- Email: [email protected]
Electrophysiological correlates of direct versus indirect semantic priming in normal volunteers
Cognitive Brain Research 8 Ž1999. 289–298 www.elsevier.comrlocaterbres
Research report
Electrophysiological correlates of direct versus indirect semantic priming in normal volunteers Matthias Weisbrod a
a,)
, Markus Kiefer b, Sabine Winkler a , Sabine Maier a , Holger Hill a , Daniela Roesch-Ely a , Manfred Spitzer b
Psychiatric Department, UniÕersity of Heidelberg, Voss-Str. 4, 69115, Heidelberg, Germany b Psychiatric Department, UniÕersity of Ulm, Ulm, Germany Accepted 22 June 1999
Abstract The N400 component of event related potentials ŽERP. was studied in 27 right-handed healthy subjects in a speeded lexical decision task. The semantic distance between primes Žalways words. and targets Ž50% words, 50% non-words. was systematically varied. Prime–target relations included directly related words Žhen–egg., indirectly related words Žlemon–sweet., and non-related words Žsofa–wing.. ERPs were recorded from 20 scalp electrodes positioned according to the international 10–20 system. The N400 reflected semantic distance with the most negative and latest N400 peak in the non-related condition, the least negative and earliest N400 peak in the direct condition and the peak in the indirect condition in-between. Hence, N400 priming effect in the indirect condition was obtained in the absence of strong sentential constraints and even though the mediating word was physically not present. q 1999 Elsevier Science B.V. All rights reserved. Keywords: Semantic priming; Lexical access; Event related potential; N400
1. Introduction Kutas and Hillyard w22x were the first to use event related potentials ŽERP. to study the physiological correlates of semantic association with a non-invasive, inexpensive, and widely available method. Their groundbreaking work extended psycholinguistic research into psychophysiology. Since then, numerous studies have demonstrated a negative ERP component peaking at about 400 ms after stimulus onset which can be reliably evoked in experiments contrasting semantically congruent with semantically incongruent sentence completions Žfor review see Ref. w24x.. This so-called N400 component has proved responsive to the manipulation of linguistic variables and has been found to be more sensitive to semantic association than to the truth-value of sentences w12,18x. It is also elicited by any stimulus with a semantic representation
) Corresponding author. Fax: q49-6221-56-5477; [email protected]
E-mail:
0926-6410r99r$19.00 q 1999 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 6 4 1 0 Ž 9 9 . 0 0 0 3 2 - 4
w23,32x and is sensitive to semantics, even when presented in different modalities w16,25,34x. In contrast, N400 is not affected by grammatical manipulations w21,44x, and is not evoked by non-linguistic symbolic mismatch w5x. With regard to sentences, N400 amplitude was shown to be modulated by the ‘‘cloze probability’’ Žthe degree of expectedness. of sentence-final words: i.e., the semantically acceptable completion in the sentence ‘‘the bill was due at the end of the hour’’ elicits a larger N400 than the predictable final word ‘‘the bill was due at the end of the month’’. In addition, associative constraints were shown to modulate N400 amplitude. Kutas and Hillyard w19x compared N400 evoked by three types of sentence completions: The first type was the most expected Žhigh cloze probability. final word Ži.e., ‘‘the pizza was to hot to eat’’., the second type was not expected but related to the most expected completion Ži.e., ‘‘the pizza was to hot to drink’’., and the third type was semantically anomalous and non-related to the most expected completion Ži.e., ‘‘the pizza was to hot to cry’’.. Anomalous endings elicited a large N400 whereas expected words elicited a small N400. The N400 to sentence completions which were anomalous but related to the most expected final word was smaller
290
M. Weisbrod et al.r CognitiÕe Brain Research 8 (1999) 289–298
than the N400 to anomalous words which were not related to the most likely completion. These results were interpreted as demonstrating that N400 is modulated Ži. by word expectancy engendered by the context, and Žii. by associations between single words of which one word is only suggested by the context and which does not need to be physically present. The majority of N400 studies have focused on words presented in sentence contexts. Interpretation of results may be difficult because of confounding influences from syntax, context, and the heavy load on semantic working memory required. This may cause problems when neuropsychiatric patients who often suffer from dysfunctional working memory are studied. The presentation of single words in a lexical decision task avoids these problems. In a lexical decision task, the subject has to decide whether a given string of characters is a word or not. If a semantically related word precedes the target word, the target word will be recognized faster, that is, semantic priming will occur w27,28x. In so far as ERPs have been recorded during semantic priming paradigms w2,3,13,15,20,33,38x, late negativities similar to sentential N400s have been observed. Words which are not repeated and semantically non-related to previous words elicit the largest N400 w3,12,31x. Kutas and Hillyard w20x showed that the degree of semantic relatedness influences N400 amplitude elicited by words presented in close succession. They instructed subjects to decide whether a target letter was present either in the first or the second member of a word pair. Target words produced the largest N400 when the word pairs were non-related, intermediate when the semantic relationship was relatively weak, and smallest when words were strongly related. Only one study examined the N400 elicited by indirectly related target words. Indirectly related words can be characterized by that the association between prime and target is only reached by an intermediate link, i.e., the prime is associated to a strong associate of the target like lemon– Žsour. –sweet. Hence, there is no way to generate from the prime word an immediate expectation which includes the target: our group w17x investigated scalp potentials evoked in healthy controls using a lexical decision task, including directly related, indirectly related and nonrelated prime–target pairs. The results in regard to the indirect condition were not conclusive. N400 for indirectly related target was larger than for directly-related words and somewhat smaller than for non-related words. However, the difference between the non-related and the indirectly-related condition did not reach significance, perhaps due to the relatively small number of subjects. Therefore, it is not safe to conclude that mediated associations do not elicit N400 effects. The present study was set up to investigate the N400 elicited by targets indirectly related to prime words on a larger group of subjects. To achieve this, ERPs were recorded while healthy subjects performed a lexical decision task which included indirect associates.
2. Materials and methods 2.1. Subjects Twenty seven right-handed subjects Ž16 male, 11 female; age: 33.2 " 11.5 years; school education: 11.4 " 1.8 years. participated in the experiment. 18 subjects participated as controls in a study which compared semantic processing between schizophrenic patients and healthy subjects w42x, the other subjects were controls in a still unpublished twin study. Hand dominance was assessed with a translated German version of the Edinburgh Inventory w35x. All subjects were healthy native German speakers with no history of psychiatric or neurological disorder. The experimental procedures were described to all subjects in detail. Written informed consent was obtained. 2.2. Design and stimulus material A lexical decision task, similar to the one previously used in purely behavioral studies w40,41x, was set up using computerized stimulus presentation and recording of subjects’ responses. Stimuli consisted of 216 pairs of letter strings Žprimes and targets.. Primes were always German words, half of the targets were real German words and the other half were legally spelled pseudowords. Subjects were told that they would first be presented with a word and subsequently with a string of characters. This second string of characters could be a word or a pseudoword. The task was simply to read the first word silently, and then to decide as quickly and accurately as possible whether the second string of letters was a real German word or not. The subjects responses consisted of pressing one of two buttons on the computer’s pointing device with either their right index finger Žreal words. or their right middle finger Žnon-words.. The 108 prime–target combinations involving real words as targets consisted of 36 non-related pairs, 36 associative indirectly related pairs, and 36 directly related pairs. Indirectly related word pairs were defined as ‘‘strongly related to a strongly related concept’’. Most of these indirectly related targets were antonyms of a strong associate Že.g., lemon– Žsour. –sweet. or a property of a strong associate Že.g., lion– Žtiger. –stripes. of the prime word and were derived from previous studies w40,41x. The stimulus material was analyzed for the association of the targets with the primes. After explaining the concept, twenty subjects were asked to classify all word pairs either as directly related, indirectly related or non-related. Only word pairs which were classified to the same category from at least 90% of the subjects were used. Special effort was undertaken to ensure that the word pairs chosen as indirectly related were not directly related. For this purpose 20 subjects were asked to produce five associates to each of the utilized prime words as quickly as possible. Only pairs in which none of the 20 subjects associated the
M. Weisbrod et al.r CognitiÕe Brain Research 8 (1999) 289–298
291
Table 1 Types of prime–target relations Žconditions. as used in the experiment Condition Žprime–target-relation.
