Short Reports Can Meaning Be Extracted from Meaningless Stimuli? JOHN O. MITTERER AND IAN BEGG McMaster University

ABSTRACT

In a levels-of-processing experiment, conventionally meaningless items (e.g. NDD) are shown to be recognized better after a decision concerning meaning than after a decision concerning pronunciation. The problematic nature of this result for theories relying upon the meaning 'extraction' metaphor is discussed, and an alternative point of view in which meaning is held to result from the interpretive activity of the receiver of a message is offered. Although it has long been assumed that different study methods result in better or worse memory, it is only recently that there has been any systematic investigation of the relation between the manner of study and the excellence of retention. Interest in the

problem has increased dramatically since the appearance of Craik and Lockhart's (1972) levels-of-processing framework. Those authors have shown, as have many others since, that words are remembered better if initial processing concerns their meaning rather than their sound or appearance. For example, the visual stimulus DOG is more likely to be remembered if it is processed to answer a question like 'Is it an animal?' than a question like 'Does it rhyme with log?.' The most obvious question raised by the results concerns why the different types of processing result in differential memorability, and this question has attracted the lion's share of debate concerning the basic results (e.g., Baddeley,

1978; Morris, Bransford, & Franks, 1977). However, the results also raise more neglected questions about the nature and locus of meaning. By the levels account, meaning is conceptualized as an all-or-none attribute of words, which either is or is not processed during study, and further, the meaning of a word is analyzed or extracted from the physical stimulus. The purpose of this paper is to advance an alternative account of meaning. In our view, verbal stimuli are merely vocal voices or visual marks, containing no meaning to extract. Rather, meaning is constructed by the cognitive activity of the receiver in the process of performing whatever task is required (cf. Begg, 1976; Begg, Upfold, & Wilton, 1978)Craik and Lockhart's (1972) position is that processing is a continuous analytical procedure in which the subject first extracts physical information from the stimulus, then proceeds to extract deeper semantic information if the task requires it. The position is consistent with the common metaphor that words are vehicles which convey meaning for receivers to extract or unpack. The extraction metaphor has been a very useful one in directing our attention towards important aspects of the communication process. However, there is a big difference between using a metaphor for strategic purposes, and believing that the metaphor is true in any fundamental sense (see Turbayne, 1971). The basic point that verbal stimuli do not contain meaning in any more than a metaphoric sense has repeatedly been made outside of psychology, both by philosophers as diverse as Russell (1940/1973), Cassirer (1953 translation), and Merleau-Ponty (1967 translation), and by linguists such as Saussure (1974 translation) and Whorf (1956). According to Whorf, the opponent 'extraction' view treats words as being analogous to units of currency, in that each has some fixed value; in his opinion, words are more like blank cheques to be filled in by the user as re-

Canad. J. Psychol./Rev. canad. Psychol., 1979,33 (3)

193

quired by contextual demands. Although we agree with the philosophers and linguists, it is our intent to bring experimental rather than argumentative evidence to bear on the issue. In the experiment reported in this paper, subjects initially processed verbal items with regard to their meaning or their pronunciation. The primary concern of the experiment is with items such as FEP and NDD, which are not conventionally meaningful. If meaning is an attribute of verbal stimuli, these stimuli contain none. Since no meaning can be extracted, there should be no retention advantage for deciding that the items are meaningless as opposed to deciding whether or not they can be pronounced. However, if meaning is constructed by the cognitive activity of the receiver, then the act of deciding that the items have no conventional meaning should be sufficient to render them meaningful in some sense, with consequent benefits for retention. The experiment also included meaningful items, such as DOG and FBI, which should benefit from meaningful processing by either account. Although the extraction notion leads quite directly to the expectation that meaningless items will not benefit from meaningful processing, there has been little research relating stimulus attributes to initial processing requirements. Seamon and Murray (1976) compared processing tasks for items differing on Noble's m, the number of verbal associates elicited in a fixed amount of time. Although they found a larger benefit for meaningful processing with high- rather than low-m items, the result could be interpreted quite easily from both the extraction position and the construction position, since all the items were conventionally meaningful words. Consequently, the issue remains in doubt. METHOD

Subjects

Sixty students participated as a course requirement.

