Developmental Perspectives on Metaphor

A Special Issue of metaphor and Symbolic Activity

Ellen Winner

• Publisher: Psychology Press
• Year: 2017
• Topic: Psychology
• eBook ISBN:
  9781317777793
• Subtopic: Cognitive Psychology & Cognition
• Index: Psychology

Research on the development of metaphor abilities in children can be dated back as far as 1960, with Asch and Nerlove's pioneering study, which concluded that children were unable to understand metaphors until middle or even late childhood. However, the study of metaphor in children did not take off until the 1970s; research continued to show metaphor as a relatively late-developing skill, based on children's inability to paraphrase correctly metaphoric sentences presented out of any situational or narrative context.

In the past decade, research into the development of figurative language has broadened considerably in scope. Efforts have been underway to demonstrate the cognitive underpinnings of the ability to make sense of figurative language and to demonstrate the role of metaphor and its cousin, analogy, in the development of cognition.

Metaphor is now considered to be a central aspect of language and thought and thus a crucial variable in cognitive development. The articles in this issue support the claim that no longer can any theory of language acquisition afford to ignore how children are able to recognize the distinction between what is said and what is meant and how they are able to grasp what is meant when people say things they do not mean.

Children’s Memory and Metaphorical Interpretation

Valerie F. Reyna

Department of Educational Psychology University of Arizona

Barbara Kiernan

Department of Speech and Hearing Services University of Arizona

Assumptions of constructivism, information-processing, and fuzzy-trace theory were tested in two experiments on memory and comprehension of metaphors with 6- and 9-year-old children. In the first experiment, verbatim memory for metaphors was compared with misrecognition of the gist of metaphorical meaning. In the second experiment, children judged metaphorical meaning, including alternative perceptual and psychological interpretations. We found that (a) children misrecognized the gist of metaphorical interpretations, especially after a delay, much like they do for literal gist (e.g., true inferences); (b) contrary to both constructivism and information-processing theory, misrecognizing metaphorical gist was independent of memory for the metaphor itself; (c) true psychological interpretations were misrecognized and judged acceptable more often than any other type at all ages; and (d) contrary to the idea that literal or perceptual interpretations are suppressed to achieve psychological interpretations, acceptances of alternative interpretations were positively dependent. Results were consistent with those obtained for literal language and for numerical information and are explained by fuzzy-trace theory.

Until recently, theorists assumed that memory and comprehension are closely connected (Bjorklund, 1989; Siegler, 1991). According to information-processing theory, comprehension depends on the ability to encode and sustain information in working memory, at least long enough to extract its meaning (Miller, 1956). The notion that memory and comprehension are related is also central to constructivism, which is the belief that understanding shapes memory. Although the roots of constructivism in cognitive development can be traced to Piaget (e.g., Piaget & Inhelder, 1973), interest was renewed in the 1970s by research on language (e.g., Bransford, Barclay, & Franks, 1972; Liben & Posnansky, 1977). Most recently, the relation between memory and comprehension was the focus of research on fuzzy-trace theory (e.g., Ackerman, 1992; Kreindler & Lumsden, 1994; Reyna, 1995, in press; Swanson, Cooney, & Brock, 1993).

In this article, we explore the relation between memory and comprehension, in particular, between the way children remember and the way they comprehend metaphors. The article proceeds as follows: In the first section, we briefly survey memory and comprehension from each of three developmental perspectives: Piagetian theory, information-processing theory, and fuzzy-trace theory. The remainder of the article is divided into two parts—one on memory and the other on comprehension—corresponding to two experiments. The first experiment concerns children’s verbatim memory for metaphors and their tendency to misrecognize the gist of metaphorical meanings. The second experiment examines an implication of the first regarding comprehension, namely, that young children entertain multiple interpretations of metaphors, including abstract psychological interpretations. In the final section, we discuss the relevance of our results to such issues as developmental differences in the interpretation of psychological metaphors (e.g., Winner, Rosenstiel, & Gardner, 1976), the relation between literal and metaphorical meanings (e.g., Glucksberg, 1991), and the indeterminacy of metaphorical meaning (e.g., Kittay, 1987; Reyna, in press).

Background

Theorists distinguish metaphors from other kinds of figurative language such as irony and divide them into three types: nominal, predicative, and sentential (Reyna, 1986; Winner, 1988). The three types differ depending on whether a noun phrase, a verb phrase, or a sentence is used metaphorically. For example, the entire sentence is used metaphorically when “The troops marched on” refers to children stubbornly refusing to cease some behavior. In the experiments we report in this article, we investigated nominal metaphors, as in “Juliet is the sun.” It should be noted, however, that developmental differences between nominal and predicative metaphors have been observed, indicating that the former are easier to interpret (Reyna, 1985, 1987).

