Adaptivity in reading literary texts;
Adaptivity in reading literary texts;
Adaptivity in reading literary texts;
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -1-<br />
FLEXIBILITY IN READING LITERARY TEXTS;<br />
DIFFERENCES BETWEEN WEAK AND STRONG ADOLESCENT READERS<br />
Tanja Janssen 1<br />
, Mart<strong>in</strong>e Braaksma 1<br />
, Gert Rijlaarsdam, 1<br />
and Huub van den Bergh 2<br />
1 Graduate School of Teach<strong>in</strong>g and Learn<strong>in</strong>g, University of Amsterdam, the Netherlands<br />
2 Institute of L<strong>in</strong>guistics, Utrecht University, the Netherlands<br />
Paper for the EARLI conference, Nicosia, Cyprus, August 23-26, 2005<br />
Contact: T.M.Janssen@uva.nl<br />
Keywords: flexibility, adaptivity, <strong>literary</strong> understand<strong>in</strong>g, reader response, read<strong>in</strong>g activities<br />
Abstract<br />
In this paper we exam<strong>in</strong>e whether weak and strong adolescent readers can be identified by differences <strong>in</strong><br />
flexibility of their read<strong>in</strong>g processes. Flexible readers, <strong>in</strong> our view, are readers who are able to adaptively change<br />
the configuration of text process<strong>in</strong>g activities (e.g., <strong>in</strong>ferenc<strong>in</strong>g, analyz<strong>in</strong>g, evaluat<strong>in</strong>g), both with<strong>in</strong> a s<strong>in</strong>gle text<br />
and between different <strong>texts</strong>. We hypothesized that successful literature readers would be more flexible <strong>in</strong> their<br />
process<strong>in</strong>g of short <strong>literary</strong> stories than less-successful readers.<br />
19 Dutch tenth-grade students from eight classes participated <strong>in</strong> a ‘known group validity’ study. Ten<br />
participants were known to be strong readers and n<strong>in</strong>e participants were known to be weak readers of literature.<br />
Each student read five different short stories under th<strong>in</strong>k aloud conditions. 92 Th<strong>in</strong>k aloud sessions were<br />
transcribed, segmented and coded. Multilevel analyses were applied to exam<strong>in</strong>e the distribution of read<strong>in</strong>g<br />
activities over the read<strong>in</strong>g process, and to compare weak and strong readers’ patterns of response.<br />
Results confirmed the flexibility hypothesis: strong readers more often changed their read<strong>in</strong>g activities<br />
over the course of the read<strong>in</strong>g process. For <strong>in</strong>stance, strong readers’ emotional responses fluctuated significantly<br />
dur<strong>in</strong>g read<strong>in</strong>g, whereas weak readers’ emotional responses rema<strong>in</strong>ed unchanged. Moreover, strong readers<br />
adapted their process<strong>in</strong>g activities to different stories, whereas weak readers tended to use monotonous patterns<br />
of response.<br />
Introduction<br />
In literature <strong>in</strong>struction, response-based approaches receive much support from teachers and<br />
researchers. There is a broad consensus that students should be encouraged to br<strong>in</strong>g forward<br />
their own personal responses to a <strong>literary</strong> text, <strong>in</strong>stead of search<strong>in</strong>g for predeterm<strong>in</strong>ed<br />
mean<strong>in</strong>gs or memoriz<strong>in</strong>g facts about the author, the <strong>literary</strong> genre or the historical period<br />
(Applebee & Purves, 1992; Janssen, 1998; Poyas, 2004; Van Schooten, 2005). However,<br />
some questions rema<strong>in</strong>: How should students develop their own personal responses to<br />
literature? And how can we encourage students to use more sophisticated strategies for<br />
<strong>literary</strong> understand<strong>in</strong>g?<br />
To beg<strong>in</strong> to answer such questions, it is helpful to know more about the <strong>literary</strong><br />
read<strong>in</strong>g process itself and about differences between <strong>in</strong>dividual students <strong>in</strong> <strong>literary</strong> process<strong>in</strong>g.<br />
Accord<strong>in</strong>g to Schunn and Reder (1998), there are three general approaches to the study of<br />
<strong>in</strong>dividual differences <strong>in</strong> cognitive skill. One approach assumes that <strong>in</strong>dividuals vary <strong>in</strong> the<br />
strategies or processes they use. For example, expert readers may use other processes than<br />
novice readers. A second approach assumes that people use the same processes but differ <strong>in</strong><br />
one or more performance parameters that affect how processes are executed (e.g., speed). For<br />
<strong>in</strong>stance, skilled readers are faster <strong>in</strong> decod<strong>in</strong>g than less-skilled readers. The third approach<br />
assumes that people differ <strong>in</strong> adaptivity. That is, although <strong>in</strong>dividuals may use the same
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -2-<br />
processes, they may differ <strong>in</strong> how well they adapt their processes to a given situation. In the<br />
present study, we adopted the last approach.<br />
<strong>Adaptivity</strong> or ‘flexibility’ (a more common term <strong>in</strong> <strong>literary</strong> studies) is theoretically<br />
considered to be important <strong>in</strong> the doma<strong>in</strong> of literature read<strong>in</strong>g. Rosenblatt (1999), for <strong>in</strong>stance,<br />
argued that ‘flexibility of m<strong>in</strong>d’ is part of the essence of literature read<strong>in</strong>g, and a fundamental<br />
goal of literature teach<strong>in</strong>g: “A poem or a novel should provide fresh <strong>in</strong>sight. Readers,<br />
therefore, must be helped to develop flexibility of m<strong>in</strong>d, a freedom from rigid emotional<br />
habits, if they are to enter <strong>in</strong>to the aesthetic experiences the artist has made possible.” (ibid.,<br />
1999: 98).<br />
Groeben (2001) related creativity and ‘flexibility’ to the <strong>literary</strong> convention of<br />
polyvalence; that is, the possibility or necessity of (actively) assign<strong>in</strong>g different mean<strong>in</strong>gs to a<br />
<strong>literary</strong> text. Flexible readers, then, would be readers who are able to generate multiple<br />
mean<strong>in</strong>gs <strong>in</strong> response to a s<strong>in</strong>gle <strong>literary</strong> text. This ability is strongly associated with <strong>literary</strong><br />
competence; <strong>in</strong> th<strong>in</strong>k aloud studies, expert readers of literature were found to be more<br />
‘flexible’ <strong>in</strong> their process<strong>in</strong>g of <strong>literary</strong> <strong>texts</strong> than novice readers, <strong>in</strong> that they searched for<br />
alternative <strong>in</strong>terpretations and were able to perceive a <strong>literary</strong> text from multiple perspectives<br />
(Andr<strong>in</strong>ga, 1995a; Earthman, 1992). Other empirical research on ‘flexibility’ <strong>in</strong> <strong>literary</strong><br />
read<strong>in</strong>g is scarce.<br />
In the present study, we mean by ‘flexibility’: the ability to spontaneously change ones<br />
read<strong>in</strong>g activities <strong>in</strong> adaptive response to different parts of a <strong>literary</strong> text and/or to different<br />
<strong>literary</strong> <strong>texts</strong>, with the purpose of mak<strong>in</strong>g mean<strong>in</strong>g. Thus, we assume that readers, <strong>in</strong> their<br />
attempt to understand a <strong>literary</strong> text, may adapt or shift their read<strong>in</strong>g activities at two levels;<br />
with<strong>in</strong> (the read<strong>in</strong>g of) a s<strong>in</strong>gle text and between different <strong>texts</strong>. For example, when read<strong>in</strong>g<br />
and <strong>in</strong>terpret<strong>in</strong>g a story, a reader may use a pattern of ‘delayed evaluation’; deferr<strong>in</strong>g<br />
evaluative responses until the end of the read<strong>in</strong>g process (Vipond & Hunt, 1984). The same<br />
reader may use a different pattern when read<strong>in</strong>g another story, for <strong>in</strong>stance, evaluat<strong>in</strong>g the<br />
story immediately at the beg<strong>in</strong>n<strong>in</strong>g of the read<strong>in</strong>g process, <strong>in</strong>stead of at the end. In other<br />
words, readers may vary the ‘orchestration’ of their mental activities over the course of the<br />
read<strong>in</strong>g process, and they may orchestrate their activities differently <strong>in</strong> response to different<br />
stories.<br />
Previous th<strong>in</strong>k aloud studies of <strong>literary</strong> read<strong>in</strong>g (e.g., Andr<strong>in</strong>ga, 1995a; Earthman,<br />
1992) have provided <strong>in</strong>dications that some readers are more capable of flexibly chang<strong>in</strong>g their<br />
patterns of response dur<strong>in</strong>g read<strong>in</strong>g than other readers of literature. In all likelihood, some<br />
readers are also more capable of chang<strong>in</strong>g their patterns of response when read<strong>in</strong>g different<br />
<strong>texts</strong>. These differences are probably related to students’ capability <strong>in</strong> the doma<strong>in</strong> of literature.<br />
Successful readers of literature may be more sensitive to differences between <strong>literary</strong> <strong>texts</strong><br />
and text parts (and adjust their process<strong>in</strong>g accord<strong>in</strong>gly) than less-successful or ‘weak’ readers.<br />
In this study, we exam<strong>in</strong>e whether adolescent readers differ <strong>in</strong> flexibility when read<strong>in</strong>g<br />
short stories, and whether strong readers are more flexible <strong>in</strong> their process<strong>in</strong>g than weak<br />
readers. Before describ<strong>in</strong>g our study, we first present an overview of empirical research,<br />
relevant to <strong>in</strong>dividual differences <strong>in</strong> <strong>literary</strong> process<strong>in</strong>g.<br />
Background<br />
In read<strong>in</strong>g research, attempts have been made to uncover the fundamental processes that are<br />
<strong>in</strong>volved <strong>in</strong> the understand<strong>in</strong>g of text (e.g., Britton & Graesser, 1996; Pressley & Afflerbach,<br />
1995; Van Dijk & K<strong>in</strong>tsch, 1983). One shared assumption is that text comprehension <strong>in</strong>volves<br />
a constructive process, <strong>in</strong> which readers are actively attempt<strong>in</strong>g to make sense of a text. A<br />
second assumption is that readers are attempt<strong>in</strong>g to build mental representations of the states,<br />
events, actions, and characters that are dealt with <strong>in</strong> the text (Van Dijk & K<strong>in</strong>tsch, 1983). It is
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -3-<br />
assumed that these representations evolve dur<strong>in</strong>g read<strong>in</strong>g; readers cont<strong>in</strong>uously extend and<br />
update their representations of a text. In this process, they draw not only on their l<strong>in</strong>guistic<br />
knowledge and world knowledge, but also on their personal knowledge and experiences.<br />