Examples as used in the experiment
Examples English transl.
Number in each stimulus-list
Word Word Word Word
Lowe–Streifen ¨ Bein–Arm Anker–Herr Apfel–Tacker ¨
lion–stripes leg–arm anchor–mister apple–kribe
36 36 36 108
— — — —
indirectly related word directly related word non-related word nonword
chosen target word given the prime word were accepted as indirectly related and implemented into the study. The length of the target words was not significantly different between stimulus conditions. Nearly all words were nouns, which frequency in German Žaccording to Ref. w37x. was not significantly different among conditions. The order of word-pairs within the stimulus list was initially randomized but then remained constant for all presentations. Examples of types of prime–target relation and the number of each type per list are shown in Table 1. All stimuli used in this study and their English translation are provided in Appendix A. 2.3. Experimental procedures All subjects were tested individually in the setting of the experimental electrophysiological laboratory at the Heidelberg Psychiatric University Hospital. Subjects were sitting in an electrically shielded dimly lit room in front of the computer screen on which the stimuli were presented. Given a viewing distance of 60 cm, each word subtended a visual angle of about 1 to 28 in width and 0.58 in height. Data collection was automatically controlled by the computer program ŽStim, Neuroscan, El Paso, TX.. Trials were presented continuously with a stimulus onset asynchrony ŽSOA. of 250 ms and an inter-trial interval of 1.5 s. A practice experiment, consisting of 10 trials similar to those of the actual experiment, was set up with no data sampled before the actual experiment was run. The sequence of one trial is displayed in Fig. 1. Fixation point, primes, and targets were displayed in the center of the screen. The entire session, consisting of electrode placement, instructions, the practice experiment, the full length experiment, and the handedness questionnaire required about 1 h and 15 min to be completed.
Fig. 1. Lexical decision task as used in the present study. Prior to the trial, the computer screen was blank. A trial started with a fixation point, followed by presentation of the prime for 250 ms which was immediately succeeded by the presentation of the target. The target was displayed for 2 s then the screen went blank for 1.5 s.
Evoked potentials were recorded from 20 sintered AGrAG-Cl scalp electrodes positioned according to the international 10–20 system. Linked-mastoids were used as reference sites and FPz as ground. Vertical EOG was recorded with supra- and infra-orbital electrodes. Electrodes on the external canthi recorded horizontal EOG. Electrode impedance was maintained below 5 k V for all recordings. The EEG was continuously collected Žlow-pass filter 70 Hz, 24 dBroctave roll-off, ArD rate 400 Hz, gain 500, range 11 mV, accuracy 0,168 mVrbit. and processed off-line. Correction for ocular artifacts was performed on the continuous-EEG-file using a regression analysis in combination with artifact averaging w39x. ERPs were averaged time locked to the target, relative to a 250 ms pre-target baseline. All trials were controlled manually for artifacts. Only artifact free trials entered further analyses. Averages for the three relevant conditions Žrelated, indirectly related, and non-related. were computed for each subject. These individual averages were digitally low-pass filtered at 16 Hz Ž24 dBroctave roll-off. for peak detection. The N400 peak was defined as the largest negative deflection at Cz in the time range from 350 to 550 ms. It was detected automatically and controlled manually. 3. Results 3.1. BehaÕioral data Priming effects were assessed by comparing mean reaction times ŽRT. and error rates ŽER. for the three real word stimulus conditions by means of two one-factor ANOVAs. The analysis of subjects’ RTs revealed a main effect of condition Ž F Ž2r52. s 109.4; p - 0.0001.. The shortest RTs were found in the directly related condition Ž662 " 122 ms., the longest in the non-related condition Ž754 " 121 ms. with the indirectly related condition in between Ž711 " 127 ms.. Post hoc analysis ŽScheffe-test ´ . showed that all three conditions were significantly different from each other Žsee Fig. 2.. Significant main effect of ERs Ž F Ž2r52. s 6.8; p s 0.002. was due to more errors made in the non-related condition Ž98% correct responses. than in both other conditions Žindirectly related: 99.5% correct responses, directly related: 99.4% correct responses.. 3.2. ERP data Since N400 was not reliably detectable at frontal electrode sites, these electrodes were not included into further
292
M. Weisbrod et al.r CognitiÕe Brain Research 8 (1999) 289–298
trodes ŽCz, Pz, Oz. as repeated measures factors were performed on peak amplitude and peak latency separately. To test for laterality two three-way ANOVAs with condition, laterality Žleftrright., and electrode site ŽC3rC4; P3rP4; T3rT4; T5rT6; O1rO2. as repeated measures factors were additionally performed on latencies and amplitudes. For all post-hoc comparisons Scheffe-test was ´ used. There were no significant N400 differences between male and female subjects, therefore, data were collapsed across groups.
Fig. 2. Reaction times Žin milliseconds; with standard deviations. for targets directly related to the prime word, indirectly related to the prime word, and not related to the prime word.
analyses. Two two-way ANOVAs with condition Ždirectly related, indirectly related, non-related. and midline elec-
3.2.1. N400 peak amplitudes As shown in Fig. 3 N400 was elicited in all three conditions. N400 amplitudes increased in the order related — indirectly related — non-related. The two-way ANOVA revealed main effects of electrode site Ž F Ž2r52. s 18.2; p - 0.0001. and condition Ž F Ž2r52. s 42.3; p - 0.0001., and a significant interaction between electrode site and condition Ž F Ž2r104. s 5.8; p - 0.001.. The most negative N400 peak was found on Cz, followed by Oz and Pz. Post hoc analysis revealed that N400 amplitudes in the directly related condition were at all electrode sites significantly smaller than in both other conditions. On Cz and Pz
Fig. 3. Grand averages from 27 subjects for targets directly related to the prime word Ž – P P – ., indirectly related to the prime word ŽP P P P P ., and not related to the prime word Ž — — . at electrode sites C3, CZ, C4 Žfirst row from left to right. and P3, P4, Pz Žsecond row from left to right.. Note that the prime word is presented 250 ms Žy250. prior to the target word Ž0 ms..
