Comprehensibility ratings of concrete and abstract sentences

Bulletin of the Psychonomic Society 1978, Vol. 11 (1) , 49-52 Comprehensibility ratings of concrete and abstract sentences EDW ARD J . ROWE Memorial University of Newfoundland, St. John's, Newfoundland, Canada A IC 5S7 and BRYCE SCHURR and DENNIS MEISINGER University of Alberta, Edmonton, Alberta, Canada T6G 2J8 Comprehensibility ratings are given for 54 concrete and 54 abstract sentences written as actives, passives, and interrogatives. Analysis of the ratings showed that the concrete sentences were easier to understand than the abstract sentences, but all three sentence forms were equally comprehensible. METHOD A number of recent investigations have examined the effect of concreteness on the retention of prose material. In agreement with the results of research using individual words, these studies suggest that concrete, easily imaged sentences or prose passages are remembered better than corresponding material of more abstract content. This difference has been attributed to the greater ease of storing concrete material in a highly memorable form as visual images (Paivio, 1971). However , Johnson , Bransford, Nyberg, and Cleary (1972) have pointed out that concrete sentences, in addition to being more imageable, are also easier to comprehend than abstract sentences. This difference in ease of comprehension, observed with both reaction time measures (Holmes & Langford, 1976; Klee & Eysenck, 1973) and subjective ratings (Johnson et al., 1972 ; Kuiper & Paivio, 1977) confounds an imagery interpretation of retention differences between concrete and abstract material. Consequently, researchers exploring the concreteness factor in sentence memory have attempted to equate concrete and abstract sentences for comprehensibility (Kuiper & Paivio, 1977; Moeser, 1974, 1975). The present report makes available a set of concrete and abstract sentences that have been scaled for comprehensibility in each of three forms : active, passive, and interrogative . While research to date has concentrated on active sentences, these materials will also provide an opportunity to examine concreteness and comprehension factors in memory for passives and interrogatives. Materials The initial pool of items consisted of 108 sentences in the active voice of the form "The (adjective) (noun) (verb) (article) (adjective) (noun)." One hundred of the sentences (SO concrete and SO abstract) were those used by Begg and Paivio (1969) , and details on their construction may be obtained from that study. Four additional concrete and abstract sentences were constructed by a similar method. Each of the sentences, which are listed in the Appendix to this report, was rewritten in the passive and interrogative voice. For example, the active sentence, "The vicious dog attacked the helpless man," was rewritten as "The helpless man was attacked by the vicious dog" in the passive voice and "Did the vicious dog attack the helpless man?" as an interrogative. Minor alterations in the wording of a few of the transformed sentences (Le., substituting "the" for "a") were necessary in order to make them more meaningful as judged by the authors. The complete set of 324 sentences is available on request. Three sets of 108 sentences each were drawn from the pool of 324, with an equal number of actives, passives, and interrogatives in each set. One-half of the sentences of each form were concrete and one-half abstract. The sentences in each set were unique, but each sentence was represented in each of its three versions across all tluee sets. The order of items within each set was randomized for presentation to the subjects, but the alternate forms of each sentence occurred in the same ordinal position in the three sets. Procedure The subjects were given the following instructions: "I'm going to read aloud a number of sentences, one at a time. For each sentence that I read, I want you to decide how easy you think the sentence is to understand or comprehend. You give your answer by circling one of the numbers in the appropriate line on the answer sheet. If you can easily understand the meaning of a sentence, circle 7 for that sentence. If you find a sentence very hard to understand, circle 1. Use the remaining numbers between 1 and 7 to indicate your reaction to sentences that are intermediate in difficulty." The sentences were read to the subjects at a rate of one every 10 sec, with a rest period of about 3 min halfway through the set. This research was supported in part by a grant to Edward J. Rowe from the National Research Council of Canada. Lee Ann Montgomery assisted in the data analysis. Bryce Schurr's present address is Headquarters Business Development Department, Bell Canada, 16th Floor, 220 Laurier, Ottawa, Canada. 49 50 ROWE, SCHURR, AND MEISINGER Subjects Each set of sentences was rated by a group of paid volunteers from undergraduate psychology classes at Memorial University, with 17, 18, and 20 subjects in the three groups. RESULTS AND DISCUSSION The mean comprehension rating for each sentence appears in the Appendix. The mean of these means was calculated for each of the six sentence types. For the concrete sentences, the overall mean ratings were 6.40 for actives, 6.40 for passives and 6.37 for interrogatives, with standard deviations of .62, .63, and .66, respectively. The abstract sentences produced ratings of 5.58 for actives, 5.43 for passives, and 5.49 for interrogatives, with standard deviations of .80, .87, and .86. The mean ratings for the 54 sentences of each type were used as the cell entries in an analysis of variance of rated comprehension. The analysis showed a significant effect of concreteness [F(I,318) = 115, p < .001, MSe = .56], but no signitlcant effect of syntactic form, and a nonsignificant interaction. Product-moment correlation coefficients were also calculated on the ratings for actives, passives, and interrogatives, separately for concrete and abstract sentences. For the concrete sentences, the ratings for actives and passives were correlated .81, actives and interrogatives were correlated .81, and passives and interrogatives .76. For the abstract sentences, the correlation between actives and passives was .68, with a correlation of .74 for actives and interrogatives and also for passives and interrogatives. In agreement with previous research (Johnson et aI., 1972; Kuiper & Paivio, 1977), the concrete sentences 'were easier to comprehend than the abstract sentences. Marschark and Paivio (1977, Experiment 3) have reported some data that seem to be discrepant with these results. They had subjects rate the "meaningfulness" or ease of understanding of a set of sentences of the same form as the abstracts used here, and found no difference between the ratings of concrete and abstract items. However, they also included a sample of anomalous sentences as part of the rated set. These items may have inflated the ratings for abstract sentences and obscured any effect of the concretene (...truncated)