Introduction

Although there has been increased emphasis placed on the importance of mathematics and science curriculum in K-12 instruction, indicators show that student achievement in mathematics and science has not been improving uniformly. While mathematics scores have risen in the last two decades, science performance has not (NSF 2010; OECD 2007; NCES 2009). Average science scores for 8th graders in 2007 in the United States were not measurably different from those of US 8th graders in 1994, despite recent efforts aimed at improving science, technology, engineering and mathematics instruction (NSF 2010). In addition, greater disparities are present in science than in mathematics because students from disadvantaged populations lag far behind their advantaged peers (NSF 2010). This project addressed student achievement in science by measuring what happens when different student-centred learning strategies were used in 7th grade science classes.

Lack of retention and comprehension of academic vocabulary is particularly acute in the science content area. One convincing argument for the failure of students to reach baseline achievement levels in science points to the long lists of unfamiliar terminology associated with the content that students must understand and retain on an annual basis (Groves 1995; Yager 1983). The use of textbooks is still central to science instruction and a summary of these textbooks reveals that more new terms are introduced in a year of science than could be expected in either a year of a foreign language in secondary school or of all other subject areas combined in one grade level in elementary school (Groves 1995; Yager 1983). Because understanding terms and definitions is a first step in learning a science, we hypothesise that inability to grasp the content vocabulary could create a barrier to further understanding of science concepts. While some academic terminology overlaps between content areas such as language arts and social studies, little of the science vocabulary is reinforced outside science class within school curriculum, making retention and understanding even more challenging in the limited class time available.

With increasing class sizes, typical school environments demand setting a pace to cover material that leaves even less regard for accurate monitoring of individual student learning or checking for understanding by teachers. In addition, the budget crises of late impacts disproportionately on courses in need of more-expensive resources, such as science laboratory materials, relative to other curricular areas. Consequently, in environments affected by tight funding and high accountability, such as science classes, reading textbooks (sometimes outdated) and listening to lectures tend to dominate instructional methods, particularly in secondary education settings (Betts 2009; Crocco and Costigan 2007).

Our specific aims were to investigate retention of science vocabulary by developing student-centred classroom strategies that could be practically implemented by teachers in public school classrooms. The strategies were based on research findings from the cognitive sciences that have been successful in improving retention of information in laboratory settings (Craik and Lockhart 1972; Conway and Gathercole 1987; Karpicke and Zaromb 2010; Metcalfe and Kornell 2007; Slamecka and Graf 1978). To truly understand scientific concepts, the students must master the terminology to the extent that they can use and apply it readily. The definition of a term when presented is a first step, but certainly is not sufficient for mastering scientific concepts. Our focus in this project was to study what happens when traditional techniques used to introduce students to science vocabulary, such as copying the definitions of science terms from the textbook, are replaced with student activities that focus attention on deeper encoding of the science vocabulary terms. We tested whether the new techniques would be comparable or even improve student retention of the course terminology. We also anticipated that the effectiveness of the student-centred strategies might vary by level of reading competence. Good readers should be able to master the vocabulary items from a textbook because they can depend on their verbal skills to understand the textbook presentation. Those with low reading skill, however, might not be as able to grasp the information directly from the textbook and might benefit from a more active or interactive strategy. They could benefit from reformatting the material prior to understanding it. When evaluating the effectiveness of specific strategies, we considered reading level as a potential moderator variable. We made a systematic effort to compare instructional exercises that were purely student-based and that minimally involved instructor differences.

There are a number of studies, particularly in the memory literature, which demonstrate that, by engaging differently with content knowledge during study, retention can be improved (Karpicke and Zaromb 2010; MacLeod et al. 2010; Rittle-Johnson and Kmicikewycz 2008). In other words, altering the conditions in which research participants study or encode information can alter retrieval rates from memory. Three sets of studies are presented below which provide the conceptual framework for the student-based exercises that we developed.