Most models of text comprehension focus on cognitive aspects of the read<strong>in</strong>g process.<br />
However, <strong>in</strong> <strong>literary</strong> read<strong>in</strong>g, feel<strong>in</strong>gs, sensations, emotional and aesthetic reactions play an<br />
important role; they are part of the process of understand<strong>in</strong>g, as shown by empirical research<br />
(e.g., Eva-Wood, 2004; Kneepkens & Zwaan, 1994). In addition, <strong>literary</strong> <strong>texts</strong> are often<br />
ambiguous, <strong>in</strong>def<strong>in</strong>ite, and open. Because <strong>literary</strong> <strong>texts</strong> raise questions about mean<strong>in</strong>gs (e.g.,<br />
by open end<strong>in</strong>gs, allusions, metaphorical language), readers are stimulated to deal with gaps,<br />
to search for connections, and to develop their own <strong>in</strong>terpretations (Andr<strong>in</strong>ga, 1995b).<br />
Andr<strong>in</strong>ga (1995b) presents a systematic overview of <strong>literary</strong> read<strong>in</strong>g processes, <strong>in</strong><br />
which components of cognitive models (e.g., Van Dijk & K<strong>in</strong>tsch, 1983) and <strong>in</strong>sights of<br />
empirical <strong>literary</strong> studies are <strong>in</strong>terwoven. She dist<strong>in</strong>guishes several types of processes:<br />
identification (e.g., determ<strong>in</strong><strong>in</strong>g the <strong>literary</strong> genre), selection (e.g., focus<strong>in</strong>g on <strong>in</strong>formation),<br />
(re)construction (e.g., fill<strong>in</strong>g <strong>in</strong> gaps), elaboration (e.g., mak<strong>in</strong>g personal associations),<br />
evaluation (judg<strong>in</strong>g, criticiz<strong>in</strong>g), emotional processes (e.g., experienc<strong>in</strong>g suspense, pity), and<br />
metacognition (e.g., reflect<strong>in</strong>g on one’s own read<strong>in</strong>g activities).<br />
Readers may vary <strong>in</strong> the extent to which they display certa<strong>in</strong> processes when they<br />
attempt to make sense of a <strong>literary</strong> text. A body of research provides evidence of differences<br />
between expert and novice readers of literature (e.g., Andr<strong>in</strong>ga, 1995a; Dorfman, 1996;<br />
Earthman, 1992; Pesk<strong>in</strong>, 1998; Thury & Friedlander, 1995; Zeitz, 1994). Some of these<br />
studies rely on questionnaires or other types of post-read<strong>in</strong>g responses, whereas others use a<br />
th<strong>in</strong>k aloud methodology to reconstruct readers’ onl<strong>in</strong>e processes.<br />
Earthman (1992), for example, conducted a th<strong>in</strong>k aloud study <strong>in</strong>volv<strong>in</strong>g eight college<br />
freshmen and eight graduate students who thought aloud while read<strong>in</strong>g two short stories and<br />
two poems. She found that the freshmen produced less elaborated responses than the graduate<br />
students. Freshmen tended to reta<strong>in</strong> their <strong>in</strong>itial view of the text, while graduate students read<br />
<strong>in</strong> a more open manner, search<strong>in</strong>g for alternative <strong>in</strong>terpretations, and assum<strong>in</strong>g vary<strong>in</strong>g<br />
perspectives.<br />
Andr<strong>in</strong>ga (1995a) compared the th<strong>in</strong>k aloud responses of three groups; adolescent<br />
readers, less experienced adult readers, and experts of literature. The 22 participants each read<br />
a Faulkner story under th<strong>in</strong>k aloud conditions. Andr<strong>in</strong>ga found that the less experienced<br />
readers focused primarily on the story events, and tended to view the story from one<br />
perspective only, whereas the expert readers were able to perceive multiple perspectives. The<br />
adolescent readers displayed a limited repertoire of text process<strong>in</strong>g activities compared to the<br />
adult readers.<br />
Similar f<strong>in</strong>d<strong>in</strong>gs were reported by Pesk<strong>in</strong> (1998), who analyzed the th<strong>in</strong>k aloud<br />
responses of eight expert readers (English PhD candidates) and eight novices (undergraduates<br />
and advanced high school students) dur<strong>in</strong>g their read<strong>in</strong>g of two poems. She found that experts<br />
produced more elaborate responses, and used more productive <strong>in</strong>terpretive strategies dur<strong>in</strong>g<br />
read<strong>in</strong>g than did the novices. Accord<strong>in</strong>g to Pesk<strong>in</strong>, the expert readers possessed a rich<br />
knowledge base, enabl<strong>in</strong>g them to recognize large and mean<strong>in</strong>gful patterns. The novices, who<br />
had only general expectations for poetry (for example, that metaphor is important <strong>in</strong> poems),<br />
had difficulty <strong>in</strong> mak<strong>in</strong>g the poems coherent.<br />
Clearly, as <strong>in</strong> other doma<strong>in</strong>s, there is a considerable gap between experts and novices.<br />
However, expert-novice studies typically do not exam<strong>in</strong>e the variability that exists with<strong>in</strong><br />
each group. Moreover, it is important to note that most novices will not atta<strong>in</strong> the expert level.<br />
In fact, the literature curriculum <strong>in</strong> secondary education is not aimed at students becom<strong>in</strong>g<br />
experts of literature. Rather, the aim is to acqua<strong>in</strong>t students with literature, and to stimulate<br />
them to become engaged and <strong>in</strong>dependent readers. So, <strong>in</strong> addition to expert-novice studies, we
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -4-<br />
need ‘novice-novice’ research; studies that may give us <strong>in</strong>sight <strong>in</strong>to the <strong>literary</strong> process<strong>in</strong>g of<br />
novice adolescent readers, and <strong>in</strong>to the variability with<strong>in</strong> this group.<br />
A number of th<strong>in</strong>k aloud studies focused on adolescent readers’ process<strong>in</strong>g of <strong>literary</strong><br />
<strong>texts</strong>. Smith (1991), for <strong>in</strong>stance, analyzed the th<strong>in</strong>k-aloud protocols of five successful and<br />
five less-successful n<strong>in</strong>th-grade students, each read<strong>in</strong>g two short stories. He found that<br />
successful readers used more processes (on average 9.4 <strong>in</strong>stead of 7.4) and made more<br />
personal responses than did the less-successful readers. However, most noteworthy were the<br />
large differences between <strong>in</strong>dividual readers. Each reader seemed to have his or her own<br />
characteristic style of response; the particular story had little effect on how a student<br />
responded.<br />
Harker (1994) exam<strong>in</strong>ed how 15 tenth-graders responded to two poems, us<strong>in</strong>g a th<strong>in</strong>kaloud<br />
method. He found that most of the students did not go beyond determ<strong>in</strong><strong>in</strong>g the “pla<strong>in</strong><br />
sense” of the poems, translat<strong>in</strong>g poetry <strong>in</strong>to prose. No effect of read<strong>in</strong>g phase was found on<br />
students’ responses; that is, students did not change their <strong>in</strong>itial <strong>in</strong>terpretations <strong>in</strong> the postread<strong>in</strong>g<br />
phase, when they had had time to reflect on the poem.<br />
These studies suggest that even successful readers <strong>in</strong> grades 9-10 lack flexibility <strong>in</strong><br />
read<strong>in</strong>g; they do not adjust their read<strong>in</strong>g activities to the particular <strong>literary</strong> text, nor do they<br />
develop their <strong>in</strong>terpretive responses after read<strong>in</strong>g. However, the limited number of<br />
participants and the limited number of <strong>literary</strong> <strong>texts</strong> that were read by each participant (just<br />
two stories or poems), do not permit generalizations. More and more different <strong>literary</strong> <strong>texts</strong><br />
should be used to be able to determ<strong>in</strong>e the effects of text; and more participants are needed to<br />
be able to exam<strong>in</strong>e the with<strong>in</strong>-subject variances.<br />
Previous studies have, <strong>in</strong> fact, focused on a ‘strategy approach’ to <strong>in</strong>dividual<br />
differences, the first approach described by Schunn and Reder (1998). The ma<strong>in</strong> assumption is<br />
that readers differ <strong>in</strong> the processes they use; successful readers are more personal, productive,<br />
elaborate and/or <strong>in</strong>terpretive <strong>in</strong> their process<strong>in</strong>g than less-successful readers. Changes <strong>in</strong> the<br />
configuration of processes over the course of read<strong>in</strong>g and over different <strong>texts</strong>, have not been<br />
taken <strong>in</strong>to account. One <strong>in</strong>herent feature of processes has been neglected: the factor of ‘time’.<br />
A process consists of two elements: (1) constitut<strong>in</strong>g cognitive and affective activities, and (2)<br />
a certa<strong>in</strong> development of time.<br />
Several studies of writ<strong>in</strong>g processes have exam<strong>in</strong>ed the factor ‘time’ (Braaksma,<br />
Rijlaarsdam, Van den Bergh & Van Hout- Wolters, 2004; Breetvelt, Van den Bergh &<br />
Rijlaarsdam, 1994). These studies have shown that able and less able student-writers may rely<br />
on the same cognitive writ<strong>in</strong>g activities, but differ from each other <strong>in</strong> the moment or phase of<br />
the writ<strong>in</strong>g process <strong>in</strong> which they perform these activities. It seems worthwhile to use such an<br />
‘adaptivity approach’ <strong>in</strong> our study of adolescents’ <strong>literary</strong> read<strong>in</strong>g processes. It would enable<br />
us to exam<strong>in</strong>e how student-readers distribute their read<strong>in</strong>g processes over moments <strong>in</strong> the<br />
read<strong>in</strong>g process, when read<strong>in</strong>g different <strong>literary</strong> <strong>texts</strong>.<br />
Present Study<br />
The aim of the present study is to get more <strong>in</strong>sight <strong>in</strong>to the (presumed lack of) flexibility of<br />
weak and strong adolescent readers of literature. The variability of read<strong>in</strong>g activities over the<br />
course of the read<strong>in</strong>g process (both with<strong>in</strong> and between stories) is taken as an <strong>in</strong>dication of<br />
‘flexibility’ <strong>in</strong> read<strong>in</strong>g. Three questions will be addressed <strong>in</strong> this study;<br />
(1) Do adolescent readers change their configuration of read<strong>in</strong>g activities over the course<br />
of their read<strong>in</strong>g process, when read<strong>in</strong>g and respond<strong>in</strong>g to short <strong>literary</strong> stories? And if so;<br />
(2a) Do strong readers more often change their read<strong>in</strong>g activities <strong>in</strong> response to a <strong>literary</strong><br />
story than their weaker peers?
