ENHANCING CONCEPTUAL LEARNING BY UNDERSTANDING LEVELS OF LANGUAGE-RICH TEACHING

by

K. David Pinkerton, Ph. D.
Curriculum and Instruction Department
University of Denver
Denver, Colorado
e-mail:  dpinkert@shhs1.smoky.org

and

Cherry Creek Public Schools
Smoky Hill High School
Aurora, Colorado

Enhanced Conceptual Learning by Understanding Levels of Language-rich Teaching

    Good lessons provide a good opportunity for students to talk. Education has experienced a major shift in the methods used to teach important concepts. The shift from telling students concepts to students generating concepts themselves, in highly interactive social environments, has become pervasive (Dufresne, Gerace, Leonard, Mestre, & Wenk, 1996). Conversation has always been the primary agent of social interaction that utilizes concepts (Bialystok & Hakuta, 1994; Vygosky, 1978) and language-rich teaching is at the center of improved teaching techniques developed over the last 10 years.

    Much of the research about language-rich teaching methods has been published in the last decade. Language-rich teaching methods include reciprocal teaching (Brown & Palincsar, 1989; Rosenshine & Meister, 1994), scaffolding (Brown & Clement, 1991), small groups, (Cohen, 1994; Webb, 1989 ), tutoring (Cohen, Kulik, & Kulik, 1982; Graesser, 1992), or reciprocal questioning (King, 1990), involve students talking and writing about important ideas.

    Language-rich methods increase students' conceptual gains by fostering talking and writing about concepts. In the domain of physics, the effort to create language-rich classrooms is exemplified by Socratic Dialogue (Hake, 1992), microcomputer-based laboratory (Thorton & Sokolov, 1990), coaching (Ludicio, 1992), questioning (Minstrell, 1991), cognitive conflict (Mazur, 1993; Meltzer & Manivannan, 1996), modeling (Halloun & Hestenes, 1987), workshop physics (Laws, 1991a), and hypermedia (Wilson, 1994).

    Language-rich teaching methods differ from traditional "lecture and listen" methods in the amount of linguistic interaction they foster during formal instructional time. But do all language-rich methods produce equivalent conceptual gains? Answering this question requires focusing on the quality or richness of the language experience.

Qualities of Richness

    Webb (1991, 1995) demonstrated that when students work in small groups, the quality of their talk improves learning. Further, the quality of language-rich interaction helps differentiate among levels of cognitive processes (Baxter, Elder, & Glasser, 1994). At least four dimensions of the quality of language-rich methods are important. Language-rich teaching techniques help students to (a) generate meaning (Wittrock, 1990), (b) elaborate on new concepts (Pressley, et al., 1994; Simpson, Olejnik, & Supattathum, 1994), (c) promote cognitive strategies (Davey & McBride, 1986), and (d) differentiate between conceptual structures of experts and novices (Chi, Glasser, & Farr, 1988; Chi & VanLehn; Cohen, 1994). Different levels of student performance in these four dimensions can result from differences in levels of language-rich teaching.

    Common language-rich science teaching methods such as hands-on experiments and small group problem solving work differ in the quality of language interaction. For example, talking about laboratory experiments is certainly different than listening to lectures. Yet when lab team members talk to each other, their conversations focus on the objects of the experiment--how to finish on time or how to get the expected answer--not the underlying school concepts (Hodson, 1993). Though students are engaged in much dialogue in hands-on settings, they tend to focus on the surface features of immediate everyday objects depicted by laboratory apparatus rather than the deeper conceptual implications which actually govern the outcome of the experiment. ¹

    Small group work is a prominent feature of many science classrooms (Hake, 1992; Laws, 1991b; Lumpe & Staver, 1995). In a pragmatic effort to take advantage of the conceptual gains associated with language-rich interaction among students, without having to dialogue with every student for extensive periods of time, teachers use small groups. When small groups are employed in science classes, gains in conceptual understanding and applications occur (Cohen, Lotan, & Leecher, 1989; Hake 1994), ability to generate rules increases (Schwarts, Black, & Strange, 1991), and improvements in attitudes, attendance, and participation occur (Cottel & Lundsford, 1995). Shachar and Sharan (1994) investigated Israeli junior high school students' verbal behavior, social interaction, and academic achievement. In all areas, small group classes produced more desirable behaviors than traditional classrooms. The authors attribute these findings to speech strategies that encourage students to focus on words as carriers of meaning. Students restructure their communication in light of their own thoughts in order to gain control over the verbal symbols and the kind of message verbal symbols express to their peers.

    Student journals can increase the quality of language-rich interaction in the classroom. Journal, if properly applied, can produce many of the same effects as one-on-one interaction. Effective journals promote students' acquisition of school concepts and require strong teacher involvement. Teachers must have active strategies to draw students' attention to reasoning and encourage them to express everyday concepts in contrast to formal school concepts (Sere, 1991). Grumbacher (1987) explains her strategies for using journals in high school physics: (a) The best problem solvers are able to relate physics to their personal experiences; (b) Writing helps students find connections between experience and theory; (c) Students will do more work when they are seeking answers to questions that they initiate; (d) Journals foster such questions; and (e) Students need time to think about new ideas. But a delicate balance between structure and open-endedness must be achieved. If teachers do nothing to structure the level of interaction among students in the classroom, they may well find that students stick to the most concrete mode of talking and writing. If teachers do too much to structure interaction, they may prevent students from thinking for themselves and gaining benefits from the interaction (Cohen, 1994, p. 22).

    The active mental processing (AMP) journal (Pinkerton, 1996) may be a tool that science teachers can use to achieve the balance that Cohen mentions. This journal keeping method depends on teacher guidance and students who are motivated to understand school concepts ². The AMP journal structures students' dialogue with each other based on compelling demonstrations of everyday misconceptions; prescribes specific note taking strategies (Chandler & Sweller, 1992; King, 1992); and requires that students pose an application and a question each day. Teachers can "shrink" classroom size by training students to pose teacher-like questions to their peers, then teachers monitor conversations around the room to diagnose inappropriate use of key concept words. In this way, the AMP journal fosters high level language-rich teaching without continual one-on-one, student-teacher linguistic interaction.

Purposes of Study

    There are two primary purposes of this study. One is to gather evidence for differences in conceptual learning among three language-rich teaching techniques in science education (hands-on, small group, and AMP journal). If differences do exists, then is there any trend? Another is to investigate whether gender plays any role in response to various levels of language-rich teaching.

Method

Design

    This study utilized a quasi-experimental, non-equivalent cohort design partitioned into three levels of treatment to foster internal validity (Cook & Campbell, 1979). Three college preparatory high school physics classes, not individual students, were randomly assigned a treatment level. This type of design carries with it several validity concerns.

    First, teacher by treatment confounds threaten internal validity (Lysynchuk et al., 1989). Therefore, the teacher-researcher instructed each class in this study.

    Second, to detect possible differences in teacher enthusiasm for a particular method, two measures were taken. At the end of the first 18 weeks of the study (one semester), a 25 question Likert-scale survey was given to all 79 students to assess differential teacher enthusiasm. No statistically significant differences were detected on the total score, F (2, 76) = 2.33, p = .10, with the order of increasing means (more enthusiastic) being AMP journal < hands-on < small group classes. Also, two administrators who were trained in teacher evaluation observed classes on a random schedule. They completed identical 12 question Likert-scale surveys designed to monitor teacher enthusiasm. No statistically significant differences were detected among classes, F (2, 9) = .50, p = .63.

    Third, differential exposure to instruction can invalidate findings. Careful experimental controls were used to ensure that differences in conceptual gain could be attributed to the level of language-rich teaching, not different instructional materials, hand-outs, or tests. All classes used the same textbook, were assigned the same homework and readings, performed the same formal laboratory experiments, solved the same chapter problems, received the same handouts, and were asked the same questions during teacher presentations. Fourth, strict measures were enacted to ensure that the amount of time spent by students on treatment and non-treatment tasks was equivalent. A nine question Likert-scale survey was administered at semester's end to detect any differences in homework time. No statistically significant differences were detected among classes, F (2,76) = .09, p = .92, M = 3.2 hours/week, SD = 1.3 hours.

