1 Situational Success at Reading Challenging Texts: Exposing the Fragile Understanding of College Students Emily Fox Patricia A. Alexander Daniel Dinsmore University of Maryland DRAFT 2 Situational Success at Reading Challenging Texts: Exposing the Fragile Understanding of College Students College undergraduates are expected to be able to comprehend and learn independently from college-level texts across relatively unfamiliar academic domains, acquiring new vocabulary and key concepts and procedures, as well as principles of reasoning (Simpson & Nist, 2002). In order to be successful in college, they need to develop understandings of how they should approach these challenging reading situations, how they should proceed while reading, and how they can tell whether they are proceeding effectively or not. How successful are undergraduates at developing these understandings? Can they make the transition to constructing their own knowledge from a variety of texts and across multiple sources, or do they continue to operate with fragile understanding of reading in how they read and in how they learn from what they read? In challenging reading situations, the fragile understandings of apparently successful readers can deteriorate, with that deterioration aligned with cognitive and motivational factors. We suggest that many undergraduates do have such fragile understandings of reading; their success in reading rests upon shaky foundations, due to a passive approach to reading, an over-reliance on background knowledge or personal experience, or a lack of metacognitive flexibility. In this investigation we will explore the factors that may be associated with undergraduates’ fragile understandings of reading as revealed when confronted with a challenging reading situation in the domains of history and of reading; factors of interest will include knowledge, interest, and on-line processing behaviors as seen in think-alouds. Theoretical Framework 3 That readers can gain new knowledge from text successfully presents a bit of a paradox; we know that constructing knowledge from text requires background knowledge, and proceeds more effectively for readers who possess such knowledge. How, then, are those who do not yet have much background knowledge supposed to acquire it by reading? Differences in the scope of readers’ relevant knowledge have been well-established as related to differences in their success in comprehending and learning from text (Ferstl & Kintsch, 1999; McKeown, Beck, Sinatra, & Loxterman, 1992). Readers’ constructive and integrative activity of necessity builds upon and connects to their relevant prior knowledge (Ferstl & Kintsch, 1999; Kintsch, 1998; Kintsch & Rawson, 2005). However, such relevant knowledge has a number of different aspects that might operate differently in reading situations. Relevant prior knowledge for reading can mean general world knowledge including familiarity with particular contexts or situations (Anderson, 2004), specific (often declarative) knowledge about a topic (Garner & Gillingham, 1991; Narvaez, 2002), and the declarative, procedural, and conditional knowledge associated with expertise in a particular domain (Alexander & Judy, 1988; Bransford, Brown, & Cocking, 1999; Feltovich, Prietula, & Ericsson, 2006; Pellegrino, Chudowsky & Glaser, 2001). In addition, relevant prior knowledge can also include knowledge of text conventions or text structures (Goldman & Rakestraw, 2000) or familiarity with the task required by the research protocol (Pressley & Afflerbach, 1995). Yet another aspect of relevant prior knowledge might include the knowledge of how to read, not in the sense of knowing how to decode or even how to comprehend a sentence successfully, but rather knowing how to approach text from a knowledge-building stance (Scardamalia & Bereiter). Such a stance characterizes those who have elsewhere been called 4 competent learners (Alexander, 2005; Alexander & Jetton, 2000) or proficient adult readers, as in this description by the RAND Research Study Group: …proficient adult readers who can read a variety of materials with ease and interest, can read for varying purposes, and can read with comprehension even when the material is neither easy to understand nor intrinsically interesting.” (2002, p. xiii) A similar characterization of mature reading was given by Chall (1983), who identified five stages of reading development. The fifth and most mature stage was usually arrived at during the college years, when readers should become able to read selectively, constructively, and critically to build their own knowledge for their own purposes. College students are expected to be able to read extensively and independently in domains to which they may still be becoming acclimated. Critical factors related to students’ success at meeting college-level reading demands may include their personal theories about the nature of reading and learning, their control of metacognitive processes, including both their knowledge of their own capabilities and also their self-regulation while reading, and their flexible and appropriate engagement in cognitive sense-making processes (Simpson & Nist, 2002). Personal theories may relate to whether students approach reading in general as active or passive participants (Alexander & Jetton, 2000; Schraw & Bruning, 1999). Students may also develop more domain-specific understandings of how to read texts appropriately for constructing knowledge within particular domains (Alexander & Jetton, 2000; Chall, 1983; Scardamalia & Bereiter, 1991; Wineburg, 1991). Personal theories call into play additional factors related to engagement, including interest and motivation for reading and for the task at hand (Schraw & Bruning, 1999), which can also be linked to their level of expertise related to the domain about 5 which they are learning (Alexander, 1997, 2003) and therefore to their knowledge as well (Alexander, 1997, 2005; Chall, 1983). Along with an appropriate theory or understanding of reading, students must have the necessary skills and knowledge available to them. They must have control of their metacognitive processing and be able to employ an appropriately flexible range of cognitive processes as they read (Simpson & Nist, 2002). The Highly Competent readers profiled by Alexander in her discussion of reading development (2005) appear to have this combination of elements. However, Alexander also identified possible profiles of readers whose success in reading rests upon a shaky foundation: Knowledge Reliant Readers and Effortful Processors. Knowledge Reliant Readers may often be able to make sense of text by connecting to and building upon their existing background knowledge, while Effortful Processors are willing to put forth considerable effort at strategic processing, while not necessarily having automated some aspects of their reading or having the metacognitive flexibility to adjust their selection of strategies in the face of difficult texts or tasks. In reading situations that challenge their competence, these apparently successful readers may be unable to perform well, while Highly Competent Readers rise to the occasion. In this study, the nature of undergraduates’ possible fragile understandings of reading was investigated. A number of factors were hypothesized to be related to students’ success in learning from text, including interest in the domain being read about, broad background knowledge of the domain as well as detailed knowledge of the topic being discussed in the text, and reading approach. Aspects of reading approach expected to be of interest included level of engagement, effectiveness of monitoring, scope of connections generated, flexibility and appropriateness of strategy use, and congruence with the text. Undergraduate participants read 6 moderately challenging