Reading between the lines - Kenniscentrum Begrijpend Lezen

Transcription

Reading between the lines - Kenniscentrum Begrijpend Lezen
10()
DEVELOPMENTAL APPROACHES
J. c.
& Goswami, U. (20()5). Reading
and skilled reading across l<UII'."'''I'.L BIll/etill, 131, :\-29. A. M., & Stone, G. 0.
Statistical
of the bidi­
Behavior Researrh
The DRC model of visual word
aloud: An pxt(:'nsion to German.
l~eading
j:C"MMMM
between the Lines
;11
C~(:«llifil'c
Proccsscs
ill Rcadill,i!, C-;olflprc/tcllsioll
PJld \'an del] Brock,
P,ll),lyiotcl Kcndcou.
M~lry Jlllt'
dlld
White, Sarah (:arison Van den Broek, P. W., White, M. J., Kendeou, P., & Carlson, S. (2009). Reading between
the lines: Developmental and individual differences in cognitive processes in reading
comprehension. In R. K. Wagner, C. Schatschneider, & C. Phythian-Sence (Eds.), Beyond
decoding: The behavioral and biological foundations of reading comprehension (pp.
107-123). New York: Guilford Press.
is ,1n c'ssential
of Ii
difficultic's read in?; and under­
to them tc'nd to suffer from
111
tor Education
2002; Neltional
is formalc'duc'tltion difficult for tht'se individuals, but the
to rdll'ct, shelre
,mel n~,lS(ln effectively are hindered in
of tl1l'ir live'S. Thus, it is csscntial that Wl' um:iersl,md lhl' processes that
hId to successful reading com prehension and thl' ways in which these
processes can fail. Such understanding can h,lVl' far-reaching implica­
tions for educational practice, specifically with respect to assessment,
diagnosis, and intervention for both good and struggling readers.
In this chapter, we discuss the cognitive processes underlying read­
and the development of these processes. The chap­
ter consists of three m<1jor sections. The first section provides a general
overview of
processes in reading comprehension drawing on
108
DEVELOPMENTAL APPROACHES
recent cognitive theories. The second section summarizes the develop_
ment of reading comprehension and difficulties experienced by strug­
readers in the context of findings from two
studies. In
the third and final section, we discuss theoretical and
implica­
tions derived fwm the results of this research.
COGNITIVE PROCESSES IN READING COMPREHENSION Extensive research h,1s l'xamined the cngni ti\'l~ processes of reading
comprehension
CernsbcKher, 1990; (;ral'sser,
& Trabasso,
Kintsch & van Diik. 1978; Trabasso & Vim den Broek, 1985; van
Piast,l, &
2006; Whitehurst & Loni­
gall,
20(6). J\ common theme that has emerged
is thilt to
is l'ssl'lltial that a )"(';)(i('r is ,1bl(' to trans­
latl' written code intoll1(',mingflll ]'lIlguage ullits imel to cornbine these
Ull i ts into a coherell t Ine'n ta I rt'PrL'S(,1l ta tion of the text. This
tion call be ,lCcesscd by the rt',lder for d iffl'rent purposes after
compldl'd: to recall inforll1,ltioll from till' tl'xt, to answer qUl'stions, to
the knowkdgl' obtailll'd from the t('xl, to drmv p<ll'dllc'ls to other
tL'xts, ,1nd so 011. Thus, it is tlw qu,llilv of the reddl'r's findl rt'pr('senta­
tion of a text th'lt ddermil1l'S thl' 1IItim'ltl' sllccess of his or lwr compre­
IWl1sioIl of th,l! tnt.
SIOIl
through ,1 tl'xl
lions once read
,llld offlinc represcntiltions "rp caus'lily rl'Lltl'd: The proccsses that
unfold during mOl11l'nt-by-monwnt H'iltiing comprchension provide
the basis for the construclilln of the olflil1(' t('xl rl'Dresent,ltion in the
re'ldcr's mind. If the online pnKl'SSt'S
so docs the final
tion.
automatic, and relatively
whereas others are slow,
relatively effortfuL The mix of automatic and
processes
differs among individual readers and, indeed, for a particular reader
acrOSS reading situations
as a function of fatigue, text difficulty, or
reading purpose). The process of interconnecting and integrating informdtion is con­ strained, on the one hand, by the limitations of human attentional capac­ ity and short-term memory and, on the other hand, by the stalldards of that a reader attempts to maintain in a particular reading situ­
ation (e.g., as a function of motivation or goal for reading, the nature
of the reading task; see \'an den Broek, Risden, &
vall de Veldc, 19R9). Readers cem use a wide range of standards
to mainh1in coherence, but two
of standards, callsal coherence
and refl'renti,ll coherence, ,lr(' prevalent in virtually all reading situa­
tions dnd an' p<lrticuiarly important for the construction of coherent
(e.g., Craessl'r & Cbrk,19R5; Kintsch,19RR; O'Brien &
Myers,19R7; Tmbasso & Sperry, 19R5). Consider, for l'x<ll1lpll', the spn­
tence pa i r:
John dropped the banana peel on the floor. Mary fell on her back.