    Fifth, process measures document what students can do and strengthen conclusions about instructional effects (Lysynchunk et al., 1989). Two process measures were used. (1) Unit tests covering from one to three chapters were designed to monitor implementation of procedures. Each test, given to all classes, featured three questions with multiple parts that drew from the four dimensions of quality mentioned earlier (generation, elaboration, acquisition of cognitive strategies, and differentiating between experts and novices). For example, one question on each unit test asked students to perform tasks associated with hands-on activities such as measuring angles or devising a procedure. Textbook problems not given as part of the small group treatment were used as small group process measures and short essay questions asking students to explain and elaborate on the physics of common experiences were used as process measures for the AMP journal class.

    Several students' tests were randomly selected, scored by the teacher-researcher and another physics teacher (interrator reliability = .81). A two way (period x gender) ANOVA was also performed. The AMP journal students performed statistically better on their process measure question than the small group or hands-on class, F (2, 79) = 5.80, p < .01, power = .87, eta²= .14. No other statistically significant differences were detected. (2) A daily teacher-activity log was kept. Entries focused on any unanticipated differences among classes and what students actually did relative to what was planned.

    Sixth, mono-operational bias can lead to lower validity since single operations under represent constructs and contain irrelevancies. To combat this problem, two methods of monitoring students' conceptual growth were used, a traditional 29 question multiple choice test named the force concept inventory (FCI) and a conceptual configuration (concept map) measure (NETSIM). These measures use vastly different approaches to monitor conceptual learning.

    Seventh, initial group differences can make it difficult to demonstrate treatment effects (Tabachnik & Fidell, 1989). In lieu of random assignment of subject to treatment level, three devices were used to account for potential group differences before instructional treatment: (1) Covariates were used to match groups statistically before treatment. For example, pretests on FCI and NETSIM were used as covariates in subsequent ANCOVAs. (2) Strict control over extraneous variables helps to ensure that differences on dependent measures are due to treatment not factors such as those mentioned as validity concerns. (3) Demographic data such as age, gender, number of years of math, number of years of science, and overall grade point average were collected. These data are summarized in the participants section.

Participants

    This study was conducted in a large (2,500 students) middle to upper middle class suburban high school in a large mountain states city. The school ethnic mix was 1% Hispanic, 3% Asian, 9% African American, and 87% Anglo. The sample for this study adequately reflected the school's ethnic distribution. Sixty-eight percent of students take more than the required two years of science. All subjects in the sample had at least mathematics through algebra 2 and one year of chemistry. There are typically ten sections of physics per year with approximately 25 students per section. A summary of demographic data is found in Table 1. A series of one way ANOVAs detected no statistically significant differences among classes on any demographic variable.

Table 1
Summary of Demographic Data

Note. Numbers in parentheses are exact counts. All other values are means.

    The teacher-researcher had 18 years of teaching experience, 11 years was in physics. The teachers who were used to create the teacher-expert composite concept map came from the same school district. Four teachers were physics majors, four were chemistry majors, and two teachers had technology backgrounds. The minimum years of service was seven years and the maximum was 25 years. All teachers were career teachers planning to stay in education until retirement.

Instruments

    Force Concept Inventory (FCI) -- The FCI is a 29 question multiple choice test of beginning concepts in physics (Hestenes, Wells, & Swackhamer, 1992). It is widely used to diagnose misconceptions in physics and to evaluate instructional effectiveness. The authors suggest that preconditions - such as students' mathematics background, socioeconomic level, and teacher's experience - have little effect on gain scores. This suggests that the FCI can be used in a wide variety of circumstances to monitor teaching methods that foster conceptual understanding.

    The Kuder-Richardson reliability has been found to be .86 for the pretest and .89 for the posttest (Halloun & Hestenes, 1985). For this study it was .54 for the pretest and .76 for the posttest. Face and content validity were determined by (a) extensive review by physics professors and graduate students, (b) preliminary testing with graduate physics students until all agreed on the answers, (c) interviews of introductory physics students who had taken the test to check for understanding, and (d) detailed analysis of student scores to find evidence of common misunderstandings. A gender by level two way ANOVA revealed no initial statistically significant differences among treatment levels on the FCI pretest, F(2, 79) = .26, p = .77, but an expected significant gender effect, F(1,79) = 19.6, p < .001.

    Factor analysis has been performed to detect multidimensionality. Huffman and Heller (1995) detected two weak factors that combined to explain 15% of the variance. A Rasch item analysis was performed in this study and no strong evidence for multidimensionality was detected. On balance, the FCI seems to monitor introductory physics concepts effectively.

    NETSIM -- Relatedness measures, which are ultimately converted to NETwork SIMilarity (NETSIM) scores, monitor conceptual learning in a vastly different fashion than the FCI. A computer algorithm called Pathfinder networks (Schvanenveldt, 1990) randomly selects two words from a list supplied by the experimenter, e.g. acceleration and gravity. Students move a cursor between high and low to indicate their sense of relatedness between the two concept words. All students rated 106 possible combination of words.

    Pathfinder generates concept maps by making links based on relatedness scores. Highly related concepts are directly linked and less related concepts are separated by two or more links . The resulting concept map captures the configural character of domain knowledge and thus represents the most salient relations among concepts (Gomez, Hadfield, & Housner, in press).

    A composite concept map using relatedness scores from 10 physics teachers was generated by the Pathfinder program. All teacher's relatedness scores correlated significantly with every other teacher. The lowest r was .35 between the oldest and youngest teacher. The highest r was .69 between this researcher and his office mate.

    Gomez, Hadfield, and Housner (in press) describe how the NETSIM score is calculated. The observed similarity between a student's concept map and the teacher-expert composite is calculated by dividing the number of links shared by both networks by the number of links in either network. The expected similarity is calculated using the probability that two networks share x links. The NETSIM score results from observed similarity minus expected similarity. NETSIM values greater than zero indicate a greater degree of similarity between two networks than is expected by chance. NETSIM values less than one suggest that the observed network similarity is less than expected by chance. Students who experience greater conceptual learning can be expected to have larger positive NETSIM values.

    The Pathfinder approach has been found to be valid in discriminating between expert and novice conceptual structures (Durso & Coggins, 1990), delineating pilot's conceptual structures for use in designing flight simulator controls (Roske-Hofstrand & Paap, 1986), and predicting examination scores .

    Learning Styles Inventory (LSI) -- The LSI (Kolb, 1985) evaluates how a person prefers to learn and how he or she deals with day-to-day situations. It features 12 sentence completion items which allow scoring along four dimensions of learning--concrete experience (CE), reflective observation (RO), abstract conceptualization (AC), and active experimentation (AE). Two learning style scores are derived from the four learning dimensions. The AC-CE score is calculated by subtracting the concrete experiencing value from the abstract conceptualization value . The AE-RO score is calculated from subtracting the reflective observation value from the active experimentation value. Cronbach's alpha for the AC-CE and AE-RO subscales were .86 and .87 respectively. Intercorrelations were used to determine that the subscales were orthogonal as intended.

Procedure

    Procedures were devised and implemented to ensure that level of treatment was the only systematic difference among classes. Within the first three days of class, grading procedures, policies, student information sheet for demographic data, FCI pretest, relatedness premeasures, and LSI were administered. Treatment ensued and fell into a two- to four-day pattern for coverage of new concepts. For each group, the pattern was teacher demonstration, teacher led discussion, teacher explanation, and activity associated with treatment. Formal laboratory experiments not done as part of treatment, reading and problem assignments, as well as unit tests were identical for each class. The activities constituting treatment for each class occurred approximately once per week. The activity for the hands-on class was a hands-on quiz lasting one class period and involved physical manipulation of laboratory equipment focused on an experimental task such as determining the acceleration of a ball down a ramp. Teams of three or four students wrote theory, hypothesis, procedure, observation, and conclusion sections for the quiz. Verbal and writing intensity was observed to be high during these hands-on quiz activities.

