The role of complexity on reading, memory, and appreciation of song lyrics and science exposition.

摘要

The primary goal of this dissertation was to examine whether variations in the complexity of texts, readers, and tasks have a systematic influence on the comprehension, memory, and appreciation of texts. A second purpose of this dissertation was to investigate whether the same relationships exists across different genres of text, specifically song lyrics and science texts.; In two experiments, variations in the complexity of three dimensions of the reading process served as independent variables. These were the number interpretations that readers were instructed to generate per text (i.e., instruction), the cognitive flexibility of the reader (cognitive flexibility), and text clarity (clarity). Three dependent measures were assessed: reading times, memory, and appreciation of the texts.; Four theoretical positions were tested: the encoding variability hypothesis, script theories, the polyvalence assumption, and the construction-integration model. Predictions were derived from these theoretical positions and tested empirically.; The two experiments used a 3 (instruction) x 2 (cognitive flexibility) x 2 (text clarity) factorial design to test the theoretical predictions. Experiment 1 was conducted with song lyrics as stimuli, whereas Experiment 2 tested scientific texts. The three levels of instruction were (1) no instructions on producing interpretations (no-interpretation condition), (2) produce one interpretation (single-interpretation condition), and (3) produce as many interpretations as possible (many-interpretation condition). There were two levels of cognitive flexibility: high versus low. Participants were categorized into high vs. low flexibility based on a composite of three measures: The Adult Torrance Test Abbreviated, Epistemological Belief Scale, and Openness portion of the NEO personality inventory (Neuroticism, Extroversion, and Openness). There were two levels of text clarity: clear versus unclear.; The results revealed that text clarity was the most predictive factor for all dependent variables. This suggests that clear texts that activate a schema are easier to read, facilitate memory by constraining the retrieval possibilities, and are appreciated more. Furthermore, different patterns of interactions were found for song lyrics vs. science texts. The results support predictions for schema-based memory models and the construction-integration model. There was no evidence supporting the predictions for the encoding variability hypothesis or the polyvalence assumption.