels. No
differences in coherence and no task effect.
Barenbaum, Compared stories under two LD group- 16 third and fifth graders and 15 seventh graders. Low
Newcomer, & conditions. Draw a picture; write achieving group- 15 third, 17 fifth, and 14 seventh graders. NA
Nodine (1987) a story group- 19 third, 19 fifth, and 17 seventh graders. LD children wrote
fewer stories and were less fluent; task effect and developmental
effect were evident. Picture drawing was not beneficial to students.
Nodine, Children were shown pictures 27 LD, 31 RD, 31 NLD; Grades 5 and 6. LD children wrote fewer
Barenbaum, & and asked to produce a story. stories and were less fluent. No differences in cohesion were noted.
Poplin, Gray, Measured writing performance 99 LD, 99 NLD; Grades 3-4, 5-6, 7-8. LD made more errors in
Larsen, on Test of Written Language spelling, style, word usage and vocabulary (for grades 7 and 8).
Banikowski, & (TOWL).
Poteet (1979) Story composition 85 LD, 124 NLD; Grades 2, 4, 6. LD children made more
mechanical errors; no difference in syntax.
Expository Berman & Participants were shown a video 20 each- 4th, 7th, 11th grade and university graduate ND students.
Written Verhoeven, (2002) depicting scenes of conflict Study was conducted in seven languages. Measures: total number of
(moral, social, physical) between words, number of words per clause, lexical diversity, and textual
people. They were asked to write coherence. Effect of age for text length was significant. Significant
a composition on the topic of effect for genre- narratives longer than expository text. Lexical
problems between people. diversity greater in expository text than narratives. Younger subjects
made more spelling errors.
Scott & Windsor Children were shown a 15- 20 LLD, 20 CA (mean age 11.5), 20 LA (mean age 8.11). Written
(2000) minute video about The Desert summaries for expository (and narratives) were shorter than oral
and asked to write a compositions. Measures of expository writing lower than measures
composition. The children were of narrative writing.
allowed 5 minutes to write.
Englert, Raphael, Measure metacognitive 46 LD, 46 NLD, 46 HA; grades 4 and 5. LD children less able to
Anderson, processes and use of various text produce text structures; insensitive to planning processes.
Gregg& Anthony structures.
Blair & Crump Compare syntactic complexity of 54 LD; grades 6,7, and 8. Essays had greater syntactic density and
(1984) stories and essays. longer T-units across grades.
Morris & Crump Compared vocabulary and 72 LD, 72 NLD; ages 9-15. LD children used less variety of
(1982) syntactic maturity in essays. vocabulary words; no significant difference in T-units across age
Moran (1981) Children were allowed to pick a 26 LD, 26 NLD; grades 7-10. Measures: Syntactic maturity (mean
topic and instructed to write a morphemes per T-unit, total number of words and morphemes),
paragraph about it. writing conventions correct (grammatical errors), and percentage of
words spelled correctly. Only spelling performance was significantly
different between two groups.
CA= Chronological age matched; ELL= English Language Learners; HA= High Achieving; LA= Language age matched; LI=
Language Impaired; LLD= Language Learning Disabled; LNIQ= Low nonverbal IQ; ND/NL/ NLD = Normally developing; NLI=
Nonspecific Language Impairment; RA = Reading age matched; SLI= Specific Language Impairment.
IRB CONSENT FORMS
1. TITLE OF PROTOCOL: Expository oral and written language samples: Tracking
developmental changes from grades 2 through 6.
2. PRINCIPAL INVESTIGATOR(s): Cynthia Puranik, MA, CCC-SLP, Clinical
Supervisor and Doctoral Candidate. University of Florida, Department of
Communication Sciences and Disorders, 354 Dauer Hall, PO Box 117420, Gainesville,
FL 32611-7420, Tel: (352) 392-2113 x 286. Fax: (352) 846-0243. E-mail:
3. SUPERVISOR (IF PI IS STUDENT): Linda J. Lombardino, Professor- Department
of Communication Sciences and Disorders, 347, Dauer Hall, PO Box 117420,
Gainesville, FL 32611-7420, Tel: (352) 392-2113 x 285. E-mail: firstname.lastname@example.org
4. DATES OF PROPOSED PROTOCOL: January 2005 to December 2005.
5. SOURCE OF FUNDING FOR THE PROTOCOL: None
6. SCIENTIFIC PURPOSE OF THE INVESTIGATION:
The analysis of spontaneously produced oral and written language samples has
played an important role in the study of children with language impairments and learning
disabilities. However there are gaps in our knowledge and limitations of current studies.
For example 1) the majority of studies have been conducted a long time ago; 2) Most
studies have focused only on a small number of variables; 3) Studies for expository text
are scant; most studies have looked at language samples using narratives; 4) No study has
examined production of narratives or expository language using the procedures intended
for this study. Both genre and modality affect production of language samples; 5)
Developmental progression has not been studied at the discourse, sentence, and word
level; 6) The role of memory in production of oral and written expository language
samples has not be examined.
In this study, we will analyze several variables at the discourse, sentence, and
word level crucial to the oral language and writing process. The primary objective of this
proposal is to produce a comprehensive understanding of the developmental progression
of children's oral and written language skills. Since this study will document the
performance of children developing normally, this knowledge will be used by teachers
and speech-language pathologists involved in working with children with written
language disabilities for assessment and intervention purposes. Specifically it will help in
1) understanding and estimating the written language capabilities of learning-disabled
children; and 2) understanding the potential pathways to provide intervention for children
at risk for written language disabilities. Secondly, this study will examine the correlations
between short-term working memory and the ability to produce oral and language
7. DESCRIBE THE RESEARCH METHODOLOGY IN NON-TECHNICAL
This research is divided into two parts. In Part I, 150 participants each from grades
2 through 6 will be chosen. Children will be read a paragraph. Following the reading,
children will be required to reproduce the paragraph in writing. Graded paragraphs will
be used for language sample collection. The written samples and vocabulary test will be
group administered and collected in the classroom. In Part II, 30 participants from each
grade will then be randomly selected from the larger group to provide an oral language
sample, complete 2 short-term (digit span, nonword repetition) and 2 working memory
tasks (listening span, digit ordering). As in the collection of the written sample, graded
paragraphs will be read to the children. Following the reading, children will reproduce
the paragraph orally. This will be audio taped by the researcher for transcription at a later
stage. After the oral sample has been collected children will be administered the memory
tasks. Oral samples and memory tasks will be collected individually in a quiet room at
8. POTENTIAL BENEFITS AND ANTICIPATED RISK.
No benefits or risks are anticipated for the participants. Most children appear to enjoy the
activities. Children will be given candy or stickers for their participation.