Most readers infer that till' infonnCltion in tlWS(' two sentences is calls­
often \sithout fl>'lli/ing that (1 causnl infe]"('ncl' h,lS
been m,lde. "IlK' fact that John dropped till' ban,m<1 pl'l'1 on tht> floor
is not sufficient to l'xpl<lin why Milry fell unless the readl'r inft'rs that
b,ll1,llld pecls <lrc slippery, ,1nd vvlwn
a person may f,ll1 (\,,111 den i3roek, Illl)()). Caus,l coherence,
is ,1tt,lil1l'd v.. hen there is sufficient inform'ltion
by tile text or
in the n,l tll I"e of
the b<lckground knowledge of tlll' rl'ildef to
('\·ents.
add ition, rderential colll'rellce occurs when the re,lcil'r connects
objects, charactt'rs, and other entities across sentences in tIll' tl'Xt. Con­
sider, for (,X,ll11 pic, the sentence p,lir:
The
COlllprehension Processes during Reading (Online)
The online construction oi d cohercnt mental
of the text
involves a complex set of processes that involve connecting and inte­
the text iniormation that the wader currently is reading
information that occurred earlier in the text as well as with infonna­
tion from background knowle-dge. Some of these proc('sses are quick,
09
Developrnel1t
gave the waiter $10. He returned to
her the change.
this pi.1ir of sentences, most readers establish rderential coherence
inferring that "he" in the second sentence rders to the "waiter" in the
sentence and that "her" refl'rs to the "lady," again usually without
eVen becoming aware of making such inferences. Interestinglv. readers
tend to infer a causal rela tion, based on their
Waiters (and restaurantst people paying waiters. and so on.
110
Thus, as the reader progresses through a text, he or she attempts
to maintain coherence by gathering information from the currently
read sentence while integrating this information with previously read
text and background knowledge. The reader does this by attempting
to identify causal, referential, and other relations. Any such attempt,
needs to take place within rather severe limitations of a per­
son's working memory or attentionallimitations; a reader is unable to
simply accumulate every piece of information provided or activated
the text but instead is subject to a continuous w,1xing and waning of
available information.
A useful way to
scape in which individual pieces of information
in their levels of activation over time (for a review, see the
model by van den l3roek, Risden, Fletcher, & Thurlow, 1996; see also
Goldman & Varma, 1(95). Only pieces of information that are activated
C,ln be connected by H1l' comprehl'nd('r. As a result, if
to-be-connected pieces of inform,ltiOll Me r('ildilv ,w,lililblt' at the same
time (e.g., if they are presl'nted close in tinll' or are in-ailablc close in
the text), comprt'lwl1siol1 is an effortless, automatic pnlCl'ss. If to-be­
connected pieCl's of information are not nc',ldily avclibbie. the execution
of strategic pron'ssl's is fl'quirl'd, resulting in !11(ll"l' dfortfuI reading or
even failure to compreI-wnd altogeHwr. Thl'rl' ,1re consideri1ble individ­
ual and developn1L'ntal diffvrel1ces in the dfectiVl'IH'SS ,mel dficiency of
thl' various pmCt'SSl'S th"t result in the Iandscapl' of activ,ltions (reflect­
ing diffl'rencl's in, e.g., background knowledge', working memory, expe­
rience, motivtltion, knowledge of
11
Reading Development
DEVELOPMENTAL APPROACHES
~
b
0
~
I
~\12
\€
o
I
(~)
FIGURE 5.1. Clllsal network
of
MOO/IS
For (1Ccurate comprehension to occur, it is neCl'ssary that one's Tl1(:'n­
tal reprt'Sent,ltioll reflects such ill1l'twork representation. This represen­
tation includes both qutlJltitati'lll' ,md qualitalive as~1l'cts of the represen­
tation. Tht'refofl" assessment of comprehension ne(:~ds to involv(:' both
measurements of childn'n's memory for (1 text
should not only provide ,1n index of how IIllltli they rCCi111 (the quantita­
tive ,lspect) but alsotu/1111 is rL'caill'd (the LJllc1litatiw aspect): Ideas that
The Mental Representation of Texts (Offline)
A 111emory represent,ltion of the h'xt <1rld rl'lev,mt background knowl­
emerges from the dyn,lmically fluctu,lting I,mdscape of dctivation,
processes, and strategies. This representation is coherent to tlw extent
that it forms a llei,POrk of relations (Craesser & Clark, 1985; O'Brien &
Myers, 1987; Trabasso & v,m den Brot'k, 1985). An L'xampk of a causal
network is presented in Figure 5.1. The numbered nodes in this network
refer to unique idl'<1s from the beginning of a narrative story, listed in
5.2. The story is about Lenore, a young princess vvho becomes
sick. In the network in Figure
the arrows between the nodes repre­
sent the causal connections between the corresponding text units. For
example, Lenore's illness (unit/node 3) causes the King to become con­
cerned about his daughter (unit/node 10), which, in turn, causes the
King to ask if his daughter needs anything (unit/node 13).