passages related to reading research and history, and thought aloud while reading, as an on-line measure of strategy use, connection to prior knowledge, interest, and effort. Passages were chosen to be accessible yet likely to be unfamiliar for participants, covering psycholinguistics (reading) and slavery in colonial Virginia (history). Pre-tests (multiple choice analogies and sentence completions) on domain and topic (passage-related) knowledge, and posttests (open-ended essay questions) on how well the students remembered and understood each passage were administered. We expected that students would approach both passages similarly, that is, that they would have a relatively stable approach to reading in this situation. We expected to see processing profiles resembling Highly Competent Readers, who would perform successfully for both passages, as well as processing profiles revealing more fragile understandings, and resembling Knowledge-Reliant Readers or Effortful Processors. These latter readers we thought would be likely to have better success on one passage than the other, either because they were able to connect more successfully to their own knowledge or experience, or because their strategic repertoire worked more appropriately in one situation than the other. We also expected that there might be some students who did not or could not engage successfully in learning from either passage. We thought interest might play a small role in students’ level of engagement as they read, bearing in mind that this was a relatively artificial reading situation. Overall, we anticipated that interest in the text and in the task, level of effortful engagement directed at global rather than local comprehension, and successful activation of relevant background knowledge would differ for students who were successful at reading both passages, for those who succeeded at one, and for those who had no success. 7 Attention to the nature of undergraduates’ understandings of reading is critical. Many undergraduates may be plodding through their required college reading as Knowledge Reliant Readers or as Effortful Processors; these approaches, while they may be situationally successful, are developmentally impoverished, and are likely to break down in more challenging reading situations. In addition, these approaches seem likely to prevent students from experiencing actual engagement while learning from text, and thus have implications for their future interest in learning from text and their ability to become lifelong and independent learners. If learning from text is always experienced as effortful and unrewarding, or as a scramble to connect to whatever knowledge is available, no matter how appropriate, why would students ever choose to do it on their own? Methods Participants Our participants were 15 college undergraduates (9 female, 6 male) drawn from three highly ranked four-year schools in mid-Atlantic states. Two participants were volunteers, and the remaining 13 participated for extra credit in their Research Methods class. (An additional two participants from the Research Methods class were excluded from this analysis based on their age; at 33 and 45 years old, they were significantly older than the remaining participants and fell outside the developmental bracket intended to be tapped by sampling undergraduates.) Participants’ ages ranged from 17 to 23 years old, with an average age of 20.5. Ten students identified themselves as European American, one as Hispanic American, two as Asian/Pacific Islander, one as White, and one did not specify an ethnicity. They had completed, on average, 2.9 years of full-time study (ranging from a low of 1 year to a high of 5 years), and had taken, on average, 1.5 college-level courses in reading (ranging from 0 to 3) and 1.8 in 8 history (ranging from 0 to 6). Their identified majors included: urban/environmental planning; liberal arts; elementary education; economics; criminal justice (2); public health (2); journalism; finance; psychology; family studies (2); and community health (2). Measures Demographics. Participants completed a demographics questionnaire giving background information on gender, ethnicity, age, current educational level, major, and number of collegelevel courses in reading and history. Interest. A 20-item interest rating questionnaire assessed interest in the academic domains of science, reading, history, and art. Responses were on a 1 to 5 scale, where 1 was “Strongly Agree” and 5 was “Strongly Disagree.” A sample item for history was, “History is personally important to me.” There were five items for each academic domain, and items were parallel in construction. Two items for each domain were negatively phrased and reverse coded. Reliability for the five-item reading interest subscale was good, with a Cronbach’s alpha of .90; reliability for the five-item history subscale was lower, but still satisfactory, with a Cronbach’s alpha of .77. An additional interest measure assessing frequency of participation in activities was also completed as part of another study, but was not used in these analyses and will not be discussed further. Knowledge. Participants completed four knowledge measures assessing the breadth of their domain knowledge and depth of their topic knowledge in reading and in history. The knowledge measures were validated by a history and a reading expert for appropriateness of domain and topic representation and accuracy of answer choices. These knowledge measures were originally designed for a broader cross-sectional study including domain experts and 9 younger participants as well, and were intended to span a wide range of possible levels of knowledge. The domain knowledge measures consisted of 20 multiple choice items in which participants were asked to choose the response best completing the given analogy. Response choices represented four possible tiers of knowledge (Alexander, Murphy, & Kulikowich, 1998): the correct response that best completed the analogy (4 points), a response that related to the topic but was not the best answer (2 points), a response that came from the domain but was not related to the topic of the analogy (1 point), and an unrelated response (0 points). A sample item for reading was: “phonology : word pronunciation :: morphology : _______ ,” where response choices were: word meaning (4 points); word identification (2 points); word attack (1 point); word processing (0 points). Reliabilities for the domain knowledge measures were low, at α = .46 for reading and α = .52 for history, but this was not unexpected given the graduated response scoring, the low levels of knowledge undergraduates might be expected to have in these domains, the relative lack of variability within this group, and the untimed test administration.. The topic knowledge measures consisted of 15 multiple choice items in which participants were asked to choose the response best completing the sentence stem. Response choices were graduated for level of knowledge represented, as in the domain knowledge measures. The topics addressed were psycholinguistics and slavery, the topics of the passages that would be read. A sample item for the history topic knowledge measure was: “In the United States Constitution of 1787, a provision concerning slavery is the ______,” where response choices were: three-fifths clause (4 points); prohibition of slave trade (2 points); XIII amendment (1 point); Westward clause (0 points). Reliabilities for the topic knowledge measures were low, at α = .23 for reading and α =.25 for history; again, this was not unexpected. One item was 10 dropped from each topic knowledge measure as poorly performing, leaving a total of 14 items that were used in these analyses. Learning outcomes. Learning outcomes in this study were assessed by two open-ended