9. DESCRIBE HOW PARTICIPANTS) WILL BE RECRUITED, THE NUMBER
AND AGE OF THE PARTICIPANTS, AND PROPOSED COMPENSATION (if
1) Following the protocol set in different counties, the appropriate administrator will
be contacted first. We will be contacting counties around Alachua.
2) After obtaining permission from the county, different schools in the county will
be contacted. If the principal of the school agrees, parental consent forms will be
sent to the parents of the children.
3) The letters will be handed to the classroom teacher who will then be requested to
send out the parental consent forms to children in his/her classroom.
4) Parental consent will be obtained for 150 children in each grade.
No monetary compensation for participation will be provided.
10. DESCRIBE THE INFORMED CONSENT PROCESS. INCLUDE A COPY OF
THE INFORMED CONSENT DOCUMENT (if applicable).
Parents of each potential participant will receive an informed consent document
describing the purpose of the study and the requirements of the participant. Classroom
teachers will first distribute the consent form to parents of these students. This
identification and consent process will take place before the investigators are made aware
of the potential participants. The document will contain information regarding the
confidentiality of those involved and statements citing that this is a completely voluntary
role with the right to withdraw without penalty. Students whose parents have agreed to
participate by signing and returning the informed consent document (attached) will then
be made known to the investigators. A participant will be part of the study pending the
signing and return of the informed consent document.
A child assent script will be used prior to each assessment and intervention session to
assure that the child is participating voluntarily (attached).
Principal Investigator's Signature
I approve this protocol for submission to the UFIRB:
Dept. Chair/Center Director
Informed Consent Letter for Parents and Guardians
I am a doctoral student in Communication Sciences and Disorders at the University of Florida,
conducting research under the supervision of Dr. Linda Lombardino, a professor of communication
processes and disorders. We are interested in tracking developmental changes in oral and written
language from grades 2 through 6. This information will be used by teachers and educators to help
children learning to read and write.
This study is divided into two parts. In Part I, written language samples will be collected. Your
child will first be read a paragraph and then asked to write what he/she remembers. This assignment is
expected to take approximately 5-10 minutes and will be conducted in the classroom. Only some children
will be randomly selected for Part II. If your child is selected for Part II, he/she will be read a paragraph
and will have to reproduce the paragraph orally and will also be given four short memory tests and a
vocabulary test. This should take between 15-20 minutes. The entire protocol will be accomplished in one
session. We also ask that you allow us to audiotape your child's oral language sample for transcription at a
later stage. Tapes and written records will be available only to an undergraduate clinical assistant, my
supervisor, and me. These will be numerically coded and will not be marked with your child's name. All
individual records will be destroyed once the study has ended. Your child's identity will be kept
confidential to the extent provided by law. No real names, initials, or other identifying information will be
used during spoken or written presentation of study results. Participation or non-participation in this study
will not affect your child's grades or placement in programs. We also ask that you give us permission to
obtain your child's FCAT scores from school files.
Your participation is voluntary. You and your child have the right to withdraw consent for your
child's participation at any time without consequence. There are no known risks in participating. Your
child will be given stickers or candy for participating. There are no direct benefits to you for participating
in this study. If you have any questions about this research protocol, please contact me at (352) 392-2113,
ext 250 or Dr. Lombardino at (352) 392-2113, ext 285. Questions or concerns about your child's rights as
a research participant may be directed to the UFIRB Office, University of Florida, Box 112250 Gainesville,
FL 32611, (352) 392-0433.
Cynthia Puranik, M.A. CCC/SLP Linda J. Lombardino, Ph. D.
Doctoral Candidate Professor, Speech Language Pathology
Parent Consent Form
(Please fill out both sides of this form)
I have read the procedure described above and I have received a copy of this form.
I voluntarily agree to allow my child to participate in this
study titled- Expository oral and written language samples: Tracking developmental
changes from grades 2 through 6.
Parent Consent Form
Child's Birthdate: Child's Grade level
With which parent does your child live? (Check one or both)
[ ] Mother or Female Guardian [ ] Father or Male Guardian
[ ] White [ ] African American [ ] Hispanic
 Asian [ ] Other
Please indicate highest level of education Mother or
[ ] Male [ ] Female
] Native American
No high school diploma or equivalent
Please specify grade level completed
High school diploma or equivalent [ ] Yes [ ] No [ ] Yes [ ] No
Attended college but no degree.
Please specify years completed.
Associate's degree or technical school [ ] Yes [ ] No [ ] Yes [ ] No
Bachelor's degree [ ] Yes [ ] No [ ] Yes [ ] No
Attended graduate school but no degree.
Please specify years completed.
Completed graduate school.
Please specify highest degree.
The primary language spoken in your home is:
[ ]English [ ]Spanish [ ]Other
Is your child taking medications prescribed by a medical doctor? [ ] Yes [ ] No
If yes, list medications
Is your child receiving services from any of the following?
Currently receiving Received in the past
Speech and Language Therapy [ ] [ ]
Resource Services for Reading [ ] [ ]
Gifted/Advance Placement [ [
Occupational Therapy  ]
Physical Therapy  
Counseling  
LEP/ESL Bilingual Education Services [ [
Child Assent Script
The following is a script that will be used prior to each session to ensure that the child
knows of his/her involvement and that he/she may choose not to participate if he/she does
not want to.
Investigator: I am going to read you a short paragraph two times. After I finish
reading, I would like you to write what you remember on a sheet of paper.
Do you want to listen to me and then write out the stories?
If the child indicates yes, the investigator will begin the session.