M. Weisbrod et al.r CognitiÕe Brain Research 8 (1999) 289–298
N400 peak in the non-related condition was more negative than in the indirectly related condition, on Pz there was a tendency Ž p - 0.1. for a more negative N400 in the non-related compared to the indirectly related condition. The three-way ANOVA revealed main effects of electrode site Ž F Ž4r104. s 7.6; p - 0.0001., laterality Ž F Ž1r26. s 30.1; p - 0.0001., and condition Ž F Ž2r52. s 31.7; p 0.0001.. Two two-way interactions welectrode-site = laterality Ž F Ž4r104. s 20.2; p - 0.0001. and electrodesite = condition ŽFŽ8r208. s 11.3; p - 0.0001.x, and the three-way interaction Ž F Ž8r208. s 2.7; p - 0.01. were also significant. In general, N400 peaks were more negative over the left than over the right hemisphere. At all electrode sites the N400 was most negative in the non-related condition, followed by the indirectly related and the directly related condition. These differences reached significance at most but not all electrode-sites.
3.2.2. Latencies N400 latencies increased in the order related — indirectly related — non-related condition. The two-way ANOVA yielded a main effect of condition Ž F Ž2r52. s 7.1; p s 0.002. and a significant interaction between electrode site and condition Ž F Ž4r104. s 3.1; p s 0.02.. Post hoc analysis revealed that at all electrode-sites N400 peaked earlier in the directly related condition than in the non-related condition. On Pz and Oz the difference between the directly related condition and the indirectly related condition reached also significance. Only on Oz N400 evoked by indirectly related targets peaked significantly earlier than N400 elicited by non-related targets. The Three-way ANOVA revealed significant main effects of electrode site Ž F Ž4r104. s 3.5; p s 0.01., laterality Ž F Ž1r26. s 4.9; p s 0.04., and condition Ž F Ž2r52. s 7.1; p s 0.002.. The interaction between electrode-site and condition Ž F Ž8r208. s 2.3; p s 0.02., and the three-way interaction reached significance Ž F Ž8r208. s 2.5; p s 0.01.. Post hoc analysis revealed that N400 peaked earlier over the left than over the right hemisphere. At all electrode sites N400 peaked first in the directly related, second in the indirectly related, and last in the non-related condition. However, not at all electrode sites differences between N400 peak latencies in the three conditions reached significance.
3.2.3. Component oÕerlap As it can be seen in Fig. 3, the N400 is followed and partially overlapped by a later positive going complex, probably a member of the P300 family. Since P300 latency may vary with semantic relationship between prime and target, its overlap with the N400 may differ depending on the prime target relationship and, therefore, may have confounded the described results especially with regard to N400 latencies. Since P300 is related to the RTs, we
293
splitted the averages in each condition according to the median RTs and compared N400 amplitudes and latencies for slow and fast responders calculating six additional ANOVAs Želectrode sites ŽCz, Pz, and Oz. as further factor.. With regard to N400 amplitudes groups were not different. N400 latency in the indirect condition peaked later in slow responders than in fast responders Ž F81r30. s 4.55; p s 0.05.. No differences for latencies were found in the other conditions. Hence, P300 overlap cannot explain our findings. Further studies, nevertheless, should use a design which allows to control for P300 overlap.
4. Discussion In this study a lexical decision task was used to elicit the N400 ERP component in healthy right-handed subjects. Primes and targets were either directly semantically related Že.g., black–white., indirectly semantically related Že.g., lemon– Žsour. –sweet. or not related Že.g., sofa–wing.. The main finding of this study consists of a significant semantic priming effect, a corresponding N400 amplitude reduction and a decrease in N400 latency for the indirectly related target words although prime and target word were only associated via an intermediate link. 4.1. BehaÕioral data The behavioral data gathered in this study reproduced the findings of earlier studies from our group, in which semantic content represented by the target word is activated by directly related primes, and — to some smaller extent — by an indirectly related prime word. Such an inverse relationship between semantic distance and the amount of activation has already been demonstrated by several authors w1,10,11,26x. 4.2. N400 The N400 component showed a pattern similar to the behavioral data: The most pronounced N400 amplitude and the latest peak latency was present in the non-related condition. Directly related targets evoked an N400 with a small amplitude and a comparatively short peak latency. The indirectly related condition elicited a smaller and earlier N400 amplitude than the non-related condition and a larger and later N400 than the related condition. Our findings bear on the question of what mechanisms are reflected by the N400 priming effects in this study. A popular functional model to date w30x posits three processes involved in semantic priming: automatic spread of activation, expectancy-induced priming, and semantic matching.
294
M. Weisbrod et al.r CognitiÕe Brain Research 8 (1999) 289–298
4.3. Expectancy induced priming Expectancy-induced priming asserts that subjects when confronted with a prime word generate an expectancy set consisting of words which are potential targets. Expectancy-induced priming is thought to be a slowly acting postlexical process. Since expectancy-induced priming needs sufficient time to build up, the short SOA of 250 ms renders it unlikely that it can account for the N400 and behavioral priming obtained in the present study. In addition, expectancy-induced priming is thought to be a controlled postlexical process. The relatively small amount of obviously related targets in our study Ž16,7% of the targets were directly-related. did not encourage controlled semantic analysis. Hence, it seems to be very unlikely that indirect associated targets were included in the expectancy set created by the prime word.
al. w4x instructed subjects to perform a graphemic judgement. Both groups found N400 being modulated by the semantic relation between prime and target indicating influential automatic processes. In contrast, Chwilla and coworkers w7x found an N400 priming effect only with a lexical decision instruction and not with an instruction requiring a pure physical discrimination. Brown and Hagoort w6x compared in a lexical decision task the N400 elicited in a masked and in an unmasked condition. Whereas reaction times showed significant priming effects in both conditions, an N400 priming effect was not obtained in the masked condition which did not allow postlexical processing. Holcomb w14x used normal and degraded targets and found no difference in the size of the ERP priming effects indicating that the process underlying the N400 priming effect operates on a postlexical level. In sum, there is an ongoing controversy, if automatic spread of activation influences N400 or not.