questions completed from memory after reading each passage: (1) Write down as many important points as you can remember from this passage; (2) Write down the argument the author is making in this passage. (Participants were also asked to complete two additional openended questions regarding what else they recalled from the passage and what they thought about the author’s argument; these questions were intended for use in another study and are not included in these analyses.) Participants were given a full blank sheet of paper for each question. The learning outcomes produced by our undergraduate participants were scored by comparison to the performance of competent adult readers. Four middle school teachers whose subject areas included mathematics, French, Spanish, and English read the reading passage and responded to the two outcome questions (with the passage available), and three of them similarly responded to the history passage. These responses were then pooled and used as the standard against which the undergraduates’ responses were evaluated, allowing a broad range of possible successful responses. The first author scored the undergraduate responses for the extent to which they accurately reproduced any important points and argument statements that were present in the standards and arrived at a global judgment of successful or unsuccessful comprehension for each passage. The outcome responses were similarly scored by the third author independently, and a 74% match between the scores resulted. Upon discussion, only outcomes that both coders had identified as demonstrating understanding of the text ended up being scored as successful learning. Texts and task 11 Passages. The reading task used in this study involved texts from the domains of reading and history that were selected and adapted to be as parallel as possible. The texts chosen were expected to address relatively unfamiliar topics for which undergraduates would not have extensive supporting background knowledge. Each of the passages chosen indirectly presents an argument. In the reading passage, the argument is that skilled reading does not occur primarily through the application of grapho-phonemic information or decoding, presented in the context of an explanation of the psycholinguistic model of reading. In the history passage, the author argues that the slavery system based on race evolved partially as an attempt to address the class struggle that was mounting in the colonies; although this interpretation is directly stated, the passage is not constructed explicitly as an argument for this thesis. The selected passages were intended to present similar levels of difficulty for readers in terms of length, structure, and reading level, and to be relatively challenging for undergraduate readers. The reading passage was taken from an article on practical applications of a psycholinguistic model of reading (Pearson, 1978) that appeared in a book targeted for reading researchers, graduate students, or teachers. In its adapted form, it was 1173 words long with two footnotes, with a Flesch-Kincaid reading level of 12.0, and a Flesch reading ease score of 40.5 (from Microsoft Word). Although not from a textbook, the history passage was taken from a text on U.S. history (Takaki, 1993) usually used at a college level, professionally written and appealing also to an adult audience interested in history. In its adapted form, it was 1393 words long with six footnotes, and had a Flesch-Kincaid reading level of 12.0 and a Flesch reading ease score of 38.5. The adaptations made were all excisions, in order to create a self-contained passage of a reasonable length. 12 Task. Participants were asked to think aloud while reading each of these passages, and their think-aloud was recorded on an audiotape. They were told to verbalize what they were thinking and doing as they read each passage, and practiced thinking aloud on a short passage before moving on to the reading and history passages. They were told that when they finished reading the passage, they would be asked to respond to open-ended questions regarding what they remembered and what they thought about the passage. These think-alouds were intended to provide data on strategy use, interest, and overall engagement with the reading activity. Coding. All 30 think-alouds (two for each participant) were coded for strategic and evaluative/monitoring behaviors by the first author. A set of 24 codes was developed and proved to be adequate to capture the range of behaviors demonstrated; these codes are given in Appendix A. After coding all 30, the quality of each participant’s processing as revealed in the pair of think-alouds was then evaluated by the first author in terms of the overall approach demonstrated for seven aspects: monitoring (effective vs. ineffective); making connections (successful or unsuccessful); strategic repertoire (flexible vs. limited); level of processing (global vs. local); literal comprehension (accurate vs. inaccurate); interpretation or elaboration (faithful vs. unfaithful); and level of engagement (high vs. low) (Pressley & Afflerbach, 1995) . The guiding questions used when making these evaluations are given in Appendix B. The third author similarly coded and evaluated six randomly selected pairs of thinkalouds, which had been were blinded to mask the participants’ ID code and therefore any link to their performance status on the outcome questions. After comparison of the codings for the overall quality of processing, two categories were eliminated, as being too unstable: making connections and monitoring. The level of agreement for the remaining categories was 97%, and the single difference on one code was resolved by discussion. The line by line codings for 13 strategic and evaluative/monitoring behaviors for three pairs of the think-alouds were compared (a total of 125 coding decisions). Agreement reached 78%, with the majority of the differences being due to consistent differences in applying two pairs of codes: questioning vs. evaluating comprehension and interpreting vs. restating the text. In this case, the first author’s codings were retained. Procedures Participants were told that their participation would involve two components. In the first part, participants were given, in order, the demographics questionnaire, interest questionnaire, activities questionnaires, domain knowledge and topic knowledge measure for either reading or history, and then the domain and topic knowledge measures for the other domain. Administration of the knowledge measures was counterbalanced such that nine undergraduates saw the reading knowledge measures first, while the other six saw history first. There were no time limits for completion. Participants were encouraged to respond to every item on each measure, and for the knowledge measures, were told that guessing was okay. The average time for completing the first part was roughly 45 minutes. They could then take a brief break before beginning the second part. In the second part, participants practiced thinking-aloud with a short passage on mosquitoes, from a popularly written science article by Marston Bates (1975) originally published in Natural History magazine. This passage was intended to be mildly interesting but not too challenging, so that participants might naturally find themselves responding to the topic and descriptions. After practice, participants then read and thought aloud for either the reading or history passage. This thinking-aloud was audiotaped. Participants were instructed that they could either read the text itself aloud or not, as they chose, while they were reading and thinking aloud. 