Smith (1979) reported an experiment in which subjects were to detect whether or not a displayed word contained a particular target letter. Her data indicated that if the word bearing the target letter was preceded by a semantically related item, the detection of the target letter was faster than it was if the preceding item was unrelated. Those results provided strong support for holistic models of word recognition, in which it is assumed that letter detection must be mediated by prior word recognition. That is, any facilitating effect of the prime on lexical access should be passed on to subsequent letter detection. The present experimental paradigm, which was very similar to (albeit different from) that of Smith, served to explore the generality of her effect, but the results did not confirm her findings. Although a lexical-decision task used in Experiments 2, 5, and 7 provided clear evidence that the priming items employed in these experiments did facilitate lexical processing, a letter-detection task used in Experiments 1, 3, 4, 6, and 8 failed to reveal any facilitating effect of semantic priming on letter detection. The conclusion is that the generality of Smith's effect is far too limited to offer support for holistic models of word recognition.Holistic models of word recognition (e.g., that of Johnson, 1977) reject the notion that component-level cognitive processing necessarily mediates the identification of a word. Such models do not assume any initial cognitive processing at the component level, and they assume that cognitive encodings of letter-level information are not immediately available for use in any decision-making process. In fact, although various models differ as to whether they assume that holistic encoding always occurs (Johnson, 1977) or occurs only under some circumstances (Healy & Drewnowski, 1983), they seem to agree on the assumption that if holistic encoding does occur, letter information must be derived from the word-level encoding (Johnson, Allen, & Strand, 1989), rather than the other way around.The key suggestion in these models is that, under normal circumstances, a usable cognitive encoding becomes available only at the pattern or word level of processing. This assumption is an integral part of the pattern-unit model (Johnson, 1975(Johnson, , 1977(Johnson, , 1979(Johnson, , 1981Johnson, Turner-Lyga, & Pettegrew, 1986), but it is also an important component of both Johnston and McClelland's (1980) hierarchical model and Healy and Drewnowski's (1983) unitization model.An important implication of holistic views of word recognition is that the cognitive encoding of word-level information precedes that of letter-level information. That Experiments I, 2, 3, 5, and 6 in this study formed T. L. B.'s masters thesis, which was submitted to Ohio State University under the name ofT. L. Strand. We are grateful for very helpful comments by Robert Proctor, Alice Healy, and two anonymous reviewers. Correspondence should be addressed to N. F. Johnson, Department of Psychology, Townshend Hall,...

No abstract

No abstract

Abstruct -In this paper, we describe the development and analy,sis of a Computer Science curricuium that fu& embraces the object-oriented (00) paradigm.The curriculum addresses ihe three compofients that the ACM has reconmended for a computer science curriculum: concepts, skills, and abilities. Ow vision for the currfculum moves beyond a scheme of integrated courses to incorporate recent research from the field of cognitive psychologv with the idea of retaining more students by bolstering their cognitive skills. Good object-oriented design (OUD) requires such critical skiIIs as strategic pluming, analogical probbm solving, and mental modeling. Past approaches have assumed either that these skill,$ are learned as an incidenial by-product of truditionui training approaches or that they are inherent to the individual student. WO suggest specific ways in which the assessment and mining of these key mefacognitive skills can be incorporated into the curriculum INTRODUCTIONRecently, basic and applied psychological research on metacognition has enjoyed a surge of interest from the education and instructional research community [16]. Educators have sought ways of improving students' metacognitive abilities as a means of enhancing critical thinking and performance in a variety of fields, particularly in the sciences [13]. The greatest difficulty seems to be in getting researchers and educators to agree upon precisely which metacognitive skills are most useful for particular domains of knowiedge and upon ways in which those skills may be enhanced. #at professionals do agree upon is that the focus of instruction should be shifted away kom merely imparting domain specific knowledge and toward the assessment and training of learning skills. According to Reif [161, due to an increasingly complex and rapidly changing world, the acquisition of facts (which are expanding exponentially) must take a back seat to the more critical abilities of applying abstract concepts and adapting to novel situations.Traditionally, hypotheses about learning outcomes in camputcr science have emphasized individual differences between the learners, such as their mathematical ability, background in computing, gender, and anxiety about computers [24j While these factors may account for some differences in achievement, a great deal of the variability in

Traditionally, hypotheses about learning outcomes in computer science have emphasized individual differences between learners in mathematical ability, background in computing, gender, and anxiety about computers. However, few models of computer science education address the individual level of metacognitive ability, such as mental modeling, analogical reasoning, and strategy selection needed for todayd object -oriented practitioner. As part of a plan to incorporate the objectoriented paradigm into an integrated computer science curriculum, we have developed a model for the assessment and enhancement of metacognitive skills of computer science students at Stetson University.The initial step in this plan was the development of the Metacognitive Skills Inventory (MSI). The MSI is intendd to assess peopleb awareness of the cognitive processes they use to solve problems and their perceived level of planning, organization, and evaluation during a problem solving task. It was loosely based upon the State Metacognitive Inventory [3], designed to assess the extent to which students were aware of the thinking skills they used in the completion of an achievement test, and the Problem Solving Inventory [2] designed to assess peoplek awareness of their own style of solving personal problems (e.g relationship conflicts, career choices). Items were selected and reworded to refer to the more general problem solving tasks encountered by students in the computer science arena.The MSI was administered to 157 undergraduate students in psychology, general business, and computer science courses. Each of the items on the inventory is rated by the students on a 4-point Likert scale, with "1" being 'Strongly agree" and "4" being 'Strongly disagree." Item analysis led to the exclusion of ten items, whih had itemtotal correlations < .30. Scores on the refined 45-item scale ranged from 69-155, with a low score indicating greater metacognitive awareness and/or skill. Reliability analysis indicated high internal consistency (alpha = .9027).A one-way analysis of variance revealed significant differences among the four groups of students, €J2,206)=2.69, p<.05. The MSI scores of upper-level computer science students &l=84.06) were significantly better than lower level psychology students & I = 95.43), and general business students M=94.26).A second set of data, comparing computer science students from the CSl course to those in the CS2 course and two upper division courses, was more revealing. A oneway analysis of variance showed a significant diffemce between student course levels on MSI scores, &2,51)=6.34, ~<.01. Post-hoc analyses (Tukeyb HSD) revealed that the CS1 students had significantly worse MSI scores M=92.79) than either the CS2 (M=81.00) or upper level (M=80.92) students. However, no such differences were found between student scores on mathematics evaluation anxiety e-4) or mathematics learning anxiety E