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -5-<br />
(2b) Do strong readers more often change their read<strong>in</strong>g activities <strong>in</strong> response to different<br />
<strong>literary</strong> stories than their weaker peers?<br />
Method<br />
Participants<br />
We <strong>in</strong>vited tenth-grade students, enrolled <strong>in</strong> eight classes of three Dutch secondary<br />
schools, to participate <strong>in</strong> a study on response to short stories. About 60 students volunteered.<br />
We then <strong>in</strong>terviewed their literature teachers, ask<strong>in</strong>g them to identify, among the volunteers,<br />
students that were particularly well motivated and competent <strong>in</strong> read<strong>in</strong>g literature and students<br />
that were unmotivated and below average <strong>in</strong> read<strong>in</strong>g literature. Dur<strong>in</strong>g our conversations, all<br />
eight teachers were able to identify some excellent and some very weak students. The<br />
teachers’ assessments were validated by other data (students’ self assessments and their<br />
grades for related school subjects).<br />
From these two extreme groups we recruited 20 participants, of whom 10 were<br />
identified as ‘strong’ and 10 as ‘weak’ readers of literature. In age and ethnic representation,<br />
the two groups were similar; students were on average 16 years old, and two students <strong>in</strong> each<br />
group were from non-Dutch (mostly Turkish or Moroccan) orig<strong>in</strong>. Because boys and girls<br />
may respond differently to <strong>literary</strong> <strong>texts</strong> (e.g., Van Schooten, 2005), we strove for an equal<br />
number of boys and girls <strong>in</strong> both groups. Eventually, one boy withdrew from the study<br />
because of personal reasons, so that 19 students participated: 10 strong (5 female) and 9 weak<br />
readers (5 female). The students participated <strong>in</strong> the study after school time, and were<br />
compensated by a small reward.<br />
The stories<br />
We used <strong>literary</strong> stories that could <strong>in</strong>vite multiple <strong>in</strong>terpretations. The stories were<br />
short (500-1000 words), so that th<strong>in</strong>k<strong>in</strong>g aloud would not take more than 20 m<strong>in</strong>utes per<br />
story. We first pre-tested ten stories, by hav<strong>in</strong>g each story read under th<strong>in</strong>k-aloud conditions<br />
by two to three volunteers (other tenth-graders). Out of these ten stories, we chose five stories<br />
that proved to be unfamiliar to the students, elicited a variety of responses, and were<br />
challeng<strong>in</strong>g without be<strong>in</strong>g too difficult. The birthday calendar by Marianne de Nooyer is a<br />
story about a boy who, out of boredom, gives marks to his family members, and writes them<br />
down on the birthday calendar. The three friends by Jeanette W<strong>in</strong>terson is a post-modern fairy<br />
tale about three friends search<strong>in</strong>g for “that which cannot be found”. In Hullay by Cees<br />
Nooteboom a boy sees his little nephew drown<strong>in</strong>g and does noth<strong>in</strong>g to help him. And then it<br />
was our turn by Kader Abdolah, is about a family suffer<strong>in</strong>g under the dictatorship of the Shah<br />
of Persia. F<strong>in</strong>ally, Primo Levi’s The <strong>in</strong>terview is a socio-critical story about a man who is<br />
be<strong>in</strong>g <strong>in</strong>terviewed by an alien about life on earth.<br />
The stories differ <strong>in</strong> <strong>literary</strong> genre; they encompass a fairy tale, a science fiction story,<br />
as well as realistic or psychological stories. Some are witty and imag<strong>in</strong>ative (e.g., The three<br />
friends), while others are serious or tragic (e.g., And then it was our turn). Three of the stories<br />
were orig<strong>in</strong>ally written <strong>in</strong> Dutch; for The three friends and The <strong>in</strong>terview we used authorised<br />
Dutch translations. For the benefit of th<strong>in</strong>k<strong>in</strong>g aloud, we divided each story <strong>in</strong>to story<br />
fragments (10 to 15), each consist<strong>in</strong>g of one or two paragraphs. We copied each fragment on a<br />
separate PowerPo<strong>in</strong>t slide.<br />
Procedure<br />
Two <strong>in</strong>dividual sessions were held with each student at the research <strong>in</strong>stitute, led by<br />
one of the researchers or an assistant. Dur<strong>in</strong>g the first session, participants watched a video of<br />
a student th<strong>in</strong>k<strong>in</strong>g aloud while try<strong>in</strong>g to solve math problems. The video was <strong>in</strong>tended to<br />
acqua<strong>in</strong>t students with the method of th<strong>in</strong>k<strong>in</strong>g aloud, without cue<strong>in</strong>g particular types of
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -6-<br />
response (Ericsson & Simon, 1993). Then, the participants received the follow<strong>in</strong>g<br />
<strong>in</strong>structions:<br />
“You are about to read a few short stories. We would like you to th<strong>in</strong>k aloud as you try to make sense of<br />
them. Try to put <strong>in</strong>to words everyth<strong>in</strong>g that is go<strong>in</strong>g on <strong>in</strong> your m<strong>in</strong>d while you are read<strong>in</strong>g. Imag<strong>in</strong>e, for<br />
<strong>in</strong>stance, a radio reporter who is giv<strong>in</strong>g an account of a football match. A reporter can see what happens<br />
<strong>in</strong> the field, but the listeners can’t. So, like a reporter, you have to tell us what’s go<strong>in</strong>g on <strong>in</strong> your m<strong>in</strong>d<br />
dur<strong>in</strong>g read<strong>in</strong>g. Don’t worry about how you say it; just share everyth<strong>in</strong>g you are th<strong>in</strong>k<strong>in</strong>g the moment it<br />
comes to your m<strong>in</strong>d.”<br />
We emphasized that it was crucial to talk as much as possible, and that there were no right or<br />
wrong responses.<br />
Participants read the stories fragment by fragment from a computer screen, two stories<br />
dur<strong>in</strong>g the first session and three stories dur<strong>in</strong>g the next session, about one week later. The<br />
stories were presented <strong>in</strong> the same order, beg<strong>in</strong>n<strong>in</strong>g with the least difficult one (The birthday<br />
calendar). For each story, we provided some contextual <strong>in</strong>formation, by show<strong>in</strong>g the actual<br />
published version on paper, a picture of the author, and some biographical details (year of<br />
birth, genres, titles of previous works).<br />
The read<strong>in</strong>g of the stories on the computer was self-paced; by press<strong>in</strong>g a key the<br />
student could advance to the next fragment, or return to a previous one. A bar <strong>in</strong>dicated the<br />
number of story fragments already processed and still to follow. When a student fell silent for<br />
more than five seconds, we used general prompts <strong>in</strong> order to stimulate a response (‘What are<br />
you do<strong>in</strong>g?’ ‘What is go<strong>in</strong>g on <strong>in</strong> your m<strong>in</strong>d?’). More specific directions were avoided, <strong>in</strong><br />
order not to cue particular responses.<br />
Students were given as much time as they wanted. The th<strong>in</strong>k aloud sessions were<br />
audio recorded on the computer, and transcripts were made of the record<strong>in</strong>gs.<br />
Protocol analysis<br />
We collected 92 th<strong>in</strong>k aloud transcripts, from the 19 students, each read<strong>in</strong>g four or five<br />
stories (three record<strong>in</strong>gs failed due to technical problems). The transcripts were segmented<br />
<strong>in</strong>to mean<strong>in</strong>gful units or statements, most often sentences. To analyze the statements, we<br />
developed a cod<strong>in</strong>g scheme of read<strong>in</strong>g activities, based on the model of Andr<strong>in</strong>ga (1995b).<br />
Two of the researchers tested the scheme on 30 transcripts collected <strong>in</strong> our pilot study, and<br />
changed some of the categories to fit our material.<br />
The process of test<strong>in</strong>g and revis<strong>in</strong>g categories resulted <strong>in</strong> a cod<strong>in</strong>g scheme of eight<br />
categories, cover<strong>in</strong>g most of the student statements <strong>in</strong> the pilot transcripts (see Table 1). Some<br />
statements did not fit <strong>in</strong>to any of the categories; for these, we added the category Other<br />
activity.<br />
<br />
To illustrate how the student responses were transcribed and coded, we present a coded<br />
transcript fragment (see Table 2). The student <strong>in</strong> this fragment responded to the first two<br />
fragments of The three friends. In Table 2, all comments of the transcriber are placed between<br />
square brackets. Cod<strong>in</strong>gs are placed between angle brackets. The abbreviations ,<br />
and so on, refer to our process<strong>in</strong>g categories.<br />
<br />
Most students first read the story fragment through (either silently or aloud), and then started<br />
to reread and th<strong>in</strong>k aloud <strong>in</strong> response to that particular fragment. Reread<strong>in</strong>g and th<strong>in</strong>k<strong>in</strong>g<br />
aloud often happened simultaneously. Sometimes students did not read the passage through,<br />
but responded immediately <strong>in</strong>-between the l<strong>in</strong>es.
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -7-<br />
Student statements were treated as <strong>in</strong>dividual cases; each statement or sentence was<br />
numbered and coded. Actual (silent) read<strong>in</strong>g by students was noted <strong>in</strong> the transcripts, as were<br />
pauses <strong>in</strong> their response, but these silences were treated as ‘default’, and hence not coded.<br />
Nonverbal responses were coded as much as possible (e.g., [Sighs], [Laughs]). Because we<br />
transcribed from audio record<strong>in</strong>gs, facial expressions and gestures could not be recorded and<br />
thus were left out.<br />
All statements <strong>in</strong> the transcripts (n = 4347) were coded by one of the researchers. In<br />
order to test the <strong>in</strong>ter-rater reliability, 10% of the student statements (n = 413) from n<strong>in</strong>e<br />
different transcripts were coded by two coders <strong>in</strong>dependently. The <strong>in</strong>ter-rater agreement was<br />
85% (Cohen’s Kappa .81).<br />
Analysis<br />
We wanted to exam<strong>in</strong>e differences between strong and weak readers <strong>in</strong> their distribution of<br />
several read<strong>in</strong>g activities over the read<strong>in</strong>g process. Therefore, we needed to model the<br />
occurrence of each read<strong>in</strong>g activity as a function of time (elapsed s<strong>in</strong>ce the start of the read<strong>in</strong>g<br />
process), and exam<strong>in</strong>e this function across strong and weak readers.<br />
To model these occurrences of read<strong>in</strong>g activities, multilevel models appear to be<br />
appropriate. Multilevel models take the hierarchy of the data <strong>in</strong>to account; the observations of<br />
the occurrence of activities are considered as nested with<strong>in</strong> story fragments, which <strong>in</strong> turn are<br />
nested with<strong>in</strong> stories and readers. Each observation ‘belongs’ to a particular reader. This<br />
quality of multilevel models makes it possible to analyze the occurrence of read<strong>in</strong>g activities<br />
for <strong>in</strong>dividual readers. There is no need to aggregate over all activities of each reader<br />
(result<strong>in</strong>g <strong>in</strong> frequencies per reader).<br />
Let Yijk be an <strong>in</strong>dicator variable, which <strong>in</strong>dicates whether fragment i of story j of<br />
<strong>in</strong>dividual k <strong>in</strong> the process is analyzed (Yijk equals 1) or not (Yijk equals 0). Now we want to<br />
show that the occurrence of a read<strong>in</strong>g activity varies dur<strong>in</strong>g read<strong>in</strong>g, and that the occurrence<br />
of a read<strong>in</strong>g activity varies between stories as well as between readers. In order to model the<br />
relation between the dichotomous dependent variable and the moment (fragment) of<br />
occurrence, a polynomial function is used.<br />
p<br />
Logit(<br />
) = β (1)<br />
Y<br />
ijk<br />
∑ = p P<br />
p=<br />
0<br />
* fr<br />
