Effects of Teaching Classification on Classification, Verbal Conceptualization, and Analogical Reasoning in Children With Developmental Language Delays

Differences Between Categorization and Classification

Both Rosch, Mervis, Gray, Johnson, and Boyes-Braem (1976) and Jacob (2004) argue that the differences between classification and categorization are essential for construction of knowledge. The failure to distinguish between them appears to stem from the misconception that they are synonymous. This misconception may be reinforced by the fact that both are mechanisms for organizing information. Categorization was defined by Jacob (2004) as the process of dividing the world into groups of entities whose members are in some way similar to each other. Recognition of resemblance across entities and the subsequent aggregation of like entities into categories lead the individual to discover order in a complex environment. “Without the ability to group entities based on perceived similarities, the individual’s experience of any one entity would be totally unique and could not be extended to subsequent encounters with similar entities in the environment” (p. 518). Unlike categorization, classification is a process involving systematic assignment of entities to one and only one class within a system of mutually exclusive and nonoverlapping classes. This process is lawful because it is carried out in accordance with an established set of principles that governs the structure of classes and class relationships. It is also systematic because it requires consistent application of these principles within the framework of a prescribed ordering of reality. Jacob identified six systemic properties for comparing systems of classification and categorization: (a) Process. Categorization is characterized by creative synthesis of entities based on context or perceived similarity, whereas classification is characterized by systematic arrangement of entities based on analysis of necessary and sufficient characteristics. (b) Boundaries. In categorization, boundaries are “fuzzy” because membership in a group is nonbinding, whereas in classification, boundaries are fixed because classes are mutually exclusive. (c) Membership. In categorization, membership is flexible, whereas in classification, it is rigorous; an entity either is or is not a member of a particular class. (d) Criteria for assignment. In categorization, the criteria for assignment is both context dependent and context independent, whereas in classification, it is predetermined by guidelines and principles. (e) Membership. In categorization, individual members of a category can be rank-ordered by typicality, whereas in classification, they are equally representative. (f) Structure. In classification, there is generally a hierarchical structure of well-defined, mutually exclusive, and nonoverlapping classes nested in a series of superordinate–subordinate systems, whereas categorization consists of variable clusters of entities that may or may not be organized in a hierarchical structure.

As children begin to conceptualize relations and to form verbal abstractions (“they are all wild animals”), the prior ability to categorize may develop into ability to classify based on conceptualized logical relations and develops into ability to classify events according to more complex and increasingly abstract criteria (Blaye, Bernard-Peyron, Paour, & Bonthoux, 2006; Blaye, Chevalier, & Paour, 2007; Bomba, 1984; Cèbe, Paour, & Goigoux, 2002; Corrigan & Denton, 1996; Mandler, 2000; Quinn & Eimas, 1996). Thus, categorization precedes classification in the developmental sequence: comparison→categorization→conceptualization→classification.

Categorization, Classification, and Language Acquisition

There is much debate over the relationship between thinking and language. According to Dromi (1986), the use of thinking categories allows children to learn new words rapidly. In this process, the acquisition of words is “horizontal,” that is, acquisition of words that are attributed to different semantic groups. Typically, a single word appears in each category. For example, the word hat is acquired under the clothing category, and the word car is acquired under the vehicle category. When the child’s vocabulary expands, there is a “vertical” acquisition, that is, acquisition of words that are related to the same category. For example, the words socks, trousers, and so forth will be added to the word hat under the clothing category. The transition from horizontal to vertical acquisition of words indicates the child’s ability to recognize similarities and differences between the items and arrange them in distinct groups or categories (Dromi & Fishelzon, 1986).

In the evolving literature, there are many approaches regarding the relationship between language and categorization among children. Diesendruck (2003) reviews three areas in which children are required to use categorization and generalization ability. When categorizing the animal domain, children’s categorization processes are not significantly affected by the processes of conceptualization and culture in relation to their categorization of objects and people. His findings indicate that interplay exists between the area of children’s specific cognitive abilities of categorization and their linguistic abilities.

While language acquisition takes place, children rely on existing knowledge. For example, Clark (2004) examined the actions of children in space, in the absence of relevant relation concepts. He found that children use existing conceptual categories for understanding new words and in construction of the relevant conceptual language. He claims that they construct their own language based on the categories that they have already acquired. The conceptual representation they have built in their first year of life for objects, relationships, property, and events provides a broad cognitive basis for their mapping of new words. In addition, cognitive research in nonlingual areas have already shown that children make generalizations and expand categories, provided that the new components possess significant and striking similarities to the components with which they are already familiar (Kotovsky & Gentner, 1996).

