The focus of our study is to clarify which role does a detailed processing of the question demands have on question-answering processes. Additionally, we are interested in how the previous reading comprehension level may mediate results. Question-answering activities have been widely studied (i.e. Goldman & Durán, 1988; Rouet, Vidal-Abarca, Bert-Erboul & Millogo, 2001). However, little attention has been paid to the effects of comprehending the questions on students’ performance.

Successful question-answering processes start by encoding of the question demands, which will allow other relevant processes to be displayed, such as strategy selection or text inspection (i.e., Graesser & Franklin, 1990; Goldman & Duran, 1988). We could argue that in the question encoding process a mental representation has to be constructed that would enable further question-answering processes. Thus, questions could be regarded as mini-texts in which similar mental processes as in usual texts would be necessary to acquire deep understanding (i.e., Kintsch, 1998).

We could aid the question-encoding process by making students display an additional processing effort to understand questions.  Previous studies may support this hypothesis (i.e.,Chi, Leew, Chiu & LaVancher, 1994). Thus, if we made students explain what the question is asking them for, this extra processing should increase answering success. Complementary, it may be expected that explaining the question should help all students at any previous comprehension level, or only those low-level comprehension students.

Having these theoretical issues in mind, we designed an experiment for 47 secondary school students (16 years old). They were previously classified into good and poor comprehenders with a standarized comprehension test. We asked students to read two texts and answer 10 comprehension questions (5 in each text) presented electronically in a software called Read&Answer. The texts were extracted from Pisa reading assessment materials (Pisa 2000, Flu and Runners). Questions were partially based on Pisa materials, but we made them all open-ended. Students were allowed to inspect the text to answer the questions. To induce question encoding, we made students explain what the question was asking them for, only in one of the texts. The text and questions to be explained were randomized.

We conducted 2x2 Anovas, with independent variables Explain (i.e. Explain question vs. No explain) and Comprehension level (Good and Poor). We analysed the effects of these two variables on: (a) percentage of success in questions; (b) a set of on-line measures reflecting overall time in experiment (in seconds),  answering process (i.e., time per word answering) and text inspection to answer the questions (i.e., percentage of relevant segments visited for the questions).

Results for percentage of success in questions yielded one significant effect for the interaction, F(1,45)= 3.81, p<.05. Interestingly, good comprehenders scored higher (M = 60.87 , S.D = 18.49) than poor comprehenders  (M = 46.29 , S.D = 20.05) when explaining the question. However, similar scores were obtained in good (M = 53.99 , S.D = 21.59)  and poor comprehenders (M = 54.02 , S.D = 16.82)  when not explaining. Therefore, it seems that explaining questions favored good comprehenders, and hindered performance for poor comprehenders.

We also found a significant effect in the two  main variables for overall time in experiment. Explaining questions made the task longer (M = 1614.48 ,S.D = 127.30)  than when not explaining, (M = 1090.93 , S.D =181.21), F(1,45)= 59.28, p<.05; and poor comprehenders needed more time (M = 1461.78, S.D = 343.25) than good comprehenders (M = 1243.63, S.D = 397.15), F(1,45)= 4.71, p<.05. In the time per word answering process measure, we found the interaction effect, F(1,45)= 4.65, p<.05. Whereas good and poor comprehenders answered their questions at a similar rate when they had explained them previously, poor comprehenders answered more slowly than good comprehenders when not explaining. Finally, we found no significant differences for text inspection results.

To analyse in more detail the characteristics of the explanations produced by students, we created a set of processing categories that we expected to appear in students’ explanations, divided each of them in utterances. These were: 1) Literal Processing (whenever  they paraphrased or copied what the question said); 2) Incomplete Processing (the paraphrase lacked information); Misunderstandings (including wrong ideas) and Inferences (including new text-based ideas or elaborations). With these variables as predictors, and percentage of success in questions as predicted variable, we performed one regression analysis. Two results were significant. First, the number of inferences was a positive predictor for success in task, (t = 2.03, p<.05). Second, the number of incomplete processing utterances was negatively associated (t = -2.85, p<.05). Consequently, this demonstrates that there is a relationship between processing activities when reading questions and success in answering.

To sum up, explaining questions favored only good comprehenders, whereas it hindered performance for poor comprehenders.  Only good comprehenders benefited from the explanation instruction and, despite they needed more time to perform the experiment, it was worth it for the final outcome. Probably, the instruction to explain activated strategies that they already know and practise when reading texts, whereas this same instruction may have overloaded poor comprehenders resources, resulting in a decreased performance. Further research should clarify this interaction and help poor comprehenders process questions at a deep level.

REFERENCES

Chi, M.T.H., de Leeuw, N., Chiu, M-H., & Lavancher, Ch. (1994). Eliciting self-explanations improves understanding. Cognitive Science, 18, 439-477.

Goldman, S.R & Durán, R.P. (1988). Answering questions from Oceanography Texts: Learner, Task and Text Characteristics. Discourse Processes, 11, 373-412.

Graesser, A.C. & Franklin, S.P. (1990). QUEST: A model of question answering. Discourse Processes, 13, 279-303.

Kintsch, W. (1998). Comprehension. A paradigm for cognition. Cambridge, Cambridge University Press.

Rouet, J.-F., Vidal-Abarca, E., Bert-Erboul, A. & Millogo, V. (2001). Effects of information search tasks on the comprehension of instructional text. Discourse Processes, 31(2), 163-186.