    On the same day that the hands-on class participated in their hands-on activity that was part of the treatment, the small group class solved chapter problems in small groups using a prescribed problem solving template. These problems were specifically chosen to broach the same concepts involved in the hands-on activity and were turned in for grading at the end of the class period. An example lesson leading up to the small group problem solving session can be found in the appendix.

    AMP journal students experienced the same teacher demonstrations and teacher-led discussions as the hands-on and small group classes. But on the same day that other classes experienced their specific treatment, AMP journal students were led by the teacher to make programmed entries into their journals. These entries were punctuated with pre and postdictions, in which the teacher prompted correct use of concept words as students recorded their thoughts. Students were asked to verbalize their journal entries to peers and make adjustments in their journals if classmates did not understand. In addition, students compared their entries to teacher comments and made additional refinements. Each day students were required to create and answer a question of the day and to record an application of the day. Emphasis was placed on applications that students drew from their personal experience. If students could not think of an appropriate application of physics for the day, they were allowed to create a metaphor that helped them understand a difficult concept. A sample AMP journal lesson can be found in the appendix.

    AMP journals were collected every three weeks and graded for completion and procedure rather than accuracy³. Optimum points for quizzes equaled those of the journal. All other graded assignments were identical for each class.

    At semester's end (18 weeks), postmeasures of relatedness scores were collected and the FCI was given as a final. After all tests were completed, the student evaluation of teacher enthusiasm was administered. On the first day of the second semester, students were shown raw FCI mean scores per class. The teacher informed students that all classes would be taught by the AMP journal method from that point on, but that formal submission of the journal would be voluntary⁴. Full compliance would yield approximately 125 journals per class per semester. The numbers of journals submitted during the voluntary system per class were: AMP journal = 47, small group = 30, and hands-on = 22. Topics for the first semester were addressed only if they became important in explaining second semester concepts. In week 36, the FCI was given as part of the second semester final.

Results

    Results can be divided into two parts: (1) first semester analyses of FCI and NETSIM data using ANCOVA, (2) and second semester FCI data using repeated measures ANOVA.

First Semester Analyses

    FCI data were screened for entry accuracy, analyzed for outliers, and checked for assumptions required by ANCOVA. Two students dropped physics in the first week of class and one student did no homework and failed to take the final. These two students were not included in analyses. No univariate or multivariate outliers were detected in the pretests, while two univariate outliers were detected in the posttests. Both cases were in the hands-on class. One was the highest score (100% correct) and the other was the lowest score (21% correct). Both cases were retained.

    ANOVA assumptions include independence, normality, and homogeneity of variance. Scores were independent, since teacher monitoring ensured that all students were tested on an individual basis. No skewness or kurtosis values were greater than 1.0 and scatter plots of standardized residuals revealed no unusual patterns. Therefore, normality was assumed. A Levine statistic for the posttest of 1.24, p = .29 suggested no significant violation of homogeneity of variance.

    The FCI pretest was used as a covariate and required that two additional assumptions be met. MANOVA revealed no significant treatment level by pretest interaction, F(2, 79) = .26, p = .77. Thus, homogeneity of regression was reasonably met. Cronbach's alpha was .54 for the pretest. This indicates error in measurement of the covariate. This issue is addressed in the limitations section.

    Descriptive statistics, as well as omnibus and planned contrasts F tests, are summarized in Tables 2 and 3, respectively. Adjusted means per gender and class period, as well unadjusted means for each treatment level, are show in Figure 1.

Table 2
Descriptive Statistics for ANCOVA of FCI Posttest with Pretest Covariate

Table 3
Analysis of Covariance for Language-rich Teaching as Measured by FCI

Note. Value in parentheses is the mean square error.
*p < .05 **p = .06

Figure 1. FCI posttest scores adjusted with pretest as covariate for three levels of language-rich teaching.

    The omnibus gender by treatment level interaction and the main effect of treatment level were significant at p = .06 and p = .05 respectively. Planned contrasts were used to probe interaction and treatment level main effects. The first contrast compared the adjusted means of the AMP journal class with a linear combination of small group and hands-on classes, p = .01, effect size = 1.25. The AMP journal adjusted class mean on the posttest was highest. The adjusted means between the small group and hands-on classes were not significantly different. The second contrast probed the gender by treatment level interaction between small group and hands-on classes only, p = .03, effect size = .60. This contrast was determined to elucidate potential gender influences on the gender main effects. Females in the small group class unexpectedly scored lower than females in the hands-on class even though the adjusted mean for the small group class as a whole was higher than the adjusted mean for the hands-on class.

    The same analyses were performed using the AC score from the LSI measure. Pretreatment measurement of AC indicated a gender difference but not a class period difference. The contrast between the adjusted means of the AMP journal class and a composite of small group and hands-on classes was significant, F (1, 79) = 5.73, p = .02. These results mirror those obtained with the FCI pretest used as a covariate.

    NETSIM data were analyzed in the same manner as FCI data. NETSIM premeasures were used as a covariate to the dependent variable--NETSIM post measure. The 2 x 3 factorial ANCOVA required appropriate data screening and assumption testing. Pre- and posttreatment kurtosis were 3.48 and 1.75, respectively, and pre- and post-treatment skewness were 1.44 and .95, receptively, but no transformations of the data were performed because ANCOVA is robust with regard to violations of normality. All four univariate outliers and both multivariate outliers were high scores. They were retained in analyses. No violation of homogeneity of variance was detected by the Levine statistic, p = .40. MANOVA revealed no statistically significant interaction between pre NETSIM and treatment level therefore, homogeneity of regression was assumed for the covariate.

    The pattern of results for NETSIM data was similar to results for FCI data. Factorial ANCOVA (gender x treatment level) results for NETSIM data are summarized in Tables 4 and 5.

Table 4
Descriptive Statistics for ANCOVA of NETSIM Post Measure with Pre Measure Covariate

Cell	n	Raw M	Adjusted M	SD	.95 CI
AMP males	19	.159	.144	.105	.022
AMP females	9	.130	.131	.105	.031
Small group males	13	.126	.117	.082	.025
Small group females	10	.047	.052	.097	.035
Hands-on males	14	.065	.068	.081	.025
Hands-on females	14	.072	.085	.062	.025

Table 5
Analysis of Covariance for Language-rich Teaching as Measured by NETSIM

Source	df	eta squared	power	F Posttest with pretest as covariate
Gender	1	.012	.17	.89
Period	2	.083	.60	3.24*
Gender x Level	2	.034	.27	1.27
Within subjects error	72	na	na	(.01)
Contrast 1: AMP vs. (SG+HO)	1	.087	.73	6.85*
Contrast 2: Gender x Level (SG vs HO)	1	.028	.29	2.04

Note . Value enclosed in parentheses is the mean square error.
*p < .05

    Once again, omnibus F tests found a statistically significant difference among levels of language-rich teaching treatment with the order of means being as hypothesized: AMP journal > small group > hands-on. A significant difference was found between the adjusted means of the AMP journal and a composite of small group and hands-on classes, p = .01, effect size = 1.26. Adjusted means between the small group and hands-on classes were not significantly different. No statistically significant gender-by-level interaction was found in the small group and hands-on classes as occurred with the FCI results, but the pattern of interaction was similar as depicted in Figure 2.

 

Figure 2. NETSIM postmeasures adjusted with premeasures as covariate for three levels of language-rich teaching.

    A variety of correlations were calculated to investigate relationships between students' performance on the FCI and NETSIM (concept maps). Pre- and postmeasures of NETSIM correlated at r = .35 while pre- and posttests for the FCI correlated at r = .86. Postmeasures of NETSIM and FCI posttests correlated at r = .39. This suggests that FCI and NETSIM capture different aspects of conceptual learning. Support for this notion follows from analyses with the raw relatedness scores. These scores were not manipulated by the pathfinder algorithm which creates a concept map of students' beliefs. Students' pre- and post raw relatedness scores correlated at r = .91. ANCOVA analyses with these data revealed no reliable differences among classes. Thus, conceptual learning differences were not detected. Creating a configural measure of students' conceptual position with pathfinder seems to be more effective at detecting conceptual learning than raw relatedness scores.

