to assess reading accuracy (measured in reading errors, one error per word misread), reading speed (measured in terms of the correct number of words read aloud per minute) and reading comprehension (scores on the 4+4 open-ended questions). All the correct answers contained at maximum 14 and 11 units of information respectively. Possible reading comprehension score for the narrative text ranged from 0-8 and for the informational text from 0-11. In statistical analyses the combined reading comprehension score, by addition, was used.
Retrospective interviews. Retrospective interviews were conducted with each of the students, after terminating their reading and responding to the reading comprehension questions. The participants were asked to report on their thoughts while they were completing the reading task (Camps, 2003; Garner, 1987; Ericsson, & Simon, 1984).
Interviews, consisted of ten (10) open-ended questions, were conducted for assessing students’ awareness of the reading tasks, the difficulties encountered in the reading process, their reading strategy use, and their perceptions on abilities and weaknesses (see Appendix 3). More specifically, the first interview questions were directly related to the text (e.g., how a participant came to the decision to choose the answer to a question) and the interviewing proceeded with more general questions (e.g., what problems were encountered in comprehending the text, what they do when they do not understand the meaning of the text during reading, what they think when they do not understand the meaning of the text etc.). What matters in the interview is how the participant came to a certain answer (the process), not her/his correct or incorrect answers (the product).
The participants were asked to comment on their strategic processes during text comprehension—successful and unsuccessful strategies employed, retrospectively (Haynes, 1993; Morrison, 1996). The immediate recall minimizes the possibility that participants may start relying on inferences rather than reporting what happened (Camps, 2003; Ericsson & Simon, 1980). Compared to the ‘on line verbal reports’ (think-aloud reports), retrospective interviews provide more generalizable information than does the concurrent think-aloud technique corresponding more to the awareness concept (Wesche, & Paribakht, 2000). Moreover, retrospective verbal reports, as off-line reports, exclude the possibility of reactivity; they do not interfere with the normal process of reading, as the online think-aloud reports do. However, the retrospective technique has some drawbacks, since participants can rationalize their behaviour after the event or they can fail to recall accurately what they were thinking during the reading (Nisbett, & Wilson, 1977; cf. Ericsson & Simon, 1980, 1984), since the reporting is not concurrent with the processes being described (Ericsson & Simon, 1984).
During the main session, which followed the session of screening test administration, each participant was tested individually in the two reading texts. After completing the first reading, the participants were given a chance to look over the text material for 30 seconds, and were subsequently asked to answer four open-ended questions for comprehension assessing purposes.
Upon the completion of reading comprehension questions, individual semi-structured retrospective interviews were conducted with the 36 students. Interviews averaged 15 minutes. The whole session including the reading tasks, reading comprehension questions, and semi-structured interviews were tape-recorded and transcribed.
Verbal data analysis
The verbal data underwent the following procedures:
a) Data reduction, which involved first and second level coding, resulted in groups of sub-categories, ‘labelled’ by a specific name (Miles, & Humberman, 1994), that were classified into two major categories: cognitive and metacognitive strategies.
b) The classification of cognitive and metacognitive strategies was based on taxonomies provided by Jacobs, & Paris (1987), O’Malley, & Chamot (1990), and Pressley, & Afflerbach (1995).
c) Furthermore, statistical analyses of the verbal data were not restricted to report mere frequencies of cognitive and metacognitive strategies as those frequency scores did not reflect the efficiency of strategies (Veenman, & Beishuizen, 2004). For this reason, following Jacobs, & Paris (1987), Lau (2006), and Pereira-Laird, & Deane (1997), we used quantified coding criteria to assess readers’ strategy efficiency. Each strategy code was rated on a scale ranging from 0 to 2 (see Table 1).