| 9 | Why are Abstract Concepts Hard to Understand? |
Most of us possess the intuitive feeling that abstract word concepts are harder to understand than concrete word concepts. If you ask people why this is do, their answer would most likely reflect the belief that abstract words are hard to understand because they lack the fairly direct sensory referents that concrete words have. In fact, in cognitive psychology, it has long been assumed that it is this difference between these two concept types that makes abstract concepts harder to understand, process, acquire, remember, and so on than concrete ones. In this chapter, I use concreteness as a generic term to refer generally to the constellation of variables that distinguish abstract words from concrete words. However, the term is used merely as a convenience for differentiating among word types and, by itself, should not be taken to imply that abstract words are harder to understand than concrete words because they lack direct sensory referents. Instead, my purpose here is to evaluate that basic explanation.
Understanding the representational, developmental, and processing differences between abstract and concrete words is critical to developing an understanding of word meanings in general. There is a considerable amount of evidence that concreteness represents a fundamental semantic distinction among words. Concreteness invariably emerges as an important variable that differentiates words in large factor analytic studies (DiVesta & Walls, 1970; Paivio, 1968; Rubin, 1980). Yet, most current theories of word meaning do not directly address this basic distinction. A full theory of word meaning must include information regarding how words may differ in their representations, how these differences are acquired developmentally, and how such variations are reflected in verbal processing. The study of concreteness effects should help us move toward this ultimate goal.
In this first section, I review the evidence that abstract words are harder to understand than concrete words in order to establish that there is, indeed, something important that a theory of word meaning must explain. As we shall see, although concreteness effects in verbal processing are ubiquitous, they are not uniformly obtained. Despite this nonuniformity, however, some regularities are discernible. Following that, I describe and evaluate various theories that have been proposed to account for why abstract words are hard to understand. Such theories are important to the development of a psychology of word meaning because of the varying claims they make with regard to representations and processes associated with the meanings of words.
ARE ABSTRACT WORDS HARDER TO UNDERSTAND THAN CONCRETE WORDS?
There are several different kinds of evidence that are relevant to deciding whether abstract words are harder to discern than concrete words. Some of this evidence revolves around the initial processing of abstract and concrete verbal materials. Other evidence involves memory for abstract and concrete verbal materials after they have already been processed and stored in long-term memory. There are several recent reviews of research on concreteness effects in verbal memory by Marschark, Richman, Yuille, and Hunt (1987) and by Paivio (1986). These reviews highlight the fact that 25 years of research has shown us that concreteness effects in verbal memory are often large and reasonably consistent, but that the cause of these concreteness effects in long-term memory is not as yet fully understood. Instead, this review focuses on studies examining the initial processing, development, and representation of abstract and concrete verbal materials and does not deal extensively with studies examining memory for abstract and concrete verbal information.
The evidence regarding concreteness effects occurring early in verbal processing takes several forms. Some of this research focuses on the timed comprehension of abstract and concrete sentences. Other studies center on lexical processing of abstract and concrete words using tasks such as lexical decision and naming. Still other research examines the acquisition of abstract and concrete words in children’s oral and reading vocabularies. I examine each of these in turn.
The Comprehension of Abstract and Concrete Sentences
If abstract words are harder to understand than concrete words, then it would seem to follow that sentences composed of abstract words should also be harder to understand than those composed of concrete words. This greater difficulty in comprehension for abstract sentences should be reflected in generally longer processing times for abstract than concrete sentences. Such evidence might be important for deciding the general issue of whether abstract concepts are more difficult to understand than concrete words.
There are two tasks that have been used to gain evidence regarding the issue of whether abstract sentences are harder to understand: (1) sentence verification, and (2) timed sentence comprehension. The most commonly used task in examining concreteness effects in sentence comprehension has been the sentence verification task. In this task, persons are asked to make some kind of timed meaningfulness judgment regarding the abstract or concrete sentence. Faster verification times are assumed to reflect faster comprehension. However, it is important to note that sentence verification actually involves both the comprehension of the sentence and a decision regarding its truth value (Glass, Millen, Beck, & Eddy, 1985). Thus, we can’t be sure whether the finding of concreteness effects in sentence verification reflect processes operating during comprehension or during the decision regarding the truth value of a sentence already comprehended. On the other hand, Lorch (1982) has noted similar semantic processing in a category verification task (which is similar to the sentence verification task in that it requires a decision) and an instance naming task (which does not involve such a decision). To some degree, then, studies on the verification of abstract and concrete sentences might provide some initial insight into the question of whether abstract verbal materials are harder to understand.
Sentence Verification. Studies comparing the verification of abstract and concrete sentences have ranged from those using simple semantic statements of the sort typically used in standard semantic memory experiments (e.g., Glass & Holyoak, 1975; Smith, 1978) to those using more complex, but ordinary sentences. For example, in a highly cited study by Jorgensen and Kintsch (1973), subjects were asked to verify high and low imagery statements such as “Truck has oil” (low imagery) and “Book has cover” (high imagery). Despite the fact that all of the content words used in all the statements were very concrete, some sentences were considered low imagery because subjects had rated the statements taken as a whole to be less imageable than their high imagery counterparts. In that study, high imagery true statements were verified 571 ms faster than the low imagery true statements. Thus, imageability was concluded to have a beneficial effect on retrieving information from semantic memory.
This early conclusion regarding the role of imagery in semantic verification has been questioned by Holyoak (1974), however, who suggested that the faster verification times of the Jorgensen and Kintsch (1973) study might be accounted for by factors other than imagery. His subjects rated the low imagery sentences used by Jorgensen and Kintsch as being less easy to comprehend, having more difficult verbs, and having more semantically related predicates than the high imagery sentences. Furthermore, because all the sentences used in the study possessed highly concrete content words, it does not completely address the issue of whether semantic verification is more difficult for abstract sentences than concrete sentences.
Glass, Eddy, and Schwanenflugel (1980, Experiment 2) attempted to address more directly the issue of whether simple semantic verifications for abstract statements are harder to make than for concrete statements by comparing the verification of abstract sentences such as “Every idea is a thought” with concrete sentences such as “Every shotgun is a weapon.” Their study attempted to minimize the influence of possible confounding variables such as perceived comprehensibility, intersubject agreement in truth value, and semantic relatedness of subjects and predicates. They found semantic verifications to be no faster for concrete than abstract statements, although accuracy was higher for the concrete statements. Thus, these studies using simple semantic verification have proven to be indeterminate with regard to the issue of whether abstract verbal materials are harder to understand than concrete ones, but are in the direction of suggesting a concreteness advantage.
Other studies have asked people to verify the general meaningfulness of more elaborate, ordinary abstract and concrete sentences in order to examine the question of whether abstract sentences are harder to understand than concrete ones. Klee and Eyesenck (1973), for example, asked subjects to decide whether concrete sentences like “The veteran soldier rode the lame horse” and abstract sentences such as “The wrong attitude caused a major loss” were meaningful, plausible sentences. They found an overall 420 ms benefit for concrete over abstract sentences, although many of the plausible abstract sentences used in that study have also been criticized for not being very plausible (Holmes & Langford, 1976; Moeser, 1974). Moreover, subjects were asked to perform the verifications with a memory load.
A study by Belmore, Yates, Bellack, Jones, and Rosenquist (1982) examined the verification of three kinds of statements derived from abstract and concrete sentences that subjects had just encoded: (1) identical statements in which subjects were represented with the original encoded sentence, (2) paraphrase statements in which subjects were presented with a paraphrase of the original encoded statement, and (3) inference statements which were reasonable from the original encoded statement. Overall, verifications took longer for abstract sentences than concrete sentences for all types of verification statements with the difference being larger in the paraphrase and inference cases.
Probably the clearest case for the position that abstract sentences are verified more slowly than concrete sentences comes from a well-controlled study by Holmes and Langford (1976). In that study, abstract and concrete target sentences of varying syntactic form that were highly plausible and meaningful were used and, yet, a 183 ms concreteness benefit was noted.
The cumulative evidence from all of these studies tells us that abstract sentences are, indeed, harder to verify than concrete ones. But this does not mean that all concrete sentences are easier to verify. Glass and his colleagues (Glass et al., 1980; Glass et al., 1985) point out that we may need to distinguish between two types of concrete sentences: (a) concrete sentences that seem to require the explicit formation of images for their verification, and (b) concrete sentences that do not require such image formation. For example, the concrete sentence “A pool table has six pockets” intuitively seems more likely to evoke imagery formation as a potential verification strategy than does an equally concrete sentence such as “A canary has wings.”