14 Participants were also told that if they remained silent for an extended period, they would be prompted to say what they were thinking. When participants indicated that they were finished reading the passage, the passage was taken away, and they were given the packet of outcome questions. They were told that they could complete the questions in any order and look back and forth between their responses on the questions if they wished. Upon completing the outcome questions, the participants were again reminded of the think-aloud directions and given the other passage, following the same procedure as outlined above. Administration of passages was counterbalanced such that those participants who saw the reading knowledge measures first also saw the reading passage first, and the same for history. There was no time limit for these tasks. Participants tended to take very nearly the same time for each passage, and the average time for the second session was again roughly 45 minutes. The think-alouds were transcribed into text documents by the first author. Results Outcomes The undergraduates’ success rates on the learning outcomes assessing their comprehension of the passages they had read were below chance. The two consistently successful participants were successful on both passages. Of the five situationally successful participants, two performed well on just the reading passage, and three on just the history passage. The remaining participants were consistently unsuccessful on both passages. Table 1 shows how participants performed by type of passage and by order of passage presentation. (All names used are pseudonyms.) In particular, their rate of success on the first passage they read, 2 out of 15, was significantly below chance, χ2 (1) = 8.07, p < .01. The five participants who 15 were situationally successful all demonstrated higher comprehension on the outcome questions for the second passage that they read. Knowledge and interest Overall, undergraduates’ scores on the domain and topic knowledge measures for both reading and history were quite low. For reading, they averaged 35.6 for domain knowledge (out of a possible 80) and 30.3 for topic knowledge (out of a possible 56). Their average scores for history were 36.7 for domain knowledge (out of a possible 80) and 29.5 for topic knowledge (out of a possible 56). Their average interest was slightly positive for both domains, at 3.32 for history and 3.59 for reading. Tables 2 and 3 show participants’ mean knowledge and interest scores overall and grouped by their performance on the outcome measures. Readers who were successful on the reading passage, on average, appeared to differ little from unsuccessful readers in terms of domain and topic knowledge and in their interest in reading as a domain, as seen in Table 2. Readers who were successful on the history passage, on average, had higher scores for domain knowledge (with a mean of 42.0) than did the unsuccessful readers (with a mean of 34.1). They also had higher interest, on average (with a mean of 3.60), than did the unsuccessful readers (with a mean of 3.18). The average scores for history topic knowledge did not differ markedly between the successful and unsuccessful readers. Examination of the knowledge and interest scores of the five participants who were successful on one passage but not the other, which are given in Table 4, reveals no consistent pattern of successful performance related to higher domain or topic knowledge or interest scores. On-line processing 16 From the detailed coding of participants’ think-aloud transcripts, it appears that overall, participants tended to adopt a relatively similar approach to both passages they read. They tended to use few strategies, with the most popular being re-reading, reading aloud, and restating the text. There appeared to be a gender difference in the adoption of the strategy of reading the entire passage aloud, with five of the six participants who read both passages aloud being female. Readers’ level of engagement and effort was relatively similar across both passages, although there were several participants who commented that they found one passage to be more interesting than the other; for most of these, the history passage was the more interesting. Several readers also noticed a connection between the earlier topic knowledge test and the passage they were reading. No readers made comments indicating that they were adjusting their processing when reading the second passage as a consequence of having completed the outcome measures for the first passage. From the evaluation of the overall quality of each reader’s processing, it appears that the two consistently successful readers differed in a number of ways from the other participants, as shown in Table 5. The consistently successful readers made global connections, had a flexible repertoire of strategies, restated the text accurately, developed faithful interpretations or elaborations of the text, and showed a high level of engagement; they resembled Alexander’s profile of the Highly Competent Reader (2005). The five situationally successful readers tended to process locally, to have limited strategies, and to have a low level of engagement. The consistently unsuccessful readers were generally similar to the situationally successful readers, with the possible additional characteristics of restating the text inaccurately or developing unfaithful interpretations or elaborations. 17 From the detailed codings of the think-alouds, it appears that the quality of the connections readers made to their personal experience and to their background knowledge also played a part in their success in constructing an understanding of what they were reading. In order to explore more closely what was going on in the processing of the situationally successful and consistently unsuccessful readers, brief descriptive reports for selected participants will be presented: a situationally successful reader (Alice), and three consistently unsuccessful readers (Ann, Dave, and Kate) who displayed different processing patterns (see Table 5). Descriptive reports Alice – connecting to experience. Alice was a 22 year old European American, who had completed three years of full-time study. Her major was in public health, and she reported having taken one college-level course in reading and three in history. Her scores for reading knowledge were close to the mean for both domain knowledge (35) and topic knowledge (28), and her reported interest in reading was very high (4.8). Her scores for history knowledge were 40 for domain knowledge and 28 for topic knowledge, and her reported interest level in history was 3.4. She read the history passage first, and did not perform successfully on the outcome questions; she did succeed in identifying several of the author’s major points and in stating a version of his argument for the second passage she saw, the reading passage. Alice chose to read both passages aloud. As she read the history passage, she made connections to visual imagery (“I’m picturing Virginia.” “I’m picturing someone with a dirty face.” “I’m picturing that.”). She also drew associations to current events (“I was thinking about, uh, Latin America.” “Which makes me think of current-day stuff.” “This makes me think of an article in the Washington Post today.”) and to her personal experiences (“Which makes me think about how I want a beach house, but it’s not possible, because all the wealthy people around here 18 are taking them.”). None of these associations appeared likely to help her construct an understanding of the passage. When she came to something she didn’t understand, she re-read it. For unfamiliar words or phrases, she either guessed the meaning or just read on. As she read the reading passage, she made a few irrelevant associations to personal experience (“Flax reminds me that I just, um, started reading about that in something else.” “I just, um, remembered reading to a school in