Second semester

    The FCI was given as part of the second semester final and repeated measures analyses performed on the pretest and two posttests. This was done to investigate whether application of high level of language-rich teaching--to groups previously taught by medium or low level methods -- experience greater conceptual learning when switched to a higher levels of language-rich teaching. Further, effects of high levels of language-rich teaching on retention and transfer could be investigated.

    No univariate or multivariate outliers were detected and no violations of normality or linearity detected for the second posttest. A Levine statistic of 3.39, p = .04 indicates a violation of the homogeneity of variance assumption for the second posttest. According to Glass and Hopkins (1984), actual a = .06 with a nominal a = .05 for the reported variances calculated for posttest 2. The homogeneity of covariance assumption for repeated measures was met since Box's M = 12.1, p = .50 and no large violation of sphericity was detected as indicated by a Greenhouse-Geiser epsilon, e = .86.
MANOVA was used in a mixed (within and between subjects) analysis of the three FCI tests. The between subjects factor was level of language-rich teaching and the repeated factor was FCI test. Only univariate tests are reported.

    Contrasts were used to investigate differences in conceptual learning among treatment levels from the first to the second posttest as measured by the FCI. A statistically significant difference among treatment levels was still evident with the order of means unchanged from the first semester data, F (2,79) = 3.81, p = .03. A difference in the two posttests was significant at p = .11, F (1, 79) = 2.61. A simple effects contrasts exposed a pattern of differential increase on the FCI score. That is, though all classes scored higher on the second posttest, the small group and hands-on classes improved more than the AMP journal class. To probe this effect further, interaction of contrasts was utilized.

    Two between subjects contrasts were tested against two within subjects contrasts (Levine, 1991). The between subjects contrasts were identical to those used with the first semester FCI data. These contrasts compared group differences between the AMP journal class and a composite of small group and hands-on classes, as well as between the small group and hands-on classes only. The within subjects contrasts followed the same pattern but investigated differences among tests. The first repeated measures contrast compared the pretest to a composite of two posttests, and the second contrast compared the two posttests.

    A significant interaction between the within and between subjects contrasts was detected. Between subjects contrast one, AMP vs. (SG + HO), interacted significantly with the first within subjects contrast, pretest vs. (posttest 1+ posttest 2), F (2, 79) = 4.5, p = .04 This result suggests that the AMP journal class maintained its superior performance on the FCI test averaged over two posttests. A significant gender by level by test interaction for the two posttests was obtained, F (2, 79) = 6.44, p < .01. That is, males in the hands-on and females in the small group class improved from posttest 1 to posttest 2 while females in the hands-on and males in the small group produced lower scores on the second posttest. Figure 3 depicts class means for the two posttests graphically.

Figure 3. Raw class means of repeated measures FCI posttests. Tests given 18 weeks apart.

Discussion

    First semester results of differential conceptual learning measured by FCI and NETSIM tell the same story, but from different perspectives. The message of the story is that when different levels of language-rich teaching are applied, differential conceptual learning occurs. In other words, effects of adjusting levels of language-rich teaching exist. Furthermore, a hierarchy of language-rich teaching levels exist in which high levels produce the greatest conceptual learning in students.

    Other studies support this conclusion. Using path analysis, Pizzini and Shepardson (1992) compared two groups of eighth-grade biology students on behaviors such as attending, responding, following, and soliciting. Groups solved problems using traditional laboratory or small group methods. Students' behavior in the small group class correlated significantly with lesson structure, while those in the traditional hands-on class did not. This results suggests that at least two levels of language-rich teaching exist and that small group interaction produces student performance more closely aligned with teachers.

    Howe et al. (1995) studied four types of peer collaborative groups in middle school physical science. Their results suggest that small groups, which were exposed to hands-on and high level language-rich teaching, produce greater conceptual learning than hands-on only. This suggests that high levels of language-rich teaching add cognitive resources to hands-on methods that result in enhanced performance. The current study investigated differential conceptual learning partitioned along three levels of language-rich teaching in order to understand better its full effect.

    An increasing number of theorists are evoking a Vygotskian perspective in science education (Bowen & Roth, 1995; Lemke, 1990; Howe, 1996; O'Loughlin, 1992). Vygotsky's perspective suggests that language mediates students' intellectual progress from every day to school concepts. A cognitive science view utilizes, on the other hand, four aspects of language: lexical, morphological, sentence, and discourse levels (Caplan, 1992). Both perspectives converge to explain why a high level language-rich treatment produced reliable increases in conceptual learning.

    Vygotskian and cognitive science explanations depend on differentiating between concept types. Concepts typically broached in school tend to be abstract. Learning them requires extensive, programmed verbal interaction with teacher-experts. Everyday concepts are primarily object oriented and to a greater extent can be encoded visually. This multimodal view of semantic processing delineates between visual and verbal concepts (Caplan, 1992; Clark & Paivio, 1991). If verbal concepts are more abstract than object concepts then language-rich interaction should foster learning school concepts and common experience will reinforce everyday concepts. In effect, when learning abstract school concepts is deemed important, both Vygotskian and cognitive science views suggest a hierarchy that supports language-rich teaching to enhance school concept learning.

    This hierarchy manifests itself in the FCI and NETSIM results of the current study. The FCI reflects more of the morphological and sentence-level aspects of language's role in conceptual learning. This is due to the structure of the multiple choice test. It uses full sentences constructed around variations of key concept words in a static document. NETSIM measures the lexical dimension of concepts using techniques related to multidimensional scaling. NETSIM probes conceptual understanding at the word level by mapping conceptual configurations built from relationships among concept words. Jointly, FCI and NETSIM present a more complete picture of conceptual learning because together they tap into a broader representation of language⁵. Results from both measures indicate that high levels of language-rich teaching surpass medium and low levels when learning school concepts is the goal.

    Perhaps situated cognition (Lave & Wenger, 1991) can help explain first semester results . Simplistically, language is the manipulation of meaning with words. Semantic features of language become more or less salient depending on context (Caplan, 1992). Thus, language is a cognitive context. Not all contexts produce similar learning results. A hierarchy of linguistic situations exists which differ due the richness of linguistic interaction evident. Different agents serve as the primary delivery mechanism of language. From high to low level of language-rich teaching, those agents are teacher, peers, and objects (tools)⁶. That is, in the high level of language-rich teaching, the teacher is the primary referent in language interaction that results in students learning school concepts. In the small group and hands-on levels, peers and objects/tasks are the referents, respectively.

    The quality of language-rich experience is different among hands-on, small group, and AMP journal classes. Unit tests were scored with a rubric specifically designed to monitor the following dimensions of quality mentioned earlier; generation, elaboration, and cognitive strategies employment. AMP journal students performed significantly better than hands-on and small group classes by generating correct conceptual solutions, elaborating answers with greater depth and breadth, and employed appropriate problem solving strategies with greater regularity. NETSIM measures of structural knowledge indicated closer alignment between students' (novices) and teachers' (experts) conceptual knowledge for AMP journal students than for small group or hands-on students. In all, the quality of language-rich teaching occurred in the following order: AMP journal > small group > hands-on. As a result, enhanced conceptual performance followed the same order.

    Anderson, Reder, and Simon (1996) argue against the requirement of complex contextual and social interaction for learning, but fail to point out any differences that might be involved in conceptual versus other types of learning. On balance, any number of studies suggest that conceptual learning is enhanced in the context of high levels of language-rich interaction (Campbell & Ramey, 1995; Hart & Risley, 1995; Landes, et el., 1995; Moje, 1995; Roth 1994, 1995).

