F5.- '59,; 2f; )1. '; #43, W at: gm”- 1- ‘ ‘, ‘ .s z . .311: 3 _ A u Ag???“ ’5‘; MEL! AFWT " '* f‘<-P;.-..j A «#4 I-‘e‘afi' .2 r... 4x L' ' ‘2} U 'L 1.25:: 2‘ “~- ' . , - . ~ 1 '. ‘ ' '- ‘ ""le ' ~ “"9. . , g, . ‘~‘ w ’ 2‘ ‘mw- ‘ i“: :«w ‘i . V ‘51,". ' .. flax-g,- ‘. ‘Jg,‘ . - .." ~n , j ' . , ,1- ‘ n , ‘ .u“__ . . ,. J r Haj “3‘ L. . “:51: 1‘13:- “31%,? "fivfl&'§€’fi:‘3§7}g& 553%. . W, W. MWWZ-.1 “A I , 7 v", 44' ‘ "7‘ ' 'w‘ ‘I ~ -_\ ~ “gig ‘u€\.~ ‘ ~ . . 35" 5w" M L-' - fl: J . '- 3-” x _.e ' 46% 3:14. writ-- ' " 4 , ,.... . ' .cc'xl- “- “"11- " ~ :4. «3-33.. :uo‘r‘tfl.‘{!. ‘1'. "L. w 1“! 7 . «X: H135 'Js - . 1 ~ $1.4“ :- . “ ”.- 1’8: W’w - _‘ I- - fi .. n ' . * ,, v -' 4:36,; mg. . - erx‘zz-r"; » vgfflr‘gnw "1-,;ku "" "3a,“: r “V 1 ..~ , W P . "w ”3",“ "‘ .« "‘\v‘« ' v” " ~ ‘ “r.- _‘ ' 1. - ~ . ¢ 7'. .9 U- ‘ Y . .4 swat v m; IHIHHlWHIll\1|\\\W|\\|H\\|\|\n\\\l|\\\|\\\‘\\|\|\\W 3 1293 01046 This is to certify that the thesis entitled Learning New Vocabulary Words: The Role of Inhibition presented by Debra Lynn Wilson has been accepted towards fulfillment of the requirements for Masters degree in Psychology We Major professor 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution LIBRARY Michigan State UnlversIIy PLACE ll RETURN BOX to removothb chockom Iron your record. TO AVOID FINES return on or before date duo. DATE DUE DATE DUE DATE DUE }[ D MSU II An Nflrmatlvo Adlai/Equal Oppommlty Imtttutlon m LEARNING NEW VOCABULARY WORDS: THE ROLE OF INHIBITION BY Debra Lynn Wilson A THESIS Submitted to Michigan State University in partial fulfillment for the requirements for the degree of MASTER OF ARTS Department of Psychology 1994 ABSTRACT LEARNING NEW VOCABULARY WORDS: THE ROLE OF INHIBITION BY Debra Lynn Wilson The possibility of an inhibitory center surround attentional mechanism aiding retrieval of newly learned vocabulary words was tested by examining priming effects in a lexical decision task. According to Dagenbach, Carr, an'i Earnhardt (1990) newly learned primes whose meanings are recognized but not recalled can lead to inhibition of related target words, whereas primes whose meanings are recalled facilitate related targets. The present experiments attempted to replicate these effects, and manipulated number of exposures to the new words to chart deployment of the attentional mechanism that supposedly produces them. Results indicated that when primes were recalled facilitation occurred and when primes were unrecalled neither facilitation nor inhibition appeared in any systematic fashion. These experiments showed that facilitatory episodic priming was present in early learning and this priming depended on subjects being able to recall the newly learned meanings, but produced no evidence for center-surround inhibition in recalling them. ACKNOWLEDGEMENTS This body of research has benefitted immensely from the support and careful guidance of Dr. Tom Carr. I am grateful to him for his patience and constant desire to challenge and inspire me throughout my graduate school years. Also due thanks are the other members of my masters committee: Dr. Rose Zacks and Dr. John Henderson. Their input and advice have greatly improved and polished this masters thesis. m TABLE OF CONTENTS LIST OF TABLES INTRODUCTION EVIDENCE OF PRIMING FRON NEWLY LEARNED INFORMATION RETRIEVING NEWLY ESTABLISHED CODES DIRECTIONS FOR FURTHER RESEARCH EXPERIMENT 1 METHOD Subjects Materials and Procedures RESULTS AND DISCUSSION EXPERIMENT 2 METHOD Subjects Materials and Apparatus Procedures RESULTS DISCUSSION EXPERIMENT 3 METHOD Subjects Procedures RESULTS DISCUSSION vi 18 21 21 21 21 21 22 24 24 24 25 28 33 36 37 37 37 38 44 GENERAL DISCUSSION APPENDIX.A: APPENDIX B: APPENDIX C: REFERENCES STIMULUS MATERIALS FOR EXPERIMENT 2 STIMULUS MATERIALS FOR EXPERIMENT 3 ADDITIONAL INFORMATION FOR EXPERIMENTS 2 AND 3 45 48 54 74 78 Table 1. Mean RT Lexical Mean RT Lexical Mean RT Lexical LIST OF TABLES (msec) and Percent Error for Positive Decisions as a Function of Prime Recall (msec) and Percent Error for Positive Decisions as a Function of Prime Recall (msec) and Percent Error for Positive Decisions Broken Down by Recall Latency vi Page 31 4O 43 Episodic and semantic memory have long served to organize our ideas about how information is stored and retrieved from long term memory. Episodic memory is generally defined as memory for events that have occurred at a certain time and place whereas semantic memory is memory for general information about the world without the temporal-spatial context attached. Many argue over the reality of two separate stores in memory. Tulving (1972) defines the two as separate with different properties. Episodic memory stores specific events that happen to a person and semantic memory stores facts and general information about the world. Since this declaration from Tulving much research has been conducted trying to prove the existence of one memory store with episodic and semantic information being two points along a continuum (Ashcraft, 1989). Information in semantic memory was described by Ashcraft (1989) as being highly related and integrated and overlearned. This seems to suggest that if certain episodic information is overlearned and practiced it will begin to become integrated into semantic memory and will also behave like semantic information. We may then wonder how much exposure to information and practice does it take to 2 integrate episodic memories into semantic ones. A criterion commonly used in the memory literature suggests that if words automatically prime semantically or associatively related words then the information is semantic in nature. It is a general phenomenon in the semantic priming literature that if a related word occurs before a target in a lexical decision trial, responses will be faster to the target than if it was preceded by an unrelated word (Meyer & Schaneveldt, 1971; Neely, 1991). Subjects will be faster to decide that butter is a word if it is preceded by bread rather than an unrelated word such as table. So, if automatic priming occurs in episodic memory then we could suggest that the information is integrated and part of semantic memory. From recent priming literature one may assume that priming is automatic when a 200 or 250 ms Stimulus Onset Asynchrony (SOA) separates the prime and target and facilitation still occurs from newly learned words that is approximately equal to the facilitation produced under the same conditions by old, well learned words. Evidence of Priming from Newly Learned Information Many researchers have attempted to find automatic priming with newly learned or episodic information. Neely (1977) told subjects that an instance from category A (body parts) would be followed by an instance from category B (buildings) in a primed lexical decision task. He wanted to see if automatic priming would occur if subjects expected 3 these two types of stimuli to come together. Priming did occur, but only at a long SOA suggesting that strategies were coming into play and that the priming was not automatic as in semantic memory. McKoon and Ratcliff (1986) also tried to find automatic priming with newly learned information. They taught subjects new associations and then tested them to see if one member of the pair automatically primed the other member. McKoon and Ratcliff claimed that newly learned associations did in fact automatically prime each other in a lexical decision task. Durgunoglu and Neely (1987) reviewed various articles which had failed to find automatic priming in episodic associations. In their paper they attempted to systematically discover what parts of McKoon and Ratcliff's design were necessary to produce their results. They manipulated SOA, the use of nonwords in the study list and the episodic and semantic associations within the study list. They tried to replicate McKoon and Ratcliff's design and realized that they only found priming at short SOAs when subjects responded "word" when they saw a studied target and "nonword" when they encountered a nonstudied target creating a definite bias in responding. Episodic priming was found only when the episodic information was useful. The basic flaw in McKoon and Ratcliff's design lessens their claim of automatic priming in episodic memory. It seems then that previous studies show that episodic associations are not very well learned and integrated with 4 limited amounts of study. Dagenbach, Horst, and Carr (1990) were interested in just how much study of new information is necessary in order for it to become well integrated and automatically accessed. They were also interested in whether an association between two unrelated words or a new vocabulary word and its definition would be easier to integrate into semantic memory. In Experiment 1 subjects learned 24 unrelated word pairs by studying each of them for 30 s on the computer screen and then completing booklets where they were given one word of the pair and had to provide the other. They then did lexical decision trials with a 200 ms SOA in order to see if automatic facilitation occurred for the association. Although a small amount of facilitation did occur it was far from significant. They then did the same procedure but used new vocabulary words and their definitions as the episodic association. They thought that perhaps it would be easier to create a new association in memory rather than create one between two words that were already well established in memory and had associations of their own. In the lexical decision trials, the studied words served as the primes and the targets were either related words, unrelated, or nonwords. Surprisingly, when a studied prime was followed by a related word, a small amount of inhibition rather than facilitation occurred, although it was also not significant. The last two experiments were similar to the first two but study time was increased to a 5 week period with both vocabulary words and 5 their definitions and unrelated word pairs. Automatic facilitation occurred in the lexical decision trials between the new vocabulary words and related words, but no facilitation existed for the previously unrelated word pairs that were studied. It seems that semantic memory can develop more easily from a new, meaningful association than from a link between two unrelated words in semantic memory, at least when the operation of semantic memory is assessed by way of a short SOA or automatic priming. ~Even for meaningful associations, however, development appears to be slow and effortful. Retrieving Newly Established Codes It is becoming quite obvious that episodic associations take a lot of over learning and practice in order to become well integrated in semantic memory. The question then becomes how do we retrieve information from episodic memory when the codes are quite new and weak? There seems to be a period of time when these codes get used but not automatically like information in semantic memory as in Neely (1977). Using this new information takes effort and attention. At this point it is important to focus on non- automatic priming, priming that occurs over a longer SOA, giving attention and other strategies time to activate the new weak code when it is in competition with stronger, related codes. A few recent studies suggest that an attentional mechanism could be playing an important role in helping new 6 codes beat out the competition and gain access into working memory. Dagenbach, Carr, and Wilhelmsen (1989) provided the first evidence for this attentional mechanism in their masked priming work. In masked priming the prime in the lexical decision trial is briefly presented and then a mask appears on the screen. The mask then disappears and the target appears. Most (though not all) researchers agree that masked visual primes can activate the meaning