(i = 1, 2, …Ijk; j = 1, 2, …,Jk; k = 1, 2, …K).<br />
p<br />
ijk<br />
In the equation above, the fragment number is <strong>in</strong>dicated with frijk. So, powers of fragment<br />
number are used to describe the occurrence of (the logit 1 ) of Yijk. That is: the occurrence of a<br />
read<strong>in</strong>g activity at a certa<strong>in</strong> moment dur<strong>in</strong>g read<strong>in</strong>g is written as a function of fr 1 + fr 2 + fr 3 …<br />
Such polynomials are extremely flexible; they can take almost any shape depend<strong>in</strong>g on the<br />
number of parameters (P) and the numerical value of the regression weights. This, however,<br />
can be considered as an empirical matter; the shape of the polynomial (i.e. the number of<br />
polynomials needed) can be estimated from the data. The model chosen has to fulfil the<br />
follow<strong>in</strong>g requirement: higher order elements are only taken <strong>in</strong>to account if all lower order<br />
elements have reached significance. That is, a second order element (fr 2 ) is only taken <strong>in</strong>to<br />
account if the first order element (fr 1 ) has reached significance.<br />
However, the model accord<strong>in</strong>g to equation (1) does not allow for differences between<br />
either stories or <strong>in</strong>dividuals. We can consider the proportion at fragment i of <strong>in</strong>dividual k as<br />
deviance from the mean occurrence of this read<strong>in</strong>g activity for all <strong>in</strong>dividuals. That is, a<br />
variance between <strong>in</strong>dividuals for each regression weight has to be estimated. This is merely
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -8-<br />
another way of say<strong>in</strong>g that for each <strong>in</strong>dividual the difference with the mean regression weight<br />
is estimated. The occurrences of this read<strong>in</strong>g activity might vary between stories as well; the<br />
regression weight for story j of <strong>in</strong>dividual k (at fragment i) is considered as a deviance from<br />
the mean regression weight for this <strong>in</strong>dividual. Hence, equation (1) has to be elaborated with<br />
residuals for <strong>in</strong>dividuals as well as with residuals for the story read.<br />
p=<br />
P<br />
∑<br />
p=<br />
0<br />
p=<br />
P<br />
p<br />
∑(γ<br />
p<br />
β = + + )<br />
(2)<br />
p=<br />
0<br />
u<br />
pjk<br />
v<br />
p0k<br />
In equation (2) residuals for stories (upjk) as well as for <strong>in</strong>dividuals (vp0k) are allowed. These<br />
residuals are assumed to be normally distributed with an expected value of zero and a<br />
variance of S 2 upjk and S 2 vp0k respectively. It is however an empirical matter which variance<br />
components are needed to describe the differences between stories (S 2 u1jk, S 2 u2jk, … ) and<br />
which are needed to describe the differences between <strong>in</strong>dividuals (S 2 v10k, S 2 v20k, …).<br />
Substitution of equation (2) <strong>in</strong> (1) shows the model to be estimated:<br />
Y<br />
∑ = p P<br />
p=<br />
0<br />
p<br />
p<br />
Logit(<br />
) = [( γ + + ) * ]<br />
(3)<br />
ijk<br />
u<br />
pjk<br />
v<br />
p0<br />
k<br />
fr<br />
ijk<br />
As the dependent variable is dichotomous, a logit transformation is used (remember, Logit<br />
(Yijk) = Ln (Yijk / [1 – Yijk)).<br />
Results<br />
Figure 1 summarizes the general course of students’ read<strong>in</strong>g activities over the read<strong>in</strong>g<br />
process, constructed from the fixed part estimates of the model (see the Appendix for the<br />
estimates). Each read<strong>in</strong>g activity is presented <strong>in</strong> a separate panel.<br />
<br />
The horizontal axis represents the factor ‘time’ <strong>in</strong> the form of story fragments. On the vertical<br />
axis the estimated mean probability of occurrence of each read<strong>in</strong>g activity is diagrammed for<br />
the two ability groups; W (weak readers) and S (strong readers). Each l<strong>in</strong>e represents the<br />
mean course of the read<strong>in</strong>g activity displayed by weak or strong readers over four to five<br />
stories.<br />
It is important to note that the scale of the vertical axis differs for the various read<strong>in</strong>g<br />
activities, due to different levels of probability of occurrence. For <strong>in</strong>stance, retell<strong>in</strong>g is a rather<br />
common activity, whereas metacognitive respond<strong>in</strong>g is relatively rare.<br />
In our discussion of Figure 1, we will first focus on the general patterns <strong>in</strong> students’<br />
activities. In the next section we will consider differences <strong>in</strong> patterns between weak and<br />
strong readers.<br />
Do readers change their configuration of activities over the course of the read<strong>in</strong>g process?<br />
As appears from Figure 1, there are four dist<strong>in</strong>ct patterns <strong>in</strong> the general course of students’<br />
read<strong>in</strong>g activities;<br />
- no change over the course of the read<strong>in</strong>g process;<br />
- an <strong>in</strong>crease of the read<strong>in</strong>g activity over the course of the read<strong>in</strong>g process;<br />
- a decl<strong>in</strong>e of the read<strong>in</strong>g activity over the course of the read<strong>in</strong>g process;
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -9-<br />
- an <strong>in</strong>crease at the beg<strong>in</strong>n<strong>in</strong>g, followed by a decl<strong>in</strong>e towards the end of the read<strong>in</strong>g<br />
process.<br />
Some read<strong>in</strong>g activities (e.g., ‘analyz<strong>in</strong>g’) are represented by straight horizontal l<strong>in</strong>es. In these<br />
cases, the mean probability of occurrence of the activity rema<strong>in</strong>s constant over the read<strong>in</strong>g<br />
process as a whole. No change over the read<strong>in</strong>g process was observed.<br />
Other read<strong>in</strong>g activities show a pattern of gradual <strong>in</strong>crease (e.g., ‘problem detect<strong>in</strong>g’).<br />
At the beg<strong>in</strong>n<strong>in</strong>g of the read<strong>in</strong>g process, the probability of occurrence of problem detect<strong>in</strong>g is<br />
less than 10%. Student-readers tend to detect relatively few problems when read<strong>in</strong>g the first<br />
few story fragments. As the story read<strong>in</strong>g progresses, the probability of occurrence <strong>in</strong>creases.<br />
At the end of the read<strong>in</strong>g process, problem detect<strong>in</strong>g reaches a peak (about 20 %). Apparently,<br />
students experience relatively many <strong>in</strong>terpretation problems towards the end of the story.<br />
The third pattern can be illustrated by ‘associat<strong>in</strong>g’. This activity significantly decl<strong>in</strong>es<br />
over the course of the read<strong>in</strong>g process. At the start of the read<strong>in</strong>g process, when students are<br />
read<strong>in</strong>g the story’s title and the first few fragments, they provide relatively many personal<br />
associations <strong>in</strong> response to the story. As the read<strong>in</strong>g progresses, fewer associations are given.<br />
The probability of occurrence drops from 10 – 20% at the start to less than 5% at the end of<br />
the read<strong>in</strong>g process.<br />
The f<strong>in</strong>al pattern is illustrated by strong readers’ emotional responses. At the<br />
beg<strong>in</strong>n<strong>in</strong>g of the read<strong>in</strong>g process, there is a probability of about 5% that strong readers<br />
provide emotional responses to the story. This probability <strong>in</strong>creases significantly dur<strong>in</strong>g<br />
read<strong>in</strong>g. For story fragment 9 the probability of occurrence has been doubled (13%). From<br />
this po<strong>in</strong>t on, the chance of emotional responses to occur dim<strong>in</strong>ishes aga<strong>in</strong>, dropp<strong>in</strong>g to about<br />
8%. Apparently, strong readers tend to provide the most emotional responses when they are<br />
read<strong>in</strong>g the middle part of stories, the part where ‘the true action’ takes place.<br />
In sum, readers tend to change the configuration of their read<strong>in</strong>g activities over time.<br />
The mean probability of occurrence fluctuates over moments <strong>in</strong> the read<strong>in</strong>g process for six out<br />
of eight read<strong>in</strong>g activities; retell<strong>in</strong>g, <strong>in</strong>ferenc<strong>in</strong>g, problem detect<strong>in</strong>g, associat<strong>in</strong>g, evaluat<strong>in</strong>g<br />
and emotional respond<strong>in</strong>g. For two activities (analyz<strong>in</strong>g and metacognitive respond<strong>in</strong>g) no<br />
changes over the course of the read<strong>in</strong>g process were observed.<br />
Do strong readers more often change their activities dur<strong>in</strong>g read<strong>in</strong>g than their weaker peers?<br />
Our second research question (2a) <strong>in</strong>volves a comparison between weak (W) and strong (S)<br />
readers, regard<strong>in</strong>g changes <strong>in</strong> their read<strong>in</strong>g activities dur<strong>in</strong>g read<strong>in</strong>g. The comparison can be<br />
made from two different perspectives; <strong>in</strong> terms of differences <strong>in</strong> occurrence, and <strong>in</strong> terms of<br />
the pattern of development of the activity over time.<br />
Figure 1 shows that for analyz<strong>in</strong>g, the l<strong>in</strong>es for weak and strong readers co<strong>in</strong>cide,<br />
form<strong>in</strong>g one straight horizontal l<strong>in</strong>e. This <strong>in</strong>dicates that weak readers engage <strong>in</strong> analyz<strong>in</strong>g as<br />
often as strong readers, and that <strong>in</strong> both groups the occurrence of analyz<strong>in</strong>g rema<strong>in</strong>s constant<br />
over the read<strong>in</strong>g process.<br />
For problem detect<strong>in</strong>g, associat<strong>in</strong>g and metacognitive respond<strong>in</strong>g, the l<strong>in</strong>es of weak<br />
and strong readers are (almost) parallel. This <strong>in</strong>dicates a similar pattern of development for<br />
both groups; an <strong>in</strong>crease of problem detect<strong>in</strong>g, a decl<strong>in</strong>e of associat<strong>in</strong>g, and a stable pattern of<br />
metacognitive responses throughout their read<strong>in</strong>g of a story. However, the extent to which<br />
weak and strong readers engage <strong>in</strong> these activities differs. In general, strong readers engage <strong>in</strong><br />
these activities more often than weak readers, throughout the read<strong>in</strong>g process.<br />
The panels of the rema<strong>in</strong><strong>in</strong>g read<strong>in</strong>g activities (retell<strong>in</strong>g, <strong>in</strong>ferenc<strong>in</strong>g, evaluat<strong>in</strong>g, and<br />
emotional respond<strong>in</strong>g) show differences <strong>in</strong> occurrence and different patterns of development<br />
for weak and strong readers. For example, compared to strong readers, weak readers provide<br />
many retell<strong>in</strong>gs of the story content dur<strong>in</strong>g read<strong>in</strong>g. Their mean pattern of response does not
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -10-<br />
change over the read<strong>in</strong>g process; the pattern is stable or monotonous. For strong readers, on<br />
the other hand, the occurrence of retell<strong>in</strong>g fluctuates, follow<strong>in</strong>g a curve l<strong>in</strong>ear pattern.<br />
In Table 3, we present a summary of the differences <strong>in</strong> mean patterns of read<strong>in</strong>g<br />
activities between weak and strong readers.<br />
<br />
For five out of eight activities, weak readers display a monotonous pattern of read<strong>in</strong>g<br />
activities <strong>in</strong> the course of their read<strong>in</strong>g process. When their read<strong>in</strong>g activities change dur<strong>in</strong>g<br />
read<strong>in</strong>g (as <strong>in</strong> problem detect<strong>in</strong>g, associat<strong>in</strong>g and evaluat<strong>in</strong>g), the changes are l<strong>in</strong>ear.<br />
Furthermore, two of these three activities (problem detect<strong>in</strong>g and associat<strong>in</strong>g) change for<br />
strong readers as well, and <strong>in</strong> the same manner. This <strong>in</strong>dicates that the fluctuations <strong>in</strong> these<br />
activities dur<strong>in</strong>g the read<strong>in</strong>g process are unrelated to readers’ ability level.<br />