Accordingly, it was found that children can make broad generalizations beyond their existing conceptual knowledge to create a new verbal category, provided that the basic similarities between the new word and the target word are significant and prominent enough. This capability is compatible with nonlinguistic research, and it is evident that language generalizations relate to the ability to make other cognitive generalizations (Goldberg, Casenhiser, & White, 2007). Beyond the linguistic implications regarding the categorization of language, this study provides evidence that knowledge acquired in various fields of cognitive psychology is able to clarify the language acquisition process in general and conceptualization processes in particular. Diesendruck’s (2003) findings also indicate that there is interplay between the area of specific cognitive abilities of categorization and linguistic abilities.

Classification and Analogical Reasoning

The foregoing observations suggest that classification is fundamental to the development of relatively higher order cognitive operations and concepts. Green, Fugelsang, and Dunbar (2006) have already shown that activation of category concepts plays a fundamental role in analogical thinking. In this study, we sought to determine the extent to which improving classification skills might promote development of or improvement in analogical reasoning. The concept of analogy refers to recognizing relations between objects or events based on their similarities and transferring them to a different field (Kaniel, 2001). Each new concept formation means stretching an analogical link between different but familiar areas. Analogical reasoning is an important element in creating expectations and dealing with problem solving (Gentner, 1983). Analogical reasoning involves categorization, classification, and learning processes, which provide thinking tools that enable explanation for discovery and creative thinking (Goswami, 1992). Analogy is a very important element for enhancing broader thinking skills such as problem solving (Novick & Basssok, 2005) and inductive thinking (Holland, Holyoak, Nisbett, & Thagard, 1986) and provides a tool for thinking and scientific discoveries (Goswami, 1992).

Sample

The sample comprised 94 Hebrew-speaking kindergarten children with developmental language delays, 4–6 years old, enrolled in kindergartens in Israel for children diagnosed with developmental language delays. The term developmental delay is applied when children do not achieve developmental milestones within the normal age range. All children in this sample had language or speech delays, which included the ability to communicate with other people effectively, express and receive information such as instructions on how to complete an assignment, and to form sentences. Some of the children had secondary developmental delays in motor, social, emotional, or intellectual development, although most were not diagnosed with secondary delays because of their young age. The children were randomly assigned to training (n = 45) and no-training (n = 49) groups. The training group, 30 boys and 15 girls, was taught in five different kindergartens by the kindergarten staff, all under the training and supervision of the same supervisor. Each staff member received 6 hr of training. The staff used the Classification unit of Bright Start, as a supplemental part of their regular integrated thematic (content-oriented) curriculum. The control group, 35 boys and 14 girls, continued receiving the thematic curriculum. Both groups received the same amount of teaching time. To examine whether the experimental and control groups differed in pretreatment variables (classification, analogical reasoning, verbal conceptualization) and background variables (age, socioeconomic status, gender), we conducted a series of preliminary analyses. The analyses revealed no differences between groups in age, t(92) = −0.32, p = .75, socioeconomic status, χ²(2) = 3.95, ns, or gender, χ−(−) = 0.71, ns.

Criterion Assessment

Procedure

All participants were given the CCPAM test, Katzenberger’s Semantic Categories Subtest, and the Classification Test, in that order, before and after the intervention program. All tests were administered individually in a quiet room in the kindergarten. The Classification unit of Bright Start was applied in 16 sessions during 2 months of the school year. Children in the training group received the Bright Start training twice a week for 20 min each time in small groups of five or six children. The program was a part of each participant’s individualized education plan. The children from the control group, instead of the classification training, continued with their regular content-oriented kindergarten curriculum. This curriculum, designed by the Ministry of Education for children with developmental language delays at kindergarten, is focused on language skills. Postintervention testing was done for both groups 2 months after the intervention ended.

Classification Skills, Verbal Conceptualization, and Analogical Reasoning in the Experimental and Control Groups

To examine for differential change in classification skills (categorization sorting, essentialness, and naming), we conducted a repeated-measures multivariate analysis of variance (MANOVA) of Treatment × Time × Type of Skill (2 × 2 × 3) where Treatment was an independent between-groups variable and the last two variables were within-subject independent variables. The dependent variables were the scores in percentages. The means and standard deviations are presented in Table 1.