If the child indicates no, the investigator will not obtain a written language sample from
Investigator: I am going to read you a short paragraph two times. After I finish
reading, I would like you to tell me what you remember in your own words. I will
also tape your story so I can type what you said when I return to work. Also I will
give you four short tests that are called memory tasks.
Do you want to listen to me and retell the story in your own words? Do you want to
do the memory tasks?
If the child indicates yes to both tasks, the investigator will begin the session.
If the child indicates no for any one task, the investigator will terminate the session.
PPVT- Grades 3 & 4
I want to find out if you know the names of some pictures. Show Practice Item 1. See,
there are four pictures on this page. Each of them is numbered. Indicate by pointing to
and saying the numbers in turn: 1, 2, 3, and 4. Then say: I will say a word; then I want
you to bubble the number on your page that best tells the meaning of the word. Let's try
one. What number best tells the meaning of hook? Good!
Practice Items: Let's try another one. What number is group? That was a good answer.
When we begin the test, you will bubble #3 on your answer sheet (point to bubble sheet)
Now let's practice one last one. What number is dripping? Good! Point to the bubble
sheet and say, 'Remember to bubble the right number'). Is everybody ready?
Test Items: 1. (point to item # on stimulus sheet) What number is
1) 61. vehicle (4) 26) 97. pedal (2)
2) 63. luggage (2) 27) 90. interviewing (1)
3) 101. inhaling (4) 28) 133. blazing (3)
4) 65. hydrant (4) 29) 67. calculator (2)
5) 75. vase (3) 30) 103. tubular (3)
6) 147. ladle (2) 31) 105. tusk (1)
7) 71. vegetable (3) 32) 107. fern (1)
8) 111. inflated (3) 33) 143. pedestrian (2)
9) 73. gigantic (2) 34) 109. solo (4)
10) 141. cornea (2) 35) 69. squash (4)
11) 137. hoisting (1) 36) 155. cultivating (1)
12) 77. towing (1) 37) 113. timer (1)
13) 79. trunk (2) 38) 115. links (4)
14) 149. abrasive (1) 39) 151. cascade (4)
15) 81. island (2) 40) 117. microscope (1)
16) 83. heart (3) 41) 119. garment (4)
17) 139. consuming (4) 42) 142. constrained (3)
18) 85. flamingo (2) 43) 121. carpenter (2)
19) 87. palm (1) 44) 123. hazardous (3)
20) 135. reprimanding (1) 45) 145. syringe (4)
21) 89. canoe (3) 46) 125. valve (3)
22) 91. clarinet (4) 47) 127. feline (2)
23) 93. pitcher (3) 48) 129. coast (4)
24) 95. polluting (3) 49) 131. foundation (4)
25) 153. detonation (2) 50) 99. bouquet (4)
PPVT- Grades 5 & 6
I want to find out if you know the names of some pictures. Show Practice Item 1. See,
there are four pictures on this page. Each of them is numbered. Indicate by pointing to
and saying the numbers in turn: 1, 2, 3, and 4. Then say: I will say a word; then I want
you to bubble the number on your page that best tells the meaning of the word. Let's try
one. What number best tells the meaning of globe? Good!
Practice Items: Let's try another one. What number is vegetable? That was a good
answer. When we begin the test, you will bubble #3 on your answer sheet (point to
bubble sheet). Now let's practice one last one. What number is frame? Good! Point to the
bubble sheet and say, 'Remember to bubble the right number'). Is everybody ready?
Test Items: 1. (point to item # on stimulus sheet) What number is ?
1) 85. flamingo (2) 26) 117. microscope (1)
2) 87. palm (1) 27) 155. cultivating (1)
3) 111. inflated (3) 28) 131. foundation (4)
4) 89. canoe (3) 29) 147. ladle (2)
5) 119. garment (4) 30) 113. timer (1)
6) 91. clarinet (4) 31) 151. cascade (4)
7) 93. pitcher (3) 32) 121. carpenter (2)
8) 159. disappointed (4) 33) 153. detonation (2)
9) 95. polluting (3) 34) 170. pilfering (2)
10) 115. links (4) 35) 165. primate (4)
11) 141. cornea (2) 36) 145. syringe (4)
12) 109. solo (4) 37) 156. aquatic (4)
13) 97. pedal (2) 38) 139. consuming (4)
14) 99. bouquet (4) 39) 143. pedestrian (2)
15) 163. periodical (2) 40) 167. talon (3)
16) 101. inhaling (4) 41) 123. hazardous (3)
17) 171. trajectory (1) 42) 179. nautical (4)
18) 103. tubular (3) 43) 125. valve (3)
19) 157. indigent (2) 44) 135. reprimanding (1)
20) 105. tusk (1) 45) 127. feline (2)
21) 133. blazing (3) 46) 161. poultry (4)
22) 107. fern (1) 47) 173. derrick (4)
23) 177. gaff (1) 48) 175. monetary (3)
24) 129. coast (4) 49) 149. abrasive (1)
25) 168.octagon (3) 50) 137. hoisting (1)
CLPT-Sentence Span Task
Instructions: We will now wear earphones and listen to a tape. The person on the tape
will tell you what to do, but I'll tell you a little bit first. You'll hear groups of sentences.
After each sentence, tell me YES if it is true or NO if it is false. The person on the tape
will ask you to tell me the last word in each sentence in the group. As we go, it gets
harder because there are more and more sentences in the groups, but just do your best.
We'll do some practice ones so you'll get an idea of what the sentences are like. Are you
ready to begin?