This distinction between sentences requiring imagery and sentences not requiring imagery for verification seems an important one for identifying processes that may become involved in verification but not comprehension. In several studies, highly concrete sentences requiring imagery formation have been found to take longer to verify than other concrete sentences even though they do not necessarily take longer to comprehend (Glass et al., 1985). Consequently, it is important to distinguish between truly abstract sentences and different kinds of concrete sentences in deciding the issue of whether abstract sentences are harder to verify than concrete sentences. However, for most ordinary concrete sentences, it does appear that concrete sentences are verified faster than abstract sentences.
Sentence Comprehension. Given that sentence verification entails extra processes beyond simple comprehension, it seems reasonable to continue to ask whether abstract sentences are specifically more difficult to comprehend than concrete sentences. The problem with asking this question directly is that comprehension is not a precisely defined concept. Can we say that comprehension has occurred when one has merely skimmed a sentence to get the vaguest sense of its gist? Or do we wish to use a more stringent criteria for determination of comprehension such as being able to remember or answer questions about the materials some time later? How we ask this question may determine whether we find a concreteness advantage in sentence comprehension.
Two experiments conducted at the extremes of this question uniquely illustrate this point. In an early study on the question of whether concreteness influences the on-line comprehension of abstract and concrete sentences, Paivio and Begg (1971) instructed subjects simply to “release the button when they understood the sentence.” In that study, the approximately 300 ms difference between abstract and concrete sentences was not significant. On the other hand, it should be noted that relatively small numbers of subjects were used and also that sentence comprehension times correlated quite highly with sentence imaging times questioning the reliability of this nonsignificant effect. At the other extreme is a study by Moeser (1974). Subjects in that study were asked to read the sentences in order to detect meaning or wording changes in an upcoming test. Comprehension times were quite long (around 10 seconds), but a significant 2850 ms concreteness benefit was found. Clearly, in this latter study, subjects were both comprehending and attempting to encode the sentences into long-term memory, leaving the precise locus of the concreteness effects unclear.
This distinction between comprehending for later recall and comprehending for possibly marginal understanding emerges clearly in a study by Marschark (1979). In that study, some subjects were simply asked to comprehend the abstract versus concrete passages and others were asked to read it to be able to recall the passages. In the cases where subjects had to read for simple comprehension, only marginally significant concreteness benefits were found (and subsequent recall was also very low). When the task was to recall the materials later, large and significant concreteness benefits were found.
Most studies have opted for some middle ground with regard to the goal of the comprehension task, ensuring that subjects read with comprehension without enforcing unusual memory encoding demands. For example, Haberlandt and Graesser (1985) asked people to be prepared to complete a multiple choice task after reading the passages and found that sentence imagery was negatively related to comprehension time. Similarly, Schwanenflugel and Shoben (1983) asked their subjects to perform an immediate verification of a simple partial paraphrase following the comprehension of abstract and concrete sentences presented in isolation. Again, they also found that abstract sentences take longer to comprehend than concrete sentences.
In sum, most studies find that persons take longer to comprehend abstract than concrete sentences. However, the degree to which this is true is probably affected by the way comprehension is defined in the experiment.
Conclusion. What conclusions can we draw from these disparate findings regarding concreteness effects in sentence verification and comprehension? Overall, I think we see a decided trend for abstract sentences to be both harder to comprehend and to verify than concrete sentences. In most cases, even when abstract sentences do not take significantly longer to comprehend, the means are almost invariably in the direction of their taking longer to comprehend. However, it is also clear that task demands will influence the degree to which these effects surface. For example, when persons deem imaging to be strategically necessary to complete verification, this additional transformational and constructive imaging may cause concrete sentences actually to take longer to verify than abstract sentences. When abstract and concrete sentences are to be encoded into long-term memory, we see that the processing of abstract sentences particularly suffers, producing especially longer comprehension times.
To the degree that we can isolate a specific comprehension stage in processing, then, it does appear that abstract sentences take longer to comprehend than concrete sentences. Whereas we might infer that abstract words are harder to understand than concrete words from this evidence, it is also important to realize that sentence comprehension is likely to engage processes that understanding the meanings of words alone does not. To the degree that these other sentence-level processes are also differentially affected by the concreteness of the lexical items within sentences, evidence regarding the comprehension of abstract and concrete sentences only serves as indirect evidence for the proposition that abstract words are harder to understand.
Concreteness and Lexical Processing
More direct evidence for the view that abstract words are hard to understand might be expected to come from studies specifically examining lexical processing of abstract and concrete words. In this section, I focus on studies using one of two target tasks: (1) the lexical decision task, and (2) the naming (sometimes called pronunciation) task. In the lexical decision task, subjects are presented with abstract or concrete target words and are asked simply to decide whether each item represents a real English word as quickly and as accurately as possible. In the naming task, subjects are also presented with abstract or concrete words, but instead are asked to name them quickly. It is important to note that neither task requires the subjects to retrieve the meaning of the words they are naming or making lexical decisions for. Yet, as we shall see, very often concreteness effects emerge in lexical decision and naming. Therefore, to the degree that such basic semantic differences surface in such tasks, it would mean that concreteness has its effects at a very early stage in lexical processing and that retrieval of word meaning occurs regardless of the explicit demands of the task.
In order to address the question of whether there are concreteness effects in early processing, I focus on evidence from studies examining the processing of abstract and concrete words presented either in isolation or in semantically neutral contexts. Focusing on such studies will provide us with the most direct case for deciding the issue of whether abstract words are generally difficult to understand. On the other hand, it is important to keep in mind that the issue is likely to turn out quite differently when studies examining the processing of abstract and concrete words in meaningful contexts are also included. However, as we shall see later, this distinction between the processing of abstract and concrete words in context or in isolation may be useful for deciding between various theories of concreteness effects.
Lexical Decision. The evidence regarding whether abstract words take longer to process than concrete words is quite mixed for lexical decision. Some researchers have found that abstract words have longer lexical decision times than concrete words (Akin & Schwanenflugel, 1990; Bleasdale, 1987; deGroot, 1989; Howell & Bryden, 1987; James, 1975, Experiments 1 and 2; Kroll & Mervis, 1986, Experiment 2; Rubin, 1980; Schwanenflugel, Harnishfeger, & Stowe, 1988; Schwanenflugel & Shoben, 1983; Whaley, 1978), whereas others have not (Gernsbacher, 1984; James, 1975, Experiments 3 and 4; Kroll & Merves, 1986, Experiment 1; Richardson, 1976). As we see below, it is not entirely clear what the contributors are to these diverse findings.
Some experimenters have taken a correlational approach to the question of whether abstract words take longer to understand than concrete ones. In these studies, subjects are presented with a random variety of words ranging on many dimensions along with concreteness and no attempt is made to select words only from the extremes of the concreteness and imagery scales. From the subjects’ point of view, the task is simply to decide whether a particular item is or is not a real English word. They are not told of the various characteristics of the items they will be making decisions about. Nor are they told that they should try to retrieve the meanings of the various words as they make their decisions. Usually, lexical decision times for particular words are averaged across subjects and these are then correlated with concreteness and imagery ratings, among other variables.
In almost all of such studies conducted to date, rated concreteness and, especially, imagery have been found to be significantly correlated with lexical decision time. Whaley (1978) noted a correlation of –.41 between imageability and lexical decision time, and a somewhat smaller correlation of –.22 for concreteness. Similarly, Rubin (1980) reported a significant correlation of –.27 between imagery and lexical decision, and a correlation of –.19 for concreteness. Howell and Bryden (1987) report correlations between imageability and lexical decision time ranging from –.19 to –.32, when word length and word frequency were partialed out of both variables. Moreover, the size of these correlations did not vary consistently as a function of the visual field in which the words are presented. Schwanenflugel, Harnishfeger, and Stowe (1988) found a significant correlation between lexical decision time and imageability (r = –. 15, p <.05), but not concreteness (r = –.11, p >.05), when word length and word frequency was partialed out. Thus, there seems to be an overall small, but significant relationship between the imageability of the concept to which the word refers and the time it takes people to make a lexical decision.
Other studies have taken an experimental approach attempting to control for or manipulate potential nuisance variables that are associated with lexical imageability. A number of both experimental and lexical variables have been implicated in the appearance and disappearance of concreteness effects in lexical decisions.
