The Effects of Coherence Explicitness on College Readers' Prose Comprehension

Text Cueing and Comprehension

This is one in a series of studies concerned with the effectiveness of instructional assistance on comprehension of scrambled text. Subjects were General Psychology students at the University of Colorado. Subjects reconstructed one of two 26-sentence passages with the aid of signalling sentences which assisted in determining the original sequence. These sentences were placed in proper order prior to the reconstruction task. A third group of subjects read the scrambled discourse without reconstruction. Our major findings were: (1) recall of idea units was assisted by the use of signalling sentences, (2) recognition of original sentences and paraphrases was also assisted by signalling, and (3) degree of concordance ( tau) between reconstructed order and the original sentence sequence was a function of passage content. The contribution of signalling alone as compared to signalling and feedback suggests that the amount of instructional assistance provided is not necessarily proportional to achievement. There appears to be a complex set of relations among achievement, content, and instructional assistance; these need to be examined further.

Signaling Effects: A Review of the Research—Part I

This article reviews previous research on the effects of certain structural cues, called signals, that affect a reader's comprehension of expository prose. It concludes that the inconsistent results of many studies may be due to inadequate methodologies that have failed to control for confounding variables, such as text length and difficulty, reader familiarity with the topic, and timing of comprehension tests. Further, accepted signal types (headings, logical connectives, and previews) have not been sufficiently examined for their individual effects, perhaps creating unidentified disordinal interactions that could preclude the possibility of researchers identifying significant effects. This article concludes with recommendations for more valid research methodology to be used in prose assessment studies. The next issue of this journal will present Part II of this article, which details a new study of signaling effects for readers of expository prose, a study that is based on the refined methodology suggested in this article.

Reconstruction of Sentence Clusters at the Paragraph Level from Scrambled Discourse Using Feedback

Subjects reconstructed paragraphs from scrambled passages There were two passages of 240 and 400 words and three conditions (read-only, reconstruct with no feedback, and reconstruct with feedback). The major dependent measures were recall of idea units, recognition of original sentences, and concordance with the original sentence in paragraph clusters. Analysis of variance yielded (1) higher recall for idea units of the shorter passage and reconstruct-no feedback condition, (2) higher recognition of original sentences for the reconstruct-no feedback condition, and (3) greater concordance for the shorter passage. Generally speaking, feedback seemed to intrude into the reconstruction process, and greater concordance did not directly lead to improved comprehension.

Effects of Signalling and Feedback on the Comprehension of Reconstructed Text

64 college students reconstructed two scrambled 26-sentence passages, one based primarily on semantic content and the other on episodic content. Assistance was provided in the form of feedback and three signalling sentences. There were statistically significant main effects favoring the semantic passage for recall and for concordance with the original order. There were no effects for signalling. However, a qualitative analysis of sentence clustering suggested that feedback and signalling were differentially more effective for the semantic passage.