Level Correct Child's T/F Word Score
Response Response Score recall
1. Children can play Y Y N
1. Apples are black N Y N
2. Ice is hot N Y N
Mice eat cheese Y Y N
1. Trees have leaves Y Y N
1. Trains can fly N Y N
2. Pumpkins are pple N Y N
Buses have wheels Y Y N
2. Boys can eat Y Y N
Bananas are blue N Y N
3. Carrots can dance N Y N
Water is dry N Y N
Sugar is sweet Y Y N
3. Buckets tell jokes N Y N
Horses have tails Y Y N
Milk is white Y Y N
4. Feathers can tickle Y Y N
Babies drive trucks N Y N
Birds can fly Y Y N
Cars build bridges N Y N
4. Snails have shells Y Y N
Chairs eat cake N Y N
Giants are small N Y N
Balloons can float Y Y N
Level Correct Child's T/F Word Score
Response Response Score recall
5. Shoes have ears N Y N
Fire bums paper Y Y N
Robins eat worms Y Y N
Cars can race Y Y N
Hotdogs can bark N Y N
5. Horses have feet Y Y N
Dishes can whistle N Y N
Fish pull wagons N Y N
Roses have thorns Y Y N
Cats can talk N Y N
6. Apples are square N Y N
Rabbits read books N Y N
Houses can jump N Y N
Pencils eat candy N Y N
Airplanes can fly Y Y N
Balls are round Y Y N
6. Fish can swim N Y N
Clouds wear slippers N Y N
Sheep eat lions N Y N
People have eyes Y Y N
Dogs can run Y Y N
Lemons are yellow Y Y N
Scoring guide: (1) One point each given for every True/False correct response for a
maximum possible of 42. (2) One point given for every word recalled for a maximum
possible of 42. Note: Sequence of word recall need not match order of presentation.
Read the number series clearly and slowly to the subject (about 1 word per second).
Their task is to repeat them back to you in numerical order from lowest to highest. Put a
check in the Score box when they get that sequence right. Discontinue when all items
have been given at a level and subject made two or more errors at that level.
"You are going to hear lists of numbers from 1 to 10 that are out of order. You need to
say the list back to me in the correct numerical order, which means from the smallest to
the biggest number. For example, if you hear 7-4, you say 4-7. It starts out easy so you
can get used to the task. Let's try some practice items first"
Practice Items: (1) 6 3 # of trials:
(2) 1 10 # of trials:
Subject #: Grade:
Item# Test sequence Response Score
1 5-2 2-5
2 3-8 3-8
3 9-4 4-9
4 8-1 1-8
5 5-2-3 2-3-5
6 7-4-10 4-7-10
7 6-2-9 2-6-9
8 3-8-5 3-5-8
9 5-4-1-7 1-4-5-7
10 8-3-9-6 3-6-8-9
11 6-2-10-4 2-4-6-10
12 8-6-3-5 3-5-6-8
13 6-10-7-4-1 1-4-6-7-10
14 2-8-3-7-6 2-3-6-7-8
15 9-1-5-3-10 1-3-5-9-10
16 3-6-5-2-9 2-3-5-6-9
17 10-1-8-2-4-7 1 -2-4-7-8-10
18 6-3-5-10-8-2 2-3-5-6-8-10
19 4-6-9-2-10-7 2- 4-6-7-9-10
20 3-6-2-8-5-1 1-2-3-5-6-8
21 4-10-3-7-2-5-1 1 -2-3-4-5-7-10
22 6- 1 -4- 8- 10-3 -5 1 -3 4- 5-6- 8- 10
23 3-9-4-2-5-10-7 2-3-4-5-7-9-10
24 8-1-7-5-3-4-9 1-3-4-5-7-8-9
Total Items correct
Written Language Stimulus Paragraph
Instruct students to write their names, grades, and date on top of the page.
I am going to read a paragraph called 'Where do people live?' When I am finished, I
would like you to write down what you remember. Read the text. Just in case you missed
some parts of the paragraph, I am going to read it to you one last time. Pay good
attention, because I won't be able to repeat it. Read text. Okay now begin writing.
People live in different places. Some people live in a city while others live in the
country. Still other people live in the areas between the city and the country, which are
People live in the city to be near their jobs. Cities have lots of factories, schools,
and offices. If people don't want to drive a long way to their jobs, they live in the city.
There are many things to do in the city such as visit museums and zoos. They also have
many movie theaters.
People live in the country to be close to their jobs, too. Many people who live in
the country are farmers. They plant crops on their land. They may sell their crops or may
use them to feed the animals that live on the farm. Farmers raise cows, pigs, and
chickens. Although life in the country is quiet, there are other things to do. You can find a
river to fish in or take walks in the woods.
Many people live in the suburbs. Some people think that people who live in the
suburbs have the best of both worlds. It doesn't take as long to get to either the city or the
country. The suburbs are more crowded than the country but less crowded than the city.
Where people live depends upon what they like most.
(Total # of words: 227)
PROTOCOL FOR LINGUISTIC ANALYSIS
Discourse Level T-unit/Sentence Level Word Level
Fluency # of Sentences: Total # of different words:
Total # of words:
# Simple correct:
Total # of ideas: Spelling Accuracy:
# Simple incorrect: # incorrect:
# Complex correct:
# Complex incorrect:
Grammatical Complexity Writing conventions
Total # of T-units: % Capitalization:
MLT-unit: % End periods:
Total # of Clauses:
Level 1: C IC
Level 2: C IC
Level 3: C IC
Level 4: C IC
Level 5: C IC
Level 6: C IC
Level 7: C IC
Grammatical Errors: Lexical Errors
Use of nonwords/
Verb or pronoun tense/ neologisms or coining.
Omitted or incorrect Nonspecific
inflection/ Omitted or vocabulary/ Lack of
substitution of referent or word
grammatical elements substitutions.
Violation of word order
Rules for coding in SALT
1) Enter utterances as T-units
A t-unit is one main clause plus any subordinate (dependent) clause or
nonclausal structure (e.g., prepositional/ verbal phrases etc.) that is
embedded in it. All coordinated clauses are separated out into T-units,
unless they contain a co-referential subject deletion in the second clause.
i. If people live in the city they don't have to drive (1 t-unit)
ii. There are people that live in the city and people that live in the
country (2 t-units)
The words 'and', 'then', 'but', 'or', and 'so' are often used to start a new
2) Must have a period (.) at the end of the T-unit.
3) Each line must have a code for
# of clauses (e.g., [Icl], [2cl], [3cl])
-correct or incorrect sentence
-simple or complex sentence (e.g., [sc], [si], [cc],[ci])
Line codes are always enclosed in square brackets [ ]
4) As needed each utterance will have a code for the following
Errors-grammatical and/or lexical ([GE] [LE])
Punctuation errors- missing capitalization, missing end period ([i-caperr]
Spelling errors ([spelerr])
5) Explanation for each code (as shown below) should appear at the top of each
written language transcript.