One experimental variable that has been associated with the appearance of concreteness effects in lexical decisions is distractor type. James (1975) has argued that, when pronounceable distractors are used, subjects will have a greater tendency to retrieve semantic information surrounding a word as a way of providing evidence for the lexical decision. He showed that the use of pronounceable nonwords as distractors is associated with the finding of significant concreteness effects, but the use of unpronounceable distractors is not. However, Richardson (1976) also compared lexical decisions for abstract and concrete words embedded in an experimental context of pronounceable or unpronounceable distractors and did not find significant concreteness effects in lexical decisions in either case. Neither was such an effect of distractor type revealed in a study by Gernsbacher (1984), who also did not find concreteness effects. On the other hand, both the Richardson and Gernsbacher studies employed very small numbers of subjects in each condition, so it is not clear how reliable their findings might be. Consequently, it is not clear what the influence of distractor pronounceability is, if any.
Another experimental factor that has been implicated in the appearance and disappearance of concreteness effects in lexical decision is whether the abstract and concrete items are presented in a blocked manner or in randomly intermixed trials. Kroll and Merves (1986) showed that, when abstract and concrete items are blocked, concreteness effects are much less evident than when they are intermixed. This is particularly true if a block of abstract words is completed first. Exactly how this blocking by concreteness works to influence the kind of information subjects retrieve for making lexical decision is not yet clear.
Several lexical factors have also been implicated in the appearance and disappearance of concreteness effects in lexical decision. Among these lexical factors is word frequency. One proposal is that concreteness effects in lexical decision are noticeable only for low frequency words. Thus, one would expect to find larger concreteness effects for low frequency than high frequency words. This basic trend has been noted in several studies. For examples, James (1975) found larger concreteness effects for low frequency words when pronounceable non-words were used. Kroll and Merves (1986) found a similar interaction between concreteness and word frequency when abstract and concrete words were presented in an intermixed experimental setting. DeGroot (1989) also found this interaction between word imageability and frequency in lexical decision. Thus, it seems that the degree to which marked concreteness effects are displayed in lexical decision is related to the frequency with which a lexical item appears in the language.
Of these lexical factors, several related semantic factors have been implicated in the appearance and disappearance of concreteness effects. One of these is experiential familiarity. In a study by Gernsbacher (1984), subjects rated a set of abstract and concrete words on their familiarity with the word’s referent in daily experience. Then, abstract and concrete words rated similarly on this variable were presented in a lexical decision study. No significant difference in lexical decision time between abstract and concrete words was found. On the other hand, there were two potential problems with this study. First, the experiments suffered from a low number of subjects which might influence the reliability of the findings. Second, controls for word frequency and word length were not reported, so we cannot be sure that the abstract words used in the study did not have higher overall actual frequency than concrete words.
Another semantic variable implicated in the appearance and disappearance of concreteness effects in lexical decisions is what has been termed context availability. In an early unpublished study conducted on this topic in my laboratory, I asked subjects to rate a large number of nouns on the ease with which they could think of a context or circumstance in which each word could appear. I believe these ratings reflect the ease with which persons are able to retrieve associated information from long-term memory for each word. These ratings showed that lexical concreteness is highly correlated with rated context availability, so that people usually find it harder to think of associated contextual information for abstract than concrete words. I, then, identified a set of the abstract and concrete words for which this confound did not seem to apply (that is, words that had been rated similarly on context availability and not differing on word length and frequency) and presented them in a lexical decision task. No concreteness effect in lexical decision surfaced in that study. However, given the somewhat spotty history of concreteness effects in lexical processing, it seemed necessary to demonstrate that concreteness effects are obtainable when concreteness also covaries with context availability, as they do for most abstract and concrete nouns. In a follow-up study (Schwanenflugel et al., 1988), we were able to show that concreteness effects appear only when lexical concreteness is confounded with rated context availability. We have since been able to demonstrate this effect for abstract and concrete words acquired early and for 10-year-old children (Akin, 1989; Akin & Schwanenflugel, 1990).
One problem with taking the approach of controlling for nuisance semantic factors associated with lexical concreteness is that, very often, by controlling for such factors, we have eliminated part of the difference that makes abstract words distinguishable from concrete words. Lexical concreteness is significantly correlated with numerous semantic variables. In fact, factor analytic studies by Rubin (1980) and Paivio (1968) show fairly conclusively that lexical concreteness reflects a constellation of semantic variables that covary. If part of the goal of studying abstract and concrete words is to discern how they differ semantically and how those semantic differences affect their processing, then examination of these correlated variables might also be used to enhance our understanding of why abstract words are harder to understand than concrete words.
All in all, then, it does appear that lexical decisions for abstract words take longer to make than concrete words. However, it seems that subtle characteristics of the experimental situation may also influence the degree to which such concreteness effects are shown.
Naming. In recent years, it has become clear that the task of naming upcoming words has somewhat different properties than lexical decision. Lexical decision times have been shown to reflect both a lexical access component and a decision component (Seidenberg, Waters, Sanders, & Langer, 1984; Tanenhaus, Carlson, & Seidenberg, 1985), whereas naming times seem to reflect primarily the lexical access component. It is possible that the concreteness effects shown in those studies are the result of the decision component of lexical decision, and not the lexical access component. For example, lexical decisions appear to be highly affected by semantic information (Balota & Chumbley, 1984; Balota & Lorch, 1986; Chumbley & Balota, 1984). Therefore, it is a reasonable possibility that lexical concreteness influences the decision component of lexical decisions rather than the lexical access component.
In fact, the evidence with regard to the issue of whether lexical concreteness affects the naming of words is quite mixed, perhaps more than it is for lexical decisions. Whereas some studies have found significant concreteness effects in naming (Bleasdale, 1987; deGroot, 1989; Rubin, 1980; Shoben, Wilson, & Schwanenflugel, 1990; Schwanenflugel & Stowe, 1989), about as many have not (Brown & Watson, 1987; Coltheart, Laxton, & Keating, 1988; Richardson, 1976; Supraner & Kroll, cited in Kroll & Merves, 1986). Moreover, when significant concreteness effects in naming are found, they are sometimes quite small. For example, the effect was only 7 ms in the deGroot (1989) study and 11 ms in the Bleasdale (1987) study. On the other hand, Shoben, Wilson, and Schwanenflugel (1990) have not found concreteness effects to vary as a function of task. It currently is not quite clear why some studies show significant concreteness benefits and others do not.
Surveying these studies, it seems that one potential contributor to the finding of concreteness effects in naming is the tendency of the experimental context to encourage semantic processing of the words that are named. For example, in many studies conducted in my laboratory, we typically present abstract and concrete words alternately in meaningful and neutral contexts. We have found concreteness benefits in naming for the abstract and concrete words presented in neutral contexts to be generally greater than 40 ms in size (Schwanenflugel & Stowe, 1989; Shoben, Wilson, & Schwanenflugel, 1990). Moreover, we have also found that the presence of unrelated or incongruous context trials seems to reduce significantly the size of the concreteness effects displayed in both naming and lexical decision (Shoben, Wilson, & Schwanenflugel, 1990). I have also noted such a trend in my previous lexical decision studies. For example, in the Schwanenflugel et al. (1988) study, a very small correlation between lexical imageability and lexical decision time was noted when words were presented in isolation (r = –.15 for Experiment 2). However, when some trials engaged subjects in meaningful sentence processing, we found the correlation between lexical imageability and lexical decision time to increase marginally for the abstract and concrete words presented in a neutral context (r = –.31 for Experiment 3). In the literature, all studies noting nonsignificant concreteness effects in naming have been ones in which words were always presented in isolation and in which no trials engaged subjects in meaningful contextual processing. Thus, the experimental context may serve to influence the degree to which people retrieve lexical meaning as part of the naming task. Consequently, we can tentatively conclude that when the experimental context encourages meaningful processing, concreteness effects in naming are more likely to be found.
Conclusion. In the majority of studies examining concreteness effects in lexical processing, it seems that most find reliable concreteness effects on processing time. However, four factors do seem implicated in the finding of significant context effects in lexical processing. First, blocking by concreteness seems to have not well understood effects on the presence of concreteness effects. Second, low word frequency seems to make the occurrence of concreteness effects more likely. Third, when subjects are engaged in processing meaningful contexts on some trials, the occurrence of concreteness effects in neutral trials is more likely. Last, concreteness effects in lexical processing seem to be more prevalent in lexical decision tasks than naming tasks.
