Journal of Psycholinguistic Research, Vol. 9, No. 1, 1980

Comprehension and Memory of the Literal and Figurative Meaning of Proverbs J. Kathryn Bock I and William F. Brewer 2 Received June 15, 1978

Proverbs were used to examine recognition memory for four types of sentence information (figurative meaning, literal meaning, lexical information, syntactic information) with two levels of comprehension. Forced-choice recognition tasks showed that subjects were successful in remembering all four types of information. Recognition scores for the figurative meaning of unfamiliar proverbs presented in isolation were above chance and were also significantly improved when procedures were employed which encouraged comprehension of this level of meaning. The results are most consistent with theories of comprehension which provide for different types of processing for figurative and literal language.

INTRODUCTION Proverbs are a particularly interesting class of materials for psycholinguistic research, since most proverbs have both a literal and a figurative level of meaning. At the literal level, a proverb such as "Run after two hares and you will catch neither" gives advice about catching rabbits, whereas at the figurative level it suggests that if you try to do too many things you may accomplish none of them. The ability to comprehend both levels of meaning is productive. For example, someone who has never heard the proverb "One who scalds his tongue upon hot soup blows upon

IDepartment of Psychology, Michigan State University, East Lansing, Michigan 48824. 2Department of Psychology, Psychology Building, University of Illinois, Champaign, Illinois 61820. 59 0090-6905/80/0100-0059503.00/0 9 1980 Plenum PublishingCorporation

60

Bock and Brewer

cold soup" can use the linguistic input as a guide to construct an interpretation at the literal level, where the proverb deals with a person who burns his tongue on hot soup. The hearer can also construct an interpretation at the figurative level, where the proverb means roughly the same thing as the proverb "He goes cautiously alongside the brook, who has slipped into the sea." Figurative language can be produced in several forms, of which metaphor is perhaps the most common. Metaphors, according to Richards (1965), are composed of two principal terms, the topic and vehicle, and the relationship between them, the ground. The topic is usually the subject of the metaphorical sentence, while the vehicle is the term being used metaphorically. The ground is created by the points or relations of similarity between the two. For example, in "Billboards are warts on the landscape" (from Verbrugge and McCarrell, 1977), "billboards" is the topic, "warts" is the vehicle and "unsightly protrusions on a surface" is the ground. Perrine (1971) has proposed a classification scheme in which proverbs are treated as a type of metaphor in which only the vehicle--the metaphorical term--is explicitly stated. Proverbs may then be viewed as a type of metaphor which requires greater contextual support or more processing than the type of metaphor in which both the topic and vehicle are explicitly stated. The ability to understand proverbs (Biihler, 1908; Honeck, 1973; Honeck et al., 1975), metaphor (Harris, 1976; Ortony et al., 1978b; Verbrugge and McCarrell, 1977), and the figurative level of language in general is of considerable theoretical interest since, until recently, most theories of meaning in psychology and linguistics dealt with only the literal level (e.g., Anderson and Bower, 1973; Katz and Fodor, 1963). Although this deficiency has been remedied in several models, none of the mechanisms suggested for the comprehension of figurative language by current theories is fully satisfactory. The approaches to figurative language in current theories fall into two general categories. The first type of approach emphasizes the continuity between the comprehension of metaphorical language and literal language (cf. Collins and Quillian, 1972; Verbrugge, 1977). Both metaphorical and literal statements are interpreted relative to their contexts of occurrence, without distinctly different types of inferences or stages of processing. A similar view is implicit in Bobrow and Norman's (1975) schema theory. A potential problem for this approach is the fact that in many cases there is clear intuitive difference between figurative and literal language (Bflhler, 1951; Kintsch, 1974). Theories which postulate no differences