Memory effects

In laboratory settings, if students are encouraged to generate answers to questions, rather than read or copy answers that they have read from a textbook, the likelihood of retention is significantly increased (Karpicke and Zaromb 2010; Metcalfe and Kornell 2007; Slamecka and Graf 1978). Further evidence gathered in a classroom test with 3rd graders suggests that students with low prior knowledge benefit more from the generation effect than those with more prior knowledge (Rittle-Johnson and Kmicikewycz 2008). So, the retention benefit when students generate answers rather than copying them could be increased when the studied material is new or unfamiliar. Other manipulations that have shown significant gains in retention include the use of imagery (Paivio 1971) and elaboration (Craik and Lockhart 1972). When participants are encouraged to develop a mental image of a stimulus or to think about the stimulus in some way other than the way in which the information is presented, it enriches the representation of the information. The recoding process produces a substantial memory advantage when compared to a baseline condition of simply reading the stimulus material (Craik and Lockhart 1972; Paivio 1971).

MacLeod et al. (2010) built on previous findings with the production effect, which shows that speaking a word out loud during study rather than silent reading can improve explicit memory of the information spoken. This effect is similar to the modality effect, with an advantage to long-term retention when words were read aloud rather than silently (Conway and Gathercole 1987; Gathercole and Conway1988). Macleod et al. (2010) demonstrates through a series of studies that the production effect results from a process during study that increases the distinctiveness of the stimulus material. They suggest further that any manipulation that produces an overt response during encoding (such as performing an action rather than reading an instruction) has a similar effect of increasing distinctiveness and, as a result, should benefit memory.

Another phenomenon referred to as the self-explanation effect documents the considerable learning advantage when students are asked to explain concepts to either themselves or to others (Williams and Lombrozo 2010). The effect has shown gains when used with study material from a number of domains such as physics (Chi et al. 1989) and biology (Chi et al. 1994) and when compared to alternative study strategies that include reading the material two times, thinking aloud or receiving feedback (Amsterlaw and Wellman 2006; Chi et al. 1994; Siegler 2002). By comparing self-explanation to describing items and thinking aloud, Williams and Lombrozo (2010) show that the effect works because of the role of explanation in discovery and generalisation of knowledge.

Format effects

One of the benefits of living in the information age is that we have the tools to present information in any number of formats and modalities. Which one we choose should be guided by our desire as scientists and educators to provide an effective means of communicating information to students. Previous work that examined format differences in the visual modality revealed better memory for pictures than for printed words when explicit memory was tested (e.g. Goolkasian and Park 1980; Kroll and Corrigan 1981; Paivio 1975, 1978; Pellegrino et al. 1977; Smith and Magee 1980) and in tests of implicit memory which ask for semantic information, such as naming members of some category (McBride and Dosher 2002; Wippich et al. 1998). Paivio’s dual coding theory (1975) and Nelson’s (1979) sensory-semantic model of encoding influenced this work and much of our later thinking with respect to picture-word differences. The differences obtained in these studies were thought to be the result of encoding. Pictures can have dual codes (Paivio 1975), more distinctive encoding (Nelson 1979) or more direct access to semantic coding (e.g. Nelson 1979) in comparison to printed words. Still another model (Larkin and Simon 1987) indicates that the picture-word difference lies in the way in which information is extracted from the different formats. Some features could be directly represented in a picture but must be inferred from a word. They suggest that picture-word differences lie in the efficiency of the search for information and differences in explicitness of the information. The emphasis in this approach is on the richness of the pictorial representation following encoding (Larkin and Simon 1987).

Other work examined auditory-visual modality differences. Better memory for auditory presentation, particularly for the last several items in a list, has been found (Cowan et al. 2002; Gardiner et al. 1983; Gathercole and Conway 1988; Greene 1985; Greene et al. 1988) and has been shown to consist of sustained superiority for these auditory items, as well as a brief advantage attributable to echoic storage that can be eliminated by other auditory input (i.e. the suffix effect; see Crowder 1972; Goolkasian and Foos 2002; Penney 1989). The long-term modality effect takes place in long-term memory and it has been found for both serial and free recall (Cowan et al. 2002; Gardiner et al. 1983; Greene and Crowder 1986).