We also differentiate fresh (or novel) from frozen metaphors. Although the “leg” of a table may have been a novel expression at one time, once such metaphorical resemblances become frozen, it is unlikely that they are processed in the same way as fresh metaphors (Reyna, 1986). Of course, metaphors differ in the degree to which they are fresh or frozen. We made an effort, however, to sample metaphors at the fresh end of the continuum, at which active reasoning processes are more likely to be observed.

Metaphor and Cognitive Development

It is widely presumed that children’s cognitive abilities constrain their interpretation of metaphors. The relation between cognitive development and metaphorical interpretation has been studied from two theoretical perspectives, Piagetian and information processing. Each of these perspectives posits general limitations on cognition, either logical or memorial, that purportedly affect the development of metaphorical interpretation.

The interpretation of novel metaphors can be viewed as a reasoning task, on analogy with classical Piagetian tasks such as class inclusion or probability judgment (e.g., Billow, 1975). For Piaget, cognitive growth was centered primarily on the development of logical reasoning abilities—the attainment of so-called operational thought (e.g., Falmagne, 1975). Thought gradually progressed through a series of stages, from prelogical intuition to formal logic in which it became less influenced by concrete perceptual factors and more influenced by abstract conceptual ones (e.g., Siegler, 1991). Although Piagetian stages have been called into question (e.g., Brainerd, 1978), the idea that “understanding metaphor is primarily a logical-analytic task” (Winner, 1988, p. 10) continues to prevail.

Specifically, current theories retain the Piagetian idea that young children are unable to appreciate abstract, as opposed to perceptual, resemblances (e.g., Keil, 1986; Winner, 1988). This inability to reason abstractly is assumed to hamper the interpretation of certain kinds of metaphors. For example, in interpreting the metaphor “The prison guard had become a hard rock,” younger children are expected to be more likely to ascribe physical attributes to the prison guard (e.g., he had hard muscles), whereas older children are more likely to ascribe abstract, psychological attributes (e.g., he was mean and unfeeling; Winner et al., 1976). In this view, which is consonant with Piagetian theory, cognition in general and language comprehension in particular are impelled by the development of reasoning.

In Piagetian theory, memory, too, is subordinated to reasoning (Piaget & Inhelder, 1973). Although memory itself was seen as relatively unimportant, its interaction with reasoning was emphasized. According to the general stance that Piaget advocated—constructivism—reasoning shapes the contents of memory. For example, children who were capable of making inferences from a text should be more likely to recall that text with the inferences (as if they had actually been part of the text) compared with children who lacked inferential ability (e.g., Brown, Smiley, Day, Townsend, & Lawton, 1977). Hence, children’s memories were assumed to reflect their level of logical reasoning (e.g., Paris & Lindauer, 1977; Perner & Mansbridge, 1983; Prawatt & Cancelli, 1976).

In contrast to Piagetian theory, information-processing theory stresses memorial limitations. Such limitations have been used to explain developmental differences in performance on a host of complex cognitive tasks (Bjorklund, 1989; Case, 1985; Daneman & Carpenter, 1980; Siegler, 1991; Trabasso, 1977). The common denominator in these accounts is the claim that memory capacity (or resources) influences the expression of higher cognitive abilities. The assumption that memorial limitations ought to influence metaphorical and other nonliteral interpretations was then incorporated into current models (e.g., Demorest, Silberstein, Gardner, & Winner, 1983; Vosniadou, 1987a; Vosniadou, Ortony, Reynolds, & Wilson, 1984; Vosniadou, Pearson, & Rogers, 1988; Winner et al., 1987). Unfortunately, the evidence on this point is equivocal.

For example, Winner et al. (1987) compared nonliteral interpretation in two conditions that differed in the demands placed on children’s memory. They found that although children performed better on one measure of nonliteral comprehension in the condition that minimized memory demands, they performed worse in that condition on the other comprehension measure. Similarly, although children performed better in the low-memory demand condition in comprehending literal falsehoods (i.e., speaker’s factual errors), they performed (nonsignificantly) worse in comprehending nonliteral falsehoods (i.e., sarcasm).