of the prime enough to have an effect on the target. The controversial issue is how long the prime must be presented before the mask in order for the prime to influence the target while still leaving the person unaware of the prime's identity. Therefore experiments of this type employ a psychophysical threshold-setting procedure to determine the prime—mask SOA that limits or prevents awareness. Then a set of primed lexical decision trials is run using prime- mask SOA's determined in the psychophysical threshold— setting procedure. In this experiment Dagenbach et al. used different kinds of judgment tasks in the psychophysical threshold— setting procedure to determine prime-mask SOA's. They reasoned that even if the retrieval attempt of the masked prime fails, the effort of trying to obtain the prime could influence responding to the target. Dagenbach et al. (1989) believed that different threshold setting tasks require accessing different types of information about the prime and should affect the retrieval strategies and processing of the 7 prime and subsequently the target. The subjects' first threshold task was to decide whether a blank field or a word was presented before the mask (detection task). Half of the trials that preceded the mask consisted of a word and half consisted of a blank field. Next, after the detection task, each subject performed a second threshold judgment task, representing one of three different variations of a word discrimination task. One variation (constrained detection) asked subjects whether a certain word (doctor) or a blank field was presented. Another (word—word discrimination) asked subjects which of two words was presented (for example, doctor or table). In a final variation of the threshold judgment called the semantic similarity judgment task, a prime word would be presented (doctor) and subjects would be asked which of two words was closer in meaning to the prime (nurse or table). Immediately after this second judgment task, subjects performed a set of primed lexical decision trials. The prime was presented at the detection- threshold SOA determined in the first threshold judgment task. Thus physical prime presentation conditions were approximately the same for all subjects, but subjects differed in which type of judgment they had last been trying to make just before they participated in the primed lexical decision trials. Dagenbach et al. found facilitation from the masked prime to the target when subjects had just been making detection judgments or word discrimination judgments. However inhibition occurred after subjects performed 8 semantic similarity judgments on the prime. In this experiment retrieval of the prime's meaning always failed because of the masking. In a control experiment with SOA's long enough for retrieval of the prime to succeed, they did find facilitation from the prime to the target. Thus when attention is directed toward the meaning of the prime and the retrieval attempt is successful facilitation occurs but if the retrieval attempt fails inhibition results. These results seem quite puzzling. Dagenbach et al. interpreted them in terms of a hypothetical attentional mechanism specialized to help in retrieving a weakly activated semantic code that might be suffering competition from other codes that are similar or related and as a result, are also partially activated. This attentional mechanism seems to work on a center surround principle in that related meanings to the prime are inhibited in order to try to allow the desired prime to rise to the surface of activation. The mask in this case serves to help make it difficult to retrieve the prime's meaning. In the non— semantic judgment subjects do not attempt to retrieve the meaning of the prime so no semantic inhibition occurs. Dagenbach et al. (1989) suggest that this center surround attentional mechanism may be working in other paradigms where attention is directed toward the desired meaning of a word and retrieval fails. Retrieval of word meanings fails in many different situations. When learning new words the codes are 9 relatively weak compared to older, well established codes in memory. This creates a similar situation to masked priming, but in masked priming activation is weak because of limited physical input and in the learning of new words, activation is weak because of limited learning time. Dagenbach, Carr, and Barnhardt (1990) investigated semantic priming effects from new vocabulary words that were learned by subjects to varying degrees in a lexical decision paradigm. They believed that words that were not learned as well may produce inhibition to a related target in a lexical decision trial, at least if the subject tried to retrieve the word meaning and failed. Subjects studied 45 vocabulary words and their short definitions for 15 3 each and then did lexical decision trials where the studied words served as the primes and subjects were supposed to report the meaning of the prime aloud if they could remember it. In 15 of the 45 lexical decision trials the prime was unrelated to the target, in 15 it was related, and in 15 the prime was followed by a pronounceable nonword. Subjects were told that sometimes the prime would be related to the target and they should use the prime to anticipate the target. Subjects hit one key on the computer if the target was a word and another if it was not. The SOA between the prime and target was either 7 or 17 s in hopes that the longer interval would give the subject enough time to retrieve the prime's meaning. Dagenbach et al. predicted that if subjects could report the meaning of the prime, facilitation 10 to the target would occur. A slight trend toward significant facilitation did occur. However Dagenbach et al. also predicted that if the prime's meaning could not be recalled the center surround attentional mechanism would inhibit codes that were related to the prime and inhibition would occur. No inhibition occurred in this experiment. The experimenters reasoned that perhaps nothing was learned in the first place since the study period was extremely short. If this was the case, then no code existed, not even a weak one, resulting in a lack of significant inhibition. Another reason for a lack of significant inhibition or facilitation could have been the unusually long SOA, allowing facilitation or inhibition to dissipate before the presentation of the target. In Experiment 2 Dagenbach et al. (1990) shortened the SOA to 2 s, which was more in line with SOA's standardly used in other studies of priming, and subjects were instructed to bring the prime's definition to mind instead of recalling it aloud. After the lexical decision trials subjects took a recognition memory test and a recall test. The purpose of the recognition test was to see whether unrecalled primes were in memory at all. Presumably if the prime's definition was recalled it was probably successfully retrieved in the lexical decision trial. However if the prime was recognized on the test but not recalled, it was probably not recalled during the lexical decision task, although correct recognition performance ensures that at 11 least a weak code exists. The prediction would be that the words that were correctly recalled would produce facilitation in the lexical decision task and words that were only recognized would produce inhibition. These weaker codes would need the center surround mechanism to push down words related to the new code in order to help the new word rise to retrievable levels of activation. On 83% of the trials the prime was at least recognized. Dagenbach et al. then broke the 83% into categories according to whether the prime had been recalled or not. Significant facilitation (64 ms) occurred when the prime and target were related and when the prime's definition had been recalled and recognized. Significant inhibition (also 64 ms coincidentally) occurred when the prime and target were related but the prime's definition could not be recalled at the time of the lexical decision——in other words, when the prime's definition was recognized on the test, but the code was not strong enough to be recalled. In Experiment 3 Dagenbach et al. tested the idea that if the strength of the codes was increased in semantic memory, inhibition would begin to turn into facilitation, although they were not sure how much extra study would be necessary. Subjects saw each word twice (10 3 each time) for a total of 20 3 rather than seeing the words once for 15 s as in Experiments 1 and 2. Subjects now recognized 91% of the words rather than 83%. Primes that were recalled and recognized, indicating a successful retrieval attempt during 12 the lexical decision trial, produced 70 ms of facilitation, replicating Experiment 2. Primes that were recognized but not recalled did not produce inhibition like in Experiment 2, but actually produced 32 ms of facilitation, though this effect was not significant. Dagenbach et al. propose that the facilitation in this case could be due to increased code strength but they also point out the fact that they could not control which primes were recalled. Perhaps some words were simply more memorable than others and the ones that were less memorable produced inhibition. The fact that the 32 ms of facilitation was not significant indicates that the data from unrecalled primes were quite noisy. This then suggests the possibility that at slightly higher levels of learning the influence of the inhibitory mechanism is beginning to decrease, resulting in a mix of trials on which its impact is seen and trials on which it was not deployed but recall still fails. In the fourth experiment Dagenbach et al. manipulated the study sessions in order to eliminate some of the variance in the previous experiments. Subjects in the "cycled' (or "long-lag") group saw 48 new vocabulary words once for 10 8 each during the first session. During the second session, 4 days later, subjects saw 48 more vocabulary words two times each. In this second session, one exposure was given to each of the words and the study list was repeated for a second cycle-hence the name "cycled". The cycled group was used with the hopes of I3 replicating experiments 1 and 2 using a within subjects design. Subjects in the "short-lag" group also did 2 sessions but half of the 48 words in each session were seen once for 15 s and half were seen twice. One exposure and two exposures were mixed together to produce an average lag between repetitions of three items. The short—lag group was used to extend the findings of the previous experiments while manipulating the exposures within a single session and shortening the lag between repeated words. Another difference occurred in the lexical decision trials. Subjects were told to bring the prime's meaning to mind, but they were not told that the prime would sometimes be related to the target, and they were not instructed to use the prime to predict the target. Also a delayed lexical decision test was given at the end of the second session, using the words subjects had studied during the first session, 4 days earlier, as primes. In this final experiment inhibition was found when the primes were recognized but not recalled, in other words, when the codes were relatively weak. This inhibition was found in both the cycled and short-lag study groups and was present during the immediate and delayed lexical decision trials. However, the facilitation when primes were recalled disappeared. Dagenbach et al. suggest that the change in instructions may have affected the facilitation. Subjects were not told to use the prime predictively and perhaps this caused a lack of facilitation, although the authors admit that this idea is speculative. 