These last three factors seem to suggest that, when meaning retrieval is required by the task, concreteness effects in lexical processing are likely to be displayed. That is, it would seem to be a reasonable strategy to attempt to retrieve some disambiguating semantic information while processing low frequency words, particularly in a lexical decision task. Similarly, when meaningful semantic processing is stressed in many of the experimental trials (as it is in studies examining context effects in lexical processing), it is likely that persons will continue to attempt to retrieve semantic information in the processing of abstract and concrete words presented in contextually neutral trials as well. Last, the tendency of persons to retrieve disambiguating semantic information in the decision component of lexical decision is widely agreed upon, and reflects itself here in the tendency of concreteness effects to surface in lexical decisions over naming. Thus, as for the discussion of concreteness effects in sentence processing, it can be concluded that concreteness effects are likely to appear in lexical processing when deeper semantic processing is required by the task.
Concreteness Effects and Vocabulary Acquisition
Another source of evidence that abstract concepts are harder to understand than concrete concepts is the observation that abstract words are acquired later than concrete words developmentally (Brown, 1957). This evidence comes from studies examining both the development of a spoken vocabulary, the acquisition of a reading vocabulary, and the understanding of basic meaning categories (called ontological categories) in children. I discuss each of these in turn below.
The Acquisition of Abstract and Concrete Words in Spoken Vocabulary. The tendency for young children to acquire words referring to concrete entities in their spoken vocabulary prior to words referring to abstract concepts is incontestible. Children’s first nouns invariably refer to concrete objects in their environment. A survey of the early infant language corpuses assembled by Gentner (1982) and by Nelson (1973) makes it clear that none of the early nouns acquired by children refer to abstract concepts of any sort.
This same tendency towards the acquisition of concrete over abstract words continues throughout early childhood. In a classic study on this topic, Brown (1957) compared the lexical productions of 1st-grade children and adults in order to discern whether children possessed a bias toward naming concrete actions and objects. To compare the overall concreteness of children’s vocabulary with adults’, he compared the 1000 words most often used by 1st-grade children as defined by the Rinsland (1945) count with the 1000 words most often used by adults as defined by the Thorndike-Lorge (1944) count. The Rinsland corpus, the only existing corpus of its kind, was derived from conversations, letters, and stories from a large number of 1st- through 8th-grade children. Because of its emphasis on children’s productions rather than children’s reading materials (which are largely written by adults), it provides a unique window into the productive vocabularies of children. In the Brown study, a word was considered concrete if its common referent had a “characteristic visual contour and size” (in the case of nouns) or named “animal (including human) movement” (in the case of verbs; p. 2). He found that a full 75% of these most often used children’s nouns and 67% of the verbs were concrete, whereas only 28% of the most often used adult nouns and 33% of the verbs were concrete. This suggests that there is a wide disparity between adults and children in their relative use of abstract and concrete words.
Unfortunately, these dramatic results do not tell us much about how long these disparities in word use between adults and children persist developmentally. In order to examine this issue, I determined the typical age of acquisition as defined by the Rinsland corpus for the 365 nouns used in the concreteness norms reported by Schwanenflugel et al. (1988). Specifically, I selected the 198 nouns that had an adult word frequency of greater than 30 per million in the Kucera and Francis (1967) adult corpus, with the view that words of such frequency are likely to be at least heard by children. These 198 nouns were compared with the Rinsland corpus for grades 1 through 8 and the word was classified as acquired if the item appeared at least twice in the corpus for each grade. Concrete words were defined as those words that had a rated imagery value of greater than 4.5 (on a 7-point scale) and abstract words were defined as those that had ratings of less than 4.5. The proportion of words acquired for each word type at each grade level can be found in Fig. 9.1.
What is clear from this analysis is that the disparity in the acquisition of abstract and concrete words persists developmentally until early adolescence. For these relatively high frequency words, the acquisition of concrete nouns is nearly 80% complete by 1st grade, showing very little vocabulary growth thereafter. On the other hand, the acquisition of abstract words does not achieve a similar level until around Grade 7.
A similar developmental trend is reported by West, Martindale, and Sutton-Smith (1985) in their study of children’s spontaneous fantasy narratives. Even though fantasy narratives would seem to encourage the use of highly imageable, concrete nouns and verbs at all ages, they found a significant correlation between age and noun and verb concreteness. That is, as children got older, the nouns and verbs they used tended to be less and less concrete. Thus, this trend toward greater lexical abstractness with age is a highly reliable and long-lasting developmental progression.
Fig. 9.1. The percentage of high frequency abstract and concrete words acquired by children in Grades 1 through 8 as defined by Rinsland (1945) child language corpus.
Acquisition of a Reading Vocabulary for Abstract and Concrete Words. Another type of evidence that abstract words are acquired later than concrete words comes from studies of the development of reading vocabularies in early elementary school children. Some of these have been controlled teaching studies in which young children were first instructed on a set of abstract and concrete words that they did not possess in their initial reading vocabulary, and then tested on the degree to which the words had entered the children’s reading vocabulary (Kiraly & Furlong, 1974; Richmond & McNinch, 1977; Yore & Ollila, 1985). In the Richmond and McNinch (1977) study, 1st-grade children were taught four abstract and four concrete words. No significant difference was found in the number of trials to criterion needed to acquire the abstract and concrete words, but the means were in the direction of a concreteness advantage. In a very similar study, Kiraly and Furlong (1974) did find a concreteness advantage in learning to read concrete over abstract words. Yore and Ollila (1985), using a larger number of words of each type, also found a significant 6.4% concreteness benefit in acquiring concrete words over abstract words in 1st-grade children’s reading vocabularies. Thus, of the few studies that have been conducted on this topic, early readers have generally found abstract words harder to learn to read than concrete words.
Other evidence suggests that this difference in reading vocabulary for common abstract and concrete words may be fairly persistent developmentally. In a study of the reading behavior of 10-year-old children, Coltheart, Laxton, and Keating (1988) noted a 6% difference in the reading accuracy of high frequency abstract and concrete words, with this concreteness benefit being larger for poor readers than good readers. Moreover, Akin and Schwanenflugel (1990; Akin, 1989) found that, even for words that children did know, concreteness effects in lexical processing time were considerably larger for 8- and 10-year-olds than adults (see also Wolf & Gow, 1986).
The Development of Basic Meaning Categories. Another source of evidence related to the proposition that children acquire the meaning of concrete words prior to abstract words comes from studies examining the development of ontological knowledge in children. Ontological knowledge is defined as “people’s beliefs about what the basic categories of existence are” (Keil, 1983, p. 104). An ontological category is distinguished by the kinds of predicates it can take. For example, the ontological category Living Things (such as birds and flowers) can take potential predicates such as “is dead,” “needs food,” which could never be true for Nonliving Things (such as houses or bricks) or Abstract Concepts (such as truth or happiness). Keil (1979) reported an interesting series of studies examining the degree to which children understood that various kinds of nouns could take particular kinds of predicates and showed that young children often confuse abstract concepts as being able to take predicates associated with object categories and living things. Consequently, it appears that children do not have the distinct ontological knowledge associated with abstract concepts as a whole and that such knowledge is acquired rather late in development.
Conclusion. There is substantial evidence that children take longer to acquire abstract concepts and words than concrete ones. Indeed, the differential acquisition of abstract and concrete words is quite large even for high frequency words and is quite persistent developmentally. This trend seems to be true for the acquisition of both spoken and reading vocabulary. Moreover, the understanding of how abstract concepts may operate as an ontological group is not appreciated until fairly late in development. Consequently, it might be inferred that abstract words are harder overall for children to understand than concrete words.
Summary
In this section, I have outlined three sources of evidence that abstract words presented in isolation are, indeed, harder to understand that concrete words. Abstract sentences take longer to understand and verify than concrete sentences. Abstract words are certainly more difficult to make lexical decisions for and probably more difficult to name than concrete words. Abstract words are acquired later in childhood than concrete words and maintain an inferiority in processing long after children have presumably acquired their meanings. Although the evidence for the proposition that abstract words are harder to understand than concrete words has been by no means unanimous for all tasks and observations, the bulk of it has suggested that the degree to which the various tasks require a deeper, more meaning-oriented processing of verbal materials is also the degree to which concreteness effects in understanding will surface.