1
2
3,
4
5,
6,
T
8,
9,
10.
11,
12,
13,
Once upon a time, there lived a Princess named Lenore
who was ten years old.
One day, Lenore became ill
from eating too many chocolate-raspberry tarts,
The royal physician was called to come
to make Lenore feel better.
He took her temperalure,
felt her pulse,
and made her stick out her tongue.
Knowing that the King was concerned,
the physician brought him in after he was finished with the examination.
When the King saw his daughter lying there,
he asked if there was anything she wanted.
FIGURE 5.2.
task.
[rum
Moon", written text used for narrative com­
12
DEVELOPMENTAL APPROACHES
Read1l1g I )evelopment
have many connections playa more central role in the causal network
structure than do ideas with few connections. Indeed, skilled compre­
henders recall highly connected ideas more often than they recall ideas
with few connections
Graesser & Clark, 1985; Trabasso, Secco, &
van den Broek, 1984;
& Sperry, 1985; van den Broek &
1993). For example, for the story in Figures 5.1 and
skilled readers
are more likely to recall that Lenore became ill (unit/node 3, with six
connections) than to recall that the princess was 10 years old
node 2, with one connection). The qua
of reca II is more indicative
of good comprehension than the quantity, for adults as well as children
(e.g., Casteel, 1993; Goldman & Varnhapt'n 1986; Trabasso et aI., 1984).
DEVELOPMENT OF READING COMPREHENSION For skilled adult rl'aclers, the construction of a col)('r('I11, meaningful
most of the tinw is r<ltlll'r errortil'ss for
Storch &
tiOllS of CI1IIL1ren's memorv tor nMratin's show th,1t chil­
d fl'l1, too, ,1 rl' s('nsi tive to the (\lllSal and rdcren ti,]1 structu re of till' events encoullter. Fven children as young as -t ycars tend to remember
idl'as with many meaningful cOlllWCtiOI1S rnon.' often th'll1 ideas with
few connections. Howl'ver,
tic ,lge d iffprt'IKcs a Iso enwrge. For
instance, older children are more sl'nsitivc to the c,ms,]I structure of a
story than are younger children (,mel adults Me l'\l'n more sensitive).
In addition, younger Child1'l'n tend to iorus on reCtllling causal relations
thnt occur within a particular episode of ,] story, whereas older children
arc better ahle to idt:'ntify CtHlsdl relations tl1,lt span across multiple
sodes and events in a story. Identifying ca usa I reId tions ,Kross
episodes of a story leads to a broader, more complex understanding
of abstract components, such as a story's overall theme.
dren also tend to emphaSize relations
the concrete events in
a storv. whereas older childr~n are able to notice rela tions involving
13
more abstract components, such as underlying intentions and l.wals of
characters (van den Broek, 1997; Williams, 1
Early Contributions to the Development
of Reading Comprehension
The fact that even preschool children engage in extensive coherence
building raises the following questions: (1) How do code-related and
oral comprehension skills develop? (2) How do these two sets of skills
combinp--or not combine-to support the development of reading
comprehension skills? To add ress these questions, we tested two groups
of preschool and kindergMten children (at the ages of 4 and 6 years,
(1nd monitored them as they transitioned from prere(lders
testing them again after 2 years (at the ages of 6
tinw, oral cornprehl'l1sion skills-the <lbility to
extract Ilw,11ling about events ,1I1d f,]ets and identify sem,mtic relations
between thost' ('vt'nts and facts and code-rel'lted litt'racy skills-till'
to tr,1I1s],1te writtell svmhols into nw,1l1ingful words-were'
assl'ssl'd. At t'<1C11 ,1gt', video ,md audio n,ur<ltivps were used to assc'ss
oral comprehension skills, in addition to receptive vocaDul<1rY ,1S mea­
sured bv the l't',]bodv Picture Vocabui<1ry Test-Ill (Dunn & Dunn,
For till' K-Yl'iu-old children, ,1 v.'ritten l1<1rrative W,lS included to ,lssess
reading comprehension. All n,uf<ltivl'S Wl'rl' ,1gC appropriat('
on Fk'sch-Kincaid scorc's) ,md were ,1nalyzl'd in accordann' with the
describc'd in till' previous section to ddl'rmine their causal
structuH'S. Outconw variables for till' cOlln,rehcllsion measures were
th" total ,1l11OLlnts of [('call of idl'<] units that w(,re
een tra I to the
G1US,1] network structure' (i.l'., with many cOnIlL'ctions), fl'presenting a
combill,ltion of quantitatin' ,111<1 l]l.wliti1tivl' aspects of the child's rep­
resc'ntatioll of each narr,ltive. Code-related skills werl' examined
m("lSlll"l'S of phonemic <lW<1reness ,me! lettt'r ,111J word idcntification
(Storch & Whitehurst
Oral comprehension and code-related skills emerged as unique
and intcrreh1Lcd factors that contribute to overall reading comprehen­
sion. With regard to the development of these skills, oral comprehen­
sion skills at one age uniquely
ural comprehension skills 2
years IcHer (i.e., from age 4 to age 6 and from age 6 to age 8). Similarly, a
child's code-related skills at one age predicted that child's code-related
skills 2 years later. These two unique sets of skills were also interre­
particularly dt the youngest age. For preschool children (age 4t
r'
114
DEVELOPMENTAL APPROACHES
oral comprehension skills strongly predicted code-rdated skills. As chil­
dren developed, the predictive relation became weaker, suggesting that
these two types of skills gradually separate as a child develops.