    Interpreting these results relative to other studies that used the FCI places these findings in an important context. Figure 4 displays the results (in the form of two regression lines) from over 50 studies and 3000 physics students that used the FCI to monitor student achievement and teaching method effectiveness (Hake, 1994). Hake suggested two groups of studies, one that utilized "interactive engagement" and the other traditional. Teaching techniques in the interactive engagement cluster used a variety of low-tech and high-tech methods to stimulate students' linguistic representations of concepts and to compare those representations to teacher-experts. All methods required a combination of language-rich method, proper implementation, and motivated students.

Figure 4. Comparison of Hake regression lines with gain scores of AMP journal, small group, and hands-on classes.

Gender Differences

    Females, progressing from low to high level of language-rich teaching, were expected to improve more than males. This was due to demonstrated gender differences in spatial and mechanical abilities (Baennenger & Newcombe, 1989; Halpern, 1992), which tend to concentrate in male-taught physics classes, and verbal abilities (Hyde & Linn, 1988) which are salient to language-rich methods. Though the AMP journal females demonstrated the greatest percent gain on the FCI and greatest alignment with teacher-expert concept maps, unexpectedly hands-on females outscored small group females.

    Dweck (1986) offers a possible explanation for this result in the entity theory . This theory predicts that females tend to focus on right answers because of perceived lack of ability in science. Females tend to direct their conversations towards obtaining correctness rather than conceptual understanding, unless prompted to do otherwise. Inadvertently, the hands-on treatment required students to generate a correct theory before collecting data for the hands-on quiz. This process forced females to negotiate conceptual meaning linguistically rather than generate conversations about lab equipment and procedures exclusively. Thus, the hands-on females accessed concept formation features of language and also generated discourse about experimental tools. Females in the small group class prompted each other for right answers during group quizzes rather than challenging each other for conceptual validation.

    The gender by treatment interaction is not surprising considering the equivocal nature of much research on the issue of gender performance issues in science. Burkam, Lee, & Smerdon (1997) report that hands-on activities are preferentially beneficial to females in science while Baker (1987) and Eccles (1989) suggest that teacher-directed and whole-group instruction is detrimental to females learning science. Further, females perform better than males on a broad spectrum of linguistic skills (Hyde & Linn, 1988) that are promoted in small group cooperative learning environments. In light of these studies, more research should be conducted in order to explicate which language-rich teaching method, if any, promotes the greatest conceptual gains in females. In this study, clearly AMP journal females gained the most, but a distinction between hands-on and small group classes on the effectiveness of enhanced conceptual learning is unwarranted.

Second semester results

    All three classes were taught in the AMP journal method during the second semester. No formal instruction occurred on first semester topics during the second semester, but most first semester concepts were integral to functioning with second semester topics. For example, the concept of force is equally significant in mechanics (first semester) as in electricity and magnetism (second semester). The AMP journal was one of many optional assignments for students.

    Performance on the second posttest followed the same pattern as the first posttest with the order of class means remaining AMP journal > small group > hands-on. Each class improved from posttest 1 to posttest 2 but the improvement was greater in the small group and hands-on classes. The lack of an increase, at a significance level of p < .05, in the class means for these two class could have been obscured by the gender by level by test interaction. Thus, the interaction masks whether changing instructional method from low or medium level to high level language-rich teaching improves conceptual learning regardless of the time of implementation. Though suggestive, these results do not reliably argue that changing from low to high level language-rich teaching in the middle of the year improves conceptual learning. The reason why only the hands-on males and small group females improved significantly from posttest 1 to posttest 2 is unclear. Further investigation is needed to determine the effect of switching instructional methods from lower to higher levels of language-rich teaching.

    Retention is enhanced by high levels of language-rich teaching. Note that the second FCI posttest demonstrated improvement on first semester concepts even though those concepts had not been taught formally for 18 weeks. This result can be explained by the role of language in long-term semantic memory. Accessing concepts from semantic memory is different depending upon whether the concept is contained as a word (abstract concept) or by an object . Object concepts use visual identification procedures and are accessed based on physical features. Abstract concepts do not share identifiable physical features and thus are accessed by words in processes other than recognition of an object (Caplan, 1992). High level language-rich learning environments demand that students link abstract concepts to words and manipulate them at the semantic, morphological, sentence, and discourse level of language. As a result, richly nested and deeply encoded concepts become part of the intellectual automaticity of linguistic interaction. Thus, long-term memory is promoted.

    Transfer is affected by the AMP journal method. An unsolicited letter to the teacher-researcher from a former student conveys this assertion.
Not only did you help me develop my skills in Chemistry and Physics, but the skills you taught me carried over into every other subject. The AMP journal helped me consolidate my note taking skills. I actually began to really think about what I was writing instead of arbitrarily and sporadically writing down facts. This process helped me, and I'm sure many others, to become a better student.

    This result agrees with Anderson, Reder, and Simon (1996), who suggest that the amount of transfer depends on the teacher's ability to cue students' sense of relevance of a given concept and to engage students in multiple examples. The required question and application of the day serve this function in the AMP journal. These features of the AMP journal are developed jointly between teacher and students so as to avoid triviality and to foster a sense of common purpose between teacher and students. The greater the extent of application and relevance as well as understandable examples that link common experience to new concepts, the greater the degree of transfer.

Limitations

    Not all threats to validity were eliminated. Random assignment of subject to treatment level was not possible in this public institution. Generalizing these results should be done cautiously. FCI pretest reliability was low which clouds its effectiveness as a covariate. Some measure of remedy was afforded by obtaining similar results with the LSI as a covariate.

    Not all learning is conceptual. These results apply primarily to conceptual learning and therefore should be limited to classroom situations in which conceptual content is the primary focus. Naturally, caution should be afforded when generalizing these results. It is important to replicate these results in classrooms from an array of different age, ability, and diversity backgrounds.

Recommendations

    A combination of the findings of this study suggest three pragmatic applications of language-rich teaching. First, utilize levels of language-rich teaching in appropriate situations. For example, full-time use of the high level may lead to boredom or fatigue. Instead, use high level language-rich methods when introducing abstract concepts or those that might be particularly resilient to change as done in the AMP journal class. Use the medium level when students need to rehearse concepts in a social context similar to the procedures enacted in the small group class. This helps students bridge the gap between teacher talk and student talk. Draw on the low level in situations that require procedural knowledge, recall, or categorization as exemplified in the hands-on class. Interaction with inanimate objects represents an important experiential foundation for subsequent high level language use.

    Such applications for the findings of this study could guide science a teacher's planning so as to maximize conceptual learning. Though conceptual learning may not be the primary goal of every teacher, those who wish to focus on conceptual learning can use the results of this study to orchestrate language-rich teaching methods. The knowledge that different teaching methods yield different conceptual gains in students will influence teachers' distributions of activities with an eye towards the strategic arrangement of levels of langauge-rich teaching. Instead of using an off the shelf, self described, hands-on curriculum which may only incorporate low levels of language-rich teaching, teachers can include medium and high levels of language-rich teaching at appropriate times with little investment in time or money. As a result, conceptual learning in students may increase.

    Second, identify major concepts of the course and design appropriate language-rich lessons that address them. Teachers who can identify salient concepts have a much better chance of teaching them. Once teachers check their own understanding of key concept words, they can mold classroom discourse in the form of accepted conceptual understanding using language-rich techniques.

    Third, many examples of successful language-rich teaching from a variety of settings have been published (Hake, 1992; Lemke, 1990; Pinkerton, 1996). Do not reinvent the wheel.

    The final recommendation involves an extension of the results of this study into ways of theorizing about teaching and learning. Language-rich teaching can unify the collage of learning theories prevalent today. If language-rich teaching subsumes any number of popular paradigms, then it could simplify teachers' thinking and prevent an unbalanced application of learning theory. Teachers do not have to belong solely to the behavioral, cognitive science, or social cognition camps of learning. They can view lesson plan design through a wider and more inclusive theoretical lens. This appealing parsimony could temper the angst teachers feel when the pendulum of reform swings by their schools. There is a certain visceral appeal to suggesting that the same language which makes us different than chimpanzees is crucial to humans learning concepts.