Comprehension and Memory of Proverbs

61

between figurative and literal language processing cannot account for this intuition or even for the fact that a conventional distinction between literal and figurative language exists. Contrasting with theories which do not incorporate different mechanisms for the interpretation of figurative and literal language are those which do. These theories require extraordinary measures for metaphor comprehension, generally proceeding from recognition of semantic anomaly at the literal level (Kintsch, 1972; Matthews, 1971; Weinreich, 1966). Although these theories have the advantage of being able to account for the intuitive difference between literal and figurative language, they have a different problem. Since proverbs and many metaphors are not prima facie anomalous at the literal level, it is not obvious how such models, under normal circumstances, could describe the derivation of the figurative interpretation without conceding that the process is in some way dependent on extrasentential context (Kintsch, 1974; Ortony et al., 1978a). Because of the problems with both of these approaches, neither is readily adaptable to the problem of understanding the figurative meaning of an unfamiliar proverb (e.g., "A cracked pot never falls off the hook") presented in isolation. The continuity models do not recognize a distinction between literal and figurative language, so an instruction that a sentence, out of context, is to be "literally" or "figuratively" interpreted should be meaningless. Models which recognize the distinction require semantic anomaly to trigger figurative-level processing, and proverbs are rarely anomalous. That people do, however, derive the figurative meaning of proverbs presented in isolation is suggested in an early study by Harrower (1933). Harrower showed that proverbs were better recalled when presented in conjunction with other proverbs having the same figurative meaning than when presented with unrelated proverbs. More recently, Honeck et al. (1978) have shown that even 7-year-old children were able to select from a pair of pictures the one which was congruent with the figurative interpretation of a proverb. One purpose of the present study was to directly investigate the ability to understand and remember both the figurative and literal meaning of unfamiliar proverbs presented in isolation. Subjects who knew they would be tested for recognition of the figurative meaning of proverbs were required to indicate which of two different proverbs had the same figurative meaning as a previously presented proverb. It was therefore necessary for the subjects to recognize the same figurative meaning in a completely different literal form. Subjects who were tested for literal

62

Bock and Brewer

meaning had to choose between a literally synonymous but syntactically and lexically different form of the previously presented proverb and a proverb with the same figurative meaning as the previously presented proverb. These subjects thus had to differentiate literal and figurative interpretations of the same proverb. Because the continuity models do not provide special processes for figurative meaning comprehension, or suggest strategic diagnostics for evaluating a potentially correlated dimension like inferential depth, these models appear to predict that recognition of the figurative or literal meaning of proverbs presented under these conditions should be at a chance level. Models which provide special processes for figurative-level comprehension require the assumption that these processes can be invoked strategically, without a literal anomaly cue, but with this assumption these models predict better-thanchance recognition. A second concern of this experiment was the possibility of improving figurative meaning memory by providing an orienting task appropriate to the processing of this level of meaning (Bransford et al., 1979; Morris et al., 1977). Experiments by Honeck and his colleagues (Honeck, 1973; Honeck et al., 1975) indicate that proverb recall can be increased by providing subjects with a paraphrase of the figurative meaning of the proverb, suggesting that better figurative meaning comprehension leads to better memory for material which permits this level of interpretation. If this improvement is specific to the figurative level of interpretation, emphasizing comprehension of figurative meaning should increase figurative meaning recognition, but not recognition for literal meaning or surface structure information. This finding would provide further support for the view that figurative meaning is processed differently than literal meaning. In addition to recognition memory for figurative and literal meaning, the present experiment also examined recognition of stylisticsyntactic information, and lexical information. If subjects remember information from a variety of linguistic "levels" after hearing sentences (Anderson, 1974) it should be possible to demonstrate better-than-chance recognition for each of these types of information. Proverbs provide a unique opportunity to examine this question.

METHOD Subjects The subjects were 80 undergraduate students at the University of Illinois who participated in partial fulfillment of the course requirement in introductory psychology.