4.4. Automatic spread of actiÕation
4.5. Semantic matching
Automatic spread of activation is thought to be a fast operating intralexical mechanism w9,28x which does not depend on attentional resources. Semantic features of words are assumed to be represented in the semantic memory as nodes in a network. During speech production and reception, these nodes become activated for a short period of time until passive decay or active inhibition occurs w8,28,29x. Automatic spread of activation further asserts that activation spreads to related nodes, thereby lowering their thresholds of activation. The results obtained in this study are parsimoniously explained within the framework of spreading activation within semantic networks. Fast automatic activation may spread to adjacent nodes. The activation of these nodes further activates adjacent nodes. Hence, the recognition of target words is facilitated. As the amount of activation decreases with the semantic distance between prime and target N400 in the indirectly related condition is less effected than N400 in the directly related condition. Findings of several studies suggest that automatic spread of activation influences N400. Holcomb w13x manipulated both the instructions given to the subjects and the proportion of related prime–target pairs in a way that one block was ideally suited for producing only automatic priming and another block for producing attentional priming. In the ‘‘automatic’’ block target words that followed a non-related word and target words that followed a neutral prime elicited pronounced N400s. However, the N400s were larger in the ‘‘attentional’’ block than in the ‘‘automatic’’ block suggesting that N400 may be sensitive to both automatic spread of activation and the allocation of attentional resources. Some authors used instructions which imply shallow processing and do not encourage deep semantic analysis of targets. Kutas and Hillyard w20x asked subjects to decide if a target letter is present either in the first or the second member of a word pair, and Besson et
According to the model of semantic matching, primes are matched with targets and the result of this matching process biases the wordrnon-word decision. Semantic matching is supposed to be a postlexical mechanism. Whether it is fast acting or whether it needs substantial time to proceed is currently a matter of debate. The indirect association between prime and target in the indirect condition enabled matching. The successful matching process may have biased the wordrnon-word decision and, hence, resulted in a reduced N400 amplitude. A further, recently very popular model was introduced by Rugg and Doyle Žfor a comprehensive overview see Ref. w36x.. In this view, the amplitude of N400 reflects the ease with which words are integrated with what went before. In the directly related condition the target word is preceded by a directly associated prime word. Hence, the integration of the target word is easy and, therefore, the N400 becomes small. In the indirectly related condition the integration of the target word is more difficult than in the directly condition, but facilitated in comparison to the non related condition, resulting in an N400 of medium size. With regard to the short SOA used in this study an additional assumption is inevitable concerning semantic matching to explain the obtained N400 priming effects: it has to be assumed that semantic matching is fast acting and does not demand substantial time to proceed even in the indirect condition in which an intermediate link between prime and target has to be mentally generated. SOA may be a critical variable which determines the processes influencing N400 priming effects. Van Petten and Kutas w43x recorded ERPs while presenting subjects a series of sentences, each ending with a polysemous word. At 200 ms SOA, the ERP elicited by targets related to the inappropriate meaning was similar to the ERP for the target related to the appropriate meaning during the first 500 ms
M. Weisbrod et al.r CognitiÕe Brain Research 8 (1999) 289–298
epoch. In contrast, at 700 ms SOA, non-related targets and targets related to the inappropriate meaning elicited similar N400s, both larger than targets related to the appropriate meaning. These findings suggest that intra- and postlexical processes may influence N400: first, a fast acting automatic spread of activation within semantic memory networks may immediately arise and decay w28x and, later, processes which address contextual constraints may take over. To conclude, our findings demonstrate that indirect associations modulate N400 in a lexical decision task, even when SOA is short Ž250 ms. and no sentential constraints are present. Expectancy induced priming clearly does not contribute to the obtained N400 priming effect. Nevertheless, our data do not unequivocally allow to decide whether automatic spread of activation Žintralexically., semantic matching Žpostlexically. or both influenced the N400. However, if one assumes that semantic matching contributed to the N400 effect in the indirectly related condition semantic matching must be a fast acting process in order to integrate the meaning of words which are only associated via an intermediate link. Therefore, we believe that automatic spread of activation is the most parsimonous mechanism to account for the present results. Further studies, i.e., experiments crossing different prime target relations with several SOAs, are clearly needed to further characterize the mechanisms underlying N400 priming effects.
Acknowledgements This work was supported by an established investigator’s award by the National Alliance for Research in Schizophrenia and Depression ŽNARSAD., a grant from the Deutsche Forschungsgemeinschaft, DFG ŽGerman Research Community. to Manfred Spitzer, and a grant from the German Research Community ŽDeutsche Forschungsgemeinschaft, DFG. to M. Weisbrod. We thank Kerstin Herwig and Sabine Meidner for her help during data acquisition.
Indirectly related pairs ŽEnglish translation in parentheses. Prime Mediating Link Žnot presented. Amboß Žanvil. Hammer Žhammer. Auge Žeye. Nase Žnose. Auster Žoyster. Perle Žpearl. Bier Žbeer. Wein Žwine. Blume Žflower. Rose Žrose. Decke Žceiling. Boden Žfloor. Gabel Žfork. Messer Žknife. Hand Žhand. Fuß Žfoot. Hart Žhard. Weich Žsoft. Hemd Žshirt. Hose Žtrousers. Heu Žhay. Gras Žgrass.
295
Appendix A Directly related pairs ŽEnglish translation in parentheses. Prime Target Ast Žbough. Außen Žoutside. Bauer Žfarmer. Bein Žleg. Blitz Žlightning. Blut Žblood. Dach Žroof. Dunst Žmist. Essen Žto eat. Ferne Ždistance. Fischer Žfisherman. Fluß Žriver. Frage Žquestion. Gast Žguest. Žprison. Gefangnis ¨ Gold Žgold. Groß Žbig. Haß Žhate. Kleid Ždress. Kopf Žhead. Kuchen Žcake. Luft Žair. Morgen Žmorning. Mutter Žmother. Nonne Žnun. Ort Žplace. Pfeil Žarrow. Salz Žsalt. Schuler ¨ Žpupil. Schwarte Žrind. Tante Žaunt. Tuch Žcloth. ¨ Žsick. Ubel Vieh Žcattle. Wurst Žsausage. Zimmer Žroom.
Zweig Žbranch. Innen Žinside. Hof Žfarm. Arm Žarm. Donner Žthunder. Rot Žred. Ziegel Žtile. Nebel Žfog. Trinken Žto drink. Nahe ¨ Žnearness. Boot Žboat. Bach Žbrook. Antwort Žanswer. Wirt Žhost. Polizei Žpolice. Silber Žsilver. Klein Žsmall. Liebe Žlove. Rock Žskirt. Hals Žthroat. Teig Ždough. Atmen Žto breathe. Abend Ževening. Vater Žfather. Žmonk. Monch ¨ Dorf Žvillage. Bogen Žbow. Zucker Žsugar. Lehrer Žteacher. Speck Žfat. Onkel Žuncle. Stoff Žmaterial. Schlecht Žbad. Stall Žbarn. Kase ¨ Žcheese. Raum Žchamber.
Target Nagel Žnail. Duft Žsmell. Kette Žnecklace. Traube Žgrape. Dorn Žthorn. Teppich Žcarpet. Schneiden Žto cut. Tritt Žkick. Wolle Žcotton. Gurtel ¨ Žbelt. Grun ¨ Žgreen.
296 Hochzeit Žwedding. Holle ¨ Žhell. Hose Žpants. Kreide Žchalk. Kunde Žcustomer. Lowe ¨ Žlion. Nest Žnest. Netz Žweb. Priester Žpriest. Rasch Žfast. Rauh Žrough. Schnur Žstring. Schwer Žheavy. Sommer Žsummer. Sport Žsport. Stier Žbull. Tag Žday. Tal Žvalley. Taube Ždove. Tee Žtea. Woche Žweek. Wolke Žcloud. Wuste ¨ Ždesert. Zahn Žtooth. Zitrone Žlemon.
M. Weisbrod et al.r CognitiÕe Brain Research 8 (1999) 289–298 Ring Žring. Himmel Žheaven. Hemd Žshirt. Weiß Žwhite. Konig ¨ Žking. Tiger Žtiger. Vogel Žbird. Spinne Žspider. Kirche Žchurch. Langsam Žslow. Glatt Žsmooth. Faden Žthread. Leicht Žlight. Winter Žwinter. Fußball Žsoccer. Kuh Žcow. Nacht Žnight. Berg Žmountain. Frieden Žpeace. Kaffee Žcoffee. Tag Žday. Regen Žrain. Sand Žsand. Žbrush. Burste ¨ Sauer Žsour.