With regard to later reading skills, oral comprehension and code­
related skills in early elementary school each independently predicted
comprehension (i.e., at age 8). Thus, oral comprehension and
code-reb ted skills jointly contribute to reading comprehension skills in
a child's early elementary school years.
In summary, well before children learn to read, oral comprehen_
sion and code-related skills arc present as separate
interrelated sets
of skills. Each of these sets of skills follows its own developmental tra­
jectory, with tht' skills at one age predicting the same skills 2 years later.
When the child eventll<lllv reads texts for comprt'hension, in the early
each of these two sets of skills independently
contribute's to reading SUCCl'SS.
tht' e'mnh,lsis in preschool education has been on
,lS a means to improve later reading
comprl'hensiol1. OUf findings support that pl'actin', but they
that oral compreht'l1sion, in addition to code-rl'lated litt>racy skills,
also should be emphasized during ('<Hly instruction if
lwnsion skills <lfe to dl'vclop by tIll' tillle i1 child is in
school. This is particularly il1lport<lnt bvcausl' dftt'r these
years, texts increasingly bt'conw thl' ml'ans for conveying content and,
IwnCt', ,1 lack of skills th,lt Me prel'l'quisitc to comprt'iwnsion likely
results in (l CilSCildl' of i)C,ldl'm ic
sllch ilS infl'H'IKt' generation and comprehension monitoring, devel­ oped in IIw eMly grades beconw more and morl' important for aca­
demic SLiccess in tIll' hig/wr grades. Ypt as naliol1dl rt'jJOl'ts continue to brgt' numlwrs of students in 1,1ter ell'llwntary school and in school compwhend what they read in school or at
skills h,w(' been mastered. Til readers in tlw latl'r grades, it is important to
kinds oi difficultip<; thdt tl1l'Y ('xperience. In doing so, it is
consider the processes
fl'ad ing, beca use tha t is w hen the major­
ityof
Later Contributions to the Development
of Reading Comprehension
We investigated how struggling and proficient readers differ as
integrate and connect infonnation during reading by observing data
Reading
115
from eye movements and think-aloud protocols. These methodologies
were chosen because they provided both quantitative and qualitative
aspects of online comprehension processes, parallel to our analysis of
comprehension processes in the resulting representation (offline) from
our study of young readers.
average, and proficient readers from fourth-, seventh-,
ninth-grade classes participated in this study. Levels of reading
comprehension ability were assessed by means of the Maze Test of Cur­
riculum Based Measurement (CBM-Maze; Deno, 1985; Espin & Foegen,
Students in each grade werl' divided into equal subgroups based
on the distribution of Maze scores; those in the lower 20'X, of the distri­
bution were identified as struggling readers and those between 40 and
6D'},,, and over 8m;) of the distribution as average and proficient rcad­
erSt fespectivt'ly. The validity of the identified subgroups was tested
comparing their scores on the standardized Cates-M,lcCinitie Read
Comprehension Tl'st (Mi)CCinitie, MacCinitie, Maria, & Dreyer, 2000).
Comprehension scores of the subgroups ilS assessed
the Cates­
MacCinitie test diffHed significcllltly, and the diffl'rences were consis­
tent with the CBM-Maze test results.
1\11 students read a variety of (age-appropri<lte) texts in both eye­
,. ,llld think-aloud tasks. The eye-tracking nwthodology reveals
the number, duration, and type of eye movements that occur when
reading a tl'xt, thus capturing quantitative and qualitative aspects of
online tomprel1l'nsion proC'l'sses. Hence', it allows (lilt' to ddermim'
how 111,lny fix,ltions and how much tinw a readt'r dt'votes to a partiClI­
lar ilrl'il in the tl'xt as well dS whether the rl'ader returned to that area of
the text ,liter hi1\ing continued rt>ading
looking back at prior text).
The think-aloud methodology reveals the inft'rential and nOllinferential
students as thev proceed through a text, thus
capturing
sinn prOCl'sses in a manner complementMY to that by eye
of processes reve'lled during think-aloud procedures range from
readers repeating text in their own words (i.e., paraphrasing) to
Specific predictions about an event that will occur later in the text
inference).