In general, the strong readers are more versatile dur<strong>in</strong>g read<strong>in</strong>g, adapt<strong>in</strong>g their read<strong>in</strong>g<br />
activities to the story fragment they are read<strong>in</strong>g. Their activities reflect different types of<br />
patterns <strong>in</strong> the course of read<strong>in</strong>g; an <strong>in</strong>crease, a decl<strong>in</strong>e, or a curve l<strong>in</strong>ear pattern.<br />
Individual differences among strong and weak readers<br />
Up to this po<strong>in</strong>t, we have focused on the mean differences between weak and strong readers,<br />
and the mean patterns of change as they emerge from the fixed part of the model. However,<br />
there may be large differences between <strong>in</strong>dividual readers with<strong>in</strong> the same ability group. The<br />
parameter estimates of the random part of the model (see the Appendix) give an <strong>in</strong>dication of<br />
the variability with<strong>in</strong> the group of weak and strong readers. The estimates represent the<br />
deviations of <strong>in</strong>dividual readers from the general mean read<strong>in</strong>g process <strong>in</strong> each group.<br />
From an <strong>in</strong>spection of the estimates we learn that all between-subject variances are<br />
statistically significant (p
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -11-<br />
one weak reader who provides relatively many associations over the course of his or her<br />
read<strong>in</strong>g process.<br />
For metacognitive respond<strong>in</strong>g, most of the l<strong>in</strong>es of <strong>in</strong>dividual readers are very close<br />
together, with strong readers produc<strong>in</strong>g more metacognitive responses than weak readers. One<br />
weak reader produces an exceptionally large number of metacognitive responses, surpass<strong>in</strong>g<br />
all other readers.<br />
To conclude, the <strong>in</strong>dividual differences are generally larger for the weak readers than<br />
for the strong readers. The number of readers responsible for the large variability between<br />
weak readers differs from activity to activity. Sometimes just one <strong>in</strong>dividual reader is<br />
responsible (e.g., associat<strong>in</strong>g, metacognitive respond<strong>in</strong>g), sometimes more readers (e.g.,<br />
retell<strong>in</strong>g). However, <strong>in</strong>dividual readers tend to deviate only <strong>in</strong> the extent to which they use an<br />
activity, not <strong>in</strong> the pattern of the activity. Individual weak readers have a stable or<br />
monotonous pattern of response <strong>in</strong> common.<br />
Do strong readers more often change their activities <strong>in</strong> response to different stories than their<br />
weaker peers?<br />
Our third research question (2b) aga<strong>in</strong> <strong>in</strong>volves a comparison between strong and weak<br />
readers, but now we focus on differences <strong>in</strong> response to different stories. The random<br />
parameter estimates provide <strong>in</strong>formation about the between-story variability with<strong>in</strong> the group<br />
of weak and strong readers (see the Appendix). For almost all weak and strong readers’<br />
activities the estimates are significant (p < .10). This <strong>in</strong>dicates that there is an effect of ‘story’<br />
on the probability of occurrence of read<strong>in</strong>g activities. In other words, students’ patterns of<br />
read<strong>in</strong>g activities vary, depend<strong>in</strong>g on the story they are read<strong>in</strong>g.<br />
For five of the eight read<strong>in</strong>g activities, the parameter estimates for the strong students<br />
are larger than the parameter estimates for the weak students. For <strong>in</strong>stance, the between-story<br />
variance of retell<strong>in</strong>g for strong readers is 0.34, whereas the estimate for weak readers is small:<br />
0.08. A similar difference is found for <strong>in</strong>ferenc<strong>in</strong>g, problem detect<strong>in</strong>g, associat<strong>in</strong>g and<br />
metacognitive respond<strong>in</strong>g. This <strong>in</strong>dicates that strong readers’ activities are more <strong>in</strong>fluenced by<br />
the particular story than the weak readers’ activities dur<strong>in</strong>g read<strong>in</strong>g.<br />
In most cases, the between-story variances for strong readers are larger than the<br />
between-subject variances for the same readers. In other words, strong readers do not differ so<br />
much from each other <strong>in</strong> the read<strong>in</strong>g activities they use, but they tend to respond differently to<br />
different stories. For weak readers it is the other way around; weak readers show large<br />
<strong>in</strong>dividual differences (see previous section), but they do not change their read<strong>in</strong>g activities<br />
much <strong>in</strong> response to different stories. Rather, weak readers tend to ma<strong>in</strong>ta<strong>in</strong> their own,<br />
personal pattern of response across different stories.<br />
To illustrate the between-story differences we present Figure 3. In this Figure the<br />
estimated probability of occurrence of retell<strong>in</strong>g is depicted. Each l<strong>in</strong>e represents one<br />
<strong>in</strong>dividual reader (W = weak, S = Strong) read<strong>in</strong>g one story. In total, the figure conta<strong>in</strong>s 92<br />
l<strong>in</strong>es: for 19 participants, each read<strong>in</strong>g four to five stories.<br />
<br />
Figure 3 shows that the l<strong>in</strong>es of the strong readers are relatively far apart, <strong>in</strong>dicat<strong>in</strong>g<br />
differences between stories. For the strong readers, the probability of occurrence of retell<strong>in</strong>g<br />
fluctuates between 0% and 40%, depend<strong>in</strong>g on the particular story, and on the particular<br />
fragment with<strong>in</strong> the story. The l<strong>in</strong>es of the weak readers are closer together. For the weak<br />
readers, the estimated probability of occurrence of retell<strong>in</strong>g varies between 13% and 26%.<br />
There are no effects of story or story fragments on the occurrence of the weak readers’<br />
retell<strong>in</strong>g activities.
Discussion<br />
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -12-<br />
Our first question was whether adolescent readers changed the configuration of their read<strong>in</strong>g<br />
activities over the course of their read<strong>in</strong>g process, when read<strong>in</strong>g and respond<strong>in</strong>g to short<br />
<strong>literary</strong> stories. We found that they did. In fact, most of their cognitive and affective read<strong>in</strong>g<br />
activities fluctuated over the course of read<strong>in</strong>g a short story. For example, the occurrence of<br />
problem detect<strong>in</strong>g gradually <strong>in</strong>creased dur<strong>in</strong>g read<strong>in</strong>g, whereas the occurrence of personal<br />
associations decreased. Such fluctuations were observed for six out of eight read<strong>in</strong>g activities.<br />
Our second question was whether strong adolescent readers more often changed the<br />
configuration of their read<strong>in</strong>g activities than weak readers, over the course of the read<strong>in</strong>g<br />
process and <strong>in</strong> response to different stories. This appeared to be the case. For example, strong<br />
readers varied their retell<strong>in</strong>gs over the course of read<strong>in</strong>g a story, and their pattern of retell<strong>in</strong>g<br />
varied depend<strong>in</strong>g on the story. This suggests that strong readers were ‘context-sensitive’ <strong>in</strong><br />
their use of the retell<strong>in</strong>g strategy. Weak readers, on the other hand, did not change their<br />
pattern of retell<strong>in</strong>g dur<strong>in</strong>g read<strong>in</strong>g, nor did they change their pattern of retell<strong>in</strong>g <strong>in</strong> response to<br />
different stories. Instead, they showed a monotonous pattern of response. Thus, we<br />
demonstrated that, although weak and strong adolescent readers use the same activities, they<br />
may differ <strong>in</strong> their orchestration of activities over the read<strong>in</strong>g process and over stories. We<br />
conclude that strong readers tend to show a more flexible read<strong>in</strong>g process than weak readers.<br />
However, we must po<strong>in</strong>t out some limitations of this study. Note, for <strong>in</strong>stance, that we<br />
dealt with the various cognitive and affective activities <strong>in</strong> an univariate way. However,<br />
read<strong>in</strong>g activities probably <strong>in</strong>teract with each other. For example, the <strong>in</strong>crease of problem<br />
detect<strong>in</strong>g may be directly related to the decrease <strong>in</strong> mak<strong>in</strong>g personal associations. A possible<br />
explanation of this pattern lies <strong>in</strong> the stories we used. Perhaps they were too difficult; dur<strong>in</strong>g<br />
read<strong>in</strong>g students <strong>in</strong>creas<strong>in</strong>gly experienced problems <strong>in</strong> mak<strong>in</strong>g sense of the text, and<br />
consequently they were no longer able to make any personal associations (or vice versa).<br />
However, the small sample size of 19 participants did not allow a simultaneous analysis of the<br />
occurrence of several read<strong>in</strong>g activities. Therefore, <strong>in</strong>teractions could not be analyzed.<br />
A further concern is whether th<strong>in</strong>k<strong>in</strong>g aloud had a different effect on the read<strong>in</strong>g<br />
activities of weak readers than on the activities of strong readers. One might argue that our<br />
f<strong>in</strong>d<strong>in</strong>gs are due to an <strong>in</strong>teraction between the type of participant and the demands of the<br />
th<strong>in</strong>k<strong>in</strong>g aloud method. Weak readers might be less confident and consequently less able to<br />
verbalize their thoughts dur<strong>in</strong>g read<strong>in</strong>g (let alone vary their response patterns) than strong<br />
readers. Yet, the weak readers <strong>in</strong> our study were not less elaborate <strong>in</strong> their responses than their<br />
strong peers. On the contrary, their th<strong>in</strong>k aloud transcripts were longer than those of strong<br />
readers. Besides, most students talked freely dur<strong>in</strong>g the th<strong>in</strong>k aloud sessions. It seems<br />
unlikely, then, that the differences we observed are attributable to differences <strong>in</strong> selfassurance<br />
or verbal ability.<br />
We recruited the participants <strong>in</strong> our study from two ‘known’ groups; weak and strong<br />
readers of literature. In do<strong>in</strong>g so, we maximized the likelihood of f<strong>in</strong>d<strong>in</strong>g differences <strong>in</strong><br />
read<strong>in</strong>g activities and <strong>in</strong> ‘flexibility’ <strong>in</strong> use of activities. However, the ability to read literature<br />
should rather be viewed on a cont<strong>in</strong>uum (vary<strong>in</strong>g from very weak to very strong) than <strong>in</strong><br />
terms of a dichotomy. A more precise <strong>in</strong>dication of students’ <strong>literary</strong> read<strong>in</strong>g skill could be<br />
derived from students’ (verbal or written) post-read<strong>in</strong>g responses to the same stories used for<br />
the th<strong>in</strong>k alouds. Post-read<strong>in</strong>g responses could be used to exam<strong>in</strong>e the relationship between<br />
the immediate outcome of a read<strong>in</strong>g process and the preced<strong>in</strong>g read<strong>in</strong>g activities. Such a study<br />
would enable us to determ<strong>in</strong>e which patterns of read<strong>in</strong>g activities are predictive of the quality<br />
of students’ post-read<strong>in</strong>g response.<br />
Despite these limitations, this study has yielded mean<strong>in</strong>gful <strong>in</strong>formation about<br />
<strong>in</strong>dividual differences <strong>in</strong> <strong>literary</strong> read<strong>in</strong>g processes, <strong>in</strong> particular, about differences <strong>in</strong> the
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -13-<br />
orchestration of read<strong>in</strong>g processes. The ‘strategy approach’, as described by Schunn and<br />
Reder (1998), cannot fully account for such differences, for it assumes that different groups of<br />
