Jazmín Cevasco1 & Paul van den Broek2 The role of adversative connectives and causal connections in the recall and recognition of written and spoken discourse Jazmín Cevasco1 & Paul van den Broek2 1National Scientific and Technical Research Council (Argentina) and 2University of Leiden 1. Introduction 3. Results Free Recall A 2 x 4 x 2 analysis of variance that considered connectives version (present-absent), causal connectivity (0-1, 2-3, 6, 13) and modality of presentation (written-oral) as independent variables was run on the recall of the 28 statements that were connected by the selected adversatives in the original transmission. Results indicated an effect of causal connectivity F1 (3, 1896) = 146.997, p < .001 η2 = .189, F2 (3, 96) 49,437, p < .001 η2 = .607, and modality of presentationF1 (1, 1896) = 48.892, p < .001 η2 = .023, F2 (1, 96) = 14,843, p < .001 η2 = .134, but not of connective presence (ps < .7). Those statements that had many connections were more recalled than those had fewer connections (ps <.001). Statements that were read were better recalled than those that were listened to. The only significant interaction was between causal connectivity and modality F1(3, 1896) = 9,988, p < .001 η2 = .016, F2 (3, 96) = 3,251, p = .025 η2 = .092. Those statements that had 6 causal connections and were read were more recalled than those same statements when they were listened to t1(136) = -3,798, p < .001, t2 (6) = -3,398 p = .015. Statements that had 13 connections and were read were also more recalled than when they were listened to t1(67) =-2.678, p = 009, but not by items t2(2) = -1,576 p =.256. Recognition Responses A second 2 x 4 x 2 analysis of variance that considered connectives version (present-absent), causal connectivity (0-1, 2-3, 6, 13) and modality of presentation (written-oral) as independent variables was run on the recognition of the 28 statements that were connected by the selected adversatives in the original transmission. Results only indicated a significant effect of causal connectivity F1 (3, 1894) = 9.486, p < .001, η2 = .015 in the participant analysis, but not in the items analysis (p = .120). Those statements that had many connections were better recognized than those that had fewer connections (ps <.005). Confidence Ratings on Correct Recognition Responses A third 2 x 4 x 2 analysis of variance that considered connectives version (present-absent), causal connectivity (0-1, 2-3, 6, 13) and modality of presentation (written-oral) as independent variables was run on the correct recognition of the 28 statements that were connected by the selected adversatives in the original transmission. Results only indicated an effect of causal connectivity F1(3, 1333) = 4,967, p = . 002, η2 = .011, but not in the items analysis (p = .191). Those statements that had many connections received higher confidence ratings than those that had fewer connections (ps <.013). Causal Connectivity One of the most consistent findings in reading comprehension has been that it involves the reader’s identification of causal connections (van den Broek, 2010). It has also been found that found that listeners tend to recall statements with many causal connections better, and to include them more often in answers to questions about spontaneous spoken materials than statements with fewer causal connections (Cevasco & van den Broek, 2008). Connectives Connectives are words or short phrases (such as because, and, but) that specify how to conceptually link two adjacent statements (Halliday & Hasan, 1976). Among them, adversatives (such as but, nevertheless) signal that the second statement contradicts an expectation raised by the first one. It has been proposed that, by signaling this, adversatives should facilitate the processing of the discontinuity between them (Murray, 1997). Studies on written discourse have provided mixed support for the facilitative effect of adversatives. Shorter reading times were observed for sentences disconfirming an expectation, when they were preceded by an adversative than when they were not (Golding et al., 1995; Haberlandt, 1982; Millis and Just, 1994). Faster responses to comprehension questions were found when an adversative was present than when it was absent (Millis & Just, 1994). Yet, other studies have found no facilitative effect of adversatives on recall (Caron, Micko, and Thuring, 1988; Murray, 1995). This Study The goal of this study was to investigate the role of adversative connectives and causal connections in the comprehension of long spontaneous spoken and written materials. Comparing the comprehension of the same material in both modalities is important, given that spoken and written discourse have differences that could lead to differences in their processing. Among them, written language allows the comprehender to apply strategies such as skimming, omitting sections, and reading in a different order than the text presented (Ferreira & Anes, 1994; Stubbs, 1980). These strategies are not possible with spoken discourse, given its rapid fading (Chafe, 1994). Hypotheses Considering the results of previous studies on written discourse, it was possible that connectives would facilitate recall and recognition of both spoken and written discourse. That is, by signaling the discontinuity between statements, adversative connectives could facilitate their integration both when statements are read and when they are listened to. It was also possible that adversatives would have a greater effect in the comprehension of spoken than written discourse. That is, given that spoken discourse requires listeners to process the discourse at the rate that it is produced and does not allow for them to be re-presented with the statements, listeners may not be as able to establish the contrast between the adjacent statements if the discontinuity is not signaled. With regard to causal connectivity, considering the results of previous studies, statements with many causal connections were expected to be better recalled and recognized than statements with fewer causal connections. It was possible that this effect would be the same for both modalities, given that the number of causal connections that a statement has been shown to exert a powerful effect on its probability of being recalled (Cevasco & van den Broek, 2008; Trabasso & Sperry, 1985). It was also possible that, considering the differences between written and spoken discourse, the effect of causal connectivity would be greater in written discourse than spoken discourse comprehension. That is, given that readers can process discourse at their own rate and re-read sections, they could be better able to establish causal connections than listeners. 