Comprehension and Memory of Proverbs

63

Materials

The procedures used to develop the materials produced proverb sets containing six members: (1) an original proverb, (2) a version of the original proverb that had undergone an optional surface transformation (optional transform item), (3) a version of the original proverb in which the content words were replaced by synonyms (synonym item), (4) a version of the original proverb which had undergone both optional transformation and synonym substitution (optional transform-synonym item); (5) a proverb with the same figurative meaning as the original proverb but with a different literal meaning (figurative meaning item), (6) a randomly selected proverb with a different literal and figurative meaning than the original proverb (random item). Twenty such sets were created as follows. Twenty pairs of proverbs, consisting of an original proverb and a pairmate of similar figurative meaning but different literal meaning (figurative meaning item) were chosen from standard works on proverbs such as Davidoff (1946) and Pullar-Strecker (1954). The pairs were chosen on the basis of two criteria: (1) the figurative meanings of the proverb pairs were rated "very similar" by each of four raters; and (2) both members of the pairs were rated "unfamiliar" by each of three raters. The 20 original proverbs were altered by an optional surface transformation to produce the 20 optional transform items. Insofar as possible, synonyms were substituted for the content words of the original proverbs to create the 20 synonym items. In seven of the 20 original proverbs all of the content words were replaced with synonym substitutions. In the remaining 13, lexical substitution was partial because it was not possible to find adequate synonyms. The 20 original proverbs contained 83 content words, and synonym substitutions were found for 63 (76%). The 20 optional transform-synonym items were developed by carrying out both optional transformation and synonym substitution on the original proverbs. The 20 random items were selected from additional unfamiliar proverbs that did not overlap with the original proverbs in either literal or figurative meaning. An example of one of the 20 sets is given Table I. An additional 18 proverbs were selected for use as filler items, and four additional sets of six items were developed for use as a warmup list. Since some of the original proverbs were expressed . in archaic vocabulary and syntax and some were not, the changes introduced by synonym substitution and optional transformation were sometimes toward more standard forms and sometimes away from more standard forms. To ensure that no bias was introduced into the experiment through

64

Bock and Brewer

Table I. Example of a Proverb Set i

i

i

Original proverb: "Out of another's purse it is easy to be generous." Optional transform item: ~'It is easy to be generous out of another's purse." Synonym item: " F r o m someone else's pocketbook it is easy to be charitable." Optional transform-synonym item: "It is easy to be charitable from someone else's pocketbook." Figurative meaning item: "Broad thongs are cut from other people's leather." Random item: "Even caviar tastes ill to him who is forced to eat it."

these changes, half the subjects received acquisition lists containing the 20 original proverbs and half received acquisition lists containing the 20 optional transform-synonym versions of the original proverbs. Both acquisition lists contained the same 18 filler proverbs. The items of each acquisition list were assigned to two 19-item experimental lists, so that ten of the items on each experimental list were original proverbs (or optional transform-synonym proverbs) and nine were filler proverbs. The filler proverbs on each experimental list were distributed so that two occurred at the beginning of each experimental list and two at the end of each experimental list. The order of the remaining items in the original proverb acquisition list was random, and the same random order was used for the optional transform-synonym acquisition list.

Recognition Memory Task A two-alternative forced-choice recognition memory procedure was used. There were four memory conditions differing with respect to the test items used in the recognition task. For subjects who received the original proverb acquisition list, memory for optional syntactic structure was tested with the original proverb vs. its optional transform item. The test for lexical memory contrasted the original proverb with its synonym item; for literal meaning, the optional transform-synonym item was tested against the figurative meaning item; and for figurative meaning, the figurative meaning item was tested against the random item. For the subjects receiving the optional transform-synonym acquisition list the respective test item pairs used were optional transform-synonym vs. synonym, optional transform-synonym vs. optional transform, original proverb vs. figurative meaning item, and figurative meaning item vs. random item. An example of the test items used in each of the four different memory conditions for the original proverb list is given in Table II.

Comprehension and Memory of Proverbs

65

Table II. Examples of Items for the Four Tests of Recognition Memory Test for Optional syntactic structure: Original proverb: "Out of another's purse it is easy to be generous." Optional transform item: "It is easy to be generous out of another's purse." Lexical information: Original proverb: ~'Out of another's purse it is easy to be generous." Synonym item: "From someone else's pocketbook it is easy to be charitable." Literal meaning: Optional transform-synonym item: "It is easy to be charitable from someone else's pocketbook." Figurative meaning item: "Broad thongs are cut from other people's leather." Figurative meaning: Figurative meaning item: "'Broad thongs are cut from other people's leather." Random item: -"Even caviar tastes ill to him who is forced to eat it."