Goolkasian and Foos (2002) examined retrieval from working memory and compared three types of presentation formats (pictures, printed words, spoken words) using a dual task. These studies found little influence of format on the processing task, but reported strong effects for item recall. Stimulus items presented as pictures and as spoken words were recalled (and recognised) equally well or better than printed words. In several follow-up studies (Foos and Goolkasian 2005), these researchers manipulated the type and difficulty of the sentence verification task to uncover the source of the presentation format effect. They hypothesised that the effect might arise from a disadvantage for printed words rather than different advantages for pictures and spoken words. To test the attention allocation hypothesis, they required participants to articulate each of three or six presented items to ensure some allocation of conscious attention to all items. Performance with articulated printed words improved performance while articulation had no influence on the recall of pictures and spoken words. The recall disadvantage of printed words lies in the fact that ordinarily they are not given full conscious attention, with forcing attention to printed words improving performance and diminished format effects. This research could have important implications for applying instructional strategies in classrooms.

Multimedia effects

Often, in instructional settings, information is conveyed in multiple formats. Mayer’s (2001, 2005) Cognitive Theory of Multimedia Learning and Sweller’s (1999, 2005) Cognitive Load theory have been useful in developing instructional materials that promote deeper learning and enhance transfer of knowledge.

Sweller and his associates (Sweller et al. 1990; Tindall-Ford et al. 1997) outlined the conditions under which problem solving with instructional materials can benefit from a dual rather than single mode of presentation. Attending to multiple sources of information requires participants to mentally integrate disparate information prior to problem solving and it can produce a split attention effect that interferes. However, when the material presented in varied formats is physically integrated, problem solving is facilitated because the load on working memory is reduced. Interference from split attention effects can also be reduced if information is presented in more than one sense modality (Mousavi et al. 1995). Following Baddeley’s (1992) description of working memory that includes two separate and independent processors—a visual-spatial sketch pad and a phonological loop for verbal materials—presentations that involve more than one sense modality have increased working memory capacity in comparison to single-format presentations. In classrooms, a student-centred strategy would entail having the students read about the word or process and then draw a representation of it.

Mayer and Sims (1994) also report an advantage for multimedia presentations with problem-solving tasks. They adapted Paivio’s (1971, 1986) dual coding theory to explain multimedia learning. When information is presented verbally and visually, representations of that information are encoded in separate verbal and visual systems within working memory, and referential connections between the two representations are also strengthened. Multimedia methods promote the formation of all connections and, as a result, are more likely to promote transfer of information in problem-solving tasks compared with single presentation methods (Mayer and Sims 1994). Data consistent with their theory were obtained when participants with high and low spatial ability viewed an animation simultaneously or successively with a narration. Among the high spatial learners only, problem solving was better when the materials were presented simultaneously rather than successively (Mayer and Sims 1994).

Current study

We worked with seventh grade science teachers to develop innovative yet practical instructional strategies that are based on the cognitive science principles identified above. Student-centred strategies that are active, constructive and interactive, when incorporated into learning episodes, can be more effective for improving comprehension and retention than strategies viewed as more passive (e.g. listening to lectures, reading from textbooks, filling in blanks on worksheets). However, as Chi (2009) clarifies, defining learning activities simply as either active or passive is not sufficient to describe the cognitive processes underlying the activity and could have bearing on mixed results according to her literature review. Research is needed to better define and categorise which specific activities or combinations of activities are associated with the accepted cognitive science principles and effects. Even more importantly, once defined and designed, these instructional strategies need to be tested in typical American classrooms with today’s teachers.

We selected seventh grade because, in the target school system, seventh-grade science did not involve a high stakes end-of-year test and we believed the teachers would be more willing to participate with our study under those conditions. Student-centred strategies were developed in collaboration with the seventh-grade science teachers and each was given a name to better communicate them to the students. ‘Dictionary’ was a traditional method that is frequently used by teachers to launch new units; it consisted of asking the students to write out the definition of the new vocabulary words from the glossary in the back of the science textbook. The second strategy, ‘pictionary’, required the students to read the definition in the back of the book and then draw pictures that represented the science words on three-by-five index cards using coloured pencils. Students were encouraged to create a picture that reminded them of the meaning of the word and not necessarily an actual picture of the word. ‘Conversation’ involved the students reading the definitions from the back of the text and developing their own explanations of the vocabulary, turning to their shoulder neighbour, and sharing their understanding of the terms in their own words, first one student and then the other. The teachers reported that ‘dictionary’ was frequently used in classrooms while the other two, ‘pictionary’ and ‘conversation’, were less common and based on cognitive science principles.