14 This set of experiments is a major breakthrough for the demonstration of the existence of an inhibitory attentional mechanism for the retrieval of semantic codes. Dagenbach et al. suggest that it is difficult to incorporate new information into semantic memory because related concepts and words have greater strength than the new code. It is more likely that related codes will be activated before the new code because of their strength. How then does new information become integrated and useful in semantic memory? The existence of an inhibitory mechanism that serves to push down or squelch the activation of related codes may be the answer. When trying to retrieve a new code in memory, the mechanism inhibits stronger, related codes and helps the new code gain access into working memory. When the code becomes stronger or more well learned it will not require as much of the mechanism's help, and eventually it will not have to be used at all. Evidence for the inhibitory mechanism appeared again in some additional research by Carr and Dagenbach (1990) using the masked prime paradigm and the threshold setting task. This research is based on their previous research in 1989, where subjects experienced inhibitory semantic priming after making semantic similarity judgments about masked primes. The goals of the 1990 experiment were to confirm the 1989 results and to suggest that a center surround mechanism is responsible for the inhibition. Experiment 1 confirmed the previous Dagenbach, Carr, and Wilhemson (1989) experiment 15 where facilitation is seen when subjects make a presence— absence judgment on the prime in a threshold setting task prior to the lexical decision trials. 'Experiment 1 also included a repetition condition where the same word was presented as a prime and then as a target. These trials were mixed in with the other lexical decision trials where targets were related, unrelated, or nonwords. Half of the related trials were repetitions of the prime and half were different words that were related to the prime. This first experiment confirmed that after presence-absence threshold setting tasks, priming is facilitatory when the prime and target are related. The repetition condition also produced about the same amount of facilitation as the semantic priming condition. Experiment 2 used the same materials as the first experiment but semantic similarity judgments were used to set the threshold. Carr and Dagenbach predicted that inhibition would occur for semantically related primes and targets in the lexical decision trials. This is what occurred in the 1989 experiments. Because the primes were masked and subjects were searching for the meaning of the prime, the inhibition mechanism should come into play to try to help access the code. This suppression of related codes would lead to inhibition of related target words. The second experiment also included a repetition condition. There are two predictions one could make about the repetition condition. If a center surround inhibition l6 mechanism exists for weak codes, then if the same weak code serves as the target and prime, repetition priming should be facilitatory. For example, if the mechanism pushes down stronger codes that are related to the prime in order to enhance its accessibility, then if the same code is repeated again as the target, it should be responded to more quickly. The other prediction that might be made about the repetition condition would be that instead of a center surround suppression, a general reduction of the accessibility of all the codes in a given region of semantic memory occurs. If one has trouble accessing the code, then repeating the same code will be inhibitory. The experiment supports the first prediction. Using the same word as the prime and target leads to facilitation for the sought for code, whereas following the same prime with a related target leads to inhibition. This is direct evidence that a center surround attentional mechanism exists to help out codes that are weak due to masking or limited perceptual input. This idea can also be generalized to Dagenbach, Carr, and Earnhardt (1990). Instead of limited perceptual input, the primes were words that were newly learned and low in strength, and needed the help of the center surround mechanism to boost their activational level. Evidence of the inhibition mechanism also occurred in another paradigm where subjects learned artificial categories of shapes called "fleps" and "gleps" (Carr et al., in press) College students practiced putting the l7 unfamiliar shapes named "fleps" and "gleps" into the proper categories after they had spent some time studying the shapes in the two categories. After learning the categories well, the subjects participated in trials where one shape served as a prime and another served as a target. The two shapes were either from the same category, from the opposite category, or on neutral trials, the prime was a cross. Across several sessions of practice in which the prime— target SOA and proportion of related prime-target pairings were manipulated, these well—learned category members produced consistent evidence of facilitatory semantic priming. Targets were classified faster following same— category primes than different category primes. In the most interesting session, Session 8, subjects studied new fleps and gleps but only for a very limited amount of time, in order to ensure that they were weak in memory compared to the old items that had already been practiced. Then a situation was set up similar to Dagenbach et al. (1990) with the learning of new vocabulary words. The old and new fleps and gleps served as primes for old target shapes, a primed classification task. The SOA was 2000 ms and subjects were instructed to try to bring the category of the prime to mind. Carr et al. (in press) predicted that the newly learned fleps and gleps would produce inhibition to related shapes from the same category, especially the weaker ones, similarly to the way recognized but unrecalled words produced inhibition to related targets. In order to l8 classify the newly learned shapes into weaker and stronger ones, they had subjects categorize the new shapes and the ones that took the longest to categorize were considered the weaker ones in memory. They assumed that the ones that were categorized quickly were already fairly well learned and probably did not need the center surround mechanism as much as the ones that were categorized slowly and were not as well learned. Facilitation was found from old primes to related targets. Facilitation was also found from the quickly categorized new primes to related targets, however inhibition occurred when the new primes were not well learned and were followed by a related target. The results of this study suggest that when retrieving a category member is difficult, the center surround mechanism steps in to help the weak items gain accessibility into working memory. This causes related targets to be responded to more slowly because they are temporarily suppressed in order to allow the weak code to rise to the surface. Directions for Further Research We have seen examples of an attentional center surround mechanism in masked priming, artificial category learning, and vocabulary acquisition paradigms. I focused my research on vocabulary acquisition and how the attentional mechanism aids in the process. However, instead of teaching real words I taught subjects an artificial vocabulary in which pseudowords were associated with carefully selected "meanings" consisting of already known words. Although I9 Dagenbach, Carr and Earnhardt (1990) found evidence of inhibition in the early stages of vocabulary learning, their paradigm has much room for error. The goal for my first experiment was two-fold. First I used pseudowords instead of real words in order to control how much contact individuals have with the things that they studied. None of the subjects have ever seen the pseudowords so they all start at the same level of familiarity with the stimuli. Another reason I used pseudowords is to control the degree of association between the definition of the pseudoword and the related target in the lexical decision trials. Dagenbach et al. (1990) wrote short definitions for the infrequent words based on their actual dictionary definitions, and then thought up another word that could be used as the related word in the lexical decision trials. The degree of the association was not controlled, which could affect the amount of priming from the prime to its related target. According to Rueckl and Olds (1993) pseudowords behave similarly to real words in a priming paradigm in which they are associated with meanings via paired associate training. This means that I can gain control over degrees of association and still study the same kinds of priming processes previously studied with real words. My second goal for the first experiment was to add another study condition to the one used by Dagenbach et al. (1990), one in which subjects saw each word five times in addition to conditions in which they saw the words only once 20 and twice. Dagenbach, Horst and Carr (1990) demonstrated that it takes a huge amount of learning (5 weeks) to incorporate the meanings of new words into semantic memory. It would be interesting to see whether the center—surround mechanism is still activated for information that is a little stronger in memory. The studies reviewed thus far suggest that the mechanism is needed when perceptual input is low or the strength of the desired code is low. But does this mechanism also work for items that are stronger? We are not sure when items need the mechanism and when they become strong enough on their own so that they no longer need it. Perhaps the mechanism is used for higher efficiency items as well but it is used earlier or quicker for these items. Incorporating a higher level of learning into the study may supply some answers to these questions. In addition, Dagenbach, Carr, and Earnhardt (1990) speculated that a change in instructions from active to passive use of prime information might increase evidence of inhibition and decrease evidence of facilitation, based on the results of their fourth experiment. However, after a careful review of the stimulus materials used in that experiment some flaws were discovered that might have been responsible for the apparent increase in inhibition. The prime—target pairings were not properly counterbalanced as they had been in the first three experiments. The same set of 45 target words was used in the unrelated condition of all the subjects' lexical decision trials, and these target 21 words were never used as related targets in any other condition as before. Experiment 1 In Experiment 1 I collected lexical decision latencies for the target stimuli of Experiment 4. Perhaps a bias existed in response times to the groups of target words themselves (regardless of the prime) and this may help explain the findings of Dagenbach et al's (1990) Experiment 4. Method Subjects. The subjects were 72 undergraduates recruited from the psychology pool at Michigan State University. They all received course credit for their participation. Materials and Procedures. The letter strings that served as related, unrelated, and nonword targets for the primed lexical decision trials in Dagenbach et al. were presented to the subjects without their primes. Dagenbach et al's. different lexical decision lists were kept together and the primes were simply removed. Subjects were instructed to press one key if the letter string was a real English word and another key it was not a real word. They were told to respond to the target as quickly and accurately as possible. Results and Discussion The average time it took to respond to the words that served as related targets (710 ms) was longer than the time 22 