Although other studies reported results that are broadly consistent with the memory-limitation hypothesis, research on reasoning (e.g., Reyna, 1992; Reyna & Brainerd, 1990) suggests that such results are subject to two challenges. First, it is crucial to verify that ostensible memory manipulations actually affect memory (see Reyna, 1991, 1995). Rabinowitz, Howe, and Lawrence (1989), for example, also compared high and low “memory load” conditions, but mathematical modeling revealed that processing rather than memory parameters were affected by this manipulation. The implication of this and other research is that inferences about the effect of manipulations on memory can only be made by measuring memory. The second challenge has to do with what such measures of memory reveal. Previous research also indicates that the relation between memory and comprehension cannot be inferred solely by measuring memory. As we discuss later, memory and comprehension must be measured and the relation between them directly assessed.

In summary, theories of cognitive development predict that memory and comprehension of metaphors should be linked. For Piaget, memory should be constructed out of children’s understanding of a metaphor’s meaning, and the level of that understanding should change with age. For the information-processing theorist, understanding should depend on the demands placed on a limited memory system, and the ability to handle those demands should also change with age. In the experiments that follow, we present two kinds of data that bear on these hypotheses: (a) false recognition rates for information that was understood but not actually presented, and (b) dependencies between such false recognitions and memory for actually presented information. However, to understand how such data bear on these hypotheses, we must briefly review the literature on memory for literal language.

False Recognition of Unpresented Sentences

Although theorists long maintained that memory is constructive, evidence favoring such claims proved elusive. Bartlett’s (1932) early studies are often cited as supportive of constructivism, but his findings were difficult to replicate (e.g., Gauld & Stephenson, 1967). In the 1970s, convincing evidence seemed at hand. In their now classic experiments on constructive memory (e.g., Bransford et al., 1972; Bransford & Franks, 1971), Bransford and colleagues presented participants with literal sentences such as Three turtles sat on a floating log, and a fish swam beneath it. After a brief delay, a recognition test was given that contained presented sentences and unpresented sentences consistent with the gist of presented sentences such as Three turtles sat on a floating log, and a fish swam beneath them, as well as false sentences. The key result was that participants erroneously claimed that gist-consistent sentences had been presented, sometimes with greater confidence than for sentences that had actually been presented. Paris and Carter (1973) subsequently obtained similar results with children. These false-recognition effects seemed to demonstrate that adults and children remembered what they understood rather than what they experienced.

However, the findings did not go unchallenged (Alba & Hasher, 1983; Fletcher, 1992). For example, Flagg (1976) showed that misrecognition of true inferences, which had been taken to be evidence for constructive memory, could be predicted by a tally model involving memory for verbatim phrases from presented sentences. (For example, given presented sentences such as The bird is in the cage and The cage is under the table, participants may accept the true inference The bird is under the table because it contains phrases that were presented.) Similar false-recognition effects were demonstrated using nonsense syllables, suggesting that understanding had little to do with these phenomena (Reitman & Bower, 1973; Small & Butterworth, 1981). Hasher and Griffin (1978) argued compellingly that there were two parallel, but contradictory, literatures on memory: One claimed that memory was “remarkably accurate—or reproductive” (p. 318) and the other claimed the opposite (see also Hasher, Attig, & Alba, 1981). This contradiction could not be explained away by differences in materials or other superficial factors.

Therefore, although textbooks routinely characterize memory as constructive, contradictory evidence has been neglected. Hasher and colleagues’ work (see also Alba & Hasher, 1983) pointed out these fundamental contradictions, although they were not resolved theoretically. The main issue dividing the two camps was and remains the validity of the false-recognition data. Although there were disputes about the data, observers on both sides accepted the premise that, if the effects were real, memory had to be constructive.

Fuzzy-trace theory introduced a new possibility, namely, that the effects were real but that memory was not constructive. Using fuzzy-trace theory, Reyna and Kiernan (1994) modified instructions and stimuli to address methodological criticisms of earlier sentence recognition experiments. The aim was to obtain unambiguous demonstrations of the classic effects—false recognition of gist—to determine whether they were explained by constructivism. If subjects systematically misrecognized unpresented true sentences, constructivism would have entailed positive dependency between memory judgments for presented sentences a...

Table of contents

• Cover
• Contents
• Introduction
• Understanding Versus Discriminating Nonliteral Utterances: Evidence for a Dissociation
• Understanding Minds and Metaphors:Insights From the Study of Figurative Language in Autism
• Analogical Reasoning in Cognitive Development
• Children’s Memory and Metaphorical Interpretation