+[lcl]: 1 clause
+[2cl]: 2 clauses
+[3cl]: 3 clauses
+[4cl]: 4 clauses
+[sc]: Simple correct
+[cc]: Complex correct
+[si]: Simple incorrect
+[ci]: Complex incorrect
+[spelerr]: Spelling error
+[i-caperr]: Initial word capitalization error
+[.err]: Missing end period
6) Print 'Standard Measures' and 'Code Summary' for data.
Rules for classifying Simple and Complex Sentences
Simple Sentence- Sentence with one main clause.
e.g., People live in different places.
Farmers raise cows, pigs, and chickens.
Complex Sentence- Sentence with one main and one or more
subordinate/embedded clauses or two main clause or one main clause and verb phrase
joined by a coordinating conjunction (clause- must have a verb). Note: A clause is a
group of related words containing a subject with a verb.
e.g., The suburbs are more crowded than the country but less crowded
than the city.
If people don't want to drive a long way to their jobs, they live in
Rules for Errors
1. Grammatical Errors
Verb or pronoun tense/agreement/case, omitted or incorrect inflection/ omitted or
substitution of grammatical elements, violation of word order.
e.g., The city have a lot of people.
People living in the country are also close to their jobs, which are
In the contrary (country) there farmers.
They grow crops and eigth (either) sell it of (off) feed it to their
2. Lexical Errors
Use of nonwords/revisions/ nonspecific vocabulary/ lack of referent or word
e.g., People plant crops grow farms.
People that live in the city useuly (usually) diniune people
The suburiers (suburbs) has meidun (medium) of people.
* Don't count a spelling error more than once, unless it is spelled incorrectly and
differently each time.
* Their/there, too/to, cities/city's was counted as a spelling error.
Number of ideas
No. Idea from stimulus paragraph
1 People live in different places.
2 People live in the city.
3 People live in the country.
4 People live in the suburbs.
5 Suburbs- place between the city and country.
6 People live in cities to be near their jobs.
7 Cities have schools/factories/offices
8 People live in cities if they don't want to drive a long way.
9 Visit museums, and zoos
10 Go to the movie theatre.
11 People live in the country to be close to their jobs
12 People who live in the country are farmers
13 They plant crops
14 They sell crops
15 Or feed crops to their animals
16 Farmers raise cows, pigs, chickens
17 Life in the country is quiet
18 Could go fishing
19 Take a walk in the woods
20 People who live in the suburbs have best of both worlds.
21 It doesn't take long to get to either the city or the country
22 The suburbs are less crowded than the city
23 The suburbs are more crowded than the country
24 People live where they like it best
25 There are many things to do in the city
Sample of a written language analysis in SALT
+[lcl]: 1 clause
+[2cl]: 2 clauses
+[3cl]: 3 clauses
+[4cl]: 4 clauses
+[sc]: Simple correct
+[cc]: Complex correct
+[si]: Simple incorrect
+[ci]: Complex incorrect
+[spelerr]: Spelling error
+[i-caperr]: Initial word capitalization error
+[.err]: Missing end period
c Suburb/s are between the country and city [Icl] [sc].
c there[i-caperr] are (alot) a lot (farnerms) farmer/s[spelerr] in[GE] suburb/s[.err] [Icl]
c people[i-caperr] live in the city for[LE] they can be close to their job [2cl] [ci] [L5-I].
c Farmer/s in the suburbs train[LE] pig/s, chickens and cow/s [Icl] [si].
c (their) there[i-caperr][spelerr] is[GE] (alot) a lot of schools, offices and factories in the
(cities) city/s [Icl] [si].
c (their) there[i-caperr] are not many farmers in the city [Icl] [sc].
c the[i-caperr] farmers grow crop/s and give them to their animals [2cl] [cc] [L2].
Total Utterances (number of T-units) 7
# MLU in Words (MLT-unit) 9.43
# No. Different Word Roots (NDW) 38
# Total Main Body Words (TNW) 66
Word Codes Utterance Codes
MAIN BODY MAZES
Anal Total Anal Total
Set Utts Set Utts
1 1 0 0
2 2 0 0
5 5 0 0
2 2 0 0
2 2 0 0
MAIN BODY MAZES
Anal Total Anal Total
Set Utts Set Utts
[1CL] 5 5 0 0
[2CL] 2 2 0 0
[CC] 1 1 0 0
[CI] 1 1 0 0
[SC] 2 2 0 0
[SI] 3 3 0 0
Altmann, L. J. P., Kempler, D., & Andersen, E. S. (2001). Speech errors in Alzheimer's
disease: Reevaluating morphosyntactic preservation. Journal of Speech, Language,
and Hearing Research, 44, 1069-1082.
American Speech Language and Hearing Association. (2001). Knowledge and skills
needed by Speech-Language Pathologists with respect to reading and writing in
children and adolescents. ASHA Desk Reference, vol.3, pp.355-386.
Anglin, J. M. (1993). Vocabulary development: A morphological analysis. Monographs
of the Society for Research in Child Development, 58.
Baddeley, A. (1986). Working memory. Oxford: Clarendon Press.
Baddeley, A. (2003). Working memory: Looking back and looking forward. Nature,
4(October 2003), 829-839.
Baddeley, A., & Hitch, G. (1974). Working memory. In G. Bower (Ed.), The psychology
of learning and motivation (pp. 47-90). New York: Academic Press.
Baker, S., Gersten, R., & Graham, S. (2003). Teaching expressive writing to students
with learning disabilities: Research-based applications and examples. Journal of
Learning Disabilities, 36(2), 109-123.
Berman, R., Verhoevan, L. (2002). Cross-linguistic perspectives on the development of
text-production abilities. Written Language and Literacy, 5(1), 1-43.
Beminger, V. W. (1999). Coordinating transcription and text generation in working
memory during composing: Automatic and constructive processes. Learning
Disability Quarterly, 22(2), 99-112.