it took to respond to unrelated targets (627 ms). The overall analysis revealed a significant effect of relatedness of the target, 5(1, 71) = 15.4. M§e=16010.1, p<.05. Targets that has appeared in the unrelated condition were recognized as words when presented in isolation 83 ms faster than the average of the targets that had appeared in the related condition. While this may not explain the interaction between prime-target relatedness and recall success or failure (in Experiment 4 there was massive inhibition for unrecalled primes and a small amount of facilitation for recalled primes), it could explain the larger amounts of inhibition and smaller amounts of facilitation in Experiment 4. This difference in base response times to the targets creates an obvious bias for Experiment 4. Because lexical decision latencies for the related targets were longer to begin with, this increased the odds of finding inhibition to related target words and decreased the chances of finding any facilitation. The materials in the present experiment were much more controlled and were fully counterbalanced to avoid this type of inherent bias in the target words. Experiment 2 The second experiment was a replication and extension of Dagenbach, Carr, and Earnhardt's (1990) Experiment 2, substituting pseudowords for real but infrequent words, and adding a two and five time learning condition. Subjects learned 45 pseudowords as best as they could, given the 23 amount of study time allowed. Different subjects saw the words once, twice, or five times. Next they participated in a primed lexical decision task with an SOA of 2000 ms, enough time to use the definition of the prime strategically to predict the target. Subjects were instructed to bring the definition of the prime to mind and use it to anticipate the target because sometimes it was related to the target. After the lexical decision trials subjects took two memory tests. The first was a recall test and the second was a recognition test. Presumably, if the subject could recall the definition of the pseudoword, then the word was incorporated to some extent into memory and was probably recalled during the lexical decision trials. However, if the subject could recognize the definition of the word but not recall it, the trace may be quite weak in memory and thus the word may have not been recalled during the lexical decision trials. The predictions for this study are as follows. Pseudowords whose definitions are recognized but not recalled are weak traces in memory and probably need the help of the center surround attentional mechanism in order for the new word's meaning to be activated enough to be recalled. In order for the mechanism to work, it has to push down stronger codes that are related to the desired code. If the subject is then presented with a related target, inhibition may occur in responding to that target, since related codes were pushed below resting level. Pseudowords whose definitions are recalled on the memory 24 test (and were presumably accessed during the lexical decision trials) are stronger codes in memory and may no longer need the help of the center surround mechanism to become activated, or perhaps they use the mechanism quicker or earlier in the retrieval attempt. Regardless, we may expect facilitation from a prime to a related target when the prime's definition is recalled. The different learning conditions will give us some idea of how long the mechanism continues to be necessary in early learning. It will probably be present in the one and two exposure learning conditions but it may no longer be used when subjects see the words five times each. Method Subjects. The subjects were 90 undergraduates. About 75% of them were recruited from the psychology pool at Michigan State University and received course credit for their participation. The other undergraduates responded to an advertisement in the university newspaper and were paid $4.00 for their participation. These two kind of subjects were mixed randomly through all of the conditions. All of the subjects were native English speakers. Materials and apparatus. The words that the subjects studied in the study phase, that ultimately served as primes in the lexical decision trials, were pronounceable nonwords obtained from several other journal articles (Sereno, 1991; Smith & Oscar Eerman, 25 1990) along with some of my own. One word definitions were chosen for each pseudoword along with a related word that would serve as a related target in the lexical decision trials. These words and their associates were chosen from Word Association Norms by Palermo and Jenkins (1964). Palermo and Jenkins had different age groups provide free associations to various words and they published the norms for many of the common responses. The word that was most frequently associated to the initial word was used most often for the present experiment. However, sometimes associates were related to other words in the list and in that case another high associate was used instead. There were two lists of 45 pseudowords and their definitions and subjects saw one or the other (see Appendix A). The stimuli were presented on a Macintosh SE computer using stimulus presentation and response collection routines created with Psychlab, an experiment construction package developed by Daniel Eub and Teren Gum at McGill University. Procedures. In the first phase of the experiment, subjects studied 45 pseudowords from either List A or List E and their definitions. The number of times subjects saw each word and definition, once, twice, or five times, was manipulated between subjects and counterbalanced across lists. The words stayed up on the computer screen for 10 3 each. Subjects were told that these words were unusual and old English words that had been use many years ago and had since 26 been dropped from the language. They were told to study these things for a later memory test. An equal number of subjects studied the pseudowords from List A and List E. In the second phase of the experiment, the studied pseudowords were used as primes in the lexical decision trials for well known target words. Subjects first did 15 practice trials to get used to the procedure where the prime was always the word "practice". The subjects were told to read the prime silently and then respond as quickly and accurately as possible by pressing one key on the keyboard if the target was a real word, and another key if it was not. The real lexical decision trials followed. Subjects were told that instead of the word "practice", the studied pseudowords would serve as the primes. The subjects were told to try to bring the meaning of the pseudoword to mind when they saw it and to use it to help them respond to the target, since sometimes it would be related to the target. Appropriate examples were then given. The Stimulus Onset Asynchrony (SOA) between the prime and the target was 2000 ms. This amount of time should have given the subjects ample Opportunity to bring the definition of the word to mind if they could recall it. In the 45 lexical decision trials, the target was related to the meaning of the prime 15 times, was unrelated 15 times and was a pronounceable nonword 15 times. These nonwords were all different from the pseudowords learned as 27 new "vocabulary" items. Unrelated prime—target pairings were created by assigning the related targets from the other list of study words to the primes from the first list (see Appendix A). Targets were moved in groups of 15 throughout the design in order to carefully control association strength between the definition of the pseudoword and its related target, along with frequency, length and number of syllables of the target words. Related targets from List A served as the unrelated targets for List E. For example, if targets 1-15 were paired with related primes in List A's lexical decision trials, the same 15 served as targets to unrelated primes in List E's lexical decision trials in order to keep prime—target pairings counterbalanced correctly. The designation of prime-target pairings to the related, unrelated, and nonword conditions was counterbalanced between subjects. Across subjects, each newly learned pseudoword occurred equally often as a related, unrelated, and nonword prime, and each target word occurred equally often as a related or unrelated target. After the lexical decision trials subjects took two memory tests to determine the degree of learning that had taken place. The first test was a recall test. Subjects saw each pseudoword they had studied on an index card and were instructed to write down the appropriate definition. The second test was a multiple choice recognition test in which subjects saw each study word along with four response choices. One of the alternatives was the actual definition 28 of the pseudoword, another was a definition of one of the other studied pseudowords, and the other two were definitions that had not been studied. Results First, the lexical decision trials were categorized according to whether the subject had at least recognized the prime's definition. Only trials on which the meaning of the prime was at least recognized were analyzed. In the condition where subjects studied the words once, 88% of the words' meanings were recognized (compared to 83% in Experiment 2, Dagenbach et al. (1990)). When subjects saw the pseudowords and definitions twice, they recognized 94% of the words' meanings (compared to 91% in Experiment 3, Dagenbach et al.), and when subjects studied each pseudoword 5 times, recognition went up to 99%. Recall percentages of the recognized words' meanings were as follows: 36% of the words' meanings were recalled in the one time study condition, 54% in the two time study condition and 81% in the five time study condition. Dagenbach et al. did not report recall percentages. For each learning condition the words that were recognized were divided into four categories according to whether the prime's meaning had been recalled and the relatedness of the prime and target: recalled and related, recalled and unrelated, unrecalled and related, and unrecalled and unrelated. Primed lexical decision data were then analyzed with the prime categories as factors. Because of the 29 procedure of the experiment, the number of responses in each category could not be predetermined. Subjects with at least 3 data points per cell were included in the analysis. This resulted in the deletion of 12 subjects from the study one time condition and 15 from the study two times condition. When subjects studied the words 5 times each they usually recalled most of the words, leaving very few observations in the unrecalled cells of the design. If the same criterion of at least 3 data points per cell was applied here virtually all of the subjects would be thrown out. As a result I conducted another analysis where only the data from trials with recalled primes were used for each study condition. Subjects who contributed at least 3 data points to each recalled cell in the design were included in this analysis. Mean correct reaction times and error rates for positive lexical decision trials are shown in Table 1. Again, for the five time study condition, all 30 subjects are included in the recalled means. For the unrecalled cells in the five time condition, a criterion of at least 1 data point per cell was set and this includes 17 subjects. Although the means for the unrecalled cells are reported, too few observations are present to make any certain conclusions. A 2x2x2 analysis of variance was done on the 33 subjects left in the one and two time learning condition and a main effect of prime-target relatedness was found, §(1, 29) = 9.59. MSG = 5297.1. p<.05. The analysis also revealed a significant interaction between 30 Table 1 Mean RT (msec) and Percent Error for Positive Lexical Decisions as a Function of Prime Recall Unrecalled Prime Recalled Prime Prime-target relatedness M obs/sub RT PE M obs/sub RT PE Pseudowords studied one time (88% recognized; n=18) 64% were unrecalled 36% were recalled Related 6.2 704 1.8 6 690 .8 Unrelated 6.4 739 1.7 6.7 755 3.2 Net effect +35 -.1 +65 +2.4 Pseudowords studied two times (94% recognized; n=15) 46% were unrecalled 54% were recalled Related 5.7 687 2.3 7.8 620 1.7 Unrelated 6.1 654 2.1 7.5 707 3.4 Net effect —33 -.2 +87 +1.7 Pseudowords studied five times (99% recognized; n=30) 19% were unrecalled 81% were recalled Related 1.8 587 O 12.8 603 .5 Unrelated 2.0 603 1.6 12.5 632 3.8 Net effect +16 +1.6 +29 +3.3 31 Table 1 (cont'd) Note. M obs/sub = Mean observations per subject; RT = response time; PE = percent error. Percent recalled was calculated from the percent recognized, not all the words. 