Across all three strategies, content was controlled. The differences were in how the students interacted with the content. Conversation and pictionary involved more than just a written presentation of the material. We hypothesised that, if the material was presented in a dual format or dual modality, it could lead to retention benefits because students would see a written presentation first and then work with the conceptual definition by transforming it into either a picture or verbal representation. Although copying definitions from the back of the book involved some interaction with the material at a behavioural level, there were no additional cognitive processes involved in this method as with the other two strategies.

We were interested in the ease with which multiple strategies could be implemented by a single instructor across different classes and the reactions of the students to the varied techniques. In testing for differences in student retention as a function of the three strategies, we also looked at whether a student’s reading level would make a difference. Would the low readers be able to enhance their understanding of the science definitions because of the interactive nature of the new student-centred strategies that incorporated such principles as the generation effect and the production effect? It is possible that the limited reading skill of the low readers might interfere with their ability to learn from the more traditional method of copying the definition from the book and that the more interactive techniques would help to promote deeper encoding of the science terms. On the other hand, finding a benefit to the dictionary method would support more traditional approaches to the learning while benefits in performance using the interactive strategies would support some of the current literature that suggests that there is a memory advantage to dual processing. In any case, we hypothesised that the learning strategies would make a difference in test performance and we were particularly interested in whether reading level would moderate that difference.

Retention of the science definitions was tested immediately after the learning event and then again 2 days later when students returned to the science classroom for follow-up instruction. Because there was no instruction by the teacher about the science definitions between strategy application and the retention tests, the test performances were timed to investigate the student learning strategies rather than the teacher’s instruction. Two retention tests administered on different days were used because research has shown deeper processing or consolidation of studied material after sleep (Fenn et al. 2009; Gais et al. 2002) and also because of the testing effect, which is an advantage towards long-term retention that is shown when material is retested rather than studied and then tested at a later time (Delaney et al. 2010; Roediger and Karpicke 2006). Roediger et al. (2011) have data to suggest that educators should test students more often over time as a way of improving their learning outcomes. Also, Tse and Pu (2012) have recently shown that the effectiveness of test-enhanced learning can also vary with cognitive abilities. We wanted to determine whether retesting would improve learning outcomes to the same degree with all three of our learning strategies and for students at all of our reading levels.

Method

Participants

The study, which took place in the spring of 2011 and spring of 2012, was conducted at two different middle schools in a large southeastern school district. Classrooms of five participating 7th grade science teachers from the two middle schools used the same curriculum. The teachers volunteered to participate and were enthusiastic about designing and implementing the student-centred strategies in their respective classrooms. They received $250 dollars for their participation in the study.

Data were collected in two separate semesters (Spring 2011, Spring 2012) from 398 students enrolled in twelve 7th grade classes. Girls were slightly more prevalent (53 %) than boys in the sample and the students were drawn from diverse ethnic groups and reading levels. Fifty-seven percent were White, 19 % African American, 10 % Hispanic and 10 % Asian. Although Table 1 shows differences in the proportion of students in each of the reading levels by ethnic groups, χ2 (398, 12) = 49.09, p < 0.001, the majority of the students in the sample (90 %) were assessed by the school system with moderate (level 3) or strong (level 4) reading skills. As explained in the following section, reading levels were measured in the previous school year by the school district during end-of-grade testing. All of the student data were coded to protect the anonymity of the participants.

Table 1 Distribution of reading levels in each of the ethnic groups
Full size table

Measures

Student-centred strategies and the retention tests were developed in collaboration with the 7th grade science teachers involved in the study. Seventh grade was selected specifically because there is presently no end-of-course test for 7th grade science in the state, and therefore the teachers were willing to experiment with alternative learning strategies. The researchers met with the teachers three times prior to the implementation of the first strategy to discuss implementation of the learning strategies and to select the three units from the common textbook to be included in the study. After the first strategy was implemented, the researchers met with the teachers between each subsequent strategy and following the study. The subject matter of the three units studied were the immune system (unit 1), patterns of heredity (unit 2) and motion and forces (unit 3). The vocabulary list at the beginning of each unit was used as the items to be studied by the students and as the basis for the retention tests. A list of the items is in the “Appendix”.