194

Materials

The memory items were of four types, with each item three letters in length. There were familiar words (e.g., SEA, INN, DOG), familiar acronyms (e.g., FBI, NHL, LSD) pronounceable non-words (e.g., FEP, URO, SIE) and non-pronounceable non-words (e.g., NDD, AAC, LRD). We began with a reasonably large pool of acronyms, both selected from published sources and generated by ourselves and several colleagues. We discarded acronyms we felt were obscure, then selected words which as a set were matched with the remaining acronyms in the likelihood that each position was occupied by a vowel. The letters from the words and acronyms were then used to generate pronounceable and non-pronounceable non-words. Each item was then printed on a separate index card, with the entire deck given to each of five colleague-volunteers who sorted the cards into the four categories mentioned earlier. Only items correctly sorted by all volunteers were used as memory items. The memory list consisted of 64 items, with equal numbers of each type. The 64 items were arranged in a 16-page booklet, with 4 items per page; except for the restriction that each block of 8 items contain 2 of each type, all assignments were random. Thus, each sequence of 8 items had 4 meaningful items, and 4 pronounceable items. The first 16 and last 8 items were included as primacy and recency buffers, and were not scored. Accordingly, the memory items consisted of 40 trigrams, 10 each of the four types. An additional 10 of each type were also selected to serve as distractors in a recognition test. Procedure

The basic study procedure involved having subjects answer 'yes' and 'no' to a processing question for each item in the study list, with 10s allowed for each 4 items. Four groups of subjects were tested, with 20 in each of 2 groups, and 10 in each of 2 others. One group of 20 subjects classified items as meaningful or not, circling the former, while the other group of 20 circled pronounceable items; the two groups defined a between-subjects manipulation of processing task. The groups of 10 subjects were used for a within-subjects manipulation; each group classified the items for meaningfulness on half the pages and pronounceability on the other half, with each group answering a different question for any page. In later analyses, erroneous responses are ignored, since the error rate was under 4% in all conditions. Following completion of the study list, subjects were administered an unexpected recognition test, with instructions to circle items which had been present

J.O. Mitterer & I. Begg

TABLE I

Recognition of different types of items after meaningful or pronunciation processing Processing question Condition

ftem type

Meaningful?

Pronounceable?

Between-subjects

Meaningful - Pronounceable Meaningful — Non-pronounceable Meaningless — Pronounceable Meaningless - Non-pronounceable

2.72(0.77) 2.99(0.98) 1.66(0.77) 0.99(1.15)

2.24(0.86) 1.94(0.75) 1.33(0.83) 0.58(0.66)

Within-subjects

Meaningful - Pronounceable Meaningful — Non-pronounceable Meaningless — Pronounceable Meaningless — Non-pronounceable

2.59(1.00) 2.91(1.09) 2.13(0.97) 1.14(1.17)

2.05(1.04) 2.17(1.58) 1.85(1.02) 0.82(1.12)

Note. All entries are of d' values, with standard deviations in parentheses; respective examples of the four item types are DOG, FBI, FEP, and NDD.

in the study list. The test was self-paced, but no subject required more than 5 minutes. RESULTS

For both the between- and within-subjects conditions, results were analyzed by 2 X 2 X 2 analyses of variance, with d' values (Elliott, 1964) as dependent measures. Stimulus meaningfulness and pronounceability were repeated factors in each analysis, with instructions either a repeated or independent factor, depending on the condition: F( 1,1 g) values correspond to the within-subjects manipulation of instructions, and F(i, 38) values correspond to the between-subjects condition, with a = .05. The means for the different conditions are presented in Table 1. It is obvious from the table that meaningful processing exceeded pronunciation for each type of item, ^(1,38)= 16.6,^(1, 19) = 6.60. It is simply not the case that the instructional effect disappears with meaningless items. Although there was no interaction between item type and processing task, the different item types did differ considerably in memorability. Meaningful items exceeded meaningless ones, F(i, 38) = 101, F(i, 19) = 356, pronounceable items exceeded non-pronounceable ones, F(i, 38) = 6.95,^(1,19) = 3.10,/? = .09, and the two Can meaning be extracted?

variables interacted, F(i, 38) = 8.84, F(i, ig) = 7.56; the interaction simply means that the advantage for pronounceable over non-pronounceable items was only present if the items were meaningless. DISCUSSION

The main result of the experiment was that meaningless items benefit as much from meaningful processing as do words and acronyms. Such an outcome simply cannot be explained on the basis of extraction of meaning from physical stimuli, because meaningless items have no meaning to extract. The finding leaves us with two questions. First, what is meaning? Second, why should a decision that NDD is meaningless give better retention than a decision that the same item is unpronounceable? Although the question regarding the nature of meaning has baffled philosophers and linguists for centuries, there are some simple psychological answers available. Considered as physical events, linguistic signals are simply vocal noises or visual marks on a page. The physical properties of the signals neither define nor convey semantic information, or DOG would be equally meaningful to speakers of English and French. However, some vocal signals have a history of use in certain non-