Louisiana, randomly.”) However, as she progressed through the passage, she began to connect the author’s description of reading behaviors with her own experiences of reading (“And I’m thinking about hard reading that I’ve had, where I don’t really understand everything, but I try to use the clues, ‘cause I think this is kind of what it’s referring to.” “Which, I don’t think I use that technique.” “Yes, I have used that technique.” “This makes me think of the times where I read, and I don’t get anything, and the times when I read, and I feel in the zone.” “Which I can relate to, with textbooks.” “And I’m thinking that this is an unfamiliar text, and that is why I am struggling with it.”). She appeared to resemble Alexander’s profile of the Knowledge Reliant Reader (2005). The difference between her processing on the two passages appears to lie in the relevance of the connections she made, rather than in a shift in the types of behaviors she engaged in. She used relatively few other strategies, primarily re-reading and guessing the meaning of words in context, and used them about equally for both passages. From what the think-alouds show, it is not possible to determine whether she was deliberately focusing her connecting activity toward relevant connections for the reading passage or whether it just happened that way. Although she made no statements of interest for either passage, her reported high interest in reading may also have been a factor in her more focused connections in processing the reading passage. 19 Ann – rereading and repeating. Ann was a 19 year old who did not identify her ethnicity. She had completed two and a half years of full-time study. Her major was in elementary education, and she reported having taken three college-level courses in reading and six in history. Her scores for reading knowledge were high for both domain knowledge (49) and topic knowledge (42), and her reported interest in reading was also relatively high (4.0). Her score for history domain knowledge was quite low (26) and her reported interest level in history was on the negative side (2.6). Her topic knowledge score for history was 36, which was quite high. She read the reading passage first, but did not perform successfully on the outcome questions for either passage. Ann chose to read both passages aloud. She underlined as she read (“I need to underline all this, ‘cause I will forget.” “I need to underline this.”) for both passages. She frequently reread, sometimes more than once, and reiterated phrases or paraphrased sentences after reading them. Although her level of engagement was low, she was reading effortfully, and resembled the profile of Alexander’s Effortful Processor (2005). She seemed to have more difficulty reading the reading passage, and became discouraged (“How much more of this is there?” “I don’t even know what this says anymore. I’m not really paying attention to it.”). When she got to the end of the reading passage, she looked back over it and skimmed through, repeating what she took to be key words and phrases. As she neared the end of the history passage, she ventured beyond restatement and twice elaborated on what she had read (“So the difference between the whites and blacks is that, what they’re saying anyway, is that the whites had a goal. They had a goal of them bettering themselves. But the blacks apparently didn’t show that goal? Or didn’t let people know that they didn’t want to be slaves?” “So, in order to avoid having a bunch of mad white indentured servants, you could possibly raise up the 20 black slaves as well and get them thinking that they need more rights, just give the whites what they want, and keep on bringing the blacks in, so you can get more and more slaves, and that way you can solve the race issue, and then you can create classes among the white people.”) Her elaborations were not faithful to the meaning of the passage, and were not likely to help her comprehend successfully. That she elaborated for the history passage and not the reading passage might have reflected her awareness that she would be asked to give the main points and argument when she had finished reading; she did not overtly state this as a goal, however, and her elaborations did not attempt to build a global understanding of the passage, but were restricted to the immediate content she had read. Dave – undirected associations. Dave was a 23 year old European American, who had completed four years of full-time study. His major was in criminal justice, and he reported having taken two college-level courses each in reading and in history. His score for reading domain knowledge was above the mean (39), but his score for reading topic knowledge was among the lowest seen (22), and his reported interest in reading was on the negative side (2.4). His scores for history knowledge were close to the mean, at 38 for domain knowledge and 30 for topic knowledge, and his reported interest level in history was 3.4. Dave read the history passage first, and did not perform successfully on the outcome questions for either passage. Dave brought to bear virtually no strategies as he read both passages; he did not re-read or restate any of the content. He commented only twice on whether he was understanding what he read (“I think I, like when they say, council of foreign plantations, like, I know there’s a lot of them, and I get them all kind of mixed up, uh, so it doesn’t really mean anything to me.” “I’m thinking, uh, in a couple of these sentences, the words don’t seem to make any sense. Sleefary and arfranned.”) 21 As he read both passages, Dave commented often on how interesting he was finding what he was reading (“It seems interesting that, uh, two, like, two, it says here two white men and a black man, uh, ran away together, and that just seems interesting to me, because I, I didn’t know that they, they were equal back then.” “It’s interesting to see that, um, there’s so much that is involved with reading, so just decoding and stuff, and other examples, but it just seems very interesting to me.”) He commented on the quality of the text for both passages, noting for the history passage that “this writer does a, uses a lot of descriptive words and adjectives,” and for the reading passage that it “has a lot of big words in it.” As he read both passages, Dave tended to generate irrelevant associations (“When it says de facto, I think of, uh, like a time of, a time of war, for some reason.” “I’m thinking, uh, in a perfect world, uh, this wouldn’t go on, but, uh, but, by no means is it perfect.” “I’m thinking of a recent movie I just saw, where the main character, uh, suffered from dyslexia.” “I’m thinking, search, search and destroy, it kind of reminds me of playing a video game when I was in high school.” “When it says b, or, when it says translate to buh, or other, or the other b, I’m thinking my bird could say, my old bird, before he died, could say b words. That’s about it. That’s all he could say. Words that started with b.”) From what can be observed in his think-alouds, Dave did not appear to be reading for understanding. Kate – constructing away from the text. Kate was a 19 year old Asian/Pacific Islander American, who had completed one and a half years of full-time study. Her major was in psychology/pre-med, and she reported having taken no college-level courses in reading and one in history. Her score for reading domain knowledge was at the mean for (35), but her score for reading topic knowledge was quite high (38); her reported interest in reading was near the mean (3.8). Her score for history knowledge (40) was above the mean, as was her score for history 22 topic knowledge (35). Her reported interest level in history was 3.4. Kate read the history passage first, and did not perform successfully on the outcome questions for either