Comprehension Task The c o m p r e h e n s i o n test developed for the original proverbs consisted o f two-alternative forced-choice items that required the comp r e h e n s i o n o f the figurative meaning of the original proverb for a correct r e s p o n s e . One o f the alternatives described a situation relevant to the figurative meaning of the p r o v e r b or paraphrased the figurative meaning o f the p r o v e r b , while the second alternative expressed a meaning that was the opposite o f or obliquely related to the figurative meaning of the original proverb. A p p r o x i m a t e l y half of the items were situational and half were p a r a p h r a s e s . For example, the comprehension test item for the p r o v e r b " H e that lies down with dogs will get up with fleas" required a choice b e t w e e n " N i x o n ' s popularity dropped drastically during Waterg a t e " and " S t . Augustine led a wild life as a young m a n . " The same c o m p r e h e n s i o n test items were used for the optional transform-synonym acquisition list. Similar c o m p r e h e n s i o n items were developed for the 18 filler p r o v e r b s and the four w a r m a p items. Design T h e r e were four types o f m e m o r y tests (optional syntactic structure, lexical information, literal meaning, figurative meaning) and two levels of c o m p r e h e n s i o n (acquisition lists either with the comprehension test or without the c o m p r e h e n s i o n test). There were eight groups of ten subjects.

66

Bock and Brewer

Each group of ten subjects was assigned to one of the eight treatment combinations. Procedure

All subjects received a four-item practice list and two 19-item experimental lists. The experimenter read the lists aloud with normal intonation, allowing 10 sec between proverbs. At the end of each list, the experimenter read four strings of seven-digit numbers, which the subjects wrote on designated pages of their answer booklets. Immediately following the digit-recall task, the subjects responded to the memory test. Approximately 1 min elapsed between the reading of the last proverb and the beginning of the memory test. Each subject received only one of the four types of recognition memory tests. Subjects receiving the test for optional syntactic structure and those receiving the test for lexical information were instructed to select the proverbs which were identical to the acquisition proverbs and were given examples of the type of test items they would receive. Subjects who received the test for literal meaning were instructed to select the proverbs with the same literal meaning as the acquisition proverbs, and those who received the test for figurative meaning were instructed to select the proverbs with the same figurative meaning. Subjects in the latter two conditions were given examples of literal and figurative meaning and examples of the type of test items they would receive. Items in the memory test were arranged in the same order as the acquisition list. However, the subjects were tested only on the 20 experimental proverbs and not on the 18 filler proverbs from the acquisition lists. This procedure was followed for subjects in each of two comprehension conditions. Subjects in the deep comprehension condition also responded to the comprehension item for the preceding proverb during the 10-sec interval between the proverbs on the acquisition list. Immediately after hearing each proverb the subjects in this condition turned to the next page in their answer booklet and had 10 sec to respond to the two-alternative forced-choice comprehension item. They were instructed that they were participating in an experiment on the comprehension and memory of proverbs, and that they would be given multiple-choice tests both for their understanding and memory of the proverbs they heard. Subjects in the normal comprehension condition were told that they were participating in an experiment on memory for proverbs and that they

Comprehension and Memory of Proverbs

67

would be given a multiple-choice test on their ability to remember the proverbs read to them. Subjects were run in small groups. The type of memory task was counterbalanced across subjects so that different subjects within the same group received different memory tests. RESULTS

The percentages of correct recognition responses in the normal comprehension and deep comprehension conditions for the four recognition memory tasks are given in Table III. Binomial tests for both items and subjects for bach type of recognition memory in each comprehension condition showed that recognition memory was significantly greater than chance (except for subjects in the literal meaning/normal comprehension treatment combination), p < 0.05 for all tests. An analysis of variance was carried out on the recognition memory responses with both subjects and sentences as random effects (Clark, 1973). Following Clark's notation, F~ refers to the test statistic appropriate to the design with subjects as the random effect and F, to the test statistic appropriate to the design with items as the random effect. The interaction between comprehension condition and type of recognition memory test was significant, F, (3, 72) = 5.10, F 2 (3, 57) = 10.84, min F" (3, 122) = 3.47, p < 0.05, so the scores for normal comprehension were compared with the scores for deep comprehension for each type of recognition test. The analysis for the figurative meaning task showed that recognition memory in the deep comprehenson condition was significantly superior to recognition memory in the normal comprehension condition, F~ (1, 18) = 7.93, F 2 (1, 19) = 16.48, min F' (1,32) = 5.35, p < 0.05. Similar analyses for the other recognition memory tasks showed no significant differences.

