Language and Mathematics - Key Factors influencing the Comprehension Process in reality-based Tasks

ABSTRACT

Solving reality-based tasks is an important goal in mathematics instruction and is anchored in education standards determined by mathematical modeling skills. These tasks demand a serious examination of the real-world as well as text comprehension to successfully solve them. Therefore, this study empirically reconstructed the comprehension process during the solution of reality-based tasks and examined how it correlates with process-, person-, and task-related attributes. Fifty-five seventh graders using the Think Aloud Method solved reality-based tasks that were varied in their linguistic and situational complexity level. Their mathematical performance as well as their reading ability were measured. Based on detailed analyses of solution processes combined with the performance data, we point out the relevance of comprehension activities and empirically identify factors that influence the comprehension process.

KEYWORDS:

• Mathematical modelling tasks
• comprehension process
• situation model
• reading ability

Notes

1. This study belongs to the SITRE project (generation of mental situation models and real models in solving mathematical modeling tasks) which is an interdisciplinary collaboration between mathematics education, language research, and empirical education research.

2. Due to adverse circumstances, some written data of two pupils are no longer available, resulting in a reduced sample size for certain analyses.

3. The tasks used are contained in Table 1 and in the appendix.

4. The obvious conjecture that solution processes with a relatively high proportion of comprehension processes are ones in which the learners broke off the task solution after the (first) comprehension process, hence making a correct solution impossible, is not warranted. Just 2% of the solution processes are identified as being broken off without finding an answer.

5. Besides the significant correlation between mathematical performance and correctness of the task solution, the correlation value of .51 shows that the used tasks provide substantially different requirements than the “dressed up“ word problems at the math performance test (see 4.5).

6. Which regressors were selected in each case can be seen from the respective arrows aimed at the model fit values.