passage. As she read, Kate’s dominant behavior for both passages was restating the text content or elaborating on it. She tended do local summaries as she went along, and appeared to be quite engaged in building an understanding of the passages at that level. However, her restatements were frequently inaccurate and her elaborations were unfaithful to the content. (“So they’re pointing out race difference between, in slavery, where English colonists first started as indentured servants, but Africans were just automatically slaves.” “So when they, the Negroes were picked up from Africa, they were taken, and they were never told how long they’d serve, but then it ended up being a life sentence.” “And then the white freemen also experienced the same prejudice as the black freemen.” “When you’re reading about meaning, it’s more useful in remembering than just reading, say, the physical appearance of the word.” “We have many schemas of thought, and so they interact in different ways, and apply themselves in different situations.” “So, even at a young age, as long as you define a word as something, you can understand the relationship between it and other things, even if it’s not real. But that’s strange, because first graders can’t understand abstract logic yet.”) Kate commented only once regarding her understanding of the text (“I’m not sure what de jure means.”), and gave no evidence of rereading, although she did skim and read aloud a few phrases in the middle of the reading passage. She did appear to be reading for understanding, but did not link the understanding she was developing to what the passage actually said. Summary. The type of connections readers made to their knowledge and experience and the degree to which they attend to the text in making restatements or elaborations appeared to matter. Relevant or purposeful connections to prior knowledge or experience appeared to be 23 more productive than irrelevant or random connections. Accurate restatements and faithful elaborations or interpretations appeared to be more helpful in building understanding than inaccurate restatements and unfaithful elaborations. Discussion This investigation targeted the understandings of undergraduates having more or less consistent success in reading challenging texts in reading and history. We anticipated that students would have variable rates of success in reading and learning from these texts. Surprisingly, the majority of the students were unsuccessful at reading both passages. Success on the outcome measures was defined as reproducing the gist of several of the important points identified by our comparison group along with successfully identifying one of the themes identified by the comparison group as part of the author’s argument. Some of the unsuccessful students were unable to repeat even one important point from the passage they had just read. Only two students were successful at reading both passages, while five students were successful at reading one of the passages. A number of factors were hypothesized to be related to students’ success in learning from text, including interest in the domain being read about, broad background knowledge of the domain as well as detailed knowledge of the topic being discussed in the text, and reading approach. We did not see consistent patterns suggesting that students’ performance differed in relation to their levels of domain interest, domain knowledge, or topic knowledge for the reading passage. It did appear that students who were successful on the history passage had higher scores on history domain knowledge, on average, and higher levels of interest in history. This might suggest that they had greater experience with history. The number of college-level history courses taken, however, was slightly higher for the unsuccessful (1.9) than the successful (1.6) 24 readers of the history passage. Of the two forms of knowledge we measured, topic knowledge would appear to be more likely to be relevant for constructing knowledge of the passage, and unsuccessful readers had slightly higher topic knowledge for history than did the successful readers. Our investigation of the role of these factors was limited by the small size of our sample; that we did not see consistent patterns could not be taken as evidence that such patterns would not emerge with more participants. We did see, and had not anticipated, that order of presentation of passages made a difference to students’ success in learning, in that all five of the students who were situationally successful succeeded on the second passage they read. We had anticipated that students would approach both passages similarly, with differential success expected to be related to whether their knowledge happened to align better with one than the other, for Knowledge Reliant Processors, or whether their suite of processing behaviors happened to work better for one than the other, for Effortful Processors. And this was pretty much what we saw. Students in general did approach both passages similarly. It was difficult to tell, from looking at the processing behaviors of the situationally successful readers, what was going on here. There were no overt behaviors or comments indicating that students had adjusted their processing to align with the outcome task, or that they were aware that the outcome task would be the same for the second passage. Knowledge and interest as measured by our pre-tests did not appear to be a factor. Here a post-reading survey or interview that asked about students’ purpose for reading might have been informative. Aspects of reading approach expected to be of interest included level of engagement, effectiveness of monitoring, scope of connections generated, flexibility and appropriateness of strategy use, and degree of congruence with the text. We did see that the two consistently 25 successful readers, Mike and Andy, differed from the rest in their high level of engagement while reading; although they were putting out effort, their processing did not seem to be experienced as effortful or laborious. They actively constructed meaning as they went, building their own connected understanding of the entire text, while remaining faithful to the author’s content and message. Mike, in particular, moved non-linearly through the text, jumping ahead and looking back with a great degree of freedom. They both used a variety of strategies fluidly and appropriately, and were not locked into a routine of re-reading or repeating what they had just read. They appeared to be monitoring effectively, and building toward a self-generated goal of comprehension; they appeared to come to the task with their own understanding of how to read and learn from a challenging and unfamiliar text, and were successful in doing so. These two readers exemplified Alexander’s Highly Competent Reading (2005), and appeared to be appropriately situated as mature, independent learners. The other participants seemed to have much less of a sense of control over how they read and were more passive and less engaged, with the exception of one consistently unsuccessful reader, Kate. Kate seemed to go a little too far in her sense of control of her reading, in that she tended to diverge from the text and ended up constructing meaning that was not there. In addition, the other participants stuck to a linear progression through the text and local level processing, going from sentence to sentence, as is typical of less-skilled readers (Coté & Goldman, 1999; Lundeberg, 1987). They had a limited repertoire of strategies to assist them in constructing meaning and resolving comprehension difficulties, most typically using rereading and restatement of what was in the text. Their strategy use appeared to be over-automated, in that they routinely just re-read, sometimes reading the same section