Tlw eye-tracking results were remarkably consistent across the
three grades. At each grade level, struggling readers proceeded
the texts \vith the same number of fixations as average and proficient
readers, but fixation times were longer for struggling readers than for
more proficient peers. In addition, struggling, average, and pro­
ficient readers looked back to the previous text with equal
~
116
DEVELOPMENTAL APPROACHES
but struggling readers were less systematic than more proficient read_
ers with
to ,uhat information they reread. For example, strug_
gling readers returned to reread parts of the text as often as did the
other readers, but instead of reading specific, informative segments, as
readers did, the struggling readers reread entire sections of
the text-and often uninformative sections at
returning to
their starting ._,--,
The cognitive processes identified in the think-aloud Protocols
revealed two distinct subgroups of struggling readers, each display_
a unique p<lttern of comprehension processes. Each subgroup pro­
duced less-than-optim,l! mental representations of the texts they
as evidenced by poor memory for the texts. With regMci to the
processes during reading, one subgroup of
inclined than the other to engage in the
enet's, inferences t h,l t d r,l\V on
they did so as
eVl'f, unlike the morl'
readers frl'qllcnt!y produced
the context of till' text. The second subgroup of stmggling n'clliers was
morl' inclined to del1end on the text. Unlikl' the other groUD of strug­
in fOfm,) tion
fL'w
of strug­
activitil's nlostlv !o tlK' textual
or repeilting the ll'xt and
n'.ldefs did not exhibit cogni­
tivl' proCl'SSl~S th'lt were elramatic<llly diffel'l'l1t from ,1H'ragl' (lnd
re,)dprs. In
struggling H.-<)ders pnKl'l'cll'd through the tl'xl in simi­
lar fashion ,lS more
in ferL'l1 tia I p ron'ssl'S t h,lt
mation into <l cohel'l'nt
ineffectual and time-cOnSlll1ling m,lllller in which struggling
readers madl' their W<ly through till' text th'lt
the more l1roficient readers
Long, Oppy, &
and profiCient readers differed in the strategies used dur­
with struggling readefs overemphasizing eith('r generat­
elaborative inferences or paraphr,)Sing the tl'Xt. Thus, struggling
readers appear to comprisl' at least two distinct subgroups, with one
subgroup going bel/om! the text to try and establish a coherent represen­
tation (but doing so ineffectively) and another staying Il'itliill the text
and establish a coherent representation, thereby limiting their
to recognize important relations. These distinct subgroups of
117
struggling readers can perhaps be described as explorers and plodders,
respectively, in analogy to subgroups of dyslexic readers in the con­
text of word processing (Olson, Kliegl, LJavidson, & Foltz, 1985). Other
subgroups may exist as well. Unfortunately, neither of these patterns
of processes help struggling readers achieve success in their effort to
comprehend what they are reading.
DISCUSSION
The described rPsp,lrch results contribute to Ollr ullderstcH1ding of the
devdopml'nt of reading comprehension ,llld to eiforts to prevent, d
nose, ,md femedv rl'cKling cOnl[)rl'ilensioll oroblems across the school
years.
Theoretical Implications
<mel l]u<llily of cognitive proCl'SSl'S that occur dur­
till' propertil's of the tt'xtU<l!
,Kross different ,1gl' groups and dif­
our uncll'fstanding of reading comprehen­
sion. TIll' results 01 the iirst study n'portl'd here indicate th,)t,
in till' prl'school \'(',HS, two distinct cluStlTS of skills rl'll'v,1llt to "lter
re,lel ing
written cOLil' (ktlL'r, vvmels) into concepts <mel 1l1l',1I1ingful lcmgwlge
and those' imol\-ing the idl'lltific,)tion of llll'aningful felatiolls and the
construction of COIll'rL'llCl' ,llllong thl' fdCtS, l'vents, and other
COl1\'l'vl'd by the I,mgu()ge. These clusters show n'm,lrkable
children lll<lturc ,md, illlport,mtly, m,lke strong, indepl'ndl'llt contribu­
tions to rL,,)ding colllprl'Ill'llsion by the time children begin to read. It
th<l t both sets of "ki lis ,lre llecessa ry for rem'! illg com prl'hen­
sion.
for many childn'll the necl'ss,)ry skills have
Me re,leiy 10 read texts for
many others struggle', even into higher elementary cllld
The results of the second study indicate that
ers tend to generate inaccurate and incomplete representations
reading. When we considered their online processes, it appeared that
the struggling readers did not differ from more proficient readers in the
types of cognitive processes in which they engaged per se but that they
differ with respect to the effectiveness and efficiency of those pro­
r
118
DEVELOPMENTAL APPROACHES
cesses. Moreover, the results suggest that struggling readers comprise
different subgroups. In our study, two such subgroups were identified
consistently for each of the three grades that were investigated
and 9th). One subgroup of struggling readers drew extensively on
background knowledge to infer connections between text units and
between text units and their prior knowledge; unfortunately, the acti­
vated background knowledge frequently was irrelevant or incorrect
in the context of the text. The second subgroup of struggling readers
tended to limit their processing to the text itself, frequently rereading
or otherwise reprocessing text units, without bringing in relevant back­
ground knowledge.
I t is imoortant to nole tha t these subgroups were indistinguish­
to their performance once reading was
For example, both subgroups differed from more proficient [(:'aders in
terms of memory for the texts to the same degree. it is only \A,lhen online
processes during reading Wt'n' c(lllsiderl'd thilt the subgroups emt'rged.