THEORIES FOR WHY ABSTRACT WORDS ARE HARD TO UNDERSTAND
Over the years, a large number of positions and theories have been put forth in an attempt to explain the general processing superiority of concrete verbal materials over abstract materials. Such theories are important for the light they may shed on the representation, development, and processing of word meaning. To be complete, theories of word meaning must also include information regarding how words vary in their representations and how these representational distinctions are related to known developmental and processing differences.
In this section, I concentrate on what I feel have been the three most successful theories or hypotheses concerning why abstract word concepts are harder to understand than concrete concepts: the Dual-Coding Theory, the Age-of-Acquisition Hypothesis, and the Context Availability Model. The Dual-Coding Theory formalizes the basic notion that abstract words are hard to understand because they lack the direct sensory referents associated with concrete words. The Age-of-Acquisition and Context Availability views posit other, nonsensory explanations for why abstract words are harder to understand than concrete words. In what follows, I discuss what each theory has to say about the representation, development, and processing of abstract and concrete words.
Dual-Coding Theory
The Dual-Coding Theory by Paivio and his colleagues (Paivio, 1971, 1986) is probably the best known of all the theories on concreteness effects in verbal processing. Its strength (and the task for which it was originally developed) is its relative ability account for the highly pervasive effects of concreteness effects in verbal memory (but see Marschark et al., 1987). The theory was specifically developed to describe the representational differences between abstract and concrete words that cause concrete verbal materials to be remembered so much better than abstract materials.
Dual-Coding Theory (Paivio, 1986) posits the existence of a structurally and functionally distinct verbal system (called the logogen system) and image system (called the imagen system). The logogen system is said to be set up as a system of logogens, corresponding roughly to words, which are interconnected in an associative network of relations. These associative networks are acquired as a direct function of associative experience, that is, as a function of the natural word-to-word associations that come about with language experience. Logogens are said to be most directly and immediately activated by written or spoken language. The imagen system, on the other hand, is a system of sensory images that maintain the characteristics of the original modalities from which they arise. The imagery system contains various configurations of part-whole relations that arise from sensory experience. The imagens are said to be most directly activated by visual-spatial information (in the visual sensory case).
These representational systems can act either together or separately in language processing. The logogen and imagen systems are tied together by cross-referential links that enable one system to activate the other. Concrete words are said to have stronger referential connections to the imagen system than abstract words have, although the verbal associative linkages are said to be similar for the two word types (Paivio, 1986, p. 128). Therefore, in reading or listening to linguistic materials, then, the logogen system must be activated, but the imagen system might also be activated in some cases, particularly by concrete verbal materials. As Paivio (1986) puts it, “precisely which images or descriptions will be activated at any moment depends upon the stimulus context interacting with the relative functional strength of the different referential connections” (p. 63).
The theory vacillates with regard to whether concreteness effects should be displayed in verbal comprehension and exactly which form they should take. Paivio and Begg’s (1971) early work on sentence comprehension suggested that concreteness effects were not to be found in the initial stages of comprehension. They reasoned that the imagen system was not involved in comprehension because imagery formation to sentences took longer to evoke than simple comprehension did. However, as noted earlier in this chapter, further research has suggested that concreteness effects do occur in comprehension and (most of the time) seem to facilitate sentence and word processing. Paivio (1986) concludes from this work that “… referential and associative imagery reactions are more likely to be part of the comprehension of concrete than abstract material” (p. 218). He further notes that “The contribution of imagery is supplemented by effects that are attributable to verbal processes … The effects of imagery and verbal processes are additive, both in a positive and negative sense” (p. 222). Thus, it seems that the involvement of the imagery system in language comprehension may either assist comprehension (causing shorter reaction times), or may slow comprehension (causing longer reactions times).
Assuming that the imagen system usually acts to assist in sentence and word processing, and that these effects are additive, it would seem that this theory would have difficulty explaining why it is that concreteness effects in comprehension usually disappear when abstract and concrete verbal materials are presented in a meaningful verbal context. For example, Schwanenflugel and Shoben (1983) showed that although there were large differences between abstract and concrete sentences in neutral contexts, the comprehension time differences between those sentences disappeared when they were presented in a meaningful paragraph context. Several other studies in my laboratory have shown similar results for abstract and concrete words presented in meaningful compared with neutral sentence contexts (Schwanenflugel et al., 1988; Schwanenflugel & Shoben, 1983; Schwanenflugel & Stowe, 1989). In meaningful, supportive contexts, abstract sentences and words are comprehended as quickly as concrete ones.
The problem for the dual-coding theory is this; The finding that abstract materials take longer to understand than concrete materials when they are presented in isolation suggests that the imagen system is involved in comprehending concrete verbal materials. If so, and if the effects of the imagery and verbal associative processes are additive, then it would seem that concreteness effects should be found both when the materials are presented in context as without. Moreover, for a number of studies, abstract materials were always presented in an abstract verbal context to rule out the possibility that the context merely served to activate the imagen system for abstract words. Thus, concrete materials always had better access to the imagen system than abstract materials both in and out of context and, yet, concreteness effects were not displayed when the verbal materials were presented in context. If the imagen system is to be involved early in the comprehension of words, then the dual-coding theory needs to have a mechanism to account for why the concreteness effects disappear in context. If the imagen system is not to be involved, then the theory needs to have another mechanism for accounting for why abstract words take longer to process than concrete words when they are presented in isolation.
The dual-coding theory does discuss the development of the representational differences between abstract and concrete words. The view is that nonverbal representations precede verbal ones and that the verbal representations arise out of nonverbal representations. This claim is not unique to the theory and can be found in most cognitive developmental theories including that of Piaget (Piaget & Inhelder, 1971), Bruner (1966), and Fischer (Fischer, Hand, & Russell, 1984). The general view is that concrete ideas are easier for children to acquire than abstract ideas because they bear a more direct relationship to their real world, sensory counterparts.
This theory is generally able to account for the finding that concrete words are acquired sooner than abstract words developmentally. However, it is unable to account for a developmental trend we have noticed in our laboratory (Akin, 1989; Akin & Schwanenflugel, 1990). Usually, younger readers find abstract words particularly difficult to make lexical decisions for. Concreteness effects for 3rd graders are large (around 250 ms) and highly reliable. However, in two experiments, we have found that, by 10 years old, children begin to differentiate between various kinds of abstract and concrete words in a manner displayed by adults in the Schwanenflugel et al. (1988) study. Specifically, lexical decisions for abstract words are made as fast as for concrete words by 10-year-olds when the abstract and concrete words are similar in context availability (as defined by either their own or adults’ ratings). This suggests that the processing difficulty usually associated with abstract words is minimized for certain abstract words over others. Currently, the dual-coding theory treats abstract and concrete words as relatively undifferentiated classes of words and, as a result, cannot account for developmental differences among various types of abstract and concrete words.
In sum, the strength of this theory is that it makes a strong statement with regard to the differing representations of abstract and concrete words. However, given that the dual-coding theory does not make strong predictions as to what the relative contributions of the verbal and imagen system will be in on-line language comprehension, it is difficult to say how well it does or does not do in accounting for concreteness effects occurring early in comprehension. The conditions under which the imagen system will necessarily be involved have never been specified. This means that, whenever concreteness effects are displayed, it can always be argued ad hoc that imagery was invoked. Similarly, when concreteness effects are not displayed, it can also be argued that imagen system was not involved. As a result, the current theory is virtually untestable with regard to issues of comprehension and lexical processing. The theory is more successful, however, in describing relative trends in the development of abstract and concrete word meanings (although it must be noted that it does so more by postulate than by any inherent feature of the theory itself).
Age-of-Acquisition Hypothesis
The age-of-acquisition hypothesis (Gilhooly & Gilhooly, 1979) was not developed as a theory of concreteness effects and is, therefore, less well developed than the dual-coding hypothesis for describing such effects. The theory is, however, designed to describe the kinds of trends that are seen in lexical decision and comprehension studies. The theory states simply that the concreteness effects typically displayed in lexical retrieval tasks are merely cumulative word frequency effects. It is argued that the usual word frequency indices that are obtained from objective measures of written and spoken English (such as the written word frequency index by Kucera and Francis, 1967, and the spoken word frequency index by Svartvik and Quirk, 1980) are inappropriate measures of the phenomenological frequency of abstract and concrete words because they do not take into account differential acquisition rates of the two word types. Because concrete words are acquired so much earlier in development than abstract words, they have effectively been experienced more often. They have correspondingly lower thresholds of activation, making them faster to retrieve associated lexical information for than abstract words. Moreover, adults apparently are able to judge with some accuracy the relative ages at which they learned various words (Gilhooly & Gilhooly, 1980), although just how they make this decision is not quite clear. It can be inferred from the hypothesis that the differences in the representation between abstract and concrete words can be isolated at the differing thresholds of activation for the two word types. Thus, lexical concreteness effects occurring early in lexical processing are said to be merely age-of-acquisition effects.