readers use different strategies or processes. We found that different groups of readers may<br />
use the same processes, but <strong>in</strong> a different configuration, depend<strong>in</strong>g on the moment <strong>in</strong> the<br />
read<strong>in</strong>g process and on the particular text. In other words, there are two miss<strong>in</strong>g l<strong>in</strong>ks <strong>in</strong><br />
previous research: the factor ‘time’ or moment <strong>in</strong> the read<strong>in</strong>g process, and the factor ‘text’ or<br />
the necessity of us<strong>in</strong>g many <strong>literary</strong> <strong>texts</strong> (<strong>in</strong>stead of just one or two, as done <strong>in</strong> most previous<br />
studies). The importance of the present study is that it has made these miss<strong>in</strong>g l<strong>in</strong>ks visible by<br />
empirical means.<br />
The f<strong>in</strong>d<strong>in</strong>gs of our study may also be of practical importance for the teach<strong>in</strong>g and<br />
learn<strong>in</strong>g of literature. To br<strong>in</strong>g weak student-readers to a higher level of expertise <strong>in</strong> literature<br />
read<strong>in</strong>g, it is clearly not enough to encourage them to br<strong>in</strong>g forward their personal responses<br />
to a <strong>literary</strong> text or to use more sophisticated read<strong>in</strong>g strategies dur<strong>in</strong>g read<strong>in</strong>g. Weak readers<br />
especially have to learn to differentiate their responses and strategies, depend<strong>in</strong>g on the phase<br />
<strong>in</strong> the read<strong>in</strong>g process, and on the particular <strong>literary</strong> text they are read<strong>in</strong>g.<br />
References<br />
Andr<strong>in</strong>ga, E. (1995a). Ontwikkel<strong>in</strong>gen <strong>in</strong> het omgaan met literaire complexiteit<br />
[Developments <strong>in</strong> deal<strong>in</strong>g with <strong>literary</strong> complexity]. Tijdschrift voor<br />
Onderwijsresearch, 20(3), 271-284.<br />
Andr<strong>in</strong>ga, E. (1995b). Strategieën bij het lezen van literatuur [Literary read<strong>in</strong>g strategies].<br />
Spiegel, 13(3), 7-33.<br />
Applebee, A.N. & Purves, A.C. (1992). Literature and the English language arts. In P.<br />
Jackson (Ed.), Handbook of curriculum research (pp. 726-748). New York:<br />
MacMillan.<br />
Braaksma, M.A.H., Rijlaarsdam, G., Van den Bergh, H., & Van Hout-Wolters, B.H.A.M.<br />
(2004). Observational learn<strong>in</strong>g and its effects on the orchestration of writ<strong>in</strong>g<br />
processes. Cognition and Instruction, 22(1), 1-36.<br />
Breetvelt, I., Van den Bergh, H., & Rijlaarsdam, G. (1994). Relations between writ<strong>in</strong>g<br />
processes and text quality: When and how? Cognition and Instruction, 12(2), 103-123.<br />
Britton, B.K., & Graesser, A.C. (Eds.). (1996). Models of understand<strong>in</strong>g text. Mahwah, New<br />
Jersey: Lawrence Erlbaum Associates.<br />
Dorfman, M.H. (1996). Evaluat<strong>in</strong>g the <strong>in</strong>terpretive community, evidence from expert and<br />
novice readers. Poetics, 23(6), 453-470.<br />
Earthman, E.A. (1992). Creat<strong>in</strong>g the virtual work: readers' processes <strong>in</strong> understand<strong>in</strong>g <strong>literary</strong><br />
<strong>texts</strong>. Research <strong>in</strong> the Teach<strong>in</strong>g of English, 26(4), 351-384.<br />
Ericsson, K. A., & Simon, H. A. (1993). Protocol analysis: Verbal reports as data (Revised<br />
edition). Cambridge: The MIT Press.<br />
Eva-Wood, A.L. (2004). Th<strong>in</strong>k<strong>in</strong>g and feel<strong>in</strong>g poetry: Explor<strong>in</strong>g mean<strong>in</strong>gs aloud. Journal of<br />
Educational Psychology, 96(1), 182-191.<br />
Groeben, N. (2001). Literary creativity. In Schram, D. & Steen, G. (Eds.), The psychology<br />
and sociology of literature: <strong>in</strong> honor of Elrud Ibsch (pp. 17-34).<br />
Amsterdam/Philadelphia: John Benjam<strong>in</strong>s.<br />
Harker, W.J. (1994). “Pla<strong>in</strong> sense” and “poetic significance”: tenth-grade readers read<strong>in</strong>g two<br />
poems. Poetics, 22, 199-218.<br />
Janssen, T. (1998). Literatuuronderwijs bij benader<strong>in</strong>g. Een empirisch onderzoek naar de<br />
vormgev<strong>in</strong>g en opbrengsten van het literatuuronderwijs Nederlands <strong>in</strong> de bovenbouw<br />
van het havo en vwo. [Approaches to literature teach<strong>in</strong>g; A study of the form and
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -14-<br />
results of <strong>literary</strong> education <strong>in</strong> Dutch secondary schools.] Amsterdam: Thesis<br />
Publishers.<br />
Kneepkens, E.W.E.M. & Zwaan, R.A. (1994). Emotions and <strong>literary</strong> text comprehension.<br />
Poetics, 23, 125-138.<br />
Pesk<strong>in</strong>, J. (1998). Construct<strong>in</strong>g mean<strong>in</strong>g when read<strong>in</strong>g poetry: An expert-novice study.<br />
Cognition and Instruction, 16(3), 235-263.<br />
Pressley, M., & Afflerbach, P.P. (1995). Verbal protocols of read<strong>in</strong>g: The nature of<br />
constructively responsive read<strong>in</strong>g. Hillsdale, New Jersey: Lawrence Erlbaum.<br />
Poyas, Y. (2004). Explor<strong>in</strong>g the horizons of the literature classroom; Reader response,<br />
reception theories and classroom discourse. L1 – Educational Studies <strong>in</strong> Language<br />
and Literature, 4(1), 63-84.<br />
Rosenblatt, L. (1999). Literature as exploration. Fifth edition. New York: The Modern<br />
Language Association of America.<br />
Schunn, C.D. & Reder, L.M. (1998). Strategy adaptivity and <strong>in</strong>dividual differences. The<br />
Psychology of Learn<strong>in</strong>g and Motivation, 38, 115-154.<br />
Smith, M.W. (1991). Construct<strong>in</strong>g mean<strong>in</strong>g from text: An analysis of n<strong>in</strong>th-grade reader<br />
responses. Journal of Educational Research, 84(5), 263-271.<br />
Thury, E.M. & Friedlander, A. (1995). The impact of experience: the role of experience <strong>in</strong><br />
read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong>. In Rusch, G. (Ed.), Empirical approaches to literature (pp.61-<br />
72). Siegen: LUMIS-Publications.<br />
Van Dijk, T.A., & K<strong>in</strong>tsch, W. (1983). Strategies of discourse comprehension. New York:<br />
Academic Press.<br />
Van Schooten, E. (2005). Literary response and attitude toward read<strong>in</strong>g fiction. (PhD-thesis).<br />
Gron<strong>in</strong>gen: Rijksuniversiteit Gron<strong>in</strong>gen.<br />
Vipond, D., & Hunt, R.A. (1984). Po<strong>in</strong>t-driven understand<strong>in</strong>g: Pragmatic and cognitive<br />
dimensions of <strong>literary</strong> read<strong>in</strong>g. Poetics, 13, 261-277.<br />
Zeitz, C.M. (1994). Expert-novice differences <strong>in</strong> memory, abstraction, and reason<strong>in</strong>g <strong>in</strong> the<br />
doma<strong>in</strong> of literature. Cognition and Instruction, 12(4), 277-312.
Appendix<br />
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -15-<br />
Parameter estimates by Read<strong>in</strong>g Activity (<strong>in</strong> logit scores), <strong>in</strong>dicat<strong>in</strong>g change <strong>in</strong> occurrence of<br />
activities dur<strong>in</strong>g the read<strong>in</strong>g process.<br />
Random model Fixed model<br />
R1 R2 R3 R4 F1 F2 F3<br />
Weak readers<br />
Retell<strong>in</strong>g 1.10 0.08 - - -1.02 - -<br />
Inferenc<strong>in</strong>g 0.16 0.21 - - -1.09 - -<br />
Problem detect<strong>in</strong>g 0.19 0.27 0.30 ns 4.57 -2.17 0.55 -<br />
Associat<strong>in</strong>g 1.92 0.06 ns 0.74 7.17 -2.62 -0.66 -<br />
Analyz<strong>in</strong>g 0.46 0.81 - - -3.19 - -<br />
Evaluat<strong>in</strong>g 0.53 0.80 -0.03 ns 9.67 -3.08 0.31 ns -<br />
Emotional respond<strong>in</strong>g 0.13 0.03 ns - - -2.86 - -<br />
Metacognit. respond<strong>in</strong>g 1.33 0.78 - - -3.98 - -<br />
Other activity 1.22 - - - -1.79 - -<br />
Good readers<br />
Retell<strong>in</strong>g 0.27 0.34 - - -2.03 0.57 -1.12<br />
Inferenc<strong>in</strong>g 0.25 0.39 - - -1.53 0.41 -<br />
Problem detect<strong>in</strong>g 0.15 ns 0.57 -0.29 ns 2.24 -1.83 0.37 -<br />
Associat<strong>in</strong>g 0.24 0.36 0.58 ns 5.85 -2.42 -1.38 -<br />
Analyz<strong>in</strong>g 0.36 0.76 - - -3.18 - -<br />
Evaluat<strong>in</strong>g 0.32 0.23 ns - - -2.06 -0.03 ns -<br />
Emotional respond<strong>in</strong>g 0.38 - - - -2.00 0.59 -1.49<br />
Metacognit.respond<strong>in</strong>g 0.48 0.80 - - -3.14 - -<br />
Other activity 0.12 0.12 ns - - -1.69 -0.61 ns 1.36<br />
Parameters that are not statistically significant are <strong>in</strong>dicated with ns; all other parameters are significant at .10<br />
level.<br />
Random model<br />
R1 = between student variance <strong>in</strong> activity <strong>in</strong> the <strong>in</strong>tercept;<br />
R2 = between story variance <strong>in</strong> activity <strong>in</strong> the <strong>in</strong>tercept;<br />
R3 = covariance between <strong>in</strong>tercept and l<strong>in</strong>ear change at student level;<br />
R4 = covariance between <strong>in</strong>tercept and quadratic change at student level.<br />
Fixed model<br />
F1 = <strong>in</strong>tercept;<br />
F2 = mean l<strong>in</strong>ear change <strong>in</strong> activity frequency;<br />
F3 = mean quadratic change <strong>in</strong> activity frequency
Table 1<br />
Cod<strong>in</strong>g scheme of students’ statements <strong>in</strong> the th<strong>in</strong>k aloud transcripts<br />
Read<strong>in</strong>g activity Description<br />
RET<br />
Retell<strong>in</strong>g<br />
INFER<br />
Mak<strong>in</strong>g<br />
<strong>in</strong>ferences<br />
PROB<br />
Detect<strong>in</strong>g<br />
problems<br />
ASSO<br />
Mak<strong>in</strong>g<br />
associations<br />
ANA<br />
Analyz<strong>in</strong>g<br />
EVA<br />
Evaluat<strong>in</strong>g<br />
EMO<br />
Respond<strong>in</strong>g<br />
emotionally<br />
META<br />
Respond<strong>in</strong>g<br />
metacognitively<br />
OTH<br />
Other activity<br />
Retell<strong>in</strong>g, paraphras<strong>in</strong>g or (almost<br />
literally) repeat<strong>in</strong>g the content of a<br />
story passage.<br />
Stat<strong>in</strong>g <strong>in</strong>formation not explicitly<br />
given by the text, fill<strong>in</strong>g <strong>in</strong> gaps,<br />
draw<strong>in</strong>g conclusions, mak<strong>in</strong>g<br />
predictive or explanatory <strong>in</strong>ferences.<br />
Detect<strong>in</strong>g a problem to <strong>in</strong>tegrate<br />
<strong>in</strong>formation; po<strong>in</strong>t<strong>in</strong>g out a knowledge<br />
gap or miscomprehension.<br />
Relat<strong>in</strong>g the story to personal<br />
experiences or knowledge of the<br />
world; giv<strong>in</strong>g associations or personal<br />
op<strong>in</strong>ions; mak<strong>in</strong>g <strong>in</strong>tertextual l<strong>in</strong>ks.<br />
Notic<strong>in</strong>g aspects of form, structure,<br />
style or genre; connect<strong>in</strong>g text parts.<br />
Provid<strong>in</strong>g positive or negative<br />
evaluative comments on (parts of) the<br />
story.<br />
Verbal and nonverbal emotional<br />
responses to (parts of) the story;<br />
exclamations.<br />
Monitor<strong>in</strong>g ones own read<strong>in</strong>g (e.g., by<br />
giv<strong>in</strong>g self-<strong>in</strong>structions) or reflect<strong>in</strong>g<br />
on ones own read<strong>in</strong>g habits or<br />
preferences.<br />
Responses that cannot be placed <strong>in</strong>to<br />
any of the previous categories.<br />
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -16-<br />
Sample responses to The three friends<br />
“So, there were three friends and they<br />
found another one.”<br />
“A palace. So they must be rich.”<br />
“I th<strong>in</strong>k someth<strong>in</strong>g strange is go<strong>in</strong>g<br />
to happen.”<br />
“I don’t understand what this means.”<br />
“What is ‘the wa<strong>in</strong>scott<strong>in</strong>g’?”<br />
“I have two friends myself. Together<br />
we went on a holiday.”<br />
“The three friends rem<strong>in</strong>d me of the<br />
three musketeers.”<br />
“It starts like a fairy tale.”<br />
“The author uses short sentences.”<br />
“I like how this is written (…).”<br />
“A bizarre story. I don’t like it at all.”<br />
Laugh<strong>in</strong>g, sigh<strong>in</strong>g, moan<strong>in</strong>g<br />
“How sad!”<br />
“Yuk!”<br />
“I’ll just have to reread this part.”<br />
“I am pretty used to read<strong>in</strong>g this k<strong>in</strong>d<br />
of stories.”<br />
E.g., read<strong>in</strong>g aloud, comment<strong>in</strong>g<br />
upon the th<strong>in</strong>k aloud task,<br />
<strong>in</strong>comprehensible statements.