Non-related pairs ŽEnglish translation in parentheses. Prime Target Anker Žanchor. ¨ Žanger. Arger Bahn Žtrack. Bitten Žto beg. Bleistift Žpencil. Dame Žlady. Durst Žthirst. Esel Ždonkey. Feld Žfield. Fels Žrock. Fenster Žwindow. Form Žshape. Frack Žtuxedo. Geschichte Žhistory. Gespann Žteam. Glas Žglas. Kamin Žfireplace. Kampf Žfight. Kasten Žbox. Klopfen Žto knock. Kuh Žcow. Leib Žbody. Neid Ženvy. Rast Žleisure. Rechnung Žinvoice. Reich Žempire. Rolle Žrole. Schach Žchess. Seele Žsoul. Spielen Žto play. Stein Žstone.
Herr Žmister. Perle Žpearl. Obst Žfruit. Hahn Žcock. Uhr Žclock. Bau Žbuilding. Kessel Žcettle. Rosa Žpink. Horen ¨ Žto hear. Kabel Žcable. Tugend Žvirtue. Henne Žhen. Reis Žrice. Moor Žmoor. Frosch Žfrog. Flagge Žflag. Waage Žscale. Most Žmust. Spott Žmockery. Strecke Ždistance. Lehrling Žapprentice. Zoll Žcustoms. Deckel Žlid. Getreide Žgrain. Stille Žsilence. Blick Žlook. Sahne Žcream. Kamm Žcomb. Žspoon. Loffel ¨ Ž Faß barrel. Lage Žposition.
Finger Žfinger. Erde Žearth. Kragen Žcollar. Schwarz Žblack. Kaiser Žemperor. Streifen Žstripes. Fliegen Žto fly. Insekt Žinsect. Turm Žtower. Schnecke Žsnail. Seide Žsilk. Nadel Žneedle. Feder Žfeather. Schnee Žsnow. Tor Žgoal. Milch Žmilk. Dunkel Ždark. Hoch Žhigh. Krieg Žwar. Bohne Žbean. Nacht Žnight. Naß Žwet. Strand Žbeach. Haar Žhair. Suß ¨ Žsweet.
Stricken Žto knit. Strom Žstream. Truhe Žvault. Unkraut Žweed. Vase Žvase.
Hopfen Žhop. Treffer Žhit. Sau Žboar. Sieger Žwinner. Mantel Žcoat.
Word non-word pairs ŽEnglish translation in parentheses. Prime Target Abt Žabbot. Adel Žnobility. Amt Žoffice. Apfel Žapple. April Žapril. Arzt Žphysician. Atom Žatom. Ausfall Žloss. Backen Žto bake. Žbaker. Backer ¨ Backstein Žbrick. Balg Žpelt. Balsam Žbalsam. Bar ¨ Žbear. Bart Žbeard. Base Žbase. Bastler Žmodeller. Blech Žtin. Brei Žbroth. Bremse Žbrake. Brezel Žbrezel. Brust Žchest. Bus Žbus. Chor Žchoire.
Rinte Bacht Fobbe ¨ Tacker ¨ Nogel Meich Schech Fiza Tenge Kapfel Wuse Lahrt Wemm Falsis Zerv Tek Keltan Gabist Schimp Winn Waler Zarr Mulse ¨ Zunste
M. Weisbrod et al.r CognitiÕe Brain Research 8 (1999) 289–298
Denken Žto think. Drama Ždrama. Dung Žmenure. Durre ¨ Ždrought. Eifrig Žeager. Elefant Želephant. Erbse Žpea. Ernte Žharvest. Erz Žore. Farn Žfern. Fell Žskin. Flut Žflood. Freund Žfriend. Fruh ¨ Žearly. Futter Žfeed. Gerust ¨ Žscaffolding. Gesetz Žlaw. Glanz Žshine. Glied Žjoint. Grad Ždegree. Gut Žgood. Haufen Žheap. Hecke Žfence. Hexe Žwitch. Himmel Žsky. Horn Žhorn. Jammer Žmisery. Jubeln Žto cheer. Kafer ¨ Žbeetle. Kammer Žchamber. Kanal Žcanal. Kappe Žcap. Kerl Žguy. Kern Žcore. Kino Žcinema. Kirche Žchurch. Kirsche Žcherry. Klee Žclover. Korb Žbasket. Kraut Žweed. Kreuz Žcross. Kur Žcure. Lang Žlong. Last Žburden. Macht Žpower. Mark Žmark. Menge Žquantity. Metall Žmetal. Mittag Žmidday. Montag ŽMonday. Mull ¨ Žgarbish. Nerv Žnerve. Note Žnote. Papier Žpaper. Pfarrer Žpriest. Plage Žplague.
Hajel Bergel Mehte Zammer Ronster Aket Wirst Zarrer Trachsar Ocktas Altim Tirm Sorb Afnet Astier Birmul Dant Stoge Trogge Neuz Ellas Nelle Agurt Stee Auner Famto Fegal Bantse Dirten Ogten Fame ¨ Gimmer Nuser Dollum Nahmel Beiner Paut Waxe ¨ Saunt ¨ Lirsche Prat Gambi Enster Schor Badel Tor ¨ Trast Onsten Pegle Gimber Asto Sart Derse Sutter Gernte Minge
Plakat Žsignpost. Radio Žradio. Rahmen Žframe. Rede Žspeech. Regel Žrule. Roß Žhorse. Ruder Žrudder. Ruine Žruin. Scherz Žjoke. Schicht Žlayer. Schimmer Žshine. Schreiner Žjoiner. See Žsea. Soldat Žsoldier. Stimme Žvoice. Suchen Žto seek. Theater Žtheater. Ton Žsound. Torf Žturf. Tropen Žtropics. Verkehr Žtraffic. Wach Žawake. Wanne Žtub. Weich Žsoft. Wein Žwine. Wert Žvalue. Wurzel Žroot. Zwei Žtwo.
297
Ettes Jehr Fersti Uht Wansker Wente Sammer Begelt Vise Gerbst Ater Mirsche Hurz Werm Pfanst Frame Linge Relt Dast Dels Daget Penti Lund Barzt Barn Kauler Anjum Faumle
References w1x D.A. Balota, R.F. Lorch Jr., Depth of automatic spreading activation: Mediated priming effects in pronunciation but not in lexical decision, Journal of Experimental Psychology: Learning, Memory, and Cognition 12 Ž1986. 336–345. w2x S. Bentin, Event-related potentials, semantic processes, and expectancy factors in word recognition, Brain Lang. 31 Ž1987. 308– 327. w3x S. Bentin, G. McCarthy, C.C. Wood, Event-related potentials, lexical decision, and semantic priming, Electroencephalography and Clinical Neurophysiology 60 Ž1985. 343–355. w4x M. Besson, I. Fischler, T. Boaz, G. Raney, Effects of automatic associative activation on explicit and implicit memory tests, Journal of Experimental Psychology: Learning, Memory, and Cognition 18 Ž1992. 89–105. w5x M. Besson, F. Macar, An event-related potential analysis of incongruity in music and other non-linguistic contexts., Psychopysiology 24 Ž1987. 14–25. w6x C.M. Brown, P. Hagoort, The processing nature of the N400: Evidence from masked priming, J. Cogn. Neurosci. 5 Ž1993. 34–44. w7x D.J. Chwilla, C.M. Brown, P. Hagoort, The N400 as a function of the level of processing, Psychophysiology 32 Ž1995. 274–285. w8x A.M. Collins, E.F. Loftus, A spreading-activation theory of semantic processing, Psychol. Rev. 82 Ž1975. 407–428. w9x A.M.B. De Groot, Primed lexical decision: combined effects of the proportion of related prime–target pairs and stimulus-onset asynchrony of prime and target, The Quarterly Journal of Experimental Psychology Ž1984. 281–297.