multiple times in an effort to understand. There was clear evidence of Effortful (but unengaged) Processors among both the 26 situationally successful and consistently unsuccessful readers. We were unable to determine whether they were less effective in monitoring, in general, because our coding was not stable. Even where we did see evidence of individual instances of monitoring and identification of comprehension difficulties, however, there was no guarantee that they would have access to appropriate strategies. Students tended to have less relevant background knowledge about slavery that could help them in constructing an understanding the history passage; instead, they tended to bring up connections to their own feelings, and made comments relating to the fairness or unfairness of the events the passage described. For the reading passage, although students did not have background knowledge of psycholinguistic theory, they did have access to relevant personal experience as readers; several of them were able to link successfully from the examples and descriptions given by the author to their own activity as readers. However, most of the participants did not know what phonics or decoding meant, and struggled with the idea of grapho-phonemic information. Only the two consistently successful readers appeared to use knowledge of text structure or of conventions of domain discourse in helping to build their understanding of the text. Andy identified important statements or paragraphs in both passages, and Mike used the headings and footnotes, as well as considering the sources used in the history passage. Interestingly, five of the consistently unsuccessful readers of the history passage, along with one situationally successful reader, commented in their response to the outcome measure that they did not think the history passage was presenting an argument. Their understanding of reading history was limited to the idea that historians tell us facts. 27 It is difficult to know whether what we captured in this study represents what students could actually do when confronted with a challenging text in a knowledge-lean situation. Their level of engagement and motivation, for instance, might be higher if there were consequences related to their level of performance, or in a situation in which they placed value on the information to be learned. However, that we did see high engagement and successful processing from at least a few students; it is likely that these characteristics represented a consistent approach to reading across all types of reading situations. It may be that an active approach to reading is not something that can be turned on and turned off, in which case it should have been apparent even in completing our research task. Fragile understanding related to reading does appear to be a possible problem for undergraduates. Such fragile understandings revealed by seemingly “good” college students while reading challenging domain-specific texts could at least partially be attributed to educational practices that fail to give adequate attention to the nature of domains or to the variable roles that texts serve in those domains. For instance, younger readers acquire little experience with critical evaluation of informational text, although such evaluation is an integral aspect of reading at higher levels of domain expertise. Students are allowed and expected to be passive readers, and trained that learning from text should be effortful and tedious. They do not construct their own understandings of what reading is and of how to approach a reading situation. Interventions in the college years, as suggested by Simpson and Nist (2002), may be too little and too late. Dismantling the non-optimal reading processes observed in this investigation may require rethinking theories of reading development and retooling our educational practices in the pre-college years. 28 29 References Alexander, P. A. (1997). Mapping the multidimensional nature of domain learning: The interplay of cognitive, motivational, and strategic forces. In M.L. Maehr & P.R. Pintrich (Eds.), Advances in motivation and achievement (Vol. 10, pp. 213-250). Greenwich, CT: JAI Press. Alexander, P. A. (2003). Profiling the developing reader: The interplay of knowledge, interest, and strategic processing. In D. L. Schallert, C. M. Fairbanks, J. Worthy, B. Maloch, & J. V. Hoffman (Eds.), 52nd yearbook of the National Reading Conference (pp. 47-65). Oak Creek, WI: National Reading Conference. Alexander, P. A. (2005). The path to competence: A lifespan developmental perspective on reading. [White paper commissioned by the National Reading Conference]. Oak Creek, WI: National Reading Conference. http://www.nrconline.org/publications/ThePathToCompetence.pdf Alexander, P. A., & Jetton, T. L. (2000). Learning from text : A multidimensional and developmental perspective. In M. L. Kamil, P. B. Mosenthal, P. D. Pearson, & R. Barr (Eds.), Handbook of reading research: Vol. III (pp. 285-319). Mahwah, NJ: Lawrence Erlbaum Associates. Alexander, P. A., & Judy, J. E. (1988). The interaction of domain-specific and strategic knowledge in academic performance. Review of Educational Research, 58, 375-404. Alexander, P. A., Murphy, P. K. & Kulikowich, J. M. (1998). What responses to domain-specific analogy problems reveal about emerging competence: A new perspective on an old acquaintance. Journal of Educational Psychology, 90, 397-406. 30 Anderson, R. C. (2004). Role of the reader’s schema in comprehension, learning, and memory. In R. B. Ruddell & N. J. Unrau (Eds.), Theoretical models and processes of reading (5th ed.), (pp. 594-606). Newark, DE: International Reading Association. Bates, M. (1975). The lady lives on blood. In A. Ternes (Ed.), Ants, Indians, and little dinosaurs (pp. 74-82). New York: Charles Scribner’s Sons. Bransford, J. D., Brown, A. L., & Cocking, R. R. (Eds.). (1999). How people learn: Brain, mind, experience, and school. Washington, DC: National Academy Press. Chall, J. S. (1983). Stages of reading development. New York: McGraw-Hill. Coté, N., & Goldman, S. R. (1999). Building representations of informational text: Evidence from children’s think-aloud protocols. In H. van Oostendorp & S. R. Goldman (Eds.), The construction of mental representations during reading (pp. 169-193). Mahwah, NJ: Lawrence Erlbaum Associates. Feltovich, P. J., Prietula, M. J., & Ericsson, K. A. (2006). Studies of expertise from psychological perspectives. In K. A. Ericsson, N. Charness, P. J. Feltovich, & R. R. Hoffman (Eds.), The Cambridge handbook of expertise and expert performance (pp. 4167). New York: Cambridge University Press. Ferstl, E. C., & Kintsch, W. (1999). Learning from text: Structural knowledge assessment in the study of discourse comprehension. In H. van Oostendorp & S. R. Goldman (Eds.), The construction of mental representations during reading (pp. 247-277). Mahwah, NJ: Lawrence Erlbaum Associates. Garner, R., & Gillingham, M. G. (1991). Topic knowledge, cognitive interest, and text recall: A microanalysis. Journal of Experimental Education, 59, 310-319. Goldman, S. R., & Rakestraw, J. A. (2000). Structural aspects of constructing meaning from text. 31 In M. L. Kamil, P. B. Mosenthal, P. D. Pearson, & R. Barr (Eds.), Handbook of reading research (Vol. 3, pp. 311-335). Mahwah, NJ: Lawrence Erlbaum Associates. Kintsch, W. (1998). Comprehension. New York: Cambridge University Press. Kintsch, W., & Rawson, K. A. (2005). Comprehension. In M. J. Snowling & C. Hulme (Eds.), The science of reading: A handbook (pp. 209-226). Oxford, England: Blackwell Publishing. Lundeberg, M. A. (1987). Metacognitive aspects of reading comprehension: Studying understanding in legal case analysis. Reading Research Quarterly, 22, 407-432. McKeown, M. G., Beck, I. L., Sinatra, G. M., & Loxterman, J. A. (1992). The contribution of prior knowledge and coherent text to comprehension. Reading Research Quarterly, 27, 79-93. Narvaez, D., van den Broek, P., & Ruiz, A. B. (1999). The influence of reading purpose on inference generation and comprehension in reading. Journal of Educational Psychology, 91, 488-496. Pearson, P. D. (1978). Some practical applications of a psycholinguistic model of reading. In S. J. Samuels (Ed.), What research has to say about reading instruction (pp. 84-97). Newark, DE: International Reading Association. Pellegrino, J. W., Chudowsky, N., & Glaser, R. (Eds.). (2001). Knowing what students know: The science and design of educational assessment. Washington, DC: National Academy Press. Pressley, M., & Afflerbach, P. (1995). Verbal protocols of reading: The nature of constructively responsive reading. Hillsdale, NJ: Lawrence Erlbaum Associates. 