In this study, two subgroups were idt'ntified. Future rt'st'dfch Illay iden­
additional subgroups or bl' ablp to make' filwr gr,lim'd distinctions
within the subgroups th,lt we idl'tltifil'd.
Practical hllpHcations
th,lt t,vo distinct clusters of skills-associated with trans­
rl'spl'cti\'clv-come
children's n',ld
sion and, moreover, that thest' clusters of skills <llrcady ,Hl'
the preschool years has imDliciltions for
,11 though tIlt'
dret1 with the as phonemic awareness and word dl'codine:, is e:clll'raliv re'cognized, comprehen­
sion skills in instruction as well. These skills C,1n be introduced
practiced in l1onrl'ading contexts sllch as telt'\'ision vit'wing or listen­
Viln den Rroek, I ,orch, & Thurlow, 1996) and should be aimed at
children's <1bilitv to crc,lte coherence. This includes aware­
ness of coherence as well as specific strategies for constructing coher­
ence, Causal coherencp and referential coherence are good starting
points because they tend to be amply present in the narrative materials
to which preschool children typically are exposed and because these
of coherence also are central to other types of information, includ­
the expository texts used in later school years. It is imnnrt:mt to
TIll' find
Reading Development
9
note mat, although these skills can be introduced as strategies, the goal
should be that they eventually become internalized and automatized
so that they can be applied without consuming excessive attentional
resources (Laberge & Samuels, 1974).
Second, early identification of potential problems in comprehen­
sion ability may open the door for interventions that ensure that chil­
dren hilve the comprehension skills necessary for reading comprehen­
sion once they enter elementary school. To identify potential problems
and to assess the effectiveness of interventions, proper comprehension
assessment tools need to be developed.
This leads to a third implication of our results: The goal of compre­
hension assessment is to determine whether a reader is able to gener­
all' a coherent and accurate representation of the presented informa­
tion. Such assessment should include both the qll<lIltity and quality of a
student's representation, focusing on the constructed (causal) network
(KendeOLl, Vim den Broek, White, & Lynch, 2007; van den Brm'k et aI.,
Most tmdition,ll assessnwnts me limited to mCilsuring quantity
(e.g., the amount of information a child recalls). At least as imporhmt
is the quality of the representation. ln the current studies, we present
one eXil111pk' of assessment of quality: g<1uging the extent to which a
child's reGll1 is sensitive to the centr<llity of information
the l'xtent
to which he or she rt'members highly connected information). Other
wilys could t'<lsily be developed: for example, asking coherence build­
,lin1l'd at different levels of inference making and using
techniques to scale the questions-and the skills
gtlllge'-lln qU<llittltivcly different dimensions.
TIlt' results rq';Mding struggling readers highlight a fourth implica­
tion for educational practice: Dit1gnosis of reading problems and inter­
Vl'ntion in struggling readers bent'fit from considemtion of the actu,11
procl'sst's in 1I,;hicb tlwse readers engage as they proceed through a text.
Struggling readers may appear similar in terms of the product of their
diffl'r when their cognitive processes are consid­
ered. Moreover, it is the online processes that provide the foundation for
successful-and failed-coherence' building, so interventions and use­
ful di<lgnosis are mosllikely to be effecIi Vt' when applied to the reader's
activities during reading and when they are based on an understandof which processes are being executed properly and which are not
Indeed, many strategy interventions in the literature address potential
causes of reading difficulties by methods that implicitly are assumed
to improve the effectiveness of processing, although such processing is
not dirpctlv assessed (e.g., Dole, Duffy, Roehler, & Pearson, 1991; Fuchs,
,.
121
Development
20
DEVELOPMENTAL APPROACHES
;'''/-'.ir(1
Fuchs, Mathes, & Simmons, 1997; PaHncsar & !3rown, 1984; Pressley,
that explicit
2000; Yuill & Oakhill, 1988). Our results
on processing, both in attempts to understand the origin of (reading)
comprehension problems and the development of interventions, is
essentiaL This, in turn, has a further implication for assessment, besides
the quantity-quality distinction, namely that assessment tools should
include process measures as well.
In conclusion, successful reading comprehension depends on
development of different skills, ranging from skills particular to trans­
lating the textual code into meaningful concepts to skills particular
to combining units 01" information into a coherent whole. All of these
skills begin their developmentill trajectory well before the school years.
Indeed, the processes imd skills in preschool children described here
themselves likely have precursors, at evpn earlier ;lgl'S
van den
1997). Although in education comprehension
it is as much ,) pme!':;:;. Educa­
in diagnOSis and intervention
tiona I
process('s th,)t h',ld re,lders to
by explicitly
Sllccess or failure.
edilion
Espin, C. A., &
A. (1996). Validity of general outcome measures for
.. ,',_._ secondary students' performance on content-area tasks. Excep­
62, 497~514.
D., Fuchs, L. S., i\1athes, P. C .. & Simmons, D. C.
Making classrooms more r<>c<'nn~i
,')4, 174-206.