This view does reasonably well in explaining concreteness effects for words occurring in isolation. In fact, Brown and Watson (1987) have shown age-of-acquisition ratings to be better predictors of naming times than lexical familiarity, imageability, and written or spoken frequency measures. Similarly, Coltheart, Laxon, and Keating (1988) found no difference in naming times between abstract and concrete words rated similarly on age-of-acquisition. My studies have found age-of-acquisition to be as good, but no worse, a predictor of lexical decision time as imageability, once objective word length and word frequency is partialed out (Schwanenflugel et al., 1988). Thus, age-of-acquisition seems to be a reasonable candidate for explaining why abstract words and sentences take longer to process than concrete materials presented in isolation.
The hypothesis has some difficulty explaining why the differences in comprehension between abstract and concrete verbal materials tend to disappear when they are presented in a supportive context. Although the hypothesis makes no statements with regard to the influence of context on lexical processing, it would seem that the influence of context would have to be treated similarly to the context and word frequency interactions that have been noted in the literature (see Stanovich & West, 1983). That is, it has often been noted that the difference in processing time between low frequency and high frequency words is minimized when such words are presented in a meaningful context. While it is true that the previous research generally supports the finding of diminished word frequency effects when words are presented in context, it does not support the finding that such frequency effects should disappear completely in context. Consequently, to the degree that context does not completely override the influence of frequency in the processing of upcoming words, the age-of-acquisition hypothesis cannot account for the absence of concreteness effects when abstract and concrete verbal materials are presented in a supportive context.
Finally, the age-of-acquisition hypothesis does not have a position on why abstract words take longer for children to acquire than concrete words. It merely asserts that, because they do, processing is slowed for such words in adulthood.
In sum, the age-of-acquisition hypothesis does reasonably well in accounting for comprehension time differences between abstract and concrete verbal materials presented in isolation. It fares less well in accounting for the elimination of concreteness effects when abstract and concrete words are presented in a supportive context and in explaining exactly why it is that abstract words are acquired later than concrete words.
Context Availability Model
The context availability model (Bransford & McCarrell, 1974; Kieras, 1978; Schwanenflugel & Shoben, 1983) is a variant of a more general model of comprehension that has been extended to account for concreteness effects in comprehension. This model views comprehension as being an interaction between the comprehender’s knowledge base and external stimulus context. According to the model, comprehension is aided fundamentally by the addition of contextual information to the materials that are to be understood. This contextual information may come either from the comprehender’s knowledge base or from the external stimulus context. This contextual information serves to enable comprehenders to make the necessary interrelations between concepts needed for subjective feeling of comprehension. If persons are unable to retrieve relevant information from prior knowledge to supplement the incoming message, then comprehension is difficult and a subjective sense of comprehension may not take place. Thus, the emphasis in the theory is placed on the relative retrievability of information from knowledge base and how the availability of such information influences comprehension.
Concreteness effects for abstract and concrete verbal materials presented in isolation are dealt with in this model by assuming that the retrieval of information from knowledge base is more difficult for some concepts than others. Generally, abstract materials are said to be more difficult to understand than concrete materials because persons find it more difficult to retrieve the associated contextual information from knowledge base needed to augment comprehension for abstract concepts than concrete concepts. Therefore, the main representational difference between abstract and concrete words is that abstract words possess weaker connections to associated contextual information in knowledge base than concrete words do.
Abstract words could possess weaker connections to associated contextual information than concrete words for a number of reasons. One way that an abstract word might come to have weaker connections to knowledge base is by appearing in a great diversity of contexts (Galbraith & Underwood, 1973). Another way is by simply being relatively unfamiliar (that is, by appearing so seldomly that the person cannot connect any contextual information at all with the word). What is important to this view is not how abstract words come to have weaker connections to contextual information from knowledge base, but only that they generally do.
Because contextual information retrieval is usually slower for abstract concepts than for concrete concepts, abstract sentences and words normally take longer to comprehend than concrete words when they are presented in isolation. Thus, for example, when persons read an abstract sentence such as “Many factors affected the crucial decision,” they may attempt to retrieve a plausible circumstance from prior knowledge in which such a sentence might be embedded (say, in a decision about which graduate school to attend) and may have difficulty doing so. On the other hand, for a similar concrete sentence like “Many sailors deserted the sinking vessel,” they may have a somewhat easier time retrieving such contextual information (say, in a war story) and would be correspondingly faster to comprehend such sentences. The finding of longer lexical decision times for abstract and concrete words would be interpreted in much the same way. The model claims that, in making a lexical decision, persons also attempt to retrieve related information from prior knowledge. The ease with which they are able to do so would be reflected in longer lexical decision times for abstract than concrete words. Similar predictions would also be made for naming to the degree to which persons process words for meaning in that task.
The model also makes explicit predictions regarding what should happen when abstract and concrete verbal materials are presented in meaningful, highly supportive contexts. The model predicts that abstract and concrete words should be equally easy to understand when presented in a supportive context because context serves to preactivate related information in prior knowledge. This preactivation of contextual knowledge helps to override the relative difficulty in retrieving such knowledge for abstract words, making them as easy to comprehend as concrete words. Concrete words, because they already have highly available contextual knowledge associated with them, do not benefit nearly as much from being presented in a supportive context as abstract words do. Thus, context effects should be larger for abstract than concrete sentences and words.
A number of studies have supported such a claim about the influence of context on abstract and concrete words. For example, Schwanenflugel and Shoben (1983) found that, when abstract and concrete sentences were presented in meaningful paragraph contexts, concreteness effects in comprehension times disappeared. (Wattenmaker & Shoben, 1987, extended this finding to sentence recall; see also Marschark, 1985). Schwanenflugel et al. (1988) and Schwanenflugel and Shoben (1983) have noted the elimination of lexical decision time differences for abstract and concrete words embedded in sentence contexts. Schwanenflugel and Stowe (1989) have also shown that naming times and meaningfulness judgment times do not differ for abstract and concrete words presented in sentence contexts. Thus, across a wide array of tasks, the processing difficulty usually associated with abstract materials has been shown to disappear when those materials are presented in highly supportive contexts.
Another finding supporting this view of concreteness effects in comprehension is that subjects’ ratings of context availability (in which subjects are asked to rate the relative difficulty of retrieving associated contextual information for isolated abstract and concrete words) are significantly correlated with ratings of imagery and concreteness. Moreover, context availability is also correlated with lexical decision (Schwanenflugel et al., 1988) and sentence comprehension times (Schwanenflugel & Shoben, 1983). In fact, context availability has been found to be a better predictor of lexical decision times than imagery, concreteness, or age-of-acquisition ratings. Furthermore, when rated context availability is controlled for, lexical decision times for abstract and concrete words do not differ (Akin, 1989; Akin & Schwanenflugel, 1990; Schwanenflugel et al., 1988).
At present, there is one exception to the general trend that the processing time differences between abstract and concrete words disappear when presented in a meaningful context. Specifically, Bleasdale (1987) found equivalent priming effects for abstract and concrete words in a word prime lexical decision task, yielding significant concreteness effects both in and out of context. We (Shoben, Wilson, & Schwanenflugel, 1990) have recently replicated Bleasdale’s basic findings. Although we did find differential priming for abstract and concrete words both in a naming and lexical decision task, concreteness effects did not completely disappear when the words were presented in a word prime context. Although we are not sure why the findings diverge for word prime contexts, at least two possibilities are plausible. First, in word prime studies, the degree of associative and semantic relatedness between word primes and their targets tend to be weaker than they are for sentence-target pairs. Second, sentence contexts in general may operate differently than word prime contexts, if for no other reason than that sentence contexts are more like normal text. Sentence contexts have been shown to produce larger contextual benefits than word primes alone (West & Stanovich, 1988). Consequently, it seems that abstract words need a considerable amount of contextual support to be processed similarly to concrete words.