Beminger, V. W., & Hooper, S. (1993). Preventing and remediating writing disabilities:
Interdisciplinary framework for assessment. School Psychology Review, 22(4), 590-
Beminger, V. W., Yates, C., Cartwright, A., Rutberg, J., Remy, E., & Abbot, R. (1992).
Lower-level developmental skills in beginning writing. Reading and Writing: An
Interdisciplinary Journal, 4, 257-280.
Beminger, V. W., Cartwright, A., Cheryl, M., Yates, C., Swanson, H. L., & Abbott, R.
(1994). Developmental skills related to writing and reading acquisition in the
intermediate grades. Reading and Writing: An Interdisciplinary Journal, 6, 161-
Beminger, V. W., Mizokawa, D., Bragg, R., Cartwright, A., & Yates, C. (1994).
Intraindividual differences in levels of written language. Reading and Writing
Quarterly, 10, 259-275.
Beminger, V. W., Yates, C., Cartwright, A., Rutberg, J., Remy, E., & Abbot, R. (1992).
Lower-level developmental skills in beginning writing. Reading and Writing: An
Interdisciplinary Journal, 4, 257-280.
Bourke, L., & Adams, A. (2003). The relationship between working memory and early
writing assessed at the word, sentence and text level. Educational and Child
Psychology, 20(3), 19-36.
Cantor, J., Engle, R., & Hamilton, G. (1991). Short-term, working memory, and verbal
abilities: How do they relate? Intelligence, 15, 229-246.
Carpenter, P., Miyake, A., & Just, M. (1994). Working memory constraints in
comprehension: Evidence from individual differences, aphasia, and aging. In M. A.
Gernsbacher (Ed.), Handbook ofPsycholinguistics (pp. 1075-1116). San Diego:
Cowan, N. (1996). Working memory and language impairment. In R. Gillam (Ed.),
Memory and language impairment in children and adults: New perspectives (pp. 3-
27). Gaithersburg, MD: Aspen Publishers.
Chall, J. S. (1983). Stages of reading development. New York: McGraw Hill.
Daneman, M., & Carpenter, P. (1980). Individual differences in working memory and
reading. Journal of Verbal Learning and Verbal Behavior, 19, 450-466.
Dunn, L., & Dunn, L. (1997). Peabody Picture Vocabulary Test-3. Circle Pines, MN:
American Guidance Service.
de Jong, P. F. (1998). Working memory deficits of reading disabled children. Journal of
Experimental Child Psychology, 70, 75-96.
Dixon, P., LeFevre, J., & Twilley, L. (1988). Word knowledge and working memory as
predictors of reading skill. Journal of Educational Psychology, 80, 465-472.
Ellis Weismer, S. (1996). Capacity limitations in working memory: The impact on lexical
and morphological learning by children with language impairment. Topics in
Language Disorders, 33-44.
Ellis Weismer, S., Evans, J, & Hesketh, L. (1999). An examination of verbal working
memory capacity in children with specific language impairment. Journal of Speech,
Language, and Hearing Research, 4, 1249-1260.
Ferreiro, E., & Kucchermaglio, C. (1996). Children's use of punctuation marks: The case
of quoted speech. In C. Pontecorvo, M. Orsolini, B. Burge & L. B. Resnick (Eds.),
Children's early text construction (pp. 177-205). Mahway, NJ: Erlbaum.
Fey, M., Catts, H., Proctor-Williams, K., Tomblin, B., & Zhang, X. (2004). Oral and
written story composition skills of children with language impairment. Journal of
Speech, Language, and Hearing Research, 47, 1301-1318.
Gambrell, L., Koskinen, P., & Kapinus, B. (1991). Retelling and the reading
comprehension of proficient and less-proficient readers. Journal of Educational
Research, 84(6), 356-362.
Gambrell, L., Pfeiffer, W., & Wilson, R. (1985). The effects of retelling upon reading
comprehension and recall of text information. Journal of Educational Research,
Gathercole, S. E., & Baddeley, A. D. (1990). Phonological memory deficits in language
disordered children: is there a causal connection? Journal of Memory and
Language, 29, 336-360.
Gathercole, S. E., & Baddeley, A. D. (1993a). Phonological working memory: a critical
building block for reading development and vocabulary acquisition? European
Journal of Psychology ofEducation, 8, 259-272.
Gathercole, S. E., & Baddeley, A. (1993b). Working memory and language. Hove, U.K.:
Lawrence Erlbaum Associates Ltd.
Gathercole, S. E., Pickering, S., Knight, C., & Stegmann, Z. (2004). Working memory
skills and educational attainment: Evidence from national curriculum assessment at
7 and 14 years of age. Applied Cognitive Psychology, 18, 1-16.
Gaulin, C., & Campbell, T. (1994). Procedure for assessing verbal working memory in
normal school-age children: Some preliminary data. Perceptual and Motor .\d//l,
Gazella, J., & Stockman, I. (2003). Children's story retelling under different modality and
task conditions: Implications for standardizing language sampling procedures.
American Journal of Speech-Language Pathology, 12, 61-72.
Gillam, R., Cowan, N., & Day, L. (1995). Sequential memory in children with and
without language impairment. Journal of Speech & Hearing Research, 38, 393-
Gillam, R., & Johnston, J. (1992). Spoken and written language relationships in
language/learning-impaired and normally achieving school-age children. Journal of
Speech and Hearing Research, 35, 1303-1315.
Graham, S., & Harris, K. (1997). It can be taught, but it does not develop naturally:
Myths and realities in writing instruction. The School Psychology Review, 26(3),
Graham, S., Harris, K., & Troia, G. (2000). Self-regulated strategy development
revisited: Teaching writing strategies to struggling writers. Topics in Language
Disorders, 20(4), 1-14.
Graves, A., Montague, M., & Wong, Y. (1990). The effects of procedural facilitation on
the story composition of learning disabled students. Learning Disabilities
Research, 5, 88-93.
Hadley, P. (1998). Language sampling protocols for eliciting text-level discourse.
Language, Speech and Hearing Services in Schools, 29, 132-147.