    Three well known learning theories report ample experimental support (Hassard, 1992), yet utilize teaching methods that are language-rich to affect treatment. To illustrate, consider behavioral, cognitive science, and social cognitive views of learning. Table 6 summarizes these theories and gives examples of language-rich teaching that fall into these camps. As teachers think about helping students learn concepts, they can focus on the language-rich features of these theories and produce concomitant achievement benefits. Designing language-rich lessons draws from the primary agent for conceptual learning used in all of the other theories.

Table 6
Summary of Learning Theories with Language-rich Teaching Examples

Theory	Metaphor	Operative Word	Model	Example from Language-rich
Behavioral science	mind is like a muscle	reinforcement	directed learning	teacher reinforcement of correct concept word use in AMP journal
Cognitive science	mind is like a computer	process	constructivism, inquiry learning	calling up students' existing concepts in verbal format to process conceptual change
Social cognition	mind is like a coffee house	interaction	cooperative learning	turn and talk interaction episodes, application of the day

    As teachers create curriculum, language-rich teaching can affect their thinking. Teachers can parse the major concept words of a course or unit of study and invent lessons that cause students to use these words in as many linguistic contexts as possible. Pinker (1995) illustrates the point with three words: man, dog, and bites. Juxtaposing these words differently can either change the meaning of the sentence or make it nonsensical. Each word order conveys a conceptual context linked to humans' innate ability to learn and use language.

    A physics teacher might form curriculum about the concept words force, motion, and implies. Rearranging these words in meaningful sentences produces vastly different cognizance in experts and novices. Teachers produce classroom protocol that promotes students' generation of meaning with words through discourse modeled by teachers. Through teacher demonstrations, peer discussions, and hands-on manipulation, students use language to anchor their conceptualizations in a context generated by them in response to the intellectual leadership of a teacher-expert.

Notes

1. The intent of the background section is to give an overview of the theoretical basis for the existence of levels of language-rich teaching and the possible trend in the effect on conceptual learning. The details of how this study manifested these levels in classrooms can be found in the methods section and appendix. An analysis of the trend can be found in the results section.

2. Motivation is not the explicit focus of this study but is doubtless an important factor in the success of this or any teaching technique. High school students who take physics tend to have a reasonably high level of self motivation.

3. The accuracy of student responses did not vary throughout the study even though full credit could be earned regardless of the correctness of the responses. This result is some indication that the AMP journal solicited sincere answers from students and that receiving credit did not affect students' thinking as demonstrated in journal entries.

4. I made use of the AMP journal voluntarily to determine the intrinsic appeal of this method. The number of journals submitted per class parallel class conceptual gains. This result may indicate the effect of levels of language-rich teaching on preferred approaches to learning by students. With the data available in this study, this suggested result is by no means conclusive.

5. Two measures of conceptual learning, FCI and NETSIM, were used to prevent mono operational bias, not as a priori measures of different aspects of conceptual learning. The analysis provided is my attempt to explain why these two measures were effective in tracking conceptual gains in this study with such a similar pattern.

6. Certainly methods can be employed that enhance peer interaction or student interaction with tools. For example, intelligent tutoring systems can mimic teacher-student interaction and performance assessments centered about the design and construction of a working device can increase students' interaction with tools. As of yet, teacher-student interaction still represents the highest form of language-rich teaching.

References

    Anderson, J., Reder, L., & Simon, H. (1996). Situated learning and education. Educational Researcher, 25 (4), 5 - 11.

    Baennenger, M., & Newcombe, N. (1989). The role of experience in spatial test performance: A meta-analysis. Sex Roles, 20, 327-343.

    Baker, D. (1987). Sex differences in classroom interactions in secondary science. Journal of Research in Science Teaching, 22 (2), 6-12.

    Bandura, A. (1986). Social foundations of thought and action: A social theory. Princeton, NJ: Prentice Hall.

    Baxter, G., Elder, A., & Glaser, R. (1994). Coginitive analysis of a achool performance assessment. Report prepared for Center for Research Evaluation Standards, and Student Testing, Graduate School of Education and Information Studies, University of California, Los Angeles.

    Burkam, D., Lee, V., & Smerdon, B. (1997). Gender and science learning early in high school: subject matter and laboratory experiences. American Educational Research Journal, 34 (2), 297-331.

    Bialystok, E. & Hakuta, K. (1994). In other words: The science and psychology of second-language acquisition . New York, NY: Basic Books.

    Bowen, M., & Roth, W. (1995). Knowing and interacting: A study of culture, practices, and resources in a grade 8 open-inquiry science classroom guided by a cognitve apprenticeship metaphor. Cognition and Instruction , 13 (1), 73-128.

    Brown, A. L., & Palincsar, A. S. (1989). Guided, cooperative learning and individual knowledge acquisition. In L. B. Resnick (Ed.), Knowing, learning, and instruction: Essays in honor of Robert Glaser . Hillsdale, NJ: Erlbaum.

    Brown, D., & Clement, J. (1991). Classroom teaching experiments in mechanics. In R. Duit, F. Goldberg, H. Niedderer (Eds.), Proceedings of an international workshop: Research in physics learning. Kiel, Germany: Institute fur die Padagokik der Natururssenschaften.

   Campbell, F., & Ramey, C. (1995). Cognitive and school outcomes for high-risk African-American students at middle adolescence: Positive effects of early intervention. American Educational Research Journal , 32 (4), 743-772.

    Caplan, D. (1992). Language: Structures, processing, and disorders. Cambridge, MA: The MIT Press.

    Chandler, P., & Sweller, J. (1992). The split-attention effect as a factor in the design of instruction. British Journal of Educational Psychology, 62, 233-246.

    Chi, M., & VanLehn, K. (1991). The content of physics self-explanations. The Journal of the Learning Sciences, 1, 69-105.

    Chi, M., Bassok, M., Lewis, M., Reinmann, P., & Glaser, R. (1989). Self-explanations: How students, study and use examples in learning to solve problems. Cognitive Science, 13 , 145-189.

    Chi, M., Glasser, R., & Farr, M. (1988). The Nature of Expertise. New Jersey: Lawrence Erlbaum Associates.

    Clark, J., & Paivio, A. (1991). Dual coding theory and education. Educational Psychology Review, 3 (3), 149-211.

    Cohen, E. (1994). Restructuring the classroom: Conditions for productive small groups. Review of Educational Research, 64 (1), 1-35.

    Cohen, E. G., Lotan, R. A., & Leechor, C. (1989). Can classrooms learn? Sociology of Education, 62 (April), 75-94.

    Cohen, P., Kulik, J., & Kulik, C. (1982). Educational outcomes of tutoring: A meta-analysis of findings. American Educational Research Journal, 19 (2), 237-248.

    Cook, T., & Campbell, D. (1979). Quasi-experimental design. Boston, MA: Houghton Mifflin Company.

    Cottle, P. D., & Lunsford, S. E. (1995). Increasing student participation in recitation. The Physics Teacher, 33, 23-25.

    Davey, B., & McBride, S. (1986). Effects of question-generation training on reading comprehension. Journal of Educational Psychology, 8 (4), 256-262.

    Durso, F. T., & Coggins, K. A. (1990). Graphs in the social and psychological science: Empirical contributions of Pathfinder. In R. W. Schvaneveldt (Ed.) Pathfinder associative networks: Studies in knowledge organization ( pp. 31-52). Norwood, NJ: Ablex Publishing.

    Dweck, C. (1986). Motivational processes affecting learning. American Psychologist , 41 , 1041-1048.

    Dufresne, R., Gerace, W., Leonard, W., Mestre, J., & Wenk, L. (1996). Classtalk: A classroom communication system for active learning. Journal of Computing in Higher Education, 7 (2), 3-47.