Tests were constructed for each of the three units under study to measure the students’ retention of the selected terms. Each test consisted of a one-page instrument with the science terms listed at the top in alphabetical order followed by definitions taken verbatim from glossaries in the textbook with no distractors. Unit 1 (study of the immune system) included seven vocabulary words while Unit 2 (genetics) had 12 words and Unit 3 (study of speed measures of position changes) had seven words. The students were instructed to fill in the blanks with the term that best matches each of the definitions. The students knew that they were participating in a study, and that the tests would be used as quizzes by the teachers. Test scores were the proportion of items answered correctly by each student on each of the unit tests.

Student reading levels were obtained from the end-of-grade test of reading comprehension collected in the previous year by the school district. The reading test consists of eight or nine reading selections with six to nine associated questions for each selection. Each student is asked to read five literary selections (two fiction, one nonfiction, two poems), three informational selections (two content and one consumer) and one embedded experimental selection (which can be fiction, nonfiction, poetry, consumer or content). The variety of selections on each form allows the assessment of reading for various purposes: for literary experience, to gain information, and to perform a task. Scores are reported on a developmental scale and converted to scale scores. Achievement levels are also produced to provide a measure of performance relative to a standard. (http://www.NCpublicschools.org/accountability/testing/NC/). We used the achievement levels, which ranged from 1 to 4, as a potential moderator variable in evaluating the effectiveness of the student-centred learning strategies on tests of retention.

Procedure

Across a semester timeframe, each class used each of the three student-centred strategies once. The dictionary strategy and conversation strategy took approximately 10–15 min of class time to administer, while the pictionary strategy took between 15 and 20 min. Application of the strategies was counterbalanced across the varied classes taught by an instructor and across instructors. In other words if Teacher A, Block 1 used dictionary for unit one, they then used pictionary for unit two and conversation for unit three. Moreover, if Teacher A had three classes, the strategy was also staggered between classes so that class one used pictionary, class two dictionary and class three conversation. At the end of the semester, we had data for each strategy from each student. Each instructor also taught different strategies during different class periods within the same unit. We were interested in the ease with which multiple strategies could be implemented by a single instructor across different classes and the reactions of the students to the varied techniques. Each strategy was introduced in approximate 3-week rotations to coincide with the introduction of a new unit in the 7th grade science textbook.

After introducing the strategy, the students were instructed to use the strategy during the class period to learn the 7–12 new science terms for the unit under study. A retention test was given at the end of the class period and then again 2 days later when the students returned to the science class for follow-up instruction. Teacher instruction on the science unit was provided only after the second retention test.

Anonymous surveys were also distributed at the end of each semester to allow the students to provide feedback about how much they liked each of the strategies, whether the strategies made it easier to learn, and if they would use the strategies in the future.

Results

Data from 175 of the students were eliminated from the analysis because they were not in class for one of the six retention tests. Test scores from the remaining 223 participants were analysed with a repeated-measures general linear model to test for the between-group effect of reading level and the within-group effects of learning strategies (conversation, pictionary and dictionary) and testing period (immediate and 2-day delay). Reading levels 1 and 2 were combined in the analysis because of the small number of participants relative to reading levels 3 and 4. The F tests that are reported included the Greenhouse–Geisser correction when necessary to protect against possible violation of the sphericity assumption, and used Type III sum of squares to adjust for the varying number of participants in each of the reading levels. A significance level of 0.05 was used for all statistical tests.

Table 2 presents the means scores on the retention test by reading levels, learning strategy and testing period. Consistent with our hypothesis, the type of learning strategy that the students used made a difference on test performance, F(2, 440) = 7.31, p = 0.001, η 2p  = 0.03, with the pictionary (M = 0.76, SD = 0.26) and the dictionary method (M = 0.75, SD = 0.26) leading to significantly higher retention of the test items than conversation (M = 0.69, SD = 0.31). The significance of the post hoc comparisons was confirmed by Bonferroni tests (at p < 0.01). The effect of learning strategies also varied by testing period, F(2, 440) = 3.84, p = 0.022, η 2p  = 0.02. Figure 1 presents the interaction, which resulted from a repeated testing advantage with the dictionary strategy, but not with the other two strategy conditions.