Hammill, D. & Larsen, S. (1988). Test of Written Language-2. Austin, TX: Pro-Ed.
Harp, B. (2000). The handbook of literacy assessment and evaluation (2nd ed.).
Norwood, MA: Christopher-Gordon Publishers, Inc.
Hayes, J., & Flower, L. (1980). Identifying the organization of writing processes. In L.
Gregg & E. Steinberg (Eds.), Cognitive processes in writing: An interdisciplinary
approach (pp. 3-30). Hillsdale, NJ: Erlbaum.
Hayes, J., & Flower, L. (1987). On the structure of the writing process. Topics in
Language Disorders, 7, 19-30.
Hitch, G., Towse, J., & Hutton, U. (2001). What limits children's working memory span?
Theoretical accounts and applications for scholastic development. Journal of
Experimental Child Psychology, 130(2), 184-198.
Hooper,S. & Montgomery, J. (1993). Prevalence of writing problems across three middle
school samples. School Psychology Review, 22(4), 610-623.
Hoskyn, M., & Swanson, H. L. (2003). The relationship between working memory and
writing in younger and older adults. Reading and Writing: An Interdisciplinary
Journal, 16, 759-784.
Houck, C., & Billingsley, B. (1989). Written expression of students with and without
learning disabilities: Differences across the grades. Journal of Learning
Disabilities, 22, 561-565.
Hunt, K. (1965). Grammatical structures written at three grade levels (No. National
Council of Teachers of English, Research Report No. 3.). Champaign, IL.
Hunt, K. (1970). Syntactic maturity in school children and adults. Society for Research in
Child Development Mongrahs No. 134, 35(1).
Hux, K., Morris-Friehe, M., & Sanger, D. (1993). Language sampling practices: A survey
of nine states. Language, Speech and Hearing Services in Schools, 24, 84-91.
Just, M., & Carpenter, P. (1992). A capacity theory of comprehension: Individual
differences in working-memory. Psychological Review, 99, 122-149.
Justice, L., Bowles, R., Kaderavek, J., Ukrainetz, T., Eisenberg, S., & Gillam, R. (2006).
The index of narrative micro-structure (INIMIS): A clinical tool for analyzing
school-aged children's narrative performances. American Journal of Speech-
Language Pathology,15, 177-191.
Kemper, S., & Sumner, A. (2001). The structure of verbal abilities in younger and older
adults. Psychology andAgi.ng. 16, 312-322.
Kibby, M., Marks, W., Morgan, S., & Long, C. (2004). Specific impairment in
developmental reading disabilities: A working memory approach. Journal of
Learning Disabilities, 37(4), 349-362.
Klecan-Aker, J. S., & Hedrick, L. D. (1985). A study of the syntactic language skills of
normal school-age children. Language, Speech, and Hearing Services in Schools,
Laughton, J., & Morris, N. (1989). Story grammar knowledge of learning disabled
students. Learning Disabilities Research, 4, 87-95.
Leslie, L., & Caldwell, J. (2001). Qualitative Reading Inventory-3. Boston, MA: Allyn &
Loban, W. (1976). Language development: Kindergarten through grade twelve (No. 18).
MacDonald, M. C., Almor, A., Henderson, V. W., Kempler, D., & Andersen, E. S.
(2001). Assessing working memory and language comprehension in Alzheimer's
disease. Brain and Language, 78, 17-42.
Mackie, C., & Dockrell, J. (2004). The nature of written language deficits in children
with SLI. Journal of Speech, Language, and Hearing Research, 47, 1469-1483.
McCutchen, D. (2000). Knowledge, processing, and working memory: implications for a
theory of writing. Educational Psychologist, 35(1), 13-23.
McCutchen, D., Covill, A., Hoyne, S., & Mildes, K. (1994). Individual differences in
writing: Implications of translating fluency. Journal of Educational Psychology,
Merrit, D. & Liles, B. (1989). Narrative analysis: Clinical applications of story generation
and story retelling. Journal of Speech and Hearing Disorders, 54, 438-447.
Miller, G. (1956). The magical number seven, plus or minus two: Some limits on our
capacity for processing information. Psychological Review, 63, 81-97.
Miller, J. (1981). Assessing language production in children (Vol. 1). Baltimore, MD:
University Park Press.
Miller, J. (1991). Quantifying productive language disorders. In J. Miller (Ed.), Research
on child language disorders: A decade ofprogress (pp. 211-220). Austin, TX: Pro-
Miller, J., & Chapman, R. (2001). Systematic Analysis of Language Transcripts (Version
7.0). Madison, WI: Waisman Center, University of Wisconsin-Madison.
Montague, M., Maddux, C., & Dereshiwsky, M. (1990). Story grammar and
comprehension and production of narrative prose by students with learning
disabilities. Journal ofLearning Disabilities, 23, 190-196.
Montgomery, J. (1995). Sentence comprehension in children with specific language
impairment: The role of phonological working memory. Journal of Speech and
Hearing Research, 38, 187-199.
Moran, M. (1981). Performance of learning disabled and low achieving secondary
students on formal features of a paragraph-writing task. Journal of Learning
Disabilities, 23, 271-280.
Morris, N. T., & Crump, W. D. (1982). Syntactic and vocabulary development in the
written language of learning disabled and non-learning disabled students at four age
levels. Learning Disability Quarterly, 5, 163-172.
Morrow, L. M. (1985). Retelling stories: A strategy for improving young children's
comprehension, concept of story structure, and oral language complexity. The
Elementary School Journal, 85(5), 646-660.
Nelson, N. W. (1998). Childhood language disorders in context: Infancy through
adolescence. (2nd ed.). Columbus, OH: Merrill.
Nelson, N. W., Bahr, C., & Van Meter, A. (2004). The writing lab approach to language
instruction and intervention. Baltimore, MD: Paul H. Brookes Publishing Co., Inc.
Nelson, N. W., & Van Meter, A. (2002). Assessing curriculum-based reading and writing
samples. Topics in Language Disorders, 22, 35-59.
