The Influence of Sentence Context on Reading Times for Abstract and Concrete Words

One variable that affects word processing is concreteness, the degree to which a word’s referent can be experienced by the senses. Concrete words tend to be processed more quickly than abstract words. The dual-coding theory states that processing is aided by image-based representations of words, whereas the context availability model holds that concrete words are processed more easily because they have stronger associated contextual information. This is the first study to test the context availability model by tracking eye movements in a reading task. Participants read abstract and concrete words in predictable and neutral sentences. The results showed a main effect of context and of concreteness. This finding contradicts the prediction of the context availability model that equivalent amounts of contextual information nullify the effect of concreteness. The results support the strategic imagery hypothesis, which states that imagery is used when it is helpful in completing a task. Sentence Context and Concreteness 3 The Influence of Sentence Context on Reading Times for Abstract and Concrete Words For humans, language is the medium through which we express our thoughts; we communicate our ideas about concepts and objects in the environment with words. It is not surprising, therefore, that the study of language is used to shed light on the ways in which representations and ideas are organized in the brain. In order to understand a written word, such as “book,” a reader must access ideas and knowledge related to “book” that are stored in her memory. If reading “book” takes significantly less time than reading another word, such as “cause,” we may infer from this difference that aspects of the word “book” make it easier to access in the mind than “cause.” In this way, studying the different factors that affect the way we read helps researchers understand the way we think. There are many different variables that affect the rate at which people access, process, and read words. One such variable, which has received a great deal of research attention, is concreteness. Concreteness is the extent to which a word’s referent can be perceived by the senses. Words are typically classified as concrete, meaning that they have direct sensory referents, or abstract, meaning that they lack direct sensory referents (Schwanenflugel, Harnishfeger, & Stowe, 1988). The word “book,” which refers to an object that is tangible and can be easily pictured in one’s mind, is highly concrete. The abstract word “cause,” on the other hand, refers to a concept that is intangible and not easily pictured. Differences in the rate at which people read abstract and concrete words may suggest that abstract and concrete Sentence Context and Concreteness 4 concepts are stored in the mind in different ways, and this difference in turn may be studied to uncover more about language and human thought. A large body of research suggests that lexical access (i.e., matching a word to its representation stored in the brain) and comprehension are easier for concrete words than for abstract words (e.g., Bleasdale, 1987; Juhasz & Rayner, 2003; Paivio, 1971). One study demonstrated that concrete words are read more quickly than abstract words by using eye movement methodology, in which the fixations and saccades (movements) of a reader’s eye are tracked as she reads (Juhasz & Rayner, 2003). Research has also shown that response times in naming (e.g., Schwanenflugel & Stowe, 1989) and recall tasks (e.g., Wattenmaker & Shoben, 1987) are shorter for concrete words than for abstract words when they are presented in neutral sentence or passage contexts. Furthermore, participants are able to think of word and image associates more quickly for concrete words than for abstract words (Paivio, 1966). This general phenomenonthat concrete words and concepts are processed more easily than abstract words and conceptswill be referred to as the concreteness effect throughout. Theories of the Concreteness Effect Pavio’s dual-coding theory explains the difference in processing times for abstract and concrete words by postulating two different systems of coding concepts in the mind: an imaginal system, in which concepts are stored as images, and a verbal system, in which concepts are stored as words (Paivio, 1971). Concrete words have the advantage of being encoded in both systems, while abstract words are only Sentence Context and Concreteness 5 encoded in the verbal system, due to their lack of imageability (i.e., they cannot be visualized). The word “book,” for example, would be encoded as a visual representation of a book in the imaginal system. The meaning of “book” and its semantic associates, such as “read,” would be encoded in the verbal system. An abstract word, such as “cause,” would only have a representation in the verbal system; it is quite difficult to think of an image that could represent “cause.” Retrieval of words that refer to concrete concepts is facilitated by the availability of two mental representations of the word. As a result, the dual-coding theory predicts that concrete words will be processed more quickly than abstract words whether presented alone, in sentence context, or in paragraph context. The context availability model (e.g., Schwanenflugel & Shoben, 1983) accounts for the concreteness effect through an entirely different mechanism from the dual-coding theory. According to this model, lexical access and comprehension of words, sentences, and passages are aided by the presence of contextual information (see also Kieras, 1978; Schwanenflugel et al., 1988). This contextual information may come from the reader’s prior knowledge (e.g., the reader may automatically associate the context of reading with the word “book”) or from verbal context added to the material to be comprehended (e.g., a predictable sentence context, such as “she went to the library to check out a book”). The context availability model accounts for the concreteness effect with the observation that concrete words tend to have more inherent contextual information than abstract words, as in the example of “book” and “reading” described above. Thus, according to this model concrete words are easier to process because they have a greater amount of readily accessible associated Sentence Context and Concreteness 6 contextual information, not because they have an additional imaginal representation in the brain. Several mechanisms have been proposed to explain why abstract words tend to have less inherent context than concrete words. According to Kieras (1978), all types of knowledge are stored in one type of representation, which is referred to as a proposition. A proposition is made up of a concept and all of its related concepts, verbal or perceptual, which are interconnected in the mind. The propositions of words that have a large amount of inherent context are easier to access than the propositions of words that have a small amount of inherent context, because words with more inherent context have more connections with information that is permanently stored in the memory. Concrete words, which have more perceptual properties, have more associated information than abstract words. This theory, though dependent on the level of imagery associated with concrete concepts, is different from the dual-coding theory because it does not propose two separate coding systems. Another explanation of the contextual differences between abstract and concrete words focuses on the number of associated contexts that abstract and concrete words tend to have and on the strength of the connection between each word and its associated context (e.g., Schwanenflugel & Shoben, 1983). This view states that abstract words are processed more slowly than concrete words because they tend to have a greater number of associated contexts, and consequently each context is weakly connected to the word. This becomes apparent when one attempts to list all possible contexts of an abstract word, such as “cause”: just cause, a lost cause, a cause celebre, the cause of a problem, of a victory, etc. Concrete words, on the other Sentence Context and Concreteness 7 hand, tend to have a smaller number of associated contexts. As a result, each context is more strongly associated to its concrete word, which makes each context easier to access. This view is supported by several studies in which participants were asked to rate abstract and concrete words on different measures of contextual information. These studies found that abstract words tend to appear in a greater diversity of contexts than concrete words (Schwanenflugel & Shoben, 1983), and that concrete words tend to have a stronger single associate than abstract words, which tend to have more numerous, weakly connected associates (Altarriba, Bauer, & Benvenuto, 1999). Measures of Contextual Information There are a variety of ways to measure and experimentally manipulate context. Context may be inherent in the reader’s prior knowledge of a word, as mentioned above. The context availability variable, which is often used as a measure of inherent context, represents how difficult or easy it is to think of a context in which a given word may appear (e.g., Schwanenflugel et al., 1988; Schwanenflugel & Shoben, 1983). Context availability values are obtained by subjective ratings; participants are asked to rate how easily they can think of a context for a given word (Schwanenflugel & Shoben, 1983). This variable is positively correlated with concreteness, which is consistent with the prediction of the context availability model that concrete words are more easily accessed due to a greater availability of inherent contextual information (Altarriba et al., 1999; Schwanenflugel et al., 1988). Other word-level variables, in addition to context availability, have been used to measure context. As mentioned above, research has shown that concreteness and Sentence Context and Concreteness 8 contex