Peer-assisted
to diversity. Amcr-
Hills­
dale, NJ: Erlbaum.
Goldm,m, S. R., & V<1rma, S. (1995). CAPping the
modl,1 of discourse comprehension. In C. A. VVeaver, S. Mannes, & C. R.
Fletcher (I'-tis.), Disc,lIlrs£' COlllprcl/l'lI~ioll: £~sl1ys ill 11011(11' of Wolter Kintscll
(pp.
M,lhwah, NJ: Erlb,)UI11.
GoldmJIl, S. R., & Varnilagcll, C. K. (19f\b). Ml·mory [pr embedded and seq lIen­
storv structures.
of i....l clIlOfll ,"',! 1111/>.'11(/</(',25, 4ll141f\.
1'. II., & Tuntl1er, W. E. (1 91l6). I
Graessl'r, A.
c., & C\,lrk,
rdllmlioJl, 7,6-1 D.
L.
r
Slrticlul'CS
J
.........
A.. v,.c· "I' jllln/il'it kllOiul-
Norwood. NI: i\bll'x.
Graessl'r, A. C, Sin~l'r, M., & Irdbasso, T. (19<)4). C(lilstruclin~ infercllces dur­
ing l1aIT,ltiV(' COfllprl'lwllsion.
/<'l'l'icw./O{, :'>71-:;9::;.
Kendl'oll, 1'., \,,1/1 ciVil BroL'K, P.. WhitL', M ..L & l.ynch, ]. (2ll07). Comprehen­
sion ill preschool and ('<HI)' l'il'nll'lltMy children: Skill dl'VL'\opml'llt and
stl'<ltl'g\ intl'l'\l'lltiolls. III D. S. McN<)mar,l
'//I('oril':;, illtcn'l'IIliolls, olld
ACKNOWLEDGMENTS
This n'sl'<lrch w,l.~ supportL·d
J)l.u:tn, L. M., & Dunn, L. rv1. (1997). Peabody Picture vocabulary
<PPVT-W). Circle Pines, MN: American Guidance Services.
gr'lIlh from Ihl' Cenll'r for
,lnd tilt' DlT,)[lllwnt of EduC,ltioll,lll's\'cilol(lgv (both
from the Cl'nrn for Ihe IrllprO\·,'nwnt of E.lr!l· Rl\ld
/\chil'\'t'fl1l'!lI, ']!lel
a
Rl'"Liing SCdll'-up grdllt (CID/\ 1l.f.3(),)) from till' U.s. Institute nf EducMioll Sci­
!',lUi \'(111 dl'n Brock rl\'('in'(\ supporl
l\l'sl',)rci1, Unil'L'r;.;itv of Mll1l1L'sot,1. l'<llhlViot,1 Kl'ndl'ou received
through" McCi11 Univl'rsit\ Rl'S(',Hciwr lund.
A CO/lstruc­
Erlhalltn.
discourse
Kintsch. W. (1 t)HK). Tlw lISl' pf
1{1'1,i('1l', 95, \6:1- H12.
model.
)971'1). Tow,)fd ,) model of text
Kintsch, \N., & \',)11 \)ijk, T. A.
I~('l'i{,ii), Wi, :'>63-3'1'L
inf(lrrnation
(1'174). TOWdrd ,1 theory of
/1,2Y3323.
d i ffNl' nccs i 11 I ht' ti nil'
l.cllrIl-
TIll'
REFERENCES
C<lin, K., Oakhill, L &
I'. E. (200.f). Childn'J1's
C oncurrl'nt pred iclioll
componcnt skills.lolll"llili of FdllCiltimlil!
Clsted, M. A. (1993). Effects of inference
,jren's inferential
\'l'rb,l! tlbilily, and
on chil-
.
Deno, S. l.. (191'15). CurriClllum-based l11l'dSlIrellwn t: The
C/lildrl'll, 52, 219-2::12.
C. C., Roehler, L. R., & Pearson, P. D. (]991). Mm'ing from the
old to the new: Research on readinl! cl)mprehension instruction. Ikuic1i' of
EduCIlliolloj /(.,·s(,I1I'(II, 61, 2,\9-26-+.
compre­
Tlworic5,
lwnsion.
illfl,},l'elltitlll~.
47-71). New York: Erlbaum.
IIlId
!"'""
22
DEVELOPMENTAL APPROACHES
Reading
Oakhill, j., Cain, K, & Bryant, P E. (2003). The dissociation of word reading and
text comprehension: Evidence from component skills. Lal1f;uagc and Cogni­
tive Processes, 18,443--468.
E. J., & Myers, J. L (1987). The role of caus,11 connections in the
of text. Memory t"'.r Cogniliol1, 15,419--427.
Olson, R K, Kliegl, R, Davidson, B. L & Foltz, G. (19Wi). Individual and deveI­
nnm,'nbl differences in
disability. In G. E. MacKinnon & 1. G.
Rmdil1S rc~earclt: Adl'(lllc('s ill IlIcory and pmclice (Vol. 4, pp,
1-(4). New York: Academic Press.