32 RAND Reading Study Group (2002). Reading for understanding: Toward an R&D program in reading comprehension. Santa Monica, CA: RAND. Scardamalia, M., & Bereiter, C. (1991). Literate expertise. In K. A. Ericsson & J. Smith (Eds.) Toward a general theory of expertise (pp. 172-194). New York, NY: Cambridge University Press. Schraw, G., & Bruning, R. (1999). How implicit models of reading affect motivation to read and reading engagement. Scientific Studies of Reading, 3, 281-302. Simpson, M. L., & Nist, S. L. (2002). Encouraging active reading at the college level. In C. C. Block & M. Pressley (Eds.), Comprehension instruction: Research-based best practices (pp. 365-379). New York: Guilford Press. Takaki, R. (1993). A different mirror: A history of multicultural America. Boston: Little, Brown, & Co. Wineburg, S. (1991a). Historical problem solving: A study of the cognitive processes used in the evaluation of documentary and pictorial evidence. Journal of Educational Psychology, 83(1), 73-87. 33 Appendix A Codes for Think-Alouds Strategic behaviors Reading aloud Re-reading Reading on (following statement of non-comprehension) [“I don’t know what those sources are, but I guess I’ll keep reading.”] Skimming (reading aloud while skipping portions) Guessing the meaning of a word in context [“ – which I’m guessing is trendy for the time, kind of typical, I don’t know.”] Predicting [It seems like he is gonna sort of explain what he’s talking about.”] Questioning [“What is de jure slavery, and how is it different from de facto slavery?”] Visualization [I’m picturing Virginia.”] Underlining [“I need to underline all this.”] Restating (paraphrase) or repeating text information - local (word, phrase, sentence level) [“So it’s better for the reader’s attention to focus on semantic processes.”] - global (paragraph, passage level) [“They thought using, by using African-Americans as slaves, that it would reduce their reliance on a lot of the whites who were angry and frustrated, who they didn’t want to use as slaves and indentured servants any more.”] Making connections 34 - to background knowledge (knew or didn’t know) [“I actually had no idea that they came from Germany or Ireland.”] - to personal experience [“That’s my people they’re talking about”] - to prior text [“That seems like pretty much the same thing he was just saying earlier.”] to topic knowledge test [“These are all terms that were used in the other survey.”] Interpreting or elaborating (a statement building beyond what is directly said in the text) [“It seems like, one of the things they had against using blacks as slaves was that, they wanted an all-white society.”] Monitoring/Evaluative behaviors Evaluating comprehension (positive or negative) [“That makes sense.”] Evaluating agreement with text (positive or negative) [“He’s saying a lot of stuff that’s true about reading.”] Evaluating text quality [“Those examples were pretty clear.”] Evaluation of interest (positive or negative) [“This is a pretty interesting article.”] Evaluation of importance of text [“It seems like an important paragraph here.”] Evaluation of task difficulty [“He’s using a lot of big words.”] Expression of empathy (sympathy or feelings felt or imputed to others) [“That’s a shame.”] Expression of surprise [“I expected it to be higher than 70 percent.”] 35 Appendix B Guiding Questions for Overall Aspects of Quality of Processing Global vs. local processing – are they trying to build a connected understanding, or going from sentence to sentence? Successful vs. unsuccessful connections – do they make relevant links to knowledge and experience, or are their connections non-strategic or distracting? Effective or ineffective monitoring – when they do have difficulty comprehending, do they take action to resolve their confusion? Flexible or limited strategic behaviors – do they demonstrate a variety of strategic behaviors, or are they limited to a few strategies such as re-reading? Accurate or inaccurate restatements – when they restate what they text has said, do they get it right or create their own version? Faithful or unfaithful interpretations or elaborations – if they go beyond what the text said, does their interpretation or elaboration remain faithful to what was given, or is it an unwarranted extension? High or low engagement – are they putting forth some degree of effort, or doing little in the way of active processing? 36 Table 1 Order and Outcomes for Passages Passage Participant First Second Mike R H Andy H R Mary h R Ted r H Alice h R Jane r H Sue r H Ann r h Bob r h Dave h r Rita r h Kate h r Jim r h Linda h r Pam r h Note: R = reading passage, successful outcome; H = history passage, successful outcome; r = reading passage, unsuccessful outcome; h = history passage, unsuccessful outcome. 37 Table 2 Reading Knowledge and Interest Means Overall and by Group Overall Successful Unsuccessful mean (sd) group mean (sd) group mean (sd) (n = 15) (n = 4) (n = 11) 35.6 (8.39) 35.5 (11.73) 35.6 (7.57) knowledge 30.3 (6.97) 31.5 (5.45) 29.9 (7.63) Interest 3.59 (.97) 3.55 (1.37) 3.60 (.87) Measure Domain knowledge Topic Note: Maximum possible score for domain knowledge = 80, for topic knowledge = 56, for interest = 5. 38 Table 3 History Knowledge and Interest Means Overall and by Group Overall Successful Unsuccessful mean (sd) group mean (sd) group mean (sd) (n = 15) (n = 5) (n = 10) 36.7 (9.51) 42.0 (11.42) 34.1 (7.72) knowledge 29.5 (6.64) 28.2 (6.65) 30.1 (6.90) Interest 3.32 (.58) 3.60 (.79) 3.18 (.44) Measure Domain knowledge Topic Note: Maximum possible score for domain knowledge = 80, for topic knowledge = 56, for interest = 5. 39 Table 4 Knowledge and Interest Scores for Situationally Successful Readers Participant Mary Ted Alice Jane Sue Measure Reading History Domain K. 30 43 Topic K. 27 17 Interest 2.0 2.6 Domain K. 24 40 Topic K. 28 25 Interest 3.2 3.6 Domain K. 35 40 Topic K. 28 27 Interest 4.8 3.4 Domain K. 42 36 Topic K. 38 34 Interest 4.2 3.8 Domain K. 30 31 Topic K. 32 20 Interest 4.4 3.6 Note: Bolded scores are for domain for which outcome was successful. 40 Table 5 Overall Processing Profiles Participant Processing level Mike Outcome group CS Restatement quality accurate Interpretation quality faithful Engagement level global Strategic repertoire flexible Andy CS global flexible accurate faithful high Mary SS local limited ------ ------ low Ted SS local limited ------ ------ low Alice SS local limited ------ ------ low Jane SS local limited ------ ------ low Sue SS local limited ------ ------ low Ann CU local limited accurate unfaithful low Bob CU local limited accurate unfaithful low Dave CU local ------ ------ ------ low Rita CU local limited ------ ------ low Kate CU local limited inaccurate unfaithful high Jim CU local limited inaccurate ------ low Linda CU local limited inaccurate unfaithful low Pam CU local limited ------ ------ low CS = Consistently successful SS = Situationally successful CU = Consistently unsuccessful high