Table 2 Average test performance for students in each of the reading levels by learning strategies and testing periods
Full size table
Fig. 1
figure 1

Strategy by testing interaction

Full size image

However, performance on the retention test also varied by reading level, F(2, 220) = 61.55, p < 0.001, η 2p  = 0.36; and reading level was found to interact with learning strategy, F(4, 440) = 3.30, p < 0.011, η 2p  = 0.03, and in a three-way interaction with learning strategy and testing period, F(4, 440) = 2.37, p < 0.052, η 2p  = 0.02. To understand the three-way effect, follow-up simple effects were tested for the effects of learning strategy and testing period at each of the reading levels.

For the students with limited reading skills (levels 1 and 2), there was a marginally significant effect of learning strategies, F(2, 38) = 3.01, p = 0.061, η 2p  = 0.14, but no effect of testing period, F < 1, and no interaction effect, F(2, 38) = 1.52, p = 0.232, η 2p  = 0.07. The moderate readers (level 3) also showed an effect of learning strategies, F(2, 188) = 7.04, p = 0.001, η 2p  = 0.07 and post hoc Bonferroni test (p = 0.001) located the difference to a higher test score in response to the dictionary rather than the conversation method. As in the previous group, there was no effect of testing period, F < 1, and no interaction effect, F(2, 188) = 1.45, p = 0.237, η 2p  = 0.02.

A much different data pattern was found with the skilled readers (reading level 4). For this group, learning strategies interacted significantly with testing period, F(2, 214) = 4.37, p = 0.019, η 2p  = 0.04, but there were no main effects of either learning strategies, F < 1, or testing period, F(1, 107) = 2.77, p = 0.099, η 2p  = 0.03. The interaction reflected the fact that the retention of the vocabulary items 2 days after the study period was higher than when tested immediately after but only for the pictionary and the dictionary method. When the conversation strategy was used, test performance was better in the immediate rather than the 2-day delayed test.

The students from all classes at both schools were surveyed at the end of the academic year and their responses are summarised in Table 3. They were asked to rank order the strategies according to which strategy they thought helped them to retain the words best and they were asked, in another question on the survey, which strategy they liked or enjoyed using most. Most students at both schools reported that they believed that dictionary was most helpful for remembering the words but they did not like or enjoy using the strategy. They shared comments such as “dictionary works best but it’s boring” and “dictionary is most familiar, I’m used to it.” Regarding the pictionary strategy, students commented that “Pictionary is fun” and “Pictionary is best.” Most students reported enjoying pictionary first and conversation second, with dictionary coming in last. The final question involved whether the students would use the strategies for future study. Eighty-nine percent indicated that they would use them again. Only 11 % responded no to that item.

Table 3 Percent of students who responded to survey questions about the learning strategies
Full size table

Discussion

Our comparison of two student-centred strategies (pictionary and conversation) with a more traditional dictionary method showed some differences when retention was measured with vocabulary test scores but, more importantly, that the effectiveness of the learning strategies varied by reading level and time of testing. Overall, science vocabulary items were better retained when learned either with the pictionary or the dictionary methods in comparison to the conversation method. However, those strategy differences in test performance were moderated by reading level, with low-skilled readers showing slightly higher scores with the pictionary method while moderate readers showed a significant benefit with the dictionary method. Highly-skilled readers, in contrast to the other readers, showed a distinctly different pattern in their test scores. Their test scores were strong irrespective of learning strategy and their test performance got better when tested 2 days after the study session with each of the strategies except the conversation method.

These findings provide some mixed support for the integration of engaging and interactive strategies into learning science to improve retention of vocabulary items. The fact that the pictionary method provided some advantage to the level 1 and 2 readers is consistent with the notion that the interactive methods would promote deeper encoding of the material in comparison to the dictionary method. Evidence to the contrary, however, can be found in the relatively low test scores obtained by the low and moderate readers with the conversation method. Across all reading levels, the dictionary method resulted in strong test performance. But, before we can lend support to this traditional method of instruction, we must carefully examine additional reasons why this result might have been obtained.

Because the retention tests in our study used the exact language that was used to define the science words in the glossary of the textbooks, and included a word box at the top with no distractors, we would expect that the dictionary strategy would produce the best results for retaining the science terminology. The dictionary strategy was also the method with which students and teachers were most familiar. Surprisingly, the dictionary strategy only provided an advantage for the moderate (level 3) readers and, in fact, the pictionary strategy provided just as much of an advantage to them as the dictionary strategy. For readers with less skill, pictionary had some marginal advantage relative to both other strategies and dictionary was not as strong. Also, with the low level readers, the conversation strategy produced no worse performance on the retention tests than dictionary.