298
M. Weisbrod et al.r CognitiÕe Brain Research 8 (1999) 289–298
w10x A.M.B. De Groot, The range of automatic spreading activation in word priming, Journal of Verbal Learning and Verbal Behavior 22 Ž1983. 417–436. w11x K. Den Heyer, K. Briand, Priming single digit numbers: automatic spread of activation dissipates as a function of semantic distance, Am. J. Psychol. 99 Ž1986. 315–339. w12x I. Fischler, P.A. Bloom, D.G. Childers, S.E. Roucos, N.W. Perry, Brain potentials related to stages of sentence verification, Psychophysiology 20 Ž1983. 400–409. w13x P.J. Holcomb, Automatic and attentional processing: An event-related brain potential analysis of semantic priming, Brain Lang. 35 Ž1988. 66–85. w14x P.J. Holcomb, Semantic priming an stimulus degradation: implications for the role of the N400 in language processing, Psychophysiology 30 Ž1993. 47–61. w15x P.J. Holcomb, H.J. Neville, Auditory and visual semantic priming in lexical decision: a comparison using event-related brain potentials, Language and Cognitive Processes 5 Ž1990. 281–312. w16x P.J. Holcomb, H.J. Neville, Natural speech processing: An analysis using event-related brain potentials, Psychobiology 19 Ž1991. 286– 300. w17x M. Kiefer, M. Weisbrod, I. Kern, S. Maier, M. Spitzer, Right hemisphere activation during indirect semantic priming: evidence from event-related potentials, Brain Lang. 64 Ž1998. 377–408. w18x S.A. Kounios, P.J. Holcomb, Structure and process in semantic memory: evidence from event-related potentials and reaction times, J. Exp. Psychol. Gen. 121 Ž1992. 460–480. w19x M. Kutas, S.A. Hillyard, Brain potentials during reading reflect word expectancy and semantic association, Nature 307 Ž1984. 1161–1163. w20x M. Kutas, S.A. Hillyard, An electrophysiological probe of incidential semantic association, J. Cogn. Neurosci. 1 Ž1989. 38–49. w21x M. Kutas, S.A. Hillyard, Event-related potentials to grammatical errors and semantic anomalies, Mem. Cogn. 11 Ž1983. 539–550. w22x M. Kutas, S.A. Hillyard, Reading senseless sentences: brain potentials reflect semantic incongruity, Science 207 Ž1980. 203–205. w23x M. Kutas, H.J. Neville, P.J. Holcomb, A preliminary comparison of the N400 response to semantic anomalies during reading, listening, and signing, Electroencephalography and Clinical Neurophysiology 39 Ž1987. 325–330, Supplement. w24x M. Kutas, C.K. Van Petten, Psycholinguistics electrified. Event-related brain potential investigations, in: M.A. Gernsbacher ŽEd.., Handbook of Psycholinguistics, Academic Press, New York, 1994, pp. 83–143. w25x W.C. McCallum, S.F. Farmer, P.V. Pocock, The effects of physical and semantic incongruities on auditory event-related potentials, Electroencephalography and Clinical Neurophysiology 59 Ž1984. 477– 488. w26x T.P. McNamara, J. Altaribba, Depth of spreading activation revisited: semantic mediated priming occurs in lexical decisions, Journal of Memory and Language 27 Ž1988. 545–559. w27x D.E. Meyer, R.W. Schvaneveldt, Facilitation in recognizing pairs of words: evidence of a dependence between retrieval operations, J. Exp. Psychol. Gen. 90 Ž1971. 227–234.
w28x J.H. Neely, Semantic priming and retrieval from lexical memory: roles of inhibitionless spreading activation and limited-capacity attention, J. Exp. Psychol. Gen. 106 Ž1977. 226–254. w29x J.H. Neely, Semantic priming effects in visual word recognition: A selective review of current findings and theories, in: D. Besner, G.W. Humphreys ŽEds.., Basic Progresses in Reading — Visual Word Recognition, Lawrence Erlbaum Associates, Hillsdale, NJ, 1991, pp. 264–333. w30x J.H. Neely, D.E. Keefe, Semantic context effects on visual word processing: A hybrid prospective–retrospective processing theory, in: G.H. Bower ŽEd.., The Psychology of Learning and Motivation: Advances in Research and Theory Vol. 24, Academic, New York, 1989, pp. 207–248. w31x H.J. Neville, M. Kutas, G. Chesney, A. Schmidt, Event-related brain potentials during the initial encoding and subsequent recognition memory of congruous and incongruous words, Journal of Memory and Language 25 Ž1986. 75–92. w32x A. Nigam, J.E. Hoffmann, R.F. Simons, N400 to semantically autonomous pictures and words, J. Cogn. Neurosci. 4 Ž1992. 15–22. w33x A.C. Nobre, G. McCarthy, Language-related ERPs: scalp distributions and modulation by word type and semantic priming, J. Cogn. Neurosci. 6 Ž1994. . w34x J.P. O’Halloran, R. Isenhart, C.A. Sandmann, L.S. Larkey, Brain responses to semantic anomaly in natural continuous speech, Int. J. Psychophysiol. 6 Ž1988. 243–254. w35x R. Oldfield, The assessment and analysis of handedness: The Edinburgh Inventory, Neuropsychologia 9 Ž1971. 97–113. w36x L. Osterhout, P.J. Holcomb, Event related potentials and language comprehension, in: M.D. Rugg, M.G.H. Coles ŽEds.., Electrophysiology of Mind, Oxford Psychology Series, Oxford Univ. Press, Oxford, 1997, pp. 171–215. w37x M.I. Posner, C.R.R. Snyder, Attention and cognitive control, in: R.L. Solso ŽEd.., Information Processing and Cognition: The Loyota Symposium, Lawrence Erlbaum Associates, Hillsdale, 1975, pp. w38x M.D. Rugg, The effects of semantic priming and word repetition on event-related potentials, Psychophysiology 22 Ž1985. 642–647. w39x H.V. Semlitsch, P. Anderer, P. Schuster, O. Presslich, A solution for reliable and valid reduction of ocular artefacts applied to the P300 ERP, Psychophysiology 23 Ž1986. 695–703. w40x M. Spitzer, U. Braun, L. Hermle, S. Maier, Associative semantic network dysfunction in thought-disordered schizophrenic patients: Direct evidence from indirect semantic priming, Biol. Psychiatry 34 Ž1993. 864–877. w41x M. Spitzer, U. Braun, S. Maier, L. Hermle, B.A. Maher, Indirect semantic priming in schizophrenic patients, Schizophr. Res. 11 Ž1993. 71–80. w42x M. Spitzer, S. Winkler, S. Maier, M. Weisbrod, Ereigniskorrelierte Potentiale bei semantischen Sprachverarbeitungsprozessen schizophrener Patienten, Nervenarzt 68 Ž1997. 212–225. w43x C. Van Petten, M. Kutas, Ambiguous words in context: An event-related potential analysis of the time course of meaning activation, Journal of Memory and Language 26 Ž1987. 188–208. w44x C. Van Petten, M. Kutas, Influences of semantic and context on open- and closed-class words, Mem. Cogn. 19 Ž1991. 95–112.