Table 2. Example of a coded transcript fragment<br />
Story fragment Student’s response<br />
1. The three friends<br />
2. Once upon a time there were two<br />
friends who found a third. Lik<strong>in</strong>g no<br />
one better <strong>in</strong> the whole world, they<br />
vowed to live <strong>in</strong> one palace, sail <strong>in</strong><br />
one ship, and fight one fight with<br />
equal arms.<br />
3. After three months they decided to<br />
go on a quest. ‘What shall we seek?’<br />
they asked each other. The first said,<br />
‘Gold.’ The second said, ‘Wives.’<br />
The third said, ‘That which cannot<br />
be found.’ They all agreed that this<br />
last was best and so they set off <strong>in</strong><br />
f<strong>in</strong>e array.<br />
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -17-<br />
[Reads aloud] The three friends.<br />
[Laughs]<br />
The three little pigs.<br />
Like those comic characters; Huey, Dewey<br />
and Louie.<br />
[Goes to the second segment]<br />
[Reads silently]<br />
[Laughs].<br />
It is really not a nice beg<strong>in</strong>n<strong>in</strong>g.<br />
[Rereads aloud]. Two friends who found a<br />
third.<br />
They are rich; a palace.<br />
I was th<strong>in</strong>k<strong>in</strong>g [<strong>in</strong>audible words].<br />
[Goes to the third segment].<br />
[Reads silently]<br />
[Et cetera].<br />
Cod<strong>in</strong>g<br />
<br />
<br />
<br />
<br />
<br />
<br />
<br />
<br />
<br />
Table 3. Summary: mean patterns of read<strong>in</strong>g activities of weak (W) and strong (S) readers<br />
No change Increase Decl<strong>in</strong>e Increase<br />
& decl<strong>in</strong>e<br />
Retell<strong>in</strong>g W S<br />
Inferenc<strong>in</strong>g W S<br />
Problem detect<strong>in</strong>g W / S<br />
Associat<strong>in</strong>g W / S<br />
Analyz<strong>in</strong>g W / S<br />
Evaluat<strong>in</strong>g S W<br />
Emotional respond<strong>in</strong>g W S<br />
Metacognitive respond<strong>in</strong>g W / S
(Estimated) Probability<br />
(Estimated) Probability<br />
(Estimated) Probability<br />
(Estimated) Probability<br />
0.30<br />
0.25<br />
0.20<br />
0.15<br />
0.10<br />
0.05<br />
0.17<br />
0.13<br />
0.09<br />
0.05<br />
0.05<br />
0.04<br />
0.03<br />
0.02<br />
0.01<br />
0.20<br />
0.15<br />
0.10<br />
0.05<br />
0.00<br />
Retell<strong>in</strong>g<br />
W W W W W W W W W W W W W W W<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S S S S S S S<br />
1 3 5 7 9 11 13 15<br />
Fragment<br />
Problem Detect<strong>in</strong>g<br />
S<br />
W<br />
S<br />
S<br />
W W<br />
S<br />
W<br />
S<br />
W<br />
S<br />
S<br />
(Estimated) Probability<br />
0.40<br />
0.33<br />
0.26<br />
0.19<br />
0.12<br />
0.05<br />
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -18-<br />
Inferenc<strong>in</strong>g<br />
W W W W W W W W W W W W W W W<br />
S S S S S S S S S S S S S S S<br />
1 3 5 7 9 11 13 15<br />
Fragment<br />
1. Retell<strong>in</strong>g 2. Inferenc<strong>in</strong>g<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W<br />
1 3 5 7 9 11 13 15<br />
Fragment<br />
Analys<strong>in</strong>g<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W<br />
(Estimated) Probability<br />
0.15<br />
0.10<br />
0.05<br />
0.00<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
W<br />
W<br />
W S<br />
W<br />
W<br />
W<br />
S<br />
W<br />
W S<br />
W<br />
WS W<br />
S W<br />
W<br />
W<br />
S<br />
W<br />
S<br />
S<br />
S<br />
1 3 5 7 9 11 13 15<br />
Fragment<br />
3. Problem Detect<strong>in</strong>g 4. Associat<strong>in</strong>g<br />
W<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W<br />
S<br />
1 3 5 7 9 11 13 15<br />
Fragment<br />
Emotional Respond<strong>in</strong>g<br />
S<br />
S<br />
S<br />
S<br />
(Estimated) Probability<br />
0.13<br />
0.11<br />
0.09<br />
0.07<br />
0.05<br />
0.03<br />
Evaluat<strong>in</strong>g<br />
S S S S S S S S S S S S S S S<br />
W<br />
W W W W W W W W W W W W W W<br />
1 3 5 7 9 11 13 15<br />
Fragment<br />
5. Analys<strong>in</strong>g 6. Evaluat<strong>in</strong>g<br />
S<br />
S<br />
S S S<br />
WS W W W W W W W W W W W W W W<br />
1 3 5 7 9 11 13 15<br />
Fragment<br />
S<br />
S<br />
S<br />
S<br />
S<br />
(Estimated) Probability<br />
0.10<br />
0.08<br />
0.06<br />
0.04<br />
0.02<br />
0.00<br />
Metacognitive Respond<strong>in</strong>g<br />
S S S S S S S S S S S S S S S<br />
W W W W W W W W W W W W W W W<br />
1 3 5 7 9 11 13 15<br />
Fragment<br />
7. Emotional Respond<strong>in</strong>g 8. Metacognitive Respond<strong>in</strong>g<br />
Figure 1. Mean probability of occurrence of read<strong>in</strong>g activities, for each ability group<br />
(W = weak readers, S = strong readers)
W<br />
(Estimated) Probability<br />
(Estimated) Probability<br />
(Estimated) Probability<br />
(Estimated) Probability<br />
0.30<br />
0.25<br />
0.20<br />
0.15<br />
0.10<br />
0.05<br />
0.25<br />
0.20<br />
0.15<br />
0.10<br />
0.05<br />
0.00<br />
0.15<br />
0.12<br />
0.09<br />
0.06<br />
0.03<br />
0.00<br />
0.5<br />
0.4<br />
0.3<br />
0.2<br />
0.1<br />
0.0<br />
Retell<strong>in</strong>g<br />
W W W W W W W WS S S S<br />
W W W W W W WS W W W W W<br />
S W W W<br />
S W W<br />
S S S S S S<br />
W W W W WS W WS S<br />
W W W W WS S<br />
W WS<br />
S<br />
S<br />
S<br />
S<br />
W W W WS W W W W W W W W W W W W<br />
W W W W<br />
S<br />
W W W W W W W W W W W<br />
W WS WS W W W W W W W W W W W W<br />
S S<br />
S<br />
S S S S<br />
S<br />
S S<br />
S<br />
S S S S S<br />
S<br />
S<br />
S<br />
S S S S S<br />
S S<br />
S<br />
S<br />
S<br />
S S S S S<br />
S S<br />
S S<br />
S S S S S<br />
S S<br />
S<br />
S<br />
S S S S S S S S<br />
S S S<br />
S S S S S S S<br />
S S<br />
S S S<br />
S S S<br />
S S S S S S<br />
S<br />
S<br />
S S S<br />
S<br />
S S<br />
S<br />
S S<br />
S<br />
S<br />
S<br />
S<br />
S S S<br />
S<br />
S S S<br />
S S S<br />
S S<br />
S<br />
1 3 5 7 9 11 13 15<br />
Fragment<br />
W<br />
(Estimated) Probability<br />
0.40<br />
0.33<br />
0.26<br />
0.19<br />
0.12<br />
0.05<br />
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -19-<br />
Inferenc<strong>in</strong>g<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
W W W W W W W W W W W W W W W<br />
W W W W W W W W W W W W W W<br />
W W W WW W W W W W W WW W W W WW<br />
WS S S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
W W W W W W W W W W W W W W<br />
WS WS S S S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
W W W W W W W W W W W W W<br />
S S S S S S S S S S S S S S S<br />
S S S S S S S S S S S S S S S<br />
S S S S S S S S S S S S S S S<br />
S<br />
S<br />
S S S S S S S S S S S S S<br />
S S S S S S S S S S<br />
S S S S S S S S S S S S S S S<br />
S<br />
S<br />
S S S S S S S S S S<br />
S S S S S S S S S S S S S<br />
1 3 5 7 9 11 13 15<br />
Fragment<br />
1. Retell<strong>in</strong>g 2. Inferenc<strong>in</strong>g<br />
Problem Detect<strong>in</strong>g<br />
W<br />
S<br />
W S<br />
W<br />
W W<br />
S<br />
S<br />
W<br />
S<br />
W<br />
W W<br />
S<br />
S<br />
W<br />
S<br />
W<br />
W W<br />
S<br />
S<br />
W<br />
S<br />
W<br />
W W S<br />
S<br />
WS S<br />
W<br />
WW W<br />
W S<br />
S<br />
W<br />
S<br />
S<br />
W<br />
S<br />
W<br />
W W W<br />
S<br />
S<br />
S<br />
S<br />
W<br />
S<br />
W<br />
W W WS<br />
S<br />
S<br />
S<br />
W WS W<br />
W<br />
W WS<br />
S<br />
S<br />
S<br />
W S<br />
W WS W<br />
W<br />
W W<br />
S<br />
S<br />
S<br />
WS<br />
W<br />
S<br />
WS W<br />
W<br />
W W<br />
S<br />
S<br />
S<br />
W<br />
S<br />
S<br />
W<br />
W<br />
W<br />
W<br />
W<br />
S<br />
S<br />
W<br />
WS<br />
S SS<br />
W WS<br />
W<br />
W<br />
W<br />
W<br />
S<br />
S<br />
S<br />