Not only did the pictionary strategy provide an advantage for the students, but they expressed that they preferred using it more from the other two strategies. They reported that it was ‘fun’ and less ‘boring’. All students were given three-by-five inch index cards and the option of using colored pencils to create their visual images. In some cases, the students tried to draw a picture of what the word itself stood for (as in a shot for the word ‘vaccine’) while, in others, the students drew some type of symbolic representation of the term that reminded them of the word. For example, one student drew a negative sign followed by a stick figure of a person to represent the term ‘antibody’. When using the pictionary strategy, the students did not have to write down the definition from the glossary of the textbook at any time. They only had to read the definition to then develop their pictures. Pictionary was the only strategy that involved minimal additional expenditure on supplies because the researchers provided the index cards and coloured pencils for this study. However, the teachers indicated that index cards and coloured pencils are included on the school supply list for students to purchase each year before school begins.

Another question to which we sought an answer was whether strategies could be developed based on cognitive science principles and implemented practically in classrooms. The teachers agreed that the strategies were all doable in their classrooms, but that pictionary took the most class time for the students to complete, which they saw as a negative. Reading the word and the definition, contemplating what to draw, how to draw it and which colours to use, and then actually creating the visual takes longer but causes multiple opportunities for encoding and recording to occur in the brain, resulting in better retention of the words relative to other strategies. The question for the teachers became whether class time can be afforded to improve retention. In discussions with teachers, using the pictionary strategy for homework rather than class work was suggested as a good approach. Because students reported a preference for drawing the pictures and even liked doing it, perhaps assigning the pictionary strategy for homework would improve the retention of the words as well as increase percentages of students completing homework in science class. This is an area for further investigation. Another area to be explored in future studies is to have students bring their pictures to class after producing them for homework, and having contests in class to see which ones can best be identified by other students in the class. This would add another level of encoding to the learning of the vocabulary as students try and identify words through pictures and add competition and engagement to the lesson. Some of the teachers reported having used these types of activities for chapter test review with the words and their definitions, but not using student drawn pictures. Dividing students into teams and competing for answers promoted collaboration within teams and made study even more active and engaging. Using this strategy for unit review, in addition to chapter review, also could increase retention and comprehension as more active engagement with the curriculum becomes more frequent throughout the course.

One puzzling finding was the failure to find an advantage with the conversation strategy. A possible explanation lies in the use of test items that were verbatim copies of the textbook definitions. Had we tested the application of the concepts, rather than mastery of the definitions, we might have found more of an advantage for the conversation method because it was the most interactive of the learning strategies that were tested. This was also the reason, we believe, that many of our level 4 readers performed so well (at ceiling) on the immediate- and post-tests regardless of method used. Our future work will correct this limitation by testing for the learning strategies effectiveness with items that test for the application, as well as the mastery of, the science concepts.

However, there was a testing effect for both the pictionary and dictionary strategies. If pictionary provides as much of an advantage to the high-level readers as dictionary, and provides more of an advantage for the low-level readers, perhaps pictionary would be a better instructional strategy for improving retention of the complex science vocabulary, replacing the more-frequently used dictionary strategy.

The findings of this study are limited by the fact that good readers showed such strong performance on the retention tests that their data suffered from ceiling effects. Differences among the learning strategies might have been evident had tests been more challenging. Also, we did not assess the accuracy of the definitions that the students were producing in the conversation method. Definitions copied and pictures drawn on index cards were collected and viewed by the teachers, but there was no practical way to ensure the accuracy of the conversations between 7th graders discussing the definitions and meanings of the vocabulary words. An incomplete understanding of the vocabulary term could have caused the student to share an incorrect understanding of the definitions with the other students and could have explained the poor retention of the information with this method.

It is hoped that employing alternative learning strategies, such as those developed and tested in this study, would lead to an increase in both student interest and understanding in the science curriculum by improving retention and comprehension of the complex terminology. Using strategies that are practical, effective and viewed as enjoyable by students should also be promoted in schools of education with teacher candidates. Further collaboration across disciplines, such as between the cognitive sciences and educational leadership as in this study, can inform the teacher training experience as it improves classroom practices for optimising student learning.