32 whether the subjects successfully recalled the meaning of the pseudowords and prime-target relatedness, F(l,29) = 5.80. MSG = 7615.8. p<.05. The interaction showed 75 ms of facilitatory priming from the related primes when the meaning was recalled, and only 4 ms of facilitation when the prime's meaning was not recalled but was recognized. The 75 ms of priming was significant when tested individually with a post hoc t test (p<.05), but the 4 ms was not significant. Additional analyses were done on each study time and the means are found in Table 1. Post hoc t tests revealed that the facilitation present in each study time for the recalled primes is significant (p<.05), whereas none of the effects for the unrecalled primes are significant. A main effect of list (either A or E) was also present, §(l, 29) = 11.0. MSG = 60495.0. p<.05. However list did not interact with any other variables although the patterns of the means did differ between the lists. I also calculated overall priming effects for all recognized items in each study condition (without the criterion of at least 3 data points per cell per subject). In the one time study condition there was 48 ms of facilitation when primes' meanings were recalled and 24 ms of facilitation when subjects did not recall the primes' meanings. For the two time study condition when primes' meanings were recalled there was 64 ms of facilitation and 11 ms of inhibition when they were not. All recognized items were used in the five time learning condition in the 33 original analysis. Another analysis of variance was done to include the five study time condition, but only the data from the primes whose meaning was recalled were analyzed for each study condition. The analysis revealed a main effect of prime— target relatedness, §(1, 73) = 29.7. MSE = 3240.0, p<.05. The amount of study time approached significance, F(2, 73) = 2.7. MSe = 32239.2, p=.07. The interaction between relatedness and study time was not significant, F(2, 73) = 1.5. MSe=3240.0, p=.23, though the amounts of priming did vary a bit for each study time. Sixty five ms of priming was found in the one time study condition, 59 ms in the two time condition, and 29 ms in the five time. All of these were significant when tested individually with one way F- tests. Discussion The results do not replicate Dagenbach, Carr and Earnhardt (1990). In the one time learning condition, which compares to Experiment 2 in Dagenbach et al., when experiments are classified by amount of exposure to primes during study, facilitation was found when the prime's meaning was recalled, but no evidence of inhibition was found when the prime's meaning was not recalled. It is interesting to note that almost identical amounts of facilitation were found for the primes whose meaning was recalled--65 ms in this study compared to 64 ms in the Dagenbach et al. (1990) paper. However they found 64 ms of 34 inhibition for the primes whose meaning was not recalled, compared to 35 ms of facilitation in the present experiment. In the two time learning condition (which compares to Experiment 3 in Dagenbach et al. when classified by amount of exposure during study) 87 ms of facilitation was found when the prime's meaning was recalled, similarly to Dagenbach et al. where 70 ms of facilitation was found. However, in Dagenbach et al. the inhibition that was found for the primes whose meaning was not recalled had disappeared and now these primes produced 32 ms of facilitation to their related targets in the lexical decision trials (though this facilitation was not significant). In the two time learning condition in the present experiment 33 ms of inhibition was found when the primes' meanings were not successfully retrieved (although this was not significant). Dagenbach et al. hypothesized that increased code strength may eliminate inhibition because in their experiment inhibition was present at the one time study condition, when the primes' meanings were not recalled, and was disappearing after two exposures to the word. This experiment does not lend support to that interpretation. In the present experiment one sees facilitation from the prime to its related target if the prime's definition is successfully retrieved. However, instead of seeing inhibition when the prime's definition is not recalled, we see nothing or a slight trend toward some inhibition although the pattern of inhibition does not 35 follow degree of learning. Here, the 33 ms of inhibition occurs after 2 exposure to the study words. If an attentional center surround mechanism was at work one would expect inhibition for the unrecalled definitions, especially at the lowest level of learning when subjects were struggling to retrieve the meaning of the prime and failed. The small amount of facilitation found in the five time study condition was surprising because presumably these codes should be the strongest in memory and if recalled, should provide a lot of facilitation. Perhaps "degree of learning" is not being captured by number of exposures to the study words, although a good alternative measure of learning is not readily apparent. Two alternate ways to measure learning are recognition rate and recall rate from the post-lexical decision memory tests. Recognition rates from all three study conditions in the present experiment were higher than the 83% observed in Experiment 2 of Dagenbach et al., and approaching or exceeding the 91% recognition rate Dagenbach et al. reported in Experiment 3. Using recognition rate as a criterion the present results of facilitation from recalled primes accompanied by insignificant and noisy effects from unrecalled primes look more similar to the findings of Dagenbach et a1. However recall rates seem like a more intuitive index of degree of learning than recognition rate. Unfortunately, Dagenbach et al. did not report overall recall rates. Although one can calculate them for primes 36 that were paired with unrelated and related targets, they fail to supply enough information to calculate the overall recall rate for all words that were at least recognized. Experiment 3 Experiment 2 did not replicate Dagenbach et al. when degree of learning was indexed by amount of exposure to the primes during study, and no significant inhibition was found in any condition. Experiment 3 tried again to replicate Dagenbach et al., but two changes were made. First, in order to eliminate the list differences indicated by the main effect of list observed in Experiment 2, more lists were constructed using a modified generation procedure. Instead of only two combinations of the 90 pseudowords and their definitions, there were 10 combinations. This reduces the chance of random differential list effects like the ones found in the first experiment. Second, the recall test was presented on the computer and the latencies of the subjects' responses were measured by a voice key apparatus. In Experiment 2 it was assumed that if subjects could recall the definition of the pseudowords at all they could probably do it in less than 2000 ms. However after looking at the small amount of facilitation for the successfully recalled primes and their related targets in the five time study condition, one could hypothesize that perhaps recall was taking longer than the time allowed in the lexical decision trials, and that for some recalled primes, the attentional retrieval mechanism, if it exists, was still in operation at 37 the time the lexical decision target appeared. If so, then priming from "recalled" primes would be a mix of trials in which recall was fast, producing facilitation and trials in which recall was slow, producing inhibition. Applying the logic of Carr et al's. (in press) "fleps" and "gleps" experiment, analyzing data from primes whose meanings were recalled as a function of recall speed gives another place in which to look for evidence of inhibition. Thus, subject's recall times in the free recall test were broken down into two categories. The first was pseudoword definitions recalled under 1800 ms--clearly less than the 2000 ms prime-target SOA during lexical decision-—and the second was pseudoword definitions recalled after 1800 ms. Lexical decision times were then analyzed for these two categories for each study time based on prime—target relatedness. Method Subjects. Subjects were 90 undergraduates from the same population as the first experiment. However, in this experiment all the subjects participated in the study for course credit. Procedures. The procedures were the same as in Experiment 2 except for the timed recall test. Subjects were instructed to look at each word on the computer screen and report the definition into the microphone if they could recall it. If 38 they did not know the meaning of the word or had no guess, they were told to say "pass". The experimenter recorded subjects' responses on a sheet of paper Latencies were collected by a voice key connected to the mouse port. The 10 lists were generated by each list having a "partner" list (A and E, C and D . . .). Forty five of the 90 words were randomly assigned to one of the lists and the rest to the partner list. For the lexical decision trials, each prime had a related target word. The related targets from the partner list served as the unrelated targets in the other list. This ensured that every target was seen equally often as a related and unrelated target. The unrelated and related targets in a particular list were controlled for frequency and each group of related targets in a list were controlled for association strength between the definition of the prime and the related target (see Appendix C). Results Subjects' recognition rates were similar to the second experiment. Recognition was 91% in the one time learning condition, 95% in the two time condition and 97% in the five time condition. Recall percentages were slightly lower than the first experiment. Subjects recalled 38% of the pseudowords' meanings in the one time condition, 46% in the two time condition and 70% in the five time condition (the 70% is based on all subjects--recall percentage for the 11 who had at least 3 data points per cell was 48%) as compared to 36%, 54%, and 81% in Experiment 2. Lexical decision 39 times were analyzed as in Experiment 2. However, the subjects in the five time condition were included in the overall analysis this time because more subjects (11) met the 3 data points per cell criterion. This is a direct result of lower recall in this experiment. Twelve out of 30 subjects in the one time condition, 21 out of 30 in the two time condition, and 11 from the five time condition contributed data to the overall analysis. Results from these 44 subjects appear in Table 2. A 3x2x2 analysis of variance was done on these 44 subjects. The interaction between recall success and prime-target relatedness approached significance, F(1,41) = 3.45. MSE = 8898.1. p = .07. This interaction reflected 45 ms of priming when the pseudowords' definitions were recalled and no effect of prime—target relatedness (mean priming = 0) when the pseudowords' definitions were not recalled. A post hoc t test found the 45 ms of priming to be significant. When the analysis was broken down by study time no significant effects were found. Post hoc t tests showed that the facilitation found for recalled definitions of the primes to related targets was not significant for any of the study times (p>.05). Also, none of the effects for the unrecalled definitions of the primes were significant. I also calculated overall priming effects for all recognized items in each study condition. In the one time study condition when primes' meanings were recalled there was 72 ms of facilitation and 29 ms of facilitation 40 Table 2 Mean RT (msec) and Percent Error for Positive Lexical Decisions as a Function of Prime Recall Unrecalled Prime Recalled Prime Prime-target relatedness M obs/sub RT PE M obs/sub RT PE Pseudowords studied one time (91% recognized; n=12) 62% were unrecalled 38% were recalled Related 7.4 705 3.3 6.3 720 2.6 Unrelated 7.6 710 2.2 5.7 784 1.4 Net effect +5 -1.1 +64 —1.2 Pseudowords studied two times (95% recognized; n=21) 54% were unrecalled 46% were recalled Related 6.4 729 .7 7.7 723 1.2 Unrelated 6.7 740 4.1 7.1 753 2.6 Net effect +11 +3.4 +30 +1.4 Pseudowords studied five times (97% recognized; n=ll) 52% were unrecalled 48% were recalled Related 6.7 736 3.9 7.5 710 1.2 Unrelated 6.8 709 0 7.0 766 4.9 Net effect —27 -3.9 +56 +3.7 41 Table 2 (cont'd) Weighted overall mean 0 +45 Note. M obs/sub = Mean observations per subject; RT - response time; PE 2 percent error. Percent recalled is taken from the percent recognized. 42 when they were not recalled. For the two time learning condition when primes' meanings were recalled there was 18 ms of facilitation and 32 ms of facilitation when they were unrecalled. For the five time condition when the primes' definitions were recalled there was 56 ms of facilitation and 1 ms of facilitation when they were not recalled. In a separate analysis the lexical decision reaction times for successfully recalled definitions were further broken down by recall latency. Lexical decision times were then analyzed for each of these categories based on prime- target relatedness. The results are in Table 3. Average recall time for primes' meanings when the pseudowords were studied once was 2156 ms. When pseudowords were studied twice each average recall time was 1861 ms, and when they were studied five times each average recall time was 1716 ms. An analysis of variance was done on the related and unrelated lexical decision times based on recall times and amount of study time. Nothing reached significance in the overall analysis although if primes' meanings were recalled under 1800 ms there was 49 ms of priming which was significant when tested with a post hoc t test (p<.05). Analyses were also done on each individual study time. The only significant effect was found in the two time study condition where a significant interaction was found between recall time and relatedness, F(1, 29) = 6.9. M§e=20235.8, p<.05. If the primes' meanings were recalled in less than 1800 ms a trend toward facilitation was found, although Table 3 Mean RT (msec) 43 and Percent Error for Positive Lexical Decisions broken down by Recall Latency Prime—target (1800 ms >1800 ms relatedness M obs/sub RT PE M obs/sub RT PE study one time condition Related 2.4 756 O 1.9 778 5.1 Unrelated 2.2 799 3. 1.5 836 2.1 Net effect +43 +3. +58 -3.0 study two times condition Related 4.3 735 3. 2.6 801 O Unrelated 4.7 794 1. 1.7 725 3.8 Net effect +59 -1. -76 +3.8 study five times condition Related 7.9 723 2.6 747 1.3 Unrelated 7.3 767 7. 2.5 800 6.3 Net effect +44 +6. +53 +5.0 Weighted overall mean +49 +12 Note. M obs/sub = mean observations per subject; RT response time; PE = percent error. 44 it was not significant when tested with a post hoc t_test. When primes' meanings were recalled after 1800 ms, 76 ms of inhibition was present which was significant (p<.05). None of the other individual effects shown in Table 3 were significant. Discussion Experiment 3 differed from Experiment 2 in a few ways. First the recall test was timed and subjects had to report the definitions into a microphone instead of writing them on a sheet of paper. The percentage of pseudoword definitions recalled in Experiment 3 was lower than Experiment 2 in the two and five time study conditions. Subjects may have been intimidated by speaking aloud or felt they were under more pressure to recall than in the second experiment. Perhaps we need a better way to chart the progress of new learning. In Experiment 3, when broken down by study time, nothing was significant. The facilitation found when pseudowords' meanings were recalled was not significant which shows how noisy these data are (also there were few subjects per group). However, the basic trend remains. When primes' definitions are recalled a trend toward facilitation exists whereas when primes' definitions are not recalled there is no overall explainable pattern, producing a net effect of 0 averaged across study time conditions. Recall latencies were than broken down into primes whose meanings were recalled before the presentation of the target and after the target. This pattern of results is 45 also difficult to comprehend. About the same amounts of facilitation exist when the primes' meanings were recalled before 1800 ms but very different patterns occur when they were recalled after 1800 ms. One would imagine that if the subjects could not access the prime's definition before the target appears, inhibition would result. Significant inhibition did occur for primes recalled after 1800 ms in the two time study condition. It is puzzling that inhibition only occurs here whereas trends toward facilitation occur for the other study conditions. General Discussion Generally it seems that when primes' meanings are recalled facilitation exists and when primes' meanings are unrecalled neither facilitation nor inhibition is present in any systematic pattern. Instead of providing evidence for a center surround attentional mechanism, this set of experiments basically shows that episodic priming exists in new learning as long as the meaning of the newly learned primes can be successfully recalled—-however the amount does not vary in a consistent pattern across study conditions. Although trends toward inhibition do not follow the amount of learning in an understandable way, they are present and do suggest that something may be going on. It is intearesting to note that the qualitative pattern in these data when measured by recognition rates does match the Pattern in Experiment 3 in Dagenbach et al. (1990) pretty well. This may be a replication or a coincidence but it 46 suggests that "degree of learning" is something that has to be more carefully measured. Subjects in each study time vary greatly according to how well they recognize and recall the words they had just seen. A key to understanding the pattern may be linked to more accurately charting new learning and the strength of new codes. As was previously mentioned, the present experiment was much more rigorously controlled than Dagenbach et al. (1990). Frequencies of the target words were controlled for related and unrelated targets as well as the length of the words and number of syllables. Dagenbach et al. did not control for any of these factors which may account for the differences in the results. Another factor that was controlled in the present experiment was the degree of association between the definitions of the pseudowords and the related targets. This should directly affect the amount of priming between the prime and target. Dagenbach et al. intuitively generated related targets for definitions of the study words. Perhaps this lack of control also contributed to the differences between these studies. The present study also used pseudowords instead of real, but infrequent words, although the subjects were probably equally unfamiliar with both. The idea of a center surround retrieval mechanism for new knowledge seems very logical and even quite possible if one looks at the results of Dagenbach et al. (1990), especially interpreted in the context of the masked priming 47 results of Dagenbach et al. (1989); Carr and Dagenbach (1990), and the category learning results of Carr et al. (in press). However, the present results do not lend much support to that interpretation. Perhaps the mechanisms that control masked priming, category learning, and vocabulary learning are very different. Although an inhibition mechanism seems to exist in both masked priming and category learning, it does not seem be working in vocabulary learning. Perhaps there are three separate mechanisms controlling each type of learning and a general "inhibition mechanism" can not be applied across all types of learning. Small amounts of inhibition do occur in some of the unrecalled cells in the present experiments, but the overall patterns of facilitation and inhibition needed to claim that a center surround mechanism exists in vocabulary acquisition are lacking. APPENDICES APPENDIX A 48 APPENDIX A Stimulus Materials for Experiment 2 List A Targets in lexical decision Pseudowords Definitions Related Unrelated Nonwords 1. trolix sweet candy law affir 2. rell lion tiger job wend 3. hevil deep dark hair roven 4. swerp take give low serd 5. dister anger mad carry kellon 6. wenny thinner fatter lamp frit 7. plaret cabbage lettuce stream bottar 8. hucus tobacco smoke round trask 9. trucent music song sour greel 10. asrow whiskey drunk then vade 11. prive bath clean older lant 12. lanodyne black white bad ruckly 13. hestim wish want cats fand 14. drime spider web steal crist 15. jeek boy girl dirt gope 16. rital sell buy fear peval 17. albing city town nearer vig 18. croxy swift fast house soleg l9. marlet citizen American flowers predab 20. skires guns shoot letters zade 21. shokets doors windows cry rive 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. parbin doil gultan olmster wape thave bour pomare jadir vupial sagad bink mivers mitney gruce danal hoat fandler femon hewer shantling woral dute chope eagle joy people clearer carpet street appear king hungry moon ocean find shoes sheep soldier always stand broader butter farther command table live short 49 bird happy crowd foggy rug road see queen food stars sea look feet wool army forever sit wider bread away order chair die tall walking sleep hand pepper train UP games plant nail hill cut ache water talk pretty mouse harder noise moth church under peace hot mind deize toinpy geets seary sude colast bimp benth chube fubbler namp rimsape katile vorst dort leb glub favig pim bram riggy clebe sarp bleam List E Pseudowords l. roaken 2. plew 3. santler 4. donter 5. calark 6. puxil 7. andle 8. tramet 9. pight 10. nacuna 11. spraw 12. fergin l3. drocking 14. sance 15. teast l6. yerule l7. shain 18. lunter l9. bummler 20. soabits 21. clasking 22. summy 23. adode Definitions salt beautiful younger now thief closer river square trouble light high head working justice lift cottage afraid bitter blossom numbers running baby dream 50 Targets in lexical decision Related pepper pretty older then steal nearer stream round bad lamp low hair job law carry house fear sour flowers letters walking cry sleep Unrelated Nonwords tiger dark candy give fatter mad lettuce smoke song drunk clean want web girl white fast buy town American shoot windows crowd bird sharm gorl nell lacat mese gresh legate poudry floom hend lunk wode sludor tasil trake flopate murp bince blash wazo glur sket ponzo 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. bettals