    Eccles, J. (1989). Brining young women to math and science. In M. Crawford, & M. Gentry (Eds.), Gender and thought: Psychological perspectives. New York: Springer-Verlag.

    Glass, G. & Hopkins, K. (1984). Statistical methods in education and psychology . Boston: Allyn and Bacon.

    Gomez, R., Hatfield, O., Houser, L. (in press). Conceptual maps and simulated teaching episodes as indicators of competence in teaching elementary mathematics. Journal of Educational Psychology .

    Graesser, A. (1992). Questioning mechanisms during complex learning. Arlington, VA: Office of Naval Research.

    Grumbacher, J. (1987). How writing helps physics students become better problem solvers. In T. Fulwiler (Ed.), The Journal Book (pp. 323-329). Portsmouth, NH: Boyton/Cook.

    Hake, R. (1992). Socratic pedagogy in the introductory physics laboratory. The Physics Teacher, 30 (9), 546-552.

    Hake, R. (1994). Toward mastery of Newtonian mechanics by the average student. AAPT Announcer, 24 (2), 55.

    Halloun, A., & Hestenes, D. (1985). The initial knowledge state of college physics students. American Journal of Physics, 53 , 1043-1055.

    Halloun, A., & Hestenes, D. (1987). Modeling instruction in mechanics. American Journal of Physics, 55, 455-462.

    Halpern, D. (1992). Sex differences in cognitive abilities (2nd Ed.). Hillsdale, NJ: Lawrence Erlbaum.

    Hart, B., & Risley, T. (1995). Meaningful differences in the everyday experiences of young American children .. Baltimore: Paul H. Brookes.

    Hassard, J. (1992). Minds on science: Middle and secondary school methods. New York: Harper Collins.

    Hestenes, D., Wells, M., & Swackhamer, S. (1992). Force concept inventory. The Physics Teacher, 30 (3), 141-158.

    Hodson, D. (1993). Re-thinking old ways: Towards a more critical approach to practical work in school science. Studies in Science Education, 22, 85-142.

    Howe, A. (1996). Development on science concepts with a Vygoskian framework. Science Education , 80 (1), 35-51.

    Howe, C. , Tolmie, A., Green, K., & Makenzie, M. (1995). Peer collaboration and conceptual growth in physics: Task influences on children's understanding of heating and cooling. Cognition and Instruction , 13 (4), 483-503.

    Huffman, D. & Heller, P. (1995). What does the force concept inventory actually measure? The Physics Teacher, 33 , 133-143.

    Hyde, J., & Linn, M. (1988). Gender differences in verbal ability: A meta-analysis. Psychological Bulletin, 104 (1), 53-69.

    King, A. (1990). Enhancing peer interaction and learning in the classroom through reciprocal questioning. American Educational Research Journal, 27( 4), 664-687.

    King, A. (1992). Comparison of self-questioning, summarizing and note-taking review as strategies for learning from lectures. American Educational Research Journal, 29 (2), 303-323.

    Klopfer, L. (1990). Learning scientific inquiry in the student laboratory. In E. Hegarty-Hazel (Ed.), The student laboratory and the science curriculum. London: Routledge.

    Kolb, D. (1985). Learning styles inventory. Boston: McBer & Company.

    Landes, S., Merrill, K., Merril, D., & Reiser, B. (1995). Tutoring: Guided learning by doing. Cognition and Instruction , 13 (3), 315-372.

    Lave, J., & Wenger, E. (1991). Situated learning: Legitimate peripheral participation. Cambridge, MA: Cambridge University Press.

    Laws, P. (1991a). Workshop physics: Learning introductory physics by doing it. Change, 23 (4), 20-27.

    Laws.P. (1991b). Calculus-based physics without lectures. Physics Today, 12 , 24-31.

    Levene, G. (1991). A guide to SPSS for analysis of variance . Hillsdale, N. J.: Lawrence Erlbaum.

    Lemke, J. (1990). Talking science. Norwood: Ablex Publishing.

    Lucido, H. (1992). Physics for mastery. The Physics Teacher, 30 (2), 96-101.

    Lumpe, A. & Staver, J. (1995). Peer collaboration and concept development: Learning about photosynthesis. Journal of Research in Science Teaching , 32 (1), 71-98.

    Lysynchuk, L. M., Pressley, M., d'Ailly, H., Smith, M., & Cake, H. (1989). A methodological analysis of experimental studies of comprehension strategy instruction. Reading Research Quarterly, 24, 458-470.

    Mazur, E. (1993). Understanding or memorization: Are we teaching the right thing? Procedings of the Resnick Conference on Introductory Physics Courses. (Rensselaer Polytechnical Institute, Troy, NY)

    Meltzer, D., & Manivannan, K. (1996). Promoting interactivity in physics lecture classes. The Physics Teacher, 34 (2), 72-76.

    Minstrell, J. (1991). Facets of student knowledge and relevant instruction. In E. Hegarty-Hazel (Ed.), The student laboratory and the science curriculum. London: Routledge.

    Moje, E. (1995). Talking about science: An interpretation of the effects of teacher talk in a high school science classroom. Journal of Research in Science Teaching, 32 (4), 349 - 371.

    O'Loughlin, M. (1992). Rethinking science education: Beyond Piagetian constructivism toward a sociocultural model of teaching and learning. Journal of Research in Science Teaching , 29 (8), 791-820.

    Pinker, S. (1995). The language instinct: How the mind creates language. New York: Harper Perennial.

    Pinkerton. K. D. (1996). Low-tech solutions, high-tech results. The Physics Teacher , 34 (1), 30-34.

    Pizzini, E., & Shepardson, D. (1992). A comparison of the classroom dynamic of a problem-solving and traditional laboratory model of instruction using path analysis. Journal of Research in Science Teaching , 29 (3), 243-258.

    Pressley, M., McDaniel, M., Tumure, J., Wood, E., & Ahmad, M. (1994). Generation and precision of elaboration: Effects on intentional and incidental learning. Journal of Experimental Psychology, 13 (2), 291-300.

    Rosenshine, B., & Meister, C. (1994). Reciprocal teaching: A review of the research. Review of Educational Research, 64 (4), 479-530.

    Roth, W. (1994). Science discourse through collaborative concept mapping: New perspectives for the teacher. International Journal of Science Education, 16 (4), 437-455.

    Roth, W. (1995). Affordances of computers in theacher-student interactions: The case of interactive physics. Journal of Research in Science Education , 32 (4), 329-347.

    Schwartz, D. L., Black, J. D., & Strange, J. (1991, April). Dyads have a fourfold advantage over individuals in inducing abstract rules. Paper presented at the annual meeting of the American Educational Research Association, Chicago.

    Sere, M. (1991). Learning by giving and receiving explanations. In E. Hegarty-Hazel (Ed.), The student laboratory and the science curriculum. London: Routledge.

    Shachar, H. & Sharan, S. (1994). Talking, relating, and achieving: Effects of cooperative learning and whole-class instruction. Cognition and Instruction, 12 (4).

    Simpson, M. L., Olejnik, S., Tam, A., & Supattathum, S. (1994). Elaborative verbal rehearsals and college students' cognitive performance. Journal of Educational Psychology, 86 (2), 267-278.

    Schvaneveldt, R. (ed.) (1990). Pathfinder networks: Studies in knowledge organization.

    Tabachnick, B., & Fidell, L. (1989). Using multivariate statistics. New York: Harper Collins.

    Thornton, R., & Sokoloff, D. (1990). Learning motion concepts using real-time microcomputer-based laboratory tools. American Journal of Physics, 58, 858-867.

    Vygotsky, L. S. (1978). Mind in society . Cambridge, MA: Harvard University Press.

    Webb, N. (1989). Peer interaction and learing in small groups. International Journal of Educational Research, 13, 21-39.

    Webb, N. (1991). Task-related verbal interaction and mathematics learning in small groups. Journal for Research in Mathematics Education, 22 (5), 366-389.

    Webb, N. (1995). Using group collaboration as a window into students cognitive processes. Report for Office of Educational Research and Improvement, Washington, DC.