The analysis revealed a significant main effect of Type of Skill, F(2, 91) = 53.11, p < .001, η_p² = .37. Sidak post hoc analyses indicated that participants achieved the highest scores on categorization sorting (M = 62.19, SE = 2.42), followed by naming (M = 48.01, SE = 2.23) and essentialness (M = 24.99, SE = 1.99). The skill types were significantly different from each other in both treatment groups (p < .001). The analysis also revealed a main effect of group, F(1, 91) = 6.77, p = .011, η_p² = .07, indicating that participants in the training group achieved higher classification scores (M = 49.51, SE = 2.47) than did those in the control group (M = 40.61, SE = 2.37). There was also a main effect of Time, F(1, 91) = 26.10, p < .001, η_p² = .22, indicating that postintervention scores on classification (M = 49.38, SE = 1.96) were higher than preintervention scores (M = 40.75, SE = 1.86). The main effects of Group and Time were moderated by an interaction of Group by Time, F(1, 91) = 14.40, p < .001, η_p² = .14. Simple effects tests with Bonferroni adjustment indicated that classification scores increased significantly in the training group (p < .001) but not in the control group. Other effects were not significant. Figure 1 portrays the significant interaction.

FIGURE 1.

Pre- and postintervention scores in classification skills.

Pre- to Postintervention Change in Verbal Conceptualization

To examine the pre- to postintervention change in verbal conceptualization (i.e., verbal retrieval time, verbal categorical retrieval time, verbal object similarity) and whether the treatment affected it differentially, we conducted repeated-measures MANOVA of Treatment by Time by Type of Skill, the last two variables being within-subject independent variables. The dependent variable was the score on these tests expressed in percentages of total possible score. The means and standard deviations are presented in Table 2.

This analysis revealed a significant main effect of Type of Skill, F(2, 91) = 73.10, p < .001, η_p² = .44. Sidak post hoc analyses indicated that participants achieved the highest scores on verbal retrieval time (M = 78.72, SE = 2.02), followed by categorical retrieval time (M = 73.04, SE = 2.24) and verbal object similarity (M = 60.50, SE = 2.42; all ps < .01). The analysis also revealed a significant main effect of Time, F(1, 91) = 36.85, p < .001, η_p² = .29, indicating that postintervention scores on verbal conceptualization (M = 74.66, SE = 1.97) were higher than preintervention scores (M = 66.85, SE = 2.12). The main effect of Time was moderated by a Treatment by Time interaction, F(1, 91) = 21.16, p < .001, η_p² = .19. Simple effects tests with Bonferroni adjustment indicated that verbal conceptualization skills significantly increased in the experimental group (p < .001) but not in the control group. Figure 2 portrays these relationships.

FIGURE 2.

Pre- and postintervention scores in verbal conceptualization skills.

Pre- to Postintervention Improvement in Analogical Reasoning

To examine the pre- to postintervention change in analogical reasoning and whether the treatment affected it differentially, we conducted repeated-measures MANOVA of Treatment by Time by Type of Analogy, the last two variables being within-subject independent variables. The dependent variable was the analogical reasoning score expressed in percentage of total possible score. The means and standard deviations are presented in Table 3.

The analysis revealed a significant main effect of Time, F(1, 91) = 40.48, p < .001, η_p² = .31, indicating that the preintervention analogical reasoning scores (M = 39.10, SE = 1.58) were lower than postintervention scores (M = 48.95, SE = 2.03). The main effect was modified by the interaction of Type of Analogy and Time, F(1, 91) = 9.72, p = .002, η_p² = .10. Simple effects test with Bonferroni adjustment indicated that perceptual analogies scores increased significantly from pre- to postintervention (p < .02), whereas no significant changes were found for conceptual analogies. The two-way interaction was further modified by the triadic interaction of Treatment by Type of Analogy by Time, F(1, 91) = 5.09, p = .026, η_p² = .05. Figure 3 portrays these effects.

FIGURE 3.

Pre- and postintervention scores in analogical reasoning.

Simple effects tests with Bonferroni adjustment revealed a significant effect of Group × Time for perceptual analogies, F(1, 91) = 6.48, p = .01, η_p² = .07, but not for conceptual analogies, F(1, 91) = 0.04, p = .95, η_p² = .00. These findings indicate that although perceptual analogies scores increased for children in both groups (experimental and control), the magnitude of increase was significantly greater in the experimental group (p < .001) than in the control group (p = .005). Other effects were not significant.

The Relation Between Verbal Conceptualization, Classification Skills, and Analogical Reasoning

To examine whether classification, conceptual analogies, and perceptual analogies abilities were systematically related to verbal conceptualization, we combined the pre- and postintervention scores for all children and conducted a series of Pearson product–moment correlations for the whole sample. The analyses revealed significant positive correlations between verbal conceptualization and classification, r(92) = .58, p < .001; conceptual analogies, r(92) = .37, p < .001; and perceptual analogies, r(92) = .31, p = .003.