Research report
Electrophysiological correlates of direct versus indirect semantic priming in normal volunteers Matthias Weisbrod a
a,)
, Markus Kiefer b, Sabine Winkler a , Sabine Maier a , Holger Hill a , Daniela Roesch-Ely a , Manfred Spitzer b
Psychiatric Department, UniÕersity of Heidelberg, Voss-Str. 4, 69115, Heidelberg, Germany b Psychiatric Department, UniÕersity of Ulm, Ulm, Germany Accepted 22 June 1999
Abstract The N400 component of event related potentials ŽERP. was studied in 27 right-handed healthy subjects in a speeded lexical decision task. The semantic distance between primes Žalways words. and targets Ž50% words, 50% non-words. was systematically varied. Prime–target relations included directly related words Žhen–egg., indirectly related words Žlemon–sweet., and non-related words Žsofa–wing.. ERPs were recorded from 20 scalp electrodes positioned according to the international 10–20 system. The N400 reflected semantic distance with the most negative and latest N400 peak in the non-related condition, the least negative and earliest N400 peak in the direct condition and the peak in the indirect condition in-between. Hence, N400 priming effect in the indirect condition was obtained in the absence of strong sentential constraints and even though the mediating word was physically not present. q 1999 Elsevier Science B.V. All rights reserved. Keywords: Semantic priming; Lexical access; Event related potential; N400
1. Introduction Kutas and Hillyard w22x were the first to use event related potentials ŽERP. to study the physiological correlates of semantic association with a non-invasive, inexpensive, and widely available method. Their groundbreaking work extended psycholinguistic research into psychophysiology. Since then, numerous studies have demonstrated a negative ERP component peaking at about 400 ms after stimulus onset which can be reliably evoked in experiments contrasting semantically congruent with semantically incongruent sentence completions Žfor review see Ref. w24x.. This so-called N400 component has proved responsive to the manipulation of linguistic variables and has been found to be more sensitive to semantic association than to the truth-value of sentences w12,18x. It is also elicited by any stimulus with a semantic representation
) Corresponding author. Fax: q49-6221-56-5477; [email protected]
E-mail:
0926-6410r99r$19.00 q 1999 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 6 4 1 0 Ž 9 9 . 0 0 0 3 2 - 4
w23,32x and is sensitive to semantics, even when presented in different modalities w16,25,34x. In contrast, N400 is not affected by grammatical manipulations w21,44x, and is not evoked by non-linguistic symbolic mismatch w5x. With regard to sentences, N400 amplitude was shown to be modulated by the ‘‘cloze probability’’ Žthe degree of expectedness. of sentence-final words: i.e., the semantically acceptable completion in the sentence ‘‘the bill was due at the end of the hour’’ elicits a larger N400 than the predictable final word ‘‘the bill was due at the end of the month’’. In addition, associative constraints were shown to modulate N400 amplitude. Kutas and Hillyard w19x compared N400 evoked by three types of sentence completions: The first type was the most expected Žhigh cloze probability. final word Ži.e., ‘‘the pizza was to hot to eat’’., the second type was not expected but related to the most expected completion Ži.e., ‘‘the pizza was to hot to drink’’., and the third type was semantically anomalous and non-related to the most expected completion Ži.e., ‘‘the pizza was to hot to cry’’.. Anomalous endings elicited a large N400 whereas expected words elicited a small N400. The N400 to sentence completions which were anomalous but related to the most expected final word was smaller
290
M. Weisbrod et al.r CognitiÕe Brain Research 8 (1999) 289–298
than the N400 to anomalous words which were not related to the most likely completion. These results were interpreted as demonstrating that N400 is modulated Ži. by word expectancy engendered by the context, and Žii. by associations between single words of which one word is only suggested by the context and which does not need to be physically present. The majority of N400 studies have focused on words presented in sentence contexts. Interpretation of results may be difficult because of confounding influences from syntax, context, and the heavy load on semantic working memory required. This may cause problems when neuropsychiatric patients who often suffer from dysfunctional working memory are studied. The presentation of single words in a lexical decision task avoids these problems. In a lexical decision task, the subject has to decide whether a given string of characters is a word or not. If a semantically related word precedes the target word, the target word will be recognized faster, that is, semantic priming will occur w27,28x. In so far as ERPs have been recorded during semantic priming paradigms w2,3,13,15,20,33,38x, late negativities similar to sentential N400s have been observed. Words which are not repeated and semantically non-related to previous words elicit the largest N400 w3,12,31x. Kutas and Hillyard w20x showed that the degree of semantic relatedness influences N400 amplitude elicited by words presented in close succession. They instructed subjects to decide whether a target letter was present either in the first or the second member of a word pair. Target words produced the largest N400 when the word pairs were non-related, intermediate when the semantic relationship was relatively weak, and smallest when words were strongly related. Only one study examined the N400 elicited by indirectly related target words. Indirectly related words can be characterized by that the association between prime and target is only reached by an intermediate link, i.e., the prime is associated to a strong associate of the target like lemon– Žsour. –sweet. Hence, there is no way to generate from the prime word an immediate expectation which includes the target: our group w17x investigated scalp potentials evoked in healthy controls using a lexical decision task, including directly related, indirectly related and nonrelated prime–target pairs. The results in regard to the indirect condition were not conclusive. N400 for indirectly related target was larger than for directly-related words and somewhat smaller than for non-related words. However, the difference between the non-related and the indirectly-related condition did not reach significance, perhaps due to the relatively small number of subjects. Therefore, it is not safe to conclude that mediated associations do not elicit N400 effects. The present study was set up to investigate the N400 elicited by targets indirectly related to prime words on a larger group of subjects. To achieve this, ERPs were recorded while healthy subjects performed a lexical decision task which included indirect associates.
2. Materials and methods 2.1. Subjects Twenty seven right-handed subjects Ž16 male, 11 female; age: 33.2 " 11.5 years; school education: 11.4 " 1.8 years. participated in the experiment. 18 subjects participated as controls in a study which compared semantic processing between schizophrenic patients and healthy subjects w42x, the other subjects were controls in a still unpublished twin study. Hand dominance was assessed with a translated German version of the Edinburgh Inventory w35x. All subjects were healthy native German speakers with no history of psychiatric or neurological disorder. The experimental procedures were described to all subjects in detail. Written informed consent was obtained. 2.2. Design and stimulus material A lexical decision task, similar to the one previously used in purely behavioral studies w40,41x, was set up using computerized stimulus presentation and recording of subjects’ responses. Stimuli consisted of 216 pairs of letter strings Žprimes and targets.. Primes were always German words, half of the targets were real German words and the other half were legally spelled pseudowords. Subjects were told that they would first be presented with a word and subsequently with a string of characters. This second string of characters could be a word or a pseudoword. The task was simply to read the first word silently, and then to decide as quickly and accurately as possible whether the second string of letters was a real German word or not. The subjects responses consisted of pressing one of two buttons on the computer’s pointing device with either their right index finger Žreal words. or their right middle finger Žnon-words.. The 108 prime–target combinations involving real words as targets consisted of 36 non-related pairs, 36 associative indirectly related pairs, and 36 directly related pairs. Indirectly related word pairs were defined as ‘‘strongly related to a strongly related concept’’. Most of these indirectly related targets were antonyms of a strong associate Že.g., lemon– Žsour. –sweet. or a property of a strong associate Že.g., lion– Žtiger. –stripes. of the prime word and were derived from previous studies w40,41x. The stimulus material was analyzed for the association of the targets with the primes. After explaining the concept, twenty subjects were asked to classify all word pairs either as directly related, indirectly related or non-related. Only word pairs which were classified to the same category from at least 90% of the subjects were used. Special effort was undertaken to ensure that the word pairs chosen as indirectly related were not directly related. For this purpose 20 subjects were asked to produce five associates to each of the utilized prime words as quickly as possible. Only pairs in which none of the 20 subjects associated the
M. Weisbrod et al.r CognitiÕe Brain Research 8 (1999) 289–298
291
Table 1 Types of prime–target relations Žconditions. as used in the experiment Condition Žprime–target-relation.