Table III. Percent Correct Recognition Responses under Two Levels of

Comprehension Recognitiontest

Normal comprehension

Deep comprehension

Figurative Literal Lexical Syntactic

77.5 76.5 95.5 80.0

89.5 85.5 91.0 70.0

68

Bock and Brewer

Examination of the responses on the comprehension task showed that 84% of the responses were correct, suggesting that subjects were reasonably accurate in interpreting the figurative meanings of the proverbs.

DISCUSSION The dual finding that memory for the figurative level of meaning was above chance in the normal comprehension condition, and was significantly improved by requiring subjects to identify instances or paraphrases of the figurative meaning during acquisition, supports the hypothesis that subjects can understand and later recognize the figurative meaning of an unfamiliar proverb presented in isolation. These results are most easily explained by theories of processing which suggest that figurative meaning comprehension is associated with qualitatively different processing than literal meaning comprehension. Such an approach to metaphor is suggested by Kintsch (1972) and Weinreich (1966). One problem for these "multiple process" accounts when applied to proverb comprehension arises from their claim that figurative meaning computation depends on the recognition of literal anomaly. The present experiment suggests that literal anomaly is not a necessary condition for the processing of figurative meaning and that such processing can be strategically controlled by subjects. A possible solution to this problem is to allow contextual anomaly or inappropriateness of the literal meaning of an expression in the context of its utterance to trigger the processing of figurative meaning. This type of solution is embodied in an account of metaphor processing called the "pragmatics approach" by Ortony et al. (1978b), which derives from Grice's (1975) work on conversation and is supported I~y an experiment on indirect speech by Clark and Lucy (1975). Kintsch (1974) has made a similar suggestion. With the assumption that figurative-level processing may also be triggered by instructions, these multiple-process models appear to be better able to explain proverb comprehension than continuity models. An approach that incorporates features of both the continuity and multiple-process views has been offered by Ortony et al. (1978b). In an experiment contrasting effects of short vs. long contexts on comprehension time, these investigators found that, with extensive contexts, the times taken to process metaphors vs. literal language did not differ significantly, but metaphorical statements took longer than literal state-

Comprehension and Memory of Proverbs

69

ments with shorter contexts. With long contexts, both metaphorical and literal statements were comprehended faster than with short contexts. This experiment therefore suggests that it is not literal vs. figurative language p e r s e that triggers the use of special comprehension processes; rather, figurative and literal language may be processed in the same way, and both may also, on occasion, require extraordinary processing. When normal processing breaks down, additional processing (as in the multiple process description) is required. The distinction between literal and figurative language, or between metaphorical and nonmetaphorical uses of language, is therefore not a processing distinction, in this view. This analysis accords well with the intuition that much figurative language is processed routinely in appropriate contexts. However, in order to account for the fact that metaphors are different from literal language in short contexts, the analysis requires an independent definition of metaphor, in addition to specification of the conditions for breakdowns of processing, and a description of the additional processing procedures employed (cf. Ortony, in press). The definition of metaphor would presumably establish criteria for metaphorical or figurative usage against which particular examples could be evaluated, while the specifications of processing failure and backup procedures should yield a taxonomy of processing operations which is imperfectly correlated with the criteria for metaphorical usage. The comprehension of unfamiliar proverbs in isolation found in the present experiment could then, in part, be explained in terms of strategic use of these processing operations. The improvement in deep comprehension recognition scores is also consistent with the "transfer-appropriate" processing views of Bransford and his colleagues (Bransford e t al., 1979; Morris et al., 1977), which emphasize the importance of appropriate acquisition experience for subsequent testing. A task requiring comprehension of figurative meaning should improve recognition for figurative meaning more than for other levels of meaning, if selective processing is possible; the results accord with this prediction. The findings of the present study differ substantially from several earlier experiments which examined memory for surface information and changes in literal meaning (Begg, 1971; Johnson-Laird and Stevenson, 1970; Perfetti and Garson, 1973; Trembath, 1972; Sachs, 1967, 1974). These studies generally found poor lexical memory, while the accuracy of lexical recognition found here was quite high. However, there are some fairly obvious differences in procedure that may account for this discrepancy. In the present experiment, lexical information was tested in a

70

Bock and Brewer

forced-choice task with the original item vs. an item in which most or all of the content words were replaced with synonyms; thus any fragmentary information about any one of the content words would allow the subject to make a correct response. The previous investigations used simple recognition for items in which only one word was replaced by a synonym, and thus may have given an underestimate of the amount of lexical information remembered by the subject for a given sentence. This conclusion is supported by the work of Hayes-Roth and Hayes-Roth (1977). The present study also found good recognition of optional syntactic information, probably because of the more sensitive forced-choice recognition procedure. It is difficult to compare this result with earlier work on recognition for literal meaning. In the current experiment, literal meaning was tested using an item in which the surface structure was greatly altered, while the literal meaning was unchanged. Other studies have used items in which the surface structure was essentially unchanged, while the literal meaning was altered. Thus the present study examined recognition of literal meaning, while the previous experiments investigated recognition of meaning change. Overall, the present findings agree with observations that subjects can remember information from a variety of processing levels (Anderson and Paulson, 1977; Kolers and Ostry, 1974). In summary, the major results of this experiment suggest that subjects can reliably identify the figurative meaning of proverbs presented in isolation, and that, if appropriate comprehension activities are required, memory for this information can be selectively improved. These findings can be best accounted for by approaches which provide for differences in the processing of figurative and literal meaning. ACKNOWLEDGMENTS

The authors wish to thank Ellen Brewer for help with the data analysis and Richard Harris, Tom Thieman, and Rose Zacks for comments on an earlier version of the manuscript. REFERENCES Anderson, J. R. (1974). Verbatim and propositional representation of sentences in immediate and long-term memory. J. Verb. Learn. Verb. Behav. 13:149-162. Anderson, J. R., and Bower, G. H. (1973). Human Associative Memory, Winston, Washington, D.C.

Comprehension and Memory of Proverbs

71

Anderson, J. R., and Paulson, R. (1977). Representation and retention of verbatim information. J. Verb. Learn. Verb. Behav. 16:439-45 l. Begg, I. (1971) Recognition memory for sentence meaning and wording. J. Verb. Learn. Verb. Behav. 19:176-181. Bobrow, D. G., and Norman, D. A. (1975) Some principles of memory schemata. In Bobrow, D. G., and Norman, D. A. (eds.), Representation and Understanding: Studies in Cognitive Science, Academic Press, New York. Bransford, J. D., Franks, J. J., Morris, C. D., and Stein, B. S. (1979). Some general constraints on learning and memory research. In Cermak, L. S., and Craik, F. I. M. (eds.), Levels of Processing and Human Memory, Erlhaum, Hillsdale, N.J. Biihler, K. (1908). Tatsachen und Probleme zu einer Psychologic der Denkvorg~inge. 11I. Uber Gedankenerinnerungen. Arch. Gesamte Psychol. 12:93-122. Biihler, K. ( 1951). [On thought connections.] In Rapaport, D. (ed. and trans.), Organization and Pathology of Thought': Selected Sources, Columbia University Press, New York. Clark, H. H. (1973). The language-as-fixed-effect fallacy: A critique of language statistics in psychological research. J. Verb. Learn. Verb. Behav. 12:335-359. Clark, H. H., and Lucy, P. (1975). Understanding what is meant from what is said: A study in conversationally conveyed requests. J. Verb. Learn. Verb. Behav. 14:56-72. Collins, A. M., and Quillian, M. R. (1972). How to make a language user. In Tulving, E., and Donaldson, W. (eds.), Organization of Memory, Academic Press, New York. Davidoff, H. (1946). A World Treasury of Proverbs, Random House, New York. Grice, H. P. (1975). Logic and conversation. In Cole, P., and Morgan, J. L. (eds.), Syntax and Semantics, Vot. 3: Speech Acts, Academic Press, New York. Harris, R. J. Comprehension of metaphors: A test of the two-stage processing model. Bull. Psychon. Soc. 8:321-324. Harrower, M. R. (1933). Organization in higher mental processes. Psychol. Forsch. 17:56-120. Hayes-Roth, B., and Hayes-Roth, F. (1977). The prominence of lexical information in memory representations of meaning. J. Verb. Learn. Verb. Behav. 16:119-136. Honeck, R. P. Interpretive versus structural effects on semantic memory. J. Verb. Learn. Verb. Behav. 12:448--455. Honeck, R. P., Riechmann, P., and Hoffman, R. R. (1975). Semantic memory for metaphor: The conceptual base hypothesis. Memory Cogn. 3:409-415. Honeck, R. P., Sowry, B. M., and Voegtle, K. (1978). Proverbial understanding in a pictorial context. Child Dev. 49:327-331. Johnson-Laird, P. N., and Stevenson, R. (1970). Memory for syntax. Nature 227:412. Katz, J. J., and Fodor, J. A. (1963). The structure of a semantic theory. Language 39:170-210. Kintsch, W. (1972). Notes on the structure of semantic memory. In Tulving, E., and Donaldson, W. (eds.), Organization of Memory, Academic Press, New York. Kintsch, W. (1974) The Representation oj Meaning in Memory, Erlbaum, Hitlsdale. Kolers, P. A., and Ostry, D. J. (1974). Time course of loss of information regarding pattern analyzing operations. J. Verb. Learn. Verb. Behav. 13:599~12. Matthews, R. J. (1971). Concerning a "linguistic theory" of metaphor. Found. Lang. 7:413-425. Morris, C. D., Bransford, J. D., and Franks, J. J. (1977). Levels of processing versus transfer appropriate processing. J. Verb. Learn. Verb. Behav. 16:519-533. Ortony, A. Beyond literal similarity. In Ortony, A. (ed.), Metaphor and Thought, Cambridge University Press, Cambridge (in press).

72

Bock and Brewer

Ortony, A., Reynolds, R. E., and Arter, J. A. (1978a). Metaphor: Theoretical and empirical research. Psychol. Bull. 85:919-943. Ortony, A., Schallert, D. L., Reynolds, R. E., and Antos, S. J. (1978b). Interpreting metaphors and idioms: Some effects of context on comprehension. J. Verb. Learn. Verb. Behav. 17:465-477. Perfetti, C. A., and Garson, B. (1973). Forgetting linguistic information after reading. J. Educ. Psychol. 65:135-139. Perrine, L. (1971). Four forms of metaphor. College Eng. 33:125-138. Pullar-Strecker, H. (1954). Proverbs for Pleasure, Christopher Johnson, London. Richards, I. A. (1965). The Philosophy of Rhetoric, Oxford University Press, New York. Sachs, J. S. (1967). Recognition memory for syntactic and semantic aspects of connected discourse. Percept. Psychophys. 2:437-442. Sachs, J. S. (1974). Memory in reading and listening to discourse. Memory Cogn. 2:95-100. Trembath, A. A. (1972). Comparison of sensitivity to the surface and deep structure of sentences in children. Lang. Speech 15:51-57. Verbrugge, R. (1977). Resemblances in language and perception. In Shaw, R., and Bransford, J. (eds.), Perceiving, Acting, and Knowing, Erlbaum, Hillsdale, N.J. Verbrugge, R. R., and McCarrell, N. S. (1977). Metaphoric comprehension: Studies in reminding and resembling. Cogn. Psychol. 9:494-533. Weinreich, U. (1966). Explorations in semantic theory. In Sebeok, T. A., (ed.), Current Trends in Linguistics, Vol. 3: Theoretical Foundations, Mouton, The Hague.