COMPARING GOOD AND POOR READERS’ COMPREHENSION
Lucilene Bender de Sousa (PUCRS – Brazil)
Lilian Cristine Hübner (PUCRS – Brazil)
Reading comprehension is a multicomponent process that requires the efficiency of text-based
representation and the construction of a situational model (KINTSCH, 1998). The first highly
relies on word recognition, linguistic knowledge and memory capacity. The second depends
more on world knowledge, integration and inference generation. A good text understanding is
reached when readers are able to represent the mainpropositions in memory but also to
integrate the text content to their previous knowledge and experience. This research aims to
investigate if good and poor readers’ comprehensiondiffers on text-based representation
and/or situational model by assessing their readingcomprehension through literal and
inferential questions in reading and listening comprehension tasks. The participants were 139
students at the 8th grade of public schools recruited from a group of 336 8th graders. They all
accomplished a reading and a listening comprehension task with 6 texts (3 oral and 3 written)
each followed by 5 multiple choice questions, summing up 30 points, 15 for each tasks. The
groups were classified according to their performance in the reading task: good readers had
scores at least 1 standard deviation above the meanand poor readers at least 1 standard
deviation below the mean. After selecting the groups, their performance on literal and
inferential questions was compared. The comparisonswere done by non-parametric statistical
tests, as the variables were not normally distributed. The results showed that both groups were
better answering literal questions as compared to inferential ones on reading (Z = - 6,97;
p = 0,001) and on listening (Z = -2,80; p = 0,001) task. Table 1 presents the means and
standard deviations of the percentage of right answers for each group. Good readers had
significant better performance in literal and inferential questions.
The results seem to confirm the hypothesis that poor readers have difficulty in inference
generation. According to Kintsch’s Construction Integration theory, building a coherent
situational model is a complex task and may be challenging even for good readers. On the
other hand, it was surprising that poor readers also have low performance on literal questions,
which reflects problems on text-based representation. This study showed that Brazilian young
readers with low reading ability even after eight years of formal education still struggle on
basic reading processes as text-based representation. It also shows that poor readers have
difficulty in generating inferences while resolvingreading and listening comprehension tasks.
These results have important educational implications pointing to the necessity of providing
poor readers with instruction on complex as well ason simple reading processes. Moreover,
instruction on inference generation may be extendedto oral text comprehension activities.
REFERENCE
Kintsch, W. (1998). Comprehension: a paradigm for cognition. Cambridge: Cambridge
University Press.
Trabalho apresentado no 3rd Apoll International Psycholinguistics Congress - anais completos disponível em
http://ipcinrio.sciencesconf.org/resource/page/id/18