sharms alyss corple glanking elent dreed vedom amony sardel onium gipe orenic awner snush raptic lonsure heese jible laper stime lepon fingers dogs whistle jump playing stem hammer mountain thirsty scissors stomach speak earth easier cheese loud butterfly priest over quiet stove memory fl hand cats train UP games plant nail hill water cut ache talk dirt harder mouse noise moth church under peace hot mind happy foggy feet rug see die queen look stars sea wool road army forever sit wider order away bread chair tall food tofe masset thag abran crug gleed lerst ched bleam tudus cheen gar rild rantag dolk latting beeth morod shern pash lany tomp 52 Strength of association between the definition of the pseudoword and its related target word in the lexical decision trials List A List B #1-15 association=29l #1-15 association=256 #16-30 association=308 #16-30 association=29l #31-45 association=285 #31-45 association=250 overall=295 overall=267 Average frequency of related target words List A List B #1-15 frequency=206 #1-15 frequency=206 #16-30 frequency=204 #16—30 frequency=208 #31-45 frequency=208 #31-45 frequency=206 overall=206 overall=207 Length of words and number of syllables List A List B length syllables length syllables Pseudowords #1-15 5.5 1.7 5.7 1.7 #16-30 5.5 1.7 6.1 1.8 #31-45 5.6 1.9 5.3 1.9 overall 5.5 1.8 5.7 1.8 Definitions #1-15 5.1 #16-30 5.2 #31-45 5.5 overall 5.3 Related Targets #1-15 4.6 #16-30 4.6 #31-45 4.5 overall 4.6 Nonword Targets #1-15 4.7 #16-30 4.7 #31-45 4.7 overall 4.7 APPENDIX B 54 APPENDIX B Stimulus Materials for Experiment 3 List A Targets in lexical decision Pseudowords Definitions Related Unrelated Nonword 1. laper quiet peace town pash 2. swerp take give sleep serd 3. gruce soldier army water dort 4. alyss whistle train hand thag 5. hevil deep dark walking roven 6. hewer farther away die bram 7. rell lion tiger forever wend 8. puxil closer nearer plant gresh 9. heese priest church hair morod 10. mivers shoes feet bread katile 11. wenny thinner fatter talk frit 12. drime spider web stream crist 13. croxy swift fast games soleg 14. lanodyne black white want ’ruckly 15. woral table chair mind clebe 16. sance justice law sit tasil l7. asrow whiskey drunk under vade l8. plew beautiful pretty food gorl 19. orenic earth dirt wool rild 20. trucent music song American greel 21. drocking working job lamp sludor 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. bour teast calark tramet shain chope hucus parbin jeek bink gultan lunter lonsure sagad corple fandler vedom doil dister donter dreed santler awner bummler appear lift thief square afraid short tobacco eagle boy find people bitter butterfly ocean jump broader mountain joy anger now hammer younger easier blossom 55 see carry steal round fear tall smoke bird girl look crowd sour moth sea UP wider hill happy mad then nail older harder flowers ache bad lettuce foggy queen clean noise buy pepper cut cats order shoot windows mouse letters hot road candy low stars cry rug house bimp trake mese poudry murp leam trask deize gope rimsape geets bince beeth namp abran favig ched toinpy kellon lacat lerst nell rantag blash List E Pseudowords 1. olmster 2. thave 3. danal 4. glanking 5. trolix 6. andle 7. onium 8. albing 9. fergin 10. stime 11. gipe 12. jible 13. rital l4. sardel 15. pomare l6. marlet l7. hestim 18. pight 19. lepon 20. jadir 21. soabits 22. vupial 23. clasking 24. dute Definitions clearer street always playing sweet river stomach city head stove speak over sell scissors king citizen wish trouble memory hungry numbers moon running live 56 Targets in lexical decision Related foggy road forever games candy stream ache town hair hot talk under buy cut queen American want bad mind food letters stars walking die Unrelated train look drunk flowers tall harder steal then UP fast nail mad web job see wider army away church peace dark happy crowd moth Nonwords seary colast leb crug affir legate cheen vig wode lany gar shern peval tudus benth predab fand floom tomp chube wazo fubbler glur sarp 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. wape shokets amony skires sharms adode nacuna plaret summy snush mitney elent shantling raptic yerule bettals hoat prive roaken spraw femon carpet doors thirsty guns dogs dream light cabbage baby cheese sheep stem command loud cottage fingers stand bath salt high butter 57 rug windows water shoot cats sleep lamp lettuce cry mouse wool plant order noise house hand sit clean PEPPer low bread smoke white sour girl nearer feet dirt bird sea pretty law order round give tiger fear hill carry song fatter chair sude rive bleam zade masset ponzo hend bottar sket dolk vorstq gleed riggy latting flopate tofe glub lant starm lunk pim List C Pseudowords 1. marlet 2. lunter 3. gruce 4. glanking 5. trolix 6. drocking 7. vedom 8. fandler 9. swerp 10. wape 11. donter 12. dreed 13. rital l4. woral 15. awner 16. nacuna 17. gipe 18. hestim l9. thave 20. mitney 21. trucent 22. raptic 23. dute 24. yerule Definitions citizen bitter soldier playing sweet working mountain broader take carpet now hammer sell table easier light speak wish street sheep music loud live cottage 58 Targets in lexical decision Related American sour army games candy job hill wider give rug then nail buy chair harder lamp talk want road wool song noise die house Unrelated stars pretty nearer pepper sea older cut order cats shoot windows letters hot cry round happy church crowd sit dark food girl moth white Nonwords predab bince dort crug affir sludor ched favig serd sude lacat lerst peval clebe rantag hend gar fand colast vorst greel latting sarp flopate 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. prive jible amony spraw bummler rell sance plaret lepon snush onium hewer albing bour shain wenny drime croxy hucus parbin pomare bath over thirsty high blossom lion justice cabbage memory cheese stomach farther city appear afraid thinner spider swift tobacco eagle king 59 clean under water low flowers tiger law lettuce mind mouse ache away town see fear fatter web fast smoke bird queen peace feet foggy dirt bad sleep train stream drunk hand look steal tall walking UP forever mad carry bread plant hair lant shern bleam lunk blash wend tasil bottar tomp dolk cheen bram vig bimp murp frit crist soleg trask deize benth List D Pseudowords l. 2. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. laper gultan summy orenic soabits corple puxil teast stime pight mivers chope adode femon jeek olmster danal alyss lonsure sagad fergin heese shantling bettals Definitions quiet people baby earth numbers jump closer lift stove trouble shoes short dream butter boy clearer always whistle butterfly ocean head priest command fingers 6O Targets in lexical decision Related peace crowd cry dirt letters UP nearer carry hot bad feet tall sleep bread girl foggy forever train moth sea hair church order hand Unrelated hill rug fatter chair house queen clean noise tiger candy want bird mouse American buy army town low sour wool give lettuce smoke road Nonwords pash geets sket rild wazo abran gresh trake lany floom katile leam ponzo pim gope seary leb thag beeth namp wode morod riggy tofe 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. shokets andle hoat roaken santler sharms bink asrow plew calark hevil skires elent vupial clasking doil tramet dister jadir lanodyne sardel doors river stand salt younger dogs find whiskey beautiful thief deep guns stem moon running joy square anger hungry black scissors 61 windows stream sit pepper older cats look drunk pretty steal dark shoot plant stars walking happy round mad food white cut fear law see song harder die talk wider away lamp job water games ache then nail under fast web flowers mind rive legate glub starm nell masset rimsape vade gorl mese roven zade gleed fubbler glur toinpy poudry kellon chube ruckly tudus List E Pseudowords 1. nacuna 2. soabits 3 orenic 4. summy 5. onium 6. drocking 7. bour 8. albing 9. puxil 10. calark 11. tramet 12. dister 13. woral l4. rital 15. pomare l6. sance l7. lonsure l8. lepon l9. lunter 20. elent 21. andle 22. vupial 23. dute 24. yerule Definitions light numbers earth baby stomach working appear city closer thief square anger table sell king justice butterfly memory bitter stem river moon live cottage 62 Targets in lexical decision Related lamp letters dirt cry ache job see town nearer steal round mad chair buy queen law moth mind sour plant stream stars die house Unrelated water drunk harder look forever flowers walking then nail bread hair talk army crowd church want dark away sit smoke under American wool feet Nonwords hend wazo rild sket cheen sludor bimp vig gresh mese poudry kellon clebe peval benth tasil beeth tomp bince gleed legate fubbler sarp flopate 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. shokets jadir adode jible lanodyne roaken laper snush alyss pight corple glanking rell fandler doil bettals prive drime croxy spraw jeek doors hungry dream over black salt quiet cheese whistle trouble jump playing lion broader joy fingers bath spider swift high boy 63 windows tall sleep under white PEPPer peace mouse train bad UP games tiger wider happy hand clean web fast low girl foggy lettuce pretty flowers noise sea older cut fear order cats shoot hot give carry road candy song hill rug fatter rive leam ponzo shern ruckly starm pash dolk thag floom abran crug wend favig toinpy tofe lant crist soleg lunk 90Pe List F Pseudowords 1. bink 2. plaret 3. gruce 4. thave 5. trucent 6. skires 7. clasking 8. vedom 9. shain 10. gipe 11. mivers 12. hoat 13. amony l4. santler 15. sharms l6. hewer 17. asrow 18. gultan 19. danal 20. sagad 21. mitney 22. raptic 23. heese 24. stime Definitions find cabbage soldier street music guns running mountain afraid speak shoes stand thirsty younger dogs farther whiskey people always ocean sheep loud priest stove Targets in lexical decision Related look lettuce army road song shoot walking hill fear talk feet sit water older cats away drunk crowd forever sea wool noise church hot Unrelated clean die windows mind town moth web games peace stream bad wider girl tiger fast ache mouse hand cry pepper low buy round letters Nonwords rimsape bottar dort colast greel zade glur ched murp gar katile glub bleam nell masset bram vade geets leb namp vorst latting morod lany 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. shantling donter olmster parbin femon sardel awner swerp hestim plew hevil trolix marlet fergin teast wape wenny jible dreed hucus bummler command now clearer eagle butter scissors easier take wish beautiful deep sweet citizen head lift carpet thinner over hammer tobacco blossom 65 order then foggy bird bread cut harder give want pretty dark candy American hair carry rug fatter under nail smoke flowers steal stars chair house UP haPPY train food see nearer job law sleep dirt white tall queen plant sour lamp mad riggy lacat seary deize pim tudus rantag serd fand gorl roven affir predab wode trake sude frit shern lerst trask blash List G Pseudowords l. sance 2. hestim 3. soabits 4. plew 5. trolix 6. hewer 7. elent 8. fergin 9. shantling 10. tramet ll. hoat 12. awner 13. jadir l4. rital 15. femon 16. bink 17. gultan 18. danal 19. alyss 20. sagad 21. glanking 22. stime 23. heese 24. yerule Definitions justice wish numbers beautiful sweet farther stem head command square stand easier hungry sell butter find people always whistle ocean playing stove priest cottage 66 Targets in lexical decision Related law want letters pretty candy away plant hair order round sit harder food buy bread look crowd forever train sea games hot church house Unrelated Nonwords water hand tall walking UP web talk white under then peace girl job smoke army stream see fast moth sour drunk American dark happy tasil fand wazo gorl affir bram gleed wode riggy poudry glub