Examples as used in the experiment
Examples English transl.
Number in each stimulus-list
Word Word Word Word
Lowe–Streifen ¨ Bein–Arm Anker–Herr Apfel–Tacker ¨
lion–stripes leg–arm anchor–mister apple–kribe
36 36 36 108
— — — —
indirectly related word directly related word non-related word nonword
chosen target word given the prime word were accepted as indirectly related and implemented into the study. The length of the target words was not significantly different between stimulus conditions. Nearly all words were nouns, which frequency in German Žaccording to Ref. w37x. was not significantly different among conditions. The order of word-pairs within the stimulus list was initially randomized but then remained constant for all presentations. Examples of types of prime–target relation and the number of each type per list are shown in Table 1. All stimuli used in this study and their English translation are provided in Appendix A. 2.3. Experimental procedures All subjects were tested individually in the setting of the experimental electrophysiological laboratory at the Heidelberg Psychiatric University Hospital. Subjects were sitting in an electrically shielded dimly lit room in front of the computer screen on which the stimuli were presented. Given a viewing distance of 60 cm, each word subtended a visual angle of about 1 to 28 in width and 0.58 in height. Data collection was automatically controlled by the computer program ŽStim, Neuroscan, El Paso, TX.. Trials were presented continuously with a stimulus onset asynchrony ŽSOA. of 250 ms and an inter-trial interval of 1.5 s. A practice experiment, consisting of 10 trials similar to those of the actual experiment, was set up with no data sampled before the actual experiment was run. The sequence of one trial is displayed in Fig. 1. Fixation point, primes, and targets were displayed in the center of the screen. The entire session, consisting of electrode placement, instructions, the practice experiment, the full length experiment, and the handedness questionnaire required about 1 h and 15 min to be completed.
Fig. 1. Lexical decision task as used in the present study. Prior to the trial, the computer screen was blank. A trial started with a fixation point, followed by presentation of the prime for 250 ms which was immediately succeeded by the presentation of the target. The target was displayed for 2 s then the screen went blank for 1.5 s.
Evoked potentials were recorded from 20 sintered AGrAG-Cl scalp electrodes positioned according to the international 10–20 system. Linked-mastoids were used as reference sites and FPz as ground. Vertical EOG was recorded with supra- and infra-orbital electrodes. Electrodes on the external canthi recorded horizontal EOG. Electrode impedance was maintained below 5 k V for all recordings. The EEG was continuously collected Žlow-pass filter 70 Hz, 24 dBroctave roll-off, ArD rate 400 Hz, gain 500, range 11 mV, accuracy 0,168 mVrbit. and processed off-line. Correction for ocular artifacts was performed on the continuous-EEG-file using a regression analysis in combination with artifact averaging w39x. ERPs were averaged time locked to the target, relative to a 250 ms pre-target baseline. All trials were controlled manually for artifacts. Only artifact free trials entered further analyses. Averages for the three relevant conditions Žrelated, indirectly related, and non-related. were computed for each subject. These individual averages were digitally low-pass filtered at 16 Hz Ž24 dBroctave roll-off. for peak detection. The N400 peak was defined as the largest negative deflection at Cz in the time range from 350 to 550 ms. It was detected automatically and controlled manually. 3. Results 3.1. BehaÕioral data Priming effects were assessed by comparing mean reaction times ŽRT. and error rates ŽER. for the three real word stimulus conditions by means of two one-factor ANOVAs. The analysis of subjects’ RTs revealed a main effect of condition Ž F Ž2r52. s 109.4; p - 0.0001.. The shortest RTs were found in the directly related condition Ž662 " 122 ms., the longest in the non-related condition Ž754 " 121 ms. with the indirectly related condition in between Ž711 " 127 ms.. Post hoc analysis ŽScheffe-test ´ . showed that all three conditions were significantly different from each other Žsee Fig. 2.. Significant main effect of ERs Ž F Ž2r52. s 6.8; p s 0.002. was due to more errors made in the non-related condition Ž98% correct responses. than in both other conditions Žindirectly related: 99.5% correct responses, directly related: 99.4% correct responses.. 3.2. ERP data Since N400 was not reliably detectable at frontal electrode sites, these electrodes were not included into further
292
M. Weisbrod et al.r CognitiÕe Brain Research 8 (1999) 289–298
trodes ŽCz, Pz, Oz. as repeated measures factors were performed on peak amplitude and peak latency separately. To test for laterality two three-way ANOVAs with condition, laterality Žleftrright., and electrode site ŽC3rC4; P3rP4; T3rT4; T5rT6; O1rO2. as repeated measures factors were additionally performed on latencies and amplitudes. For all post-hoc comparisons Scheffe-test was ´ used. There were no significant N400 differences between male and female subjects, therefore, data were collapsed across groups.
Fig. 2. Reaction times Žin milliseconds; with standard deviations. for targets directly related to the prime word, indirectly related to the prime word, and not related to the prime word.
analyses. Two two-way ANOVAs with condition Ždirectly related, indirectly related, non-related. and midline elec-
3.2.1. N400 peak amplitudes As shown in Fig. 3 N400 was elicited in all three conditions. N400 amplitudes increased in the order related — indirectly related — non-related. The two-way ANOVA revealed main effects of electrode site Ž F Ž2r52. s 18.2; p - 0.0001. and condition Ž F Ž2r52. s 42.3; p - 0.0001., and a significant interaction between electrode site and condition Ž F Ž2r104. s 5.8; p - 0.001.. The most negative N400 peak was found on Cz, followed by Oz and Pz. Post hoc analysis revealed that N400 amplitudes in the directly related condition were at all electrode sites significantly smaller than in both other conditions. On Cz and Pz
Fig. 3. Grand averages from 27 subjects for targets directly related to the prime word Ž – P P – ., indirectly related to the prime word ŽP P P P P ., and not related to the prime word Ž — — . at electrode sites C3, CZ, C4 Žfirst row from left to right. and P3, P4, Pz Žsecond row from left to right.. Note that the prime word is presented 250 ms Žy250. prior to the target word Ž0 ms..
M. Weisbrod et al.r CognitiÕe Brain Research 8 (1999) 289–298
N400 peak in the non-related condition was more negative than in the indirectly related condition, on Pz there was a tendency Ž p - 0.1. for a more negative N400 in the non-related compared to the indirectly related condition. The three-way ANOVA revealed main effects of electrode site Ž F Ž4r104. s 7.6; