Maintaining text coherence during reading
Hutchens, Scott Alan
In order for a reader to maintain a coherent representation of a text, it is necessary to reactivate previously read information (i.e., backgrounded information). Three experiments investigated how comprehension goals affect the process of accessing backgrounded information during reading. Recently, there has been a considerable debate concerning the process by which readers access backgrounded information. According to two major models (i.e., minimalist and constructionist), a low-level reactivation process or a high-level problem-solving process of search-after-meaning are each different ways in which individuals may gain access to backgrounded information. The current study investigated the two models. Comprehension questions were used to test for differences in reactivation as a function of the comprehension goals of the reader. The results indicated that readers adjusted their comprehension strategies based on the requirements of the reading situation. Readers processed the same text differently according to the type of question that they anticipated they would be required to answer. Readers relied more on low-level reactivation processes based on overlapping memory traces to reactivate backgrounded information in a detail question condition. That is, when readers were reminded of the goal by low-level reactivation through overlapping memory traces, they read sentences describing completion of the goal more quickly (i.e., facilitation effect) and sentences describing the incompletion of the goal more slowly (i.e., inconsistency effect). Probe word recognition times confirmed that goal information was made more available by low-level reactivation through overlapping memory traces (i.e., probe word recognition times were the fastest in the overlapping memory trace condition). However, readers relied more on high-level problem-solving process of search-after-meaning to gain access to backgrounded information in a event-sequence question condition. That is, readers read sentences describing completion of the goal more quickly than sentences describing the incompletion of the goal regardless of the presence of low- evel reactivation through overlapping memory traces. Probe word recognition times confirmed that goal information was equally available regardless of the presence of overlapping memory traces (i.e., probe word recognition times were the same in both the overlapping and non-overlapping memory trace conditions). Future research should focus on the unresolved question brought up in the present study. An experiment should be conducted to test and control for the effect that cognitive load may have had on the participants in Experiment 3. By doing so, the proposed experiment may obtain a more precise measurement of goal availability. In sum, the current study has shown that both models are correct under certain reading circumstances. Future research should focus on determining the everyday reading situations in which each model is true. Future implications are considered.