W<br />
W<br />
S<br />
W WS<br />
W<br />
W<br />
W<br />
W<br />
S<br />
S<br />
W<br />
W<br />
S<br />
W WS<br />
W<br />
S<br />
W<br />
W<br />
S<br />
S<br />
W<br />
S<br />
W<br />
W WS<br />
S<br />
S<br />
WS<br />
W<br />
W<br />
1 3 5 7 9 11 13 15<br />
Fragment<br />
Analys<strong>in</strong>g<br />
(Estimated) Probability<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0.0<br />
W W W W W W W W W W W W W W W<br />
S S<br />
S<br />
S<br />
S<br />
S S<br />
S S S S<br />
S S S S S<br />
S S S S S<br />
S S S S S<br />
S S S<br />
S<br />
S S S S S<br />
S S S S<br />
S S S S<br />
S S S S<br />
S S<br />
S S S<br />
S S S<br />
S S S<br />
S<br />
S S<br />
W<br />
S<br />
W<br />
S S<br />
S<br />
W<br />
S S S S<br />
S S<br />
W W W S S<br />
W S S<br />
S S<br />
S<br />
S<br />
S S<br />
W<br />
W<br />
W W S S SS S S S<br />
S<br />
W<br />
W<br />
W<br />
W<br />
W W W W W W W S<br />
S<br />
W W W W W W W S<br />
W<br />
W<br />
W<br />
W<br />
W<br />
W<br />
W<br />
W<br />
W W W W S<br />
W W<br />
W<br />
W W WS W S W S<br />
W<br />
W<br />
W W W W W W S<br />
S<br />
W<br />
W<br />
W W<br />
W W W<br />
W W W<br />
1 3 5 7 9 11 13 15<br />
Fragment<br />
3. Problem Detect<strong>in</strong>g 4. Associat<strong>in</strong>g<br />
S S S S S S S S S S S S S S S<br />
S S S S S S S S S S S S<br />
W W W W W W W W W W W W W W W<br />
W W W W W W W W W W W W W W W<br />
W<br />
W<br />
S<br />
W S W<br />
S<br />
W<br />
S<br />
W<br />
W<br />
S<br />
W S W<br />
S<br />
W<br />
S<br />
W<br />
W<br />
S<br />
W S W<br />
S<br />
W<br />
S<br />
W<br />
W<br />
S<br />
W S W<br />
S<br />
WW S<br />
W<br />
W<br />
S<br />
W S W<br />
S<br />
W<br />
S<br />
W<br />
W<br />
S<br />
W S W<br />
S<br />
WW S<br />
W<br />
W<br />
S<br />
W S W<br />
S<br />
W<br />
S<br />
W<br />
W<br />
S<br />
W S W<br />
S<br />
W<br />
S<br />
W<br />
W<br />
S<br />
W S W<br />
S<br />
W<br />
S<br />
W<br />
W<br />
S<br />
W S W<br />
S<br />
W<br />
S<br />
W<br />
W<br />
S<br />
W S W<br />
S<br />
W WS W<br />
S<br />
W S W<br />
S<br />
W WS S<br />
W<br />
W<br />
S<br />
W S W<br />
S<br />
W WS W<br />
W<br />
S<br />
W S<br />
W<br />
S<br />
W<br />
W<br />
W<br />
S<br />
W S<br />
W<br />
S<br />
W<br />
S<br />
1 3 5 7 9 11 13 15<br />
Fragment<br />
Emotional Respond<strong>in</strong>g<br />
(Estimated) Probability<br />
0.25<br />
0.20<br />
0.15<br />
0.10<br />
0.05<br />
0.00<br />
Evaluat<strong>in</strong>g<br />
S S S S S S S S S S S S S S<br />
S S S S S S S S S S S S S S<br />
S S S S S S S S S S S S S S S<br />
S S S S S S S S S S S S S S S<br />
WS WS S WS S WS S WS S WS S WS S WS S WS S WS S WS S WS WS S WS WS<br />
S<br />
S<br />
WS W<br />
W<br />
S<br />
WS W<br />
W<br />
S<br />
WS W<br />
W<br />
S<br />
WS W<br />
W<br />
S<br />
WS W<br />
W<br />
S<br />
WS W<br />
W<br />
S<br />
WS W<br />
W<br />
S<br />
WS W<br />
W<br />
S<br />
WS W<br />
W<br />
S<br />
WS W<br />
W<br />
S<br />
WS W<br />
W<br />
S<br />
WS W<br />
W<br />
S<br />
WS W<br />
W<br />
S<br />
WS W<br />
W<br />
S<br />
WS<br />
W<br />
W<br />
W W W W W W W W W W W W W W<br />
W W W W W W W W W W W W W W W<br />
W W W W W W W W W W W W W W W<br />
1 3 5 7 9 11 13 15<br />
Fragment<br />
5. Analys<strong>in</strong>g 6. Evaluat<strong>in</strong>g<br />
S<br />
S<br />
S S S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
W<br />
W<br />
S<br />
W WS WS S<br />
S<br />
S<br />
W S<br />
W WS S<br />
W S<br />
S<br />
S<br />
S<br />
W<br />
S<br />
WWS S<br />
WW W S<br />
WS S<br />
S<br />
W<br />
S<br />
S<br />
W W S<br />
W<br />
S<br />
W<br />
S<br />
S<br />
S<br />
W S<br />
W S<br />
W<br />
S<br />
W<br />
S<br />
S<br />
S<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W W<br />
S<br />
S<br />
S<br />
S<br />
W S<br />
W WS W W<br />
S<br />
S<br />
S<br />
S<br />
W S<br />
W W W<br />
S<br />
S<br />
S<br />
S<br />
W S<br />
W W W<br />
S<br />
S<br />
S<br />
S W S<br />
W W W<br />
S<br />
S<br />
S<br />
S W WS S<br />
W W W<br />
S<br />
S<br />
S<br />
S<br />
W S<br />
W WS W W<br />
S<br />
S<br />
S<br />
S<br />
W<br />
S<br />
W<br />
S<br />
W W S<br />
S<br />
S<br />
W<br />
W WSS W<br />
S<br />
W S<br />
S<br />
W<br />
S<br />
S<br />
W<br />
W S<br />
W<br />
S<br />
1 3 5 7 9 11 13 15<br />
Fragment<br />
(Estimated) Probability<br />
0.3<br />
0.2<br />
0.1<br />
0.0<br />
Metacognitive Respond<strong>in</strong>g<br />
W W W W W W W W W W W W W W W<br />
S S S S S S S S S S S S S S S<br />
S S S S S S S S S S S S S S S<br />
S S S S S S S S S S S S S S<br />
S S S S S S S SS S S S S SS S S<br />
W W W W W W W W W W W W W W W<br />
S<br />
S S S S S S S S S S S S S S S S S S S S S S S SS S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S S S S S<br />
S<br />
S<br />
W W W<br />
W<br />
W W<br />
W<br />
W W<br />
W<br />
W W<br />
W<br />
W W<br />
W<br />
W W<br />
W<br />
W W<br />
W<br />
W W<br />
W<br />
W W<br />
W<br />
W W<br />
W<br />
W<br />
W<br />
W W<br />
W<br />
W W<br />
W<br />
W<br />
W<br />
W<br />
1 3 5 7 9 11 13 15<br />
Fragment<br />
7. Emotional Respond<strong>in</strong>g 8. Metacognitive Respond<strong>in</strong>g<br />
Figure 2. Mean probability of occurrence of read<strong>in</strong>g activities for <strong>in</strong>dividual readers<br />
(W = weak readers, S = strong readers)
W<br />
Flexibility <strong>in</strong> read<strong>in</strong>g <strong>literary</strong> <strong>texts</strong> -20-<br />
Figure 3. Mean probability of occurrence of retell<strong>in</strong>g for <strong>in</strong>dividual readers, each read<strong>in</strong>g<br />
four to five stories (W = weak reader, S = strong reader).<br />
(Estimated) Probability<br />
0.40<br />
0.33<br />
0.26<br />
0.19<br />
0.12<br />
0.05<br />
Retell<strong>in</strong>g<br />
S<br />
S<br />
S S S<br />
WS W W<br />
W<br />
W WS S<br />
W W WS W W WS W<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
W<br />
S<br />
W<br />
W W<br />
W<br />
S<br />
W WS W W<br />
S<br />
W W W<br />
S<br />
WS S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S S<br />
S<br />
W<br />
WS W W<br />
S<br />
W<br />
W<br />
WS W WS W W WS W WS W W<br />
WS S<br />
S<br />
S<br />
S<br />
S<br />
S S<br />
S<br />
WS W<br />
S<br />
W S<br />
W<br />
W<br />
WS W WS S<br />
W W WS W W S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
WS WS W S<br />
W<br />
W<br />
WS W WS W W<br />
S<br />
W W S<br />
W<br />
S<br />
S<br />
S S<br />
S<br />
S<br />
S S S<br />
S<br />
WS WS W W WS W<br />
WS W WS W W W W<br />
S<br />
W<br />
S<br />
W WS W<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S S<br />
S<br />
S<br />
S<br />
W<br />
S<br />
W<br />
W WS W WS W<br />
S<br />
W<br />
W W<br />
S<br />
W W<br />
S<br />
W W<br />
S<br />
S S<br />
S S<br />
S<br />
S<br />
S<br />
S S<br />
S<br />
S<br />
WS WS W W<br />
S<br />
W<br />
WS W<br />
S<br />
W WS W W<br />
S<br />
W W<br />
S<br />
S<br />
S S S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
WS WS W S<br />
W<br />
W<br />
WS W WS W WW W<br />
S<br />
W W W W<br />
WS S<br />
S S S<br />
S<br />
S S<br />
S<br />
S S<br />
S<br />
S<br />
S<br />
WS<br />
WS W<br />
S<br />
W<br />
W<br />
WS<br />
W WS W<br />
S<br />
W<br />
W S<br />
W<br />
W<br />
W<br />
S<br />
S<br />
S S S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
S<br />
WS W<br />
W<br />
S<br />
WS W<br />
W W<br />
WS S S S<br />
S<br />
S<br />
WS W<br />
WS S<br />
W<br />
W W<br />
W<br />
S<br />
S<br />
S<br />
S<br />
S S<br />
W<br />
W WS W<br />
WS W<br />
W S<br />
S<br />
S<br />
S<br />
S<br />
W<br />
WS W W<br />
W<br />
S<br />
S WS<br />
W<br />
S<br />
W<br />
W<br />
W<br />
S<br />
S<br />
S<br />
1 3 5 7 9 11 13 15<br />
Fragment