Nelson, N. W., & Van Meter, A. (2003, June). Measuring written language abilities and
change through the elementary years. Poster session presented at the annual
meeting of the Symposium for Research in Child Language Disorders, Madison,
Newcomer, P., Barenbaum, E., & Nodine, B. (1988). Comparison of the story production
of LD, normally achieving, and low-achieving children under two modes of
production. Learning Disability Quarterly, 11, 82-96.
Nippold, M., Ward-Lonergan, J., & Fanning, J. (2005). Persuasive writing in children,
adolescents, and adults: A study of syntactic, semantic, and pragmatic
development. Language, Speech, and Hearing Services in Schools, 36,125-138.
Nodine, B., Barenbaum, E., & Newcomer, P. (1985). Story composition by learning
disabled, reading disabled, and normal children. Learning Disability Quarterly, 8,
O'Donnell, R. C., Griffin, W. J., & Norris, R. D. (1967). Syntax of kindergarten and
elementary school children: A transformational analysis (No. 8). Champaign, IL:
National Council of Teachers of English.
Paul, R. E. (2001). Language Disorders from Infancy through Adolescence: Assessment
andIntervention (2nd ed.). St. Louis, MS: Mosby.
Poplin, M., Gray, R., Larsen, S., Banikowski, A., & Mehring, T. (1980). A comparison of
components of written expression abilities in learning disabled and non-learning
disabled children at three grade levels. Learning Disability Quarterly, 3, 46-53.
Poteet, J. A. (1979). Characteristics of written expression of learning-disabled and non
learning-disabled elementary school students. Diagnostic, 4, 60-74.
Puranik, C., Lombardino, L., & Altmann, L. (in press). Writing through retellings: An
exploratory study of language-impaired and dyslexic populations. Reading and
Writing: An Interdisciplinary Journal.
Rosenberg, S. (1974). Linguistic maturity and language development in elementary
school-age children. Chicago: Computer Psychometrics Affiliates.
Rosenberg, S., & Abbeduto, L. (1987). Indicators of linguistic competence in peer group
conversational behavior of mildly retarded adults. AppliedPsycholinguistics, 8, 19-
Scott, C. (1988). Spoken and written syntax. In M. Nippold (Ed.), Later language
development: Ages nine through nineteen. (pp. 49-95). San Diego: CA: College-
Scott, C. (1989). Problem writers: Nature, assessment, and intervention. In A. Kamhi &
H. Catts (Eds.), Reading Disabilities: A developmental language perspective (pp.
303-344). Boston:MA: Allyn & Bacon.
Scott, C. (1994). A discourse continuum for school-age students: Impact of modality and
genre. In G. Wallach & K. Butler (Eds.), Langauge learning disabilities in school-
age children and adolescents (pp. 219-252). New York: Merill.
Scott, C. (1999). Learning to write. In H. Catts & A. Kamhi (Eds.), Language and
reading disabilities (pp. 224-258). Needham Heights, MA: Allyn & Bacon.
Scott, C. & Stokes, S. (1995). Measures of syntax in school-age children and adolescents.
Language, Speech, and Hearing Services in Schools, 26, 309-317.
Scott, C., & Windsor, J. (2000). General language performance measures in spoken and
written discourse produced by school-age children with and without language
learning disabilities. Journal of Speech, Language, and Hearing Research, 43, 324-
Silliman, E. R., Jimerson, T., & Wilkinson, L. (2000). A dynamic systems approach to
writing assessment in students with language learning problems. Topics in
Language Disorders, 20, 45-64.
Simone, R. (1996). Reflections on the comma. In M. O. C. Pontecorvo, B. Burge, & L.B.
Resnick (Ed.), Children's early text construction (pp. 165-175). Mahway, NJ:
Singer, B., & Bashir, A. (2004). Developmental variations in writing composition skills.
In C. A. Stone, E. R. Silliman, B. J. Ehren & K. Apel (Eds.), Handbook of
language and literacy development and disorders. NY: Guilford Press.
Swanson, H. L. (1993). Individual differences in working memory: A model testing and
subgroup analysis. Intelligence, 17, 285-332.
Swanson, H. L., & Berninger, V. W. (1996). Individual differences in children's working
memory and writing skill. Journal ofExperimental Child Psychology, 63, 358-385.
Taylor, B. (1980). Children's memory for expository text after reading. Reading Research
Quarterly, 3, 399-407.
Torgesen, J. K. (1985). Memory processes in reading disabled children. Journal of
Learning Disabilities, 18, 350-357.
Towse, J., Hitch, J., & Hutton, U. (1998). A reevaluation of working memory capacity in
children. Journal of Memory and Language, 39, 195-217.
Treiman, R. & Kessler, B. (2005). Writing systems and spelling development. In M.J.
Snowling & C. Hulme (Eds.), The science of reading: A handbook (pp. 120-134).
Malden, MA: Blackwell Publishing.
Turner, M., & Engle, R. (1989). Is working memory capacity task dependent? Journal of
Memory and Language, 28, 27-154.
U.S. Department of Education (2003). The nation's report card: Writing highlights 2002.
[Electronic version] Retrieved December 18, 2005, from
Vallecorsa, A., & Garriss, C. (1990). Story composition skills of middle-grade students
with learning disabilities. Exceptional Children, 57, 48-53.
Wagner, R., Torgesen, J., & Rashotte, C. (2001). Comprehensive Test of Phonological
Processing. Circle Pines, MN: American Guidance Service.
Cynthia Puranik was born and raised in Mumbai, India, where she completed
most of her former schooling. She began her professional career at University of Florida,
Gainesville, where she earned a Master of Arts degree in 1996. Following her master's,
Cynthia has been working as a Speech-Language Pathologist in Tampa, Florida, in
various clinical settings including rehabilitation centers, private practice and the public
schools. She returned to the University of Florida in 2002 to pursue her doctoral degree.
She hopes to pursue a career in research and teaching at the University level.
Cynthia's teaching and clinical experiences are in the areas of language
development and disorders, with a focus on literacy and reading disabilities. Her specific
areas of research interest include examining the relationship between written language
and aspects of cognitive functioning, exploring similarities and differences between
language impairment and dyslexia, and investigating the effect of grammatical deficits on