    Wilson, J. (1994). The CUPLE physics studio, The Physics Teacher, 32, 518-523.

    Wittrock, M. C. (1990). Generative processes of comprehension. Educational Psychologist, 24 (4), 345-376.

---

About the author... Dave Pinkerton has been teaching and researching for twenty years. He is involved in several curriculum and research projects ranging from development of a high school level internet chemistry course to investigating a variety of cognitive science approaches to science curriculum. His degrees include B.A. and M.B.S. in chemistry from the University of Colorado, a M.S. in chemical engineering from the Colorado School of Mines, and a Ph.D. from the University of Denver in Curriculum and Instruction. His research interests include conceptual change processes, assessment of teaching effectiveness, and cognitive science approaches to curriculum design.

---

Appendix

Sample Lessons Active Mental Processing Journal

Previous Knowledge: Students know the definition of acceleration, velocity, displacement and vectors. They have introductory skills with vectors.

Concept to be Taught: Independent nature of vertical and horizontal motion.

Script: Day 1
Each year the US. supplies hundreds of tons of food to starving people, sometimes in war torn areas. Of necessity, airplanes drop this relief. The pilots must do this accurately lest they invite disaster. Luckily, physics concepts are part of the answer.

Let me demonstrate a concept that will help you drop the food exactly on target. (I hold up a demonstration device that supports two identical steel balls. It has a spring loaded rod that will hit one ball horizontally at the same instant that it releases the other ball to fall straight down.) Please sketch this device in your AMP journal. (Give students sketching time. Walk around the room and monitor the drawings.) The sketches look good. Now elaborate your drawings by labeling the following important features; rod, spring, ball that falls straight down, ball that is hit horizontally, and both balls begin moving at the same time.

Please predict what you will hear as the balls hit the ground. Will you hear this (Clap hands once.) or this (Clap hands twice.)? Write your prediction in your journal and don't forget to say why (I walk about the room to monitor students answers.).

Now turn and talk to your neighbor. Read your prediction to them and explain your ideas. When you are done, allow them to do the same. (I listen to the groups.)
I am going to perform the demonstration. Please write exactly what you hear and see in your AMP journal. (The balls hit at the same time. Allow time to write observations.)

Why did this happen? (Someone usually says that it is due to the balls having the same vertical acceleration. What are the characteristics of the horizontal motion? (A student usually says that it constant because of inertia.) When does the horizontal motion of the ball hit sideways stop? (Students respond, when the vertical motion stops.) Now while it is fresh in your head, write a note to yourself in your journal. There are at least three important points to make. Please do it in this form:
What I saw What it means

1. 1.
   
2. 2.
   
3. 3.
   

(I walk around the room to monitor the writing.)
Let's relate this to a real world situation. Remember the relief food that I mentioned at the first of the class period? In your AMP journals, draw a sketch of a plane carrying food under its fuselage and traveling at a constant velocity and height above the ground. At what point in the flight path would the pilot release the food in order for it to hit a target? Include the path of the food and the position of the plane at the moment the food hits the target. (Walk around and monitor the answers.)
We have only three minutes remaining in the period. Please write your question of the day in your journals.

Day 2
Please take out your AMP journals and review your sketch of the food's path. Write yourself a note explaining whether the food will hit the ground behind, directly below, or ahead the plane. (Monitor their responses) Turn and talk to your neighbor. Explain to him or her what you have written. Please allow for equal time.
Let's relate the two ball demonstration to the food drop. We'll do it this way. Sketch a graph of distance versus time, velocity versus time, and acceleration versus time for the straight down ball and the vertical motion of the food. (Walk around the room and help to correct mistakes.)
Please go beyond answering my question and elaborate your notes by writing a caption of explanation on each graph. Tell yourself the meaning of each graph. (Monitor this activity.)

Now, do the same thing for the horizontal motion of the food and the ball hit horizontally. Don't forget to elaborate your sketches.
Let's summarize what we've learned. Horizontal and vertical motion are independent of one another. That is, the motion of a projectile hit horizontally close to the surface of the earth can be analyzed in two components, horizontal and vertical. What governs the horizontal motion is inertia and it is completely independent of the constant force of gravity which governs the vertical motion.

Oh look! We have only five minutes left in the period. Please write your application of the day while it is fresh in your mind.

Small Group
sample lesson

Previous Knowledge: Students know the definition of acceleration, velocity, displacement and vectors. They have introductory skills with vectors.
Concept to be Taught: Independent nature of vertical and horizontal mation.

Script: Day 1
Each year the US. supplies hundreds of tons of food to starving people, sometimes in war torn areas. Of necessity, airplanes drop this relief. The pilots must do this accurately lest they invite disaster. Luckily, physics concepts are part of the answer.

Let me demonstrate a concept that will help you drop the food exactly on target. (I hold up a demonstration device that supports two identical steel balls. It has a spring loaded rod that will hit one ball horizontally at the same instant that it releases the other ball to fall straight down.) Notice that the device will release one ball to fall straight down at exactly the same instant it hits the other ball horizontally.

Please predict what you will hear as the balls hit the ground. Will you hear this (Clap hands once.) or this (Clap hands twice.)? Let's let nature tell us the answer. (I perform the demonstration. Both balls hit at the same time.)

Why did this happen? (Someone usually says that it is due to the balls having the same vertical acceleration. What are the characteristics of the horizontal motion? (A student usually says that it constant because of inertia.) When does the horizontal motion of the ball hit sideways stop? (Students respond, when the vertical motion stops.)

Let's relate this to a real world situation. Remember the relief food that I mentioned at the first of the class period? At what point in the flight path would the pilot release the food in order for it to hit a target? You are right! The food hits exactly under the plane because the food and the food have the same horizontal motion until the food hits the ground. The amount of time the food is in the air depends on the release altitude and the acceleration of gravity. Therefore the food can travel forward only for the amount of time that the food is falling toward the earth.

Let's relate the two ball demonstration to the food drop. We'll do it this way. I am going to use our graphing skills to help explain why the balls hit at the same instant and why the food hits directly under the plane. (I draw and explain the shapes of displacement versus time, velocity versus time, and acceleration versus time in both horizontal and vertical components of motion for the two balls and the food.)

Let's summarize what we've learned. Horizontal and vertical motion are independent of one another. That is, the motion of a projectile hit horizontally close to the surface of the earth can be analyzed in two components, horizontal and vertical. What governs the horizontal motion is inertia and it is completely independent of the constant force of gravity which governs the vertical motion.

Okay, tomorrow we will have a quiz. You will have one class period to answer questions on what you have learned today.

Day 2
Today is your quiz. Remember that you may work in your small groups on the answers only after everyone has attempted all the questions. When you do start helping each other, please remember to use the talk aloud methods that I taught you. Good luck and here is your quiz. (I pass out the following questions on a separate piece of paper.)

1. How many "clicks" do you hear when the coins hit the floor? Please explain why this happens.
2. Sketch the horizontal and vertical velocity versus time graphs for each coin. Explain any differences and similarities.
3. Suppose an airplane is flying at a constant horizontal velocity with a large lead ball attached to its bottom. The ball suddenly falls. Will the lead ball fall behind, directly beneath, or in front of the plane? Provide a sketch and include the path of the ball as well as the position of the plane exactly at the instant the ball hits the ground.

---

About the author...

Dave has been teaching and researching for twenty years.  He is involved in several curriculum and research projects ranging from development of a high school level internet chemistry course to investigating a variety of cognitive science approaches to science curriculum.  His degrees include B.A. and M.B.S. in chemistry from the University of Colorado, a M.S. in chemical engineering from the Colorado School of Mines, and a Ph.D. from the University of Denver in Curriculum and Instruction.  His research interests include conceptual change processes, assessment of teaching effectiveness, and cognitive science approaches to curriculum design.

---

To get to the top of this page, click here.

To get back to the current issue of the EJSE, click here.

To get back to the EJSE's Archive page, click here.