Currently, the context availability model makes no statements with regard to the development of the representational differences between abstract and concrete words. However, I have compared the relative contributions of imageability and context availability on the acquisition of abstract and concrete words. Specifically, I correlated the typical age of acquisition as defined by the Rinsland corpus (1945) and for the 198 nouns used in the analysis featured in Fig. 9.1 with adult ratings of context availability. This analysis showed context availability to be as good a predictor of age of acquisition as rated imagery. That is, words that adults rate as easy to think of contextual information for are typically acquired earlier than words they find difficult to retrieve associated contextual information for. Thus, early acquired words tend not only to be concrete and imageable, but are also relatively easy to retrieve contextual information for in adulthood.
On the other hand, lexical imageability may be important early in language development beyond the contributions of context availability. In a recent study, Akin and Schwanenflugel (1990) found that 8-year-olds (but not 10-year-olds and adults) still took longer to make lexical decisions for acquired abstract words than acquired concrete words even when the two word types were controlled on rated context availability. For 8-year-olds, imageability ratings were better predictors of reaction times than context availability times, although the reverse was true for adults. Consequently, it does seem that younger readers may attempt to retrieve sensory, imaginal information to make lexical decisions.
In sum, the context availability model describes the representational differences between abstract and concrete words in terms of differences in strength of connections between concepts and associated knowledge base. However, some abstract words may be relatively strongly connected to their associated contextual information and some concrete words less so and, therefore, their lexical decision times should vary accordingly. The model does reasonably well in accounting for comprehension times, lexical decision, and naming times for words appearing in isolation. It also accounts for the disappearance of concreteness effects when the items are presented in a supportive context. It currently does not have any systematic position regarding how the representational differences between the two word types come about.
CONCLUSION
In this chapter, I have outlined the current status of concreteness effects occurring early in the comprehension of abstract and concrete verbal materials. In general, it has been found that greater concreteness is associated with faster processing times for both sentences and words presented in isolation. However, these concreteness effects in comprehension typically disappear when the verbal materials are presented in a meaningful, supportive context. Furthermore, concrete words are acquired earlier developmentally than abstract words, and maintain their superiority in processing long after their meaning is acquired.
Currently, none of the existing theories is able to explain all the related findings regarding concreteness effects in comprehension. The dual-coding theory does reasonably well in describing the processing of abstract and concrete words presented in isolation and in accounting for developmental trends in their acquisition. It also makes the strongest statements regarding the representational distinctions between the two word types. The age-of-acquisition hypothesis does very well in accounting for the processing of isolated abstract and concrete words, but does less well in describing their development and their processing in meaningful contexts. The context availability model, on the other hand, accounts reasonably well for findings regarding comprehension of abstract and concrete verbal materials both in and out of context. However, it currently lacks mechanisms for describing how and why the representational differences between abstract and concrete words come about developmentally.
On the other hand, I feel the context availability model has an important advantage over the other views because it is in basic agreement with current discussions about the nature of concepts and word meanings in general. Specifically, these recent views (e.g., Barsalou, 1987; Brooks, 1987; Roth & Shoben, 1983) have focused on the episodic, contextually based character of word meaning. There is a general movement away from thinking of concepts as having fixed, definitional cores that become activated on all instances of word use. The context availability model, with its emphasis on the retrieval of contextual information from prior knowledge, also emphasizes this contextually based character of word meanings. The demonstration that even the processing of words in isolation requires the retrieval of contextual information from prior knowledge further supports this general view of the nature of word meanings. Moreover, it also suggests that word meanings may never become completely decontextualized from their contexts of prior use.
These views of word meaning (Barsalou, 1982; Greenspan, 1986) point out that not all knowledge associated with concepts is equally accessible during every instance of word use. Some knowledge is said to be context-dependent (accessible only in particular contexts) and other knowledge is said to be context-independent (accessible in all contexts of word use). It is possible that words rated low in context availability largely possess context-dependent knowledge which is relatively inaccessible when the words are presented in isolation. However, when such words are presented in supportive contexts, this context-dependent information becomes highly available for deriving meaning, eliminating potential differences in comprehension between abstract and concrete words.
Although none of the theories can currently account for all the phenomena related to concreteness effects in comprehension, I have highlighted the kinds of information that a theory of concreteness effects will have to account for in order to be a comprehensive description of concreteness and word meaning. It is clear that a comprehensive theory of concreteness effects in word meaning will be a necessary feature of a complete theory of the structures and processes engaged in a psychology of word meanings.
ACKNOWLEDGMENTS
I thank Mike Martin, Alan Roth, and Ed Shoben for their helpful comments on an earlier draft of this chapter.
REFERENCES
Akin, C. E. (1989). Context availability and the developmental use of imagery. Unpublished doctoral dissertation, University of Georgia, Athens, GA.
Akin, C. E., & Schwanenflugel, P. J. (1990). Context availability and developmental trends in the processing of abstract and concrete words. Unpublished manuscript, University of Georgia, Athens, GA.
Balota, D. A., & Chumbley, J. I. (1984). Are lexical decisions a good measure of lexical access? The role of word-frequency in the neglected decision stage. Journal of Experimental Psychology: Human Perception and Performance, 10, 340–357.
Balota, D. A., & Lorch, R. F., Jr. (1986). Depth of automatic spreading activation: Mediated priming effects in pronunciation but not in lexical decision. Journal of Experimental Psychology: Learning, Memory, & Cognition, 12, 336–345.
Barsalou, L. W. (1982). Context-independent and context-dependent information in concepts. Memory & Cognition, 11, 629–654.
Barsalou, L. W. (1987). The instability of graded structure: Implications for the nature of concepts. In U. Neisser (Ed.), Concepts and conceptual development: Ecological and intellectual factors in categorization (pp. 101–140). New York: Cambridge University Press.
Belmore, S. M., Yates, J. M., Bellack, D. R., Jones, S. N., & Rosenquist, S. E. (1982). Drawing inferences from concrete and abstract sentences. Journal of Verbal Learning and Verbal Behavior, 21, 338–351.
Bleasdale, F. A. (1987). Concreteness dependent associative priming: Separate lexical organization for concrete and abstract words. Journal of Experimental Psychology: Learning, Memory, and Cognition, 13, 582–594.
Bransford, J. D., & McCarrell, N. S. (1974). A sketch of a cognitive approach to comprehension: Some thoughts on what it means to comprehend. In W. Weimer & D. Palermo (Eds.), Cognition and the symbolic processes (pp. 189–230). Hillsdale, NJ: Lawrence Erlbaum Associates.
Brooks, L. (1987). Decentralized control of categorization: The role of prior processing episodes. In U. Neisser (Ed.), Concepts and conceptual development: Ecological and intellectual factors in categorization (pp. 141–174). New York: Cambridge University Press.
Brown, G. D. A., & Watson, F. L. (1987). First in, first out: Word learning age and spoken word frequency as predictors of word familiarity and word naming latency. Memory & Cognition, 15, 208–216.
Brown, R. W. (1957). Linguistic determinism and the part of speech. Journal of Abnormal and Social Psychology, 55, 1–5.
Bruner, J. S. (1966). Toward a theory of instruction. New York: W. W. Norton.
Coltheart, V., Laxon, V. J., & Keating, C. (1988). Effects of word imageability and age of acquisition on children’s reading. British Journal of Psychology, 79, 1–11.
DeGroot, A. M. B. (1989). Representational aspects of word imageability and word frequency as assessed through word association. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15, 824–845.
DiVesta, F. J., & Walls, R. T. (1970). Factor analysis of the semantic attributes of 487 words and some relationships to the conceptual behavior of fifth-grade children. Journal of Educational Psychology Monograph, 61(6, Pt. 2).
Fischer, K. W., Hand, H. H., & Russell, S. (1984). The development of abstractions in adolescence and adulthood. In M. L. Commons, F. A. Richards, & C. Armon (Eds.), Beyond formal operations: Late adolescent and adult cognitive development. New York: Prager.
Galbraith, R. C., & Underwood, B. J. (1973). Perceived frequency of concrete and abstract words. Memory & Cognition, 1, 56–60.
Gentner, D. (1982). Why nouns are learned before verbs: Linguistic relativity versus natural partitioning. Center for the Study of Reading: Technical Report No. 257. Urbana, IL: University of Illinois at Urbana-Champaign.
Gernsbacher, M. A. (1984). Resolving 20 years of inconsistent interactions between lexical familiarity and orthography, concreteness, and polysemy. Journal of Experimental Psychology: General, 113, 256–281.
Gilhooly, K. J., & Gilhooly, M. L. (1979). Age-of-acquisition effects in lexical and episodic tasks. Memory & Cognition, 7, 214–224.