Differences Among Classification Abilities

The most difficult component of the Classification Test was essentialness, followed by naming and categorization sorting (see Table 1 and Figure 1). The most demanding and abstract mental activity required in classification is finding the essential feature that is common for the four elements and verbalizing it correctly. Children might have an intuitive understanding of the category and would sort and categorize elements of a group correctly but might have difficulty understanding the abstract unique essence of the grouping, hence the lower scores in this component in both the trained and nontrained groups. The relatively higher scores on the naming component (third stage of the test) could be related to the fact that it was given after the activity of finding the essential characteristic of the group and the fact that it is much easier to provide a name to a group than to explain the essence of the commonality. It is also interesting to note that whereas the nontrained group showed an increase in performance in categorization sorting and in naming from pre- to postintervention, it showed a decrease in the essentialness component, whereas the trained group showed an increase. Thus, the Bright Start training was most effective in the component that is, empirically, the most difficult. The difference between the two groups can be attributed to the abstract classification activities given in the training.

Goldstone and Hendrickson (2010) reported that both object-to-label associations (considered as low-level cognitive functions) and decision making (considered as a high-level cognitive function) are influenced by category learning. Similarly, Goldberg et al. (2007) have shown that language generalizations relate to the ability to make other cognitive generalizations as well as making broad generalizations beyond their existing conceptual knowledge. Paour et al. (1992) applied two Bright Start curriculum units, Self-Regulation and Classification, with preschool and kindergarten children, most of them from immigrant families in the south of France, and assessed the effects on tests of classification and several other cognitive and academic performance variables. They found marked positive effects on the children’s performance on classification tasks, indicating that classification is responsive to training in young children, as well as a dramatic positive effect on the children’s reading of novel words.

Children who received classification training showed higher pre- to posttraining improvement than did the nontrained group on perceptual analogies but not on conceptual analogies. A possible explanation for the differential effects of classification training on the two types of analogy is to be found in the characteristics of the analogies. Conceptual analogies, by definition, are more abstract than are perceptual analogies and require understanding of higher order concepts, principles, and relations (Lifshitz et al., 2005; Lifshitz et al., 2010; Tzuriel, 2007; Vakil et al., 2011). The short-term intervention in this study with kindergarten children with developmental language delays was sufficient to improve performance of concrete perceptual analogies but was not transferred to the abstract conceptual analogies. Nevertheless, these children transcended the barriers of young age and cognitive delays and improved their analogical thinking. Furthermore, a normative sample of typically developing children (n = 64) of the same age (M = 5:6) in Tzuriel’s (2007) study achieved a score of 35.95% in perceptual posttesting analogies as compared with 54.03% of the current sample of children with developmental language delays participating in the intervention. The higher analogical thinking of these children can be attributed to the teaching and learning of classification. Another possible explanation for the difference in effects on perceptual and conceptual analogies is the observation that classification, as contrasted with categorization, entails practice in conceptualizing and specifying the logical relations that govern construction and expansion of classes. It is likely that the Classification unit of Bright Start emphasizes categorization more heavily than classification, leaving responsibility for conceptualization and explicitation (see Bruderlein, 2004, for a discussion of the role of explicitation) of such relations largely in the hands of the teachers. Thus, effects of the classification training might quite reasonably be principally at the perceptual level.

We wished further to examine the associations between language and higher level thinking. We found significant positive correlations between verbal conceptualization and classification skills and analogical reasoning. These findings are supported by contemporary research indicating the existence of interplay between the cognitive abilities of categorization, classification, and linguistic abilities (Diesendruck, 2003) as well as between categorization and analogical reasoning (Gentner & Christie, 2010). A possible implication of this finding is that intervention for classification may transfer to other cognitive domains of language and higher order thinking skills such as analogical reasoning.

Although the objectives of the study were achieved and have led to the conclusion that the Bright Start Classification unit yields gains in the cognitive and linguistic functioning of young children with developmental language delays and led to near, medium, and far transfer, there are some limitations to this study that must be addressed. One limitation is the lack of follow-up of the effects of the intervention program. Another limitation is the relatively brief duration and mild intensity of the program—16 sessions of 20–30 min, thus a maximum of 8 hr of training. It might well be that the effects of the program will fade out with time. We suggest therefore in future research to implement the program with different levels of length and intensity and carry out follow-up assessment at different time intervals. Further studies may examine the continuous development of the preschoolers as they enter first grade. Follow-up is needed to determine whether lessons taught in the Classification unit of Bright Start have lasting effects on children’s cognitive and language development, as previous researchers have found (e.g., Cèbe & Paour, 2000; Paour et al., 2000).

In addition to these considerations, it is important to examine whether the findings can be generalized to typical preschool children as well as other children with special educational needs, such as those formally diagnosed with SLI, autism spectrum disorders, intellectual disabilities, and sensory, especially hearing, impairment.