rantag chube peval pim rimsape geets leb thag namp crug lany morod flopate 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. dister shokets albing adode woral sharms olmster plaret lepon swerp trucent dreed vupial teast fandler calark wape mivers roaken santler bummler anger doors city dream table dogs clearer cabbage memory take music hammer moon lift broader thief carpet shoes salt younger blossom 67 mad windows town sleep chair cats foggy lettuce mind give song nail stars carry wider steal rug feet pepper older flowers lamp die wool ache dirt bad queen clean nearer noise tiger bird cut fear shoot mouse low cry road hill fatter kellon rive vig ponzo clebe masset seary bottar tomp serd greel lerst fubbler trake favig mese sude katile starm nell blash List H Pseudowords 1. laper 2. thave 3. lunter 4. summy 5. mitney 6. andle 7. dute 8. gipe 9. vedom 10. lonsure 11. bettals 12. jible 13. hucus l4. sardel 15. jeek 16. marlet l7. gruce l8. hevil 19. shain 20. drocking 21. orenic 22. onium 23. rell 24. raptic Definitions quiet street bitter baby sheep river live speak mountain butterfly fingers over tobacco scissors boy citizen soldier deep afraid working earth stomach lion loud 68 Targets in lexical decision Related peace road sour cry wool stream die talk hill moth hand under smoke cut girl American army dark fear job dirt ache tiger noise Unrelated mad forever harder nail away town hair wider bread plant want flowers mind sit sleep food foggy feet look church steal train games lettuce Nonwords pash colast bince sket vorst legate sarp gar ched beeth tofe shern trask tudus 90Pe predab dort roven murp sludor rild cheen wend latting 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. puxil donter chope amony spraw parbin nacuna asrow corple snush pight skires bour clasking doil wenny prive drime croxy lanodyne pomare closer now short thirsty high eagle light whiskey jump cheese trouble guns appear running joy thinner bath spider swift black king 69 nearer then tall water low bird lamp drunk UP mouse bad shoot see walking happy fatter clean web fast white queen crowd pretty buy pepper song stars candy round older law carry rug chair sea hot letters house give order windows cats gresh lacat leam bleam lunk deize hend vade abran dolk floom zade bimp glur toinpy frit lant crist soleg ruckly benth List I Pseudowords 1. marlet 2. asrow 3. thave 4. summy 5. hevil 6. mitney 7. bour 8. clasking 9. stime 10. pomare 11. jible 12. mivers l3. chope 14. hucus 15. femon l6. bink 17. plaret 18. danal 19. dute 20. trolix 21. andle 22. vupial 23. vedom 24. shantling Definitions citizen whiskey street baby deep sheep appear running stove king over shoes short tobacco butter find cabbage always live sweet river moon mountain command 70 Targets in lexical decison Related American drunk road cry dark wool see walking hot queen under feet tall smoke bread look lettuce forever die candy stream stars hill order Unrelated sleep town flowers train harder steal fast mad then web plant army want hair peace church sour job wider moth happy nail away white Nonwords predab vade colast sket roven vorst bimp glur lany benth shern katile leam trask pim rimsape bottar leb sarp affir legate fubbler ched riggy 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. bettals hoat prive jadir santler sardel nacuna gultan plew lepon sagad corple rell glanking gipe yerule amony parbin jeek sharms wape fingers stand bath hungry younger scissors light people beautiful memory ocean jump lion playing speak cottage thirsty eagle boy dogs carpet 71 hand sit clean food older cut lamp crowd pretty mind sea UP tiger games talk house water bird girl cats rug foggy ache nearer noise buy pepper law windows mouse fear shoot letters low bad give carry round chair dirt song fatter tofe glub lant chube nell tudus hend geets gorl tomp namp abran wend crug gar flopate bleam deize gope masset sude List J Pseudowords 1. gruce 2. swerp 3. lunter 4. teast 5. trucent 6. drocking 7. onium 8. fandler 9. shain 10. raptic 11. doil 12. wenny 13. croxy 14. adode 15. bummler 16. olmster 17. pight 18. alyss l9. lonsure 20. hewer 21. elent 22. orenic 23. calark 24. puxil Definitions soldier take bitter lift music working stomach broader afraid loud joy thinner swift dream blossom clearer trouble whistle butterfly farther stem earth thief closer 72 Targets in lexical decision Related army give sour carry song job ache wider fear noise happy fatter fast sleep flowers foggy bad train moth away plant dirt steal nearer Unrelated cut cats order hot smoke queen under mind lamp feet wool cry pretty road tiger food older American see candy hill girl bird clean Nonwords dort serd bince trake greel sludor cheen favig murp latting toinpy frit soleg ponzo blash seary floom thag beeth bram gleed rild mese gresh 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. drime donter roaken dreed spraw lanodyne laper hestim sance snush skires soabits albing fergin heese tramet dister shokets woral awner rital spider now salt hammer high black quiet wish justice cheese guns numbers city head priest square anger doors table easier sell 73 web then pepper nail low white peace want law mouse shoot letters town hair church round mad windows chair harder buy stars lettuce house rug water drunk hand look sea tall walking UP talk games dark stream crowd sit die bread forever crist lacat starm lerst lunk ruckly pash fand tasil dolk zade wazo vig wode morod poudry kellon rive clebe rantag peval 74 APPENDIX C Additional Information for Experiments 2 and 3 (Pseudowords and the associative strength between their definitions and related targets in the lexical decision trials and the frequency of the targets.) Pseudowords l. adode 2. albing 3. alyss 4. amony 5. andle 6. asrow 7. awner 8. bettals 9. bink 10. bour ll. bummler 12. calark l3. chope l4. clasking 15. corple l6. croxy l7. danal 18. dister l9. doil Definitions dream city whistle thirsty river whiskey easier fingers find appear blossom thief short running jump swift always anger joy Targets sleep town train water stream drunk harder hand look see flowers steal tall walking UP fast forever mad happy Association 480 232 106 432 154 110 419 341 83 218 630 264 411 218 159 450 135 352 260 Frequency 97 281 86 486 61 26 14 717 910 1513 54 39 55 54 712 45 39 38 97 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. dreed donter drime drocking dute elent fandler femon fergin gipe glanking gruce gultan heese hestim hevil hewer hoat hucus jadir jeek jible lanodyne lepon lonsure lunter laper hammer now spider working live stem broader butter head speak playing soldier people priest wish deep farther stand tobacco hungry boy over black memory butterfly bitter quiet 75 nail then web job die plant wider bread hair talk games army crowd church want dark away sit smoke food girl under white mind moth sour peace 450 375 378 200 232 200 232 575 208 252 100 149 141 225 120 170 190 383 475 413 700 378 580 117 125 150 53 20 1348 302 183 182 17 41 160 275 52 152 63 451 631 160 458 314 33 198 374 685 334 350 198 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. marlet mitney mivers nacuna olmster onium orenic parbin pight plaret plew pomare puxil prive raptic rell rital roaken sagad sance santler sardel shain shantling sharms shokets skires citizen sheep shoes light clearer stomach earth eagle trouble cabbage beautiful king closer bath loud lion sell salt ocean justice younger scissors afraid command dogs doors guns 76 America wool feet lamp foggy ache dirt bird bad lettuce pretty queen nearer clean noise tiger buy pepper sea law older cut fear order cats windows shoot 71 187 358 82 58 159 143 576 107 140 150 651 247 380 231 206 560 408 155 45 525 675 280 255 680 358 200 128 10 283 24 11 43 83 134 41 51 58 43 162 13 124 387 93 245 280 416 17 53 117 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. skires swerp snush soabits spraw stime summy teast thave tramet trolix trucent vedom vupial wape wenny woral yerule guns take cheese numbers high stove baby lift street square sweet music mountain moon carpet thinner table cottage 77 shoot give mouse letters low hot cry carry road round candy song hill stars rug fatter chair house 200 116 106 190 518 226 120 245 118 315 159 164 213 230 310 321 700 265 117 1264 20 113 147 130 64 304 262 32 18 129 119 26 17 89 662 REFERENCES References Ashcraft, M. H. (1989). Human memory and cognition. New York: Harper Collins. Carr, T. H. & Dagenbach, D. (1990). Semantic priming and repetition priming from masked words: Evidence for a center-surround attentional mechanism in perceptual recognition. Journal of Experimental Psychology: Learning, Memory and Cognition, 16, 341—350. Carr, T. H., Dagenbach, D., Van Wieren, D., Carlson— Radvansky, L. A., Alejano, A. R., & Brown, J. S. (in press). Acquiring general knowledge from specific episodes of experience. In C. Umilta' & nonconscious information processing. Dagenbach, D., Carr, T. H., & Earnhardt, T. M. (1990). Inhibitory semantic priming of lexical decisions due to failure to retrieve weakly activated codes. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16, 328-340. Dagenbach, D., Carr, T. H., & Wilhelmsen, A. (1989). Task—induced strategies and near-threshold priming: Conscious effects on unconscious perception. Journal of Memory and Language, 28, 412-443. 78 79 Dagenbach, D., Horst, S., & Carr, T. H. (1991). Priming studies of learning in semantic memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16, 581-591. Durgonoglu, A., & Neely, J. H. (1987). On obtaining episodic priming in a lexical decision task following paired-associate learning. Journal of Experimental Psychology: Learning, Memory, and Cognition, 13, 206- 222. McKoon, G., & Ratcliff, R. (1986). Automatic activation of episodic and semantic memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 12, 108-115. Meyer, D. E., & Schaneveldt, R. W. (1971). Facilitation in recognizing pairs of words: Evidence of a dependence between retrieval operations. Journal of Experimental Psychology, 90, 227-234. Neely, J. H. (1977). Semantic priming and retrieval from lexical memory: Role of inhibitionless spreading activation and limited-capacity attention. Journal of Experimental Psychology: General, 106, 226—254. Neely, J. H. (1991). Semantic priming effects in visual word recognition: A selective review of current findings and theories. In D. Eesner & G. Humphreys (Eds.), Basic processes in reading: Visual word recognition. Hillsdale, N. J.: Erlbaum. 80 Palermo, D. S., & Jenkins, J. J. (1964). Word association norms: Grade school through college. Minneapolis, MN: Lund Press Inc. Posner, M. I., & Snyder, C. R. R. (1975). Attention and cognitive control. In R. L. Solso (Ed.), Information processing and cognition: The Loyola symposium. Hillsdale, N. J.: Lawrence Erlbaum. Rueckl, J. G. & Olds, E. M. (1993). When pseudowords acquire meaning: Effect of semantic associations on pseudoword repetition priming. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16 515-527. Sereno, J. A. (1991). Graphemic, associative, and syntactic priming effect at a brief stimulus onset asynchrony in lexical decision and naming. Journal of Experimental Psychology: Learning, Memory, and Cognition, 11, 459-477. Smith, M. E., & Oscar-Eerman, M. (1990). Repetition priming of words and pseudowords in divided attention and in amnesia. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16, 1033-1042. Tulving, E. (1972). Episodic and semantic memory. In E. Tulving & W. DonaldsOn (Eds.), Organization of memory (pp 381-403). New York: Academic Press. 81 Tulving, E., Hayman, C. A. G., & MacDonald, C. A. (1991). Long-lasting perceptual priming and semantic learning in amnesia: A case experiment. Journal of Experimental Psychology: Learning, Memory, and Cognition, 11, 595-617. "‘tillnil