Palincsar, A. S., & Brown, A. L (19H4). Reciprocal teaching of comprehension:
activities.
Ill1d In~ll'IIctioll, 1,117-175.
narrative comprehension in young
38,36-76.
str,ltegies instruction. In J. Osborn & F.
all: IS';lle,.; ill icadlillg tll/d
(pp. 1
New York: Cuilford Press.
M. (2000). Whilt should comprehension instruction bl'
of! In M. I.. Kamil, 1'. B. MOSl'nthal, 1'. D. l'ear'iOll, & R. B,lrr
n',warcJl (Vol. 3, pp. 545-586). 1'v1.1hw<1h, NJ: Frlbaum.
C. J. (2002). Ordll,lIlgu,lge ,111<1 code-rel,ltl'd precur­
sors to rmding: Rvidt'no' from d longitutiin,11 structllr,ll I11ndel. Deuclop­
!J:';l/clwlogy, 38, Sn4-947.
Trab,lsso, T., Secco, T., & V<lll den nroek, 1'. (1l/84). CaLls<11 cohesion and story
coherl'TKl'. In II. M,1l1dL N. L Stein, & T. Tr<1b,l;iso (Etls.), Lel/millx lind
(Pt" 8,1-111). M,lll\v,lh, NJ: [rlb,lL1f11.
L L. (1985). Cw,;al rel,1tclil1L'ss ,md import-mel' of sh)ry
],J., 595-611.
Tr,lbnsso, T, & V,1tl dl'll Bfllek, I'. (1985). elL1S,ll
and tlw I'l'presentation
of l1arr,lti\'l'l'Vl'nts.
2-1,612-(,30.
Vim dl'n !3n)l'k, I'. (1l)l)O).
infl'rl'ncl' n1,1kcr: 1l)\V,lrds a process model
of i nfC'I'l'lll'l'
In ()_ 1\. n,llota, C. B. Flores
d' Arcais, & K.
423­
445). Hillsd'llc, NJ: Erlb,lUll1.
van den Bnwk, I'. (1994)
,llld nW!110I'V of n,lfr,1ti Vl' texts. In M.
A. Cl'rnsbilchcr (Ed.), Hlllldh,Jok
539-588). Lnndon:
AC,1dl'lllic ('ress.
v(ln den Bnwk, I'.
Disc(lVl'ring the Cl'nwnt of the lInhersp: The develop­
nwnt of ('vent
from childhood to ,ldulthood. In I' \'<1n den
Brock, P
SpillL'
in
van den
aL
van den Broek, P, &
of causal
23
recogni­
relations in memory for narrative texts: Evidence from
tion. Discollrsl' Processes, 16,75-98.
van den Broek, E, Lorch, E. P, & Thurlow, K (1996). Children's and adults'
memory for television stories: The role of causal factors, story-grammar
categories, and hierarchicallevcL Child Developlllcl1t, 67, 301 ()-3028.
vanden Brock, P, Risden, K., Fletcher, CR., & Thurlow, R (1996). A "landscape"
vicw of reading: Fluctuating patterns of activation and the construction of
a stable memory representation. In B. K. Britton & A. C Graesser
Mode!:;
lexl (pp. 165--187). Mahwah, NJ: Erlbaum.
van den Broek, 1'., Risden, K., & HlIsebye-Hartmann, E.
The role of
readers' standards for Cl)herence in the generation of inferences during
R. F. Lorch & E. J. O'Brien
SOl/rees of CO!ICrCI1CC ill rL'm/iIlS
(pp. 353-373). Hillsdale, NJ: Erlballl1l.
van de Ve1de, R. G. (1989). Man, verbal text, inferencing, and coherencc_ In
W. Hevdrich, F Nl'ub<1ul'r, J. S. ('etofi, & E. Sozer (Eds.), COllllcxill/lIl1Ll
cO/ICn'l;o': /\l/tlll/sis of/c.-a alld di;;COllr~l' (pp. 174-217). New York: Walter ell'
Wagncr, R. K., Piast,l, S. B., &
J. K.
Tr,LXler & M. A. Gerllsbill'her
pp.llll 11·12). New Yllrk: Flscvil'L
Whitehurst, C. L & Llll1ig,1I1, C. l (1998). Child
i111d en1l'rgl'n t
Child Dn'clol'lIIt'1/1, 69, 841\-872.
of studpilts with ilnd without
dis,lbiliti<'s: Id('ntifi(".ltillll of Ildrrative themes ,1l1d
tic text rl'pre­
sell t,l tions. /011 /'/I1l1 [)f SIJ('citll Edl/[IIlio/l, 25, 135- 154.
Effl'cts of infL'rl'ncl' aW<1feness
on poor
2,,1345.
comprelll'ns!(>n.
ZW,l,1Il, K A., & 1{,lpP, D. N.
Discourse comprehension. In M. J, 'lhLX­
In & M. A. Ccrnsb,lclwr
lalldbook 0(
(2nd ed., pp_
New York: EIsC'\'il'r.