Gilhooly, K. J., & Gilhooly, M. L. (1980). The validity of age-of-acquisition ratings. British Journal of Psychology, 71, 105–110.
Gilhooly, K. J., & Watson, F. L. (1981). Word age-of-acquisition effects: A review. Current Psychological Reviews, 1, 269–286.
Glass, A. L., Eddy, J. K., & Schwanenflugel, P. J. (1980). The verification of high and low imagery sentences. Journal of Experimental Psychology: Human Learning and Memory, 6, 692–704.
Glass, A. L., & Holyoak, K. J. (1975). Alternative conceptions of semantic memory. Cognition, 3, 313–319.
Glass, A. L., Millen, D. R., Beck, L. G., & Eddy, J. K. (1985). Representation of images in sentence verification. Journal of Memory and Language, 24, 539–557.
Greenspan, S. L. (1986). Semantic flexibility and referential specificity of concrete nouns. Journal of Memory and Language, 25, 539–557.
Haberlandt, K. F., & Graesser, A. C. (1985). Component processes in text comprehension and some of their interactions. Journal of Experimental Psychology: General, 114, 357–375.
Holmes, V. M., & Langford, J. (1976). Comprehension and recall of abstract and concrete sentences. Journal of Verbal Learning and Verbal Behavior, 15, 559–566.
Holyoak, K. J. (1974). The role of imagery in the evaluation of sentences: Imagery or semantic factors. Journal of Verbal Learning and Verbal Behavior, 13, 163–166.
Howell, J. R., & Bryden, M. P. (1987). The effects of word orientation and imageability on visual half-field presentations with a lexical decision task. Neuropsychologia, 25, 527–538.
James, C. T. (1975). The role of semantic information in lexical decisions. Journal of Experimental Psychology: Human Perception and Performance, 104, 130–136.
Jorgensen, C. C., & Kintsch, W. (1973). The role of imagery in the evaluation of sentences. Cognitive Psychology, 4, 110–116.
Keil, F. C. (1979). Semantic and conceptual development: An ontological perspective. Cambridge, MA: Harvard University Press.
Keil, F. C. (1983). Semantic inferences and the acquisition of word meaning. In T. Seiler & W. Wannenmacher (Eds.), Concept development and the development of word meaning (pp. 103–124). Berlin: Springer-Verlag.
Kieras, D. (1978). Beyond pictures and words: Alternative information processing models for imagery effects in verbal memory. Psychological Bulletin, 85, 532–554.
Kiraly, J., & Furlong, A. (1974). Teaching words to kindergarten children with picture, configuration, and initial sound cues in a prompting procedure. Journal of Educational Research, 67, 295–298.
Klee, H., & Eyesenck, M. W. (1973). Comprehension of concrete and abstract sentences. Journal of Verbal Learning and Verbal Behavior, 12, 522–529.
Kroll, J. F., & Merves, J. S. (1986). Lexical access for concrete and abstract words. Journal of Experimental Psychology: Learning, Memory, and Cognition, 12, 92–107.
Kucera, H., & Francis, W. H. (1967). Computational analysis of present-day American English. Providence, RI: Brown University Press.
Lorch, R. F. (1982). Priming and search processes in semantic memory: A test of three models of spreading activation. Journal of Verbal Learning and Verbal Behavior, 21, 468–492.
Marschark, M. (1979). The syntax and semantics of comprehension. In G. Prideaux (Ed.), Perspectives in experimental linguistics. Amsterdam: John Benjamins B. V.
Marschark, M. (1985). Imagery and organization in the recall of high and low imagery prose. Journal of Memory and Language, 24, 734–745.
Marschark, M., Richman, C. L., Yuille, J. C., & Hunt, R. R. (1987). The role of imagery in memory: On shared and distinctive information. Psychological Bulletin, 102, 28–41.
Moeser, S. D. (1974). Memory for meaning and wording in concrete and abstract sentences. Journal of Verbal Learning and Verbal Behavior, 13, 682–697.
Nelson, K. (1973). Structure and strategy in learning to talk. Monographs of the Society for Research in Child Development, 38 (Serial No. 149).
Paivio, A. (1968). A factor-analytic study of word attributes and verbal learning. Journal of Verbal Learning and Verbal Behavior, 7, 41–49.
Paivio, A. (1971). Imagery and verbal processes. New York: Holt, Rinehart, & Winston.
Paivio, A. (1986). Mental representations: A dual coding approach. Oxford, England: Oxford University Press.
Paivio, A., & Begg, I. (1971). Imagery and comprehension latencies as a function of sentence concreteness and structure. Perception & Psychophysics, 10, 408–412.
Piaget, J., & Inhelder, B. (1971). Mental imagery in the child. New York: Basic Book.
Prior, M. R., Cumming, G., & Hendy, J., (1984). Recognition of abstract and concrete words in a dichotic listening paradigm. Cortex, 20, 149–157.
Richardson, J. T. E. (1976). The effects of stimulus attributes upon latency of word recognition. British Journal of Psychology, 67, 315–325.
Richmond, M. G., & McNinch, G. (1977). Word learning: Concrete vs. abstract acquisition. Perceptual and Motor Skills, 45, 292–294.
Rinsland, H. D. (1945). A basic vocabulary of elementary school children. New York: Macmillan.
Roth, E. H., & Shoben, E. J. (1983). The effect of context on the structure of categories. Cognitive Psychology, 15, 346–378.
Rubin, D. C. (1980). 51 properties of 125 words: A unit analysis of verbal behavior. Journal of Verbal Learning and Verbal Behavior, 19, 736–755.
Schwanenflugel, P. J., Harnishfeger, K. K., & Stowe, R. W. (1988). Context availability and lexical decisions for abstract and concrete words. Journal of Memory and Language, 27, 499–520.
Schwanenflugel, P. J., & Shoben, E. J. (1983). Differential context effects in the comprehension of abstract and concrete verbal materials. Journal of Experimental Psychology: Learning, Memory, & Cognition, 9, 82–102.
Schwanenflugel, P. J., & Stowe, R. W. (1989). Context availability and the processing of abstract and concrete words. Reading Research Quarterly, 24, 114–126.
Seidenberg, M. S., Waters, G. S., Sanders, M., & Langer, P. (1984). Pre- and post-lexical loci of contextual effects on word recognition. Memory & Cognition, 12, 313–328.
Shoben, E. J., Wilson, T., & Schwanenflugel, P. J. (1990). The influence of lexical primes on the processing of abstract and concrete words. Unpublished data, University of Georgia, Athens, GA.
Smith, E. E. (1978). Theories of semantic memory. In W. K. Estes (Ed.), Handbook of learning and cognitive processes (Vol. 6). Hillsdale, NJ: Lawrence Erlbaum Associates.
Stanovich, K. E., & West, R. F. (1983). On priming by a sentence context. Journal of Experimental Psychology: General, 112, 1–36.
Svartvik, J., & Quirk, R. (1980). A corpus of English conversation. Lund, Sweden: Gleerup.
Tanenhaus, M. K., Carlson, G. N., & Seidenberg, M. S. (1985). Do listeners compute linguistic representations? In A. Zwicky, L. Kartunnen, & D. Dowty (Eds.), Natural language parsing: Psycholinguistic, theoretical, and computational perspectives (pp. 359–408). Cambridge: Cambridge University Press.
Thorndike, E. L., & Lorge, I. (1944). The teacher’s word book of 30,000 words. New York: Teachers College, Columbia University.
Wattenmaker, W. D., & Shoben, E. J. (1987). Context and the recallability of concrete and abstract sentences. Journal of Experimental Psychology: Learning, Memory, and Cognition, 13, 140150.
West, A., Martindale, C., & Sutton-Smith, B. (1985). Age trends in the content of children’s spontaneous fantasy narratives. Genetic Psychology Monographs, 111, 389–405.
West, R. F., & Stanovich, K. E. (1988). How much of sentence priming is word priming? Bulletin of the Psychonomic Society, 26, 1–4.
Whaley, C. P. (1978). Word-nonword classification times. Journal of Verbal Learning and Verbal Behavior, 17, 143–154.
Wolf, M., & Gow, D. (1986). A longitudinal investigation of gender differences in language and reading development. First Language, 6, 81–110.
Yore, L. D., & Ollila, L. O. (1985). Cognitive development, sex, and abstractness in grade one word recognition. Journal of Educational Research, 78, 242–247.