2. Method 4. Discussion Participants Sixty-nine undergraduate students from the University of Buenos Aires participated in the study. Materials A 9:30 min segment of a radio transmission on the topic ‘The image of old people in the Middle Ages’. Fourteen adversative connectives in the original version (connectives-present) were eliminated to create an edited version (connectives-absent). Selected adversatives included but, on the contrary and nevertheless. A causal network of the transmission was developed following Trabasso & Sperry (1985) to identify the number of causal connections of each of the originally connected statements. Procedures Each participant was assigned to the connectives-present or the connectives-absent condition, and to the reading or the listening condition. After listening or reading, participants performed a free recall task, and a recognition task with confidence ratings (1= not very confident in recognition response, 3= very confident on recognition response). Consistent with previous studies, these results suggest that spoken and read statements that have more causal connections make a greater contribution to the comprehender’s discourse representation than less causally connected statements. Also, readers seem to perform better than listeners. It is possible that readers’ comprehension is facilitated by the text’s availability, which allows them to process at their own pace and re-read sections, whereas the rapid fading of speech forces listeners to process statements as they are produced (Chafe, 1994; Stubbs, 1980). Given that readers are able to process discourse more thoroughly, they may have more resources available to establish causal connections than listeners have. Adversative connectives did not facilitate recall or recognition of statements in this study. Thus, we cannot establish at this point whether adversatives contribute to the construction of a coherent discourse representation by facilitating the processing of the local connection among the adjacent statements that they link. In summary, causal connectivity and modality of presentation influence discourse comprehension in both reading and listening contexts. Further studies should be run to explore whether these effects would be observed through online measures, and during the comprehension of other spoken & written discourse materials. Excerpt of the Radio Transmission ‘The Image of Old People in the Middle Ages’ in the two Connective Presence Versions (Connectives Present Version) ‘It could be suggested that Christians were very compassionate towards old people in the Middle Ages. In fact, they gave them shelter in their monasteries. NEVERTHELESS, Christian thinkers believed that old people were the representation of sin. The decrepitude of old age was considered the image of sin. Christian writers always consider old age allegorically. The old man was the sinner that needed to be redeemed through penitence. ON THE CONTRARY, young people represented the salvation of Christ. In consequence, sin was as disgusting as old people, and old age was associated to death.’ (Connectives Absent Version) ‘It could be suggested that Christians were very compassionate towards old people in the Middle Ages. In fact, they gave them shelter in their monasteries. Christian thinkers believed that old people were the representation of sin. The decrepitude of old age evoked the image of sin. The old man was the sinner that needed to be redeemed through penitence. Young people’ represented the salvation of Christ. In consequence, sin was as disgusting as old people, and old age was associated to death.’ 5. References Caron, J., Micko, HC & Thüring, M. (1988). Conjunctions and the recall of composite sentences. Journal of Memory & Language, 27, 309-323. Cevasco, J., van den Broek, P. (2008). The importance of causal connections in the comprehension of spontaneous spoken discourse. Psicothema, 20, 801-806. Chafe, W. (1994). Discourse, Consciousness and Time. Chicago: University of Chicago Press. Golding, J.M., Millis, K.K., Hauselt, J., & Sego, S.A. (1995). The effect of connectives and causal relatedness on text comprehension. In R.F. Lorch &E.J. O’Brien (Eds.), Sources of coherence in reading (pp 127-143). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Halliday, M. and Hasan, R. (1976). Cohesion in English. Longman, London. Millis, K. K., Graesser, A. C., & Haberlandt, K. (1993). The Impact of Connectives on the Memory for Expository Texts, 7, Applied cognitive psychology, 317–339. Murray, J. D. (1995). Logical connectives and local coherence. In F. Lorch & D. O’Brien (Eds.), Sources of coherence in reading (pp. 75-94). Hillsdale, NJ: Lawrence Erlbaum Associates. Murray, J. D. (1997). Connectives and narrative text: the role of continuity. Memory and Cognition, 25(2), 227-236. Stubbs, M. (1980). Language and Literacy. London: Routledge & Kegan Paul Trabasso, T. & van den Broek, P. (1985). Causal thinking and the representation of narrative events. Journal of Memory and Language (24), 612-630 van den Broek, P. (2010). Using texts in science education: cognitive processes and knowledge representation. Science, 328, 453-456.