195

linguistic circumstances, with the result that members of a linguistic community learn to interpret those signals in similar fashion. In this way, signals like DOG and FBI become conventionally meaningful, which is a metaphoric shorthand for stating that people interpret the signals in specifiable and shared ways. In short, the meaning of an item in a particular circumstance can be defined as the particular interpretation of that item in that context. On this functional view, any stimulus can be rendered meaningful, albeit non-conventionally, by the cognitive activity of a creative human receiver. Why then should NDD benefit from meaningful processing? In order to decide that NDD has no conventional meaning, one may need to contrast it with known acronyms such as NDP, or to construct plausible referents, such as National Dentists' Delegation. Thus the event in which NDD was processed may be meaningful as an event, with a memory trace containing information sufficient to distinguish NDD as a test item from unpresented items. In contrast, an attempt to pronounce the item can be accomplished by peripheral articulatory attempts, with such attempts less distinguishable from each other than are more cognitive interpretive events. Regardless of the particular explanation one prefers regarding why meaningful processing is ever beneficial, it is clear that the locus of the effect is in the cognitive activity of people, not in the extent to which available physical information allows extraction of meaning. The extraction metaphor is not unique to the levels-of-processing conception. Indeed, it is quite common to regard language as a 'vehicle' which 'conveys' meaning, as a train conveys goods. By this metaphor, the receiver of a signal simply 'unpacks' the meaning from the message. This metaphor is simple, and it does direct us towards meaningful parsing of communicational sequences. Therefore, it may often be a useful strategy to treat meaning as if it were

196

extracted from physical stimuli. However, it is not obvious that all, or even most, contemporary theorists use the extraction notion in such a pragmatic way, or whether extraction is actually taken to be the case. We have progressed beyond the conception that the human is simply a receiver for introjected or intruded physical stimuli. This progression is towards more meaningful investigation of how we learn to interpret events, how those interpretations are accomplished, and what will be the memorial consequences for some types of interpretation as opposed to others. RESUME

Une experience sur les niveaux de traitement d'un stimulus montre que des items conventionnellement depourvus de signification (v.g. NDD) sont mieux reconnus lorsque la decision porte sur la signification plutot que sur la prononciation. La discussion souligne le caractere problematique d'un tel resultat pour les theories reposant sur la metaphore 'extraction de signification' et presente une perspective de rechange suivant Iaquelle la signification resulterait de l'activite interpretative de celui qui recoit le message. REFERENCES BADDELEY, A.D. The trouble with levels: A reexamination of Craik and Lockhart's framework for memory research. Psychol. Rev., 1978, 85, 139-152 BEGG, 1. Acquisition and transfer of meaningful function by meaningless sounds. Canad.J. Psychol., 1976, 30, 178-186 BEGG, I., UPFOLD, D., & WILTON, T.D. Imagery in verbal

communication. J. Ment. Image., 1978, s, 165—168 CASSIRER, E. Language and myth. New York: Dover, 1953 CRAIK, F.I.M., & LOCKHART, L.S. Levels of processing: A

framework for memory research.y. verb. Learn, verb. Behav., 1972, 11,671—684 ELLIOTT, P.B. Tables of a". In J.A. SWETS (Ed.), Signal detection and recognition by human observers. New York:

Wiley, 1964, pp. 651-684 MERLEAU-PONTY, M. The structure of behavior. Boston: Beacon, 1967 MORRIS, CD., BRANSFORD, J.D., & FRANKS, J.J. Levels o f

processing versus transfer appropriate processing. J. verb. Learn, verb. Behav., 1977, 16, 519—533

J.O. Mitterer & I. Begg

RUSSELL, B. An inquiry into meaning and truth. Norwich, UK: Penguin, 1973 SAUSSURE, F. DE. Course in general linguistics. Glasgow: Fontana Collins, 1974

TURBAYNE, CM. The myth of metaphor. Columbia, sc: University of South Carolina Press, 1971 WHORF, B.L. Language, thought and reality. J.B. CARROLL (Ed.). Cambridge, Mass: M.I.T. Press, 1956

SEAMON, j . c , & MURRAY, p. Depth of processing in

recall and recognition memory: Differential effects of stimulus meaningfulness and serial position. J. exp. Psychol.: Human Learn. Mem., 1976, 2, 680—687

Can meaning be extracted?

(First received 5 January 1979) (Date accepted 2 / March 1979)

197

Can meaning be extracted from meaningless stimuli?

Short Reports Can Meaning Be Extracted from Meaningless Stimuli? JOHN O. MITTERER AND IAN BEGG McMaster University ABSTRACT In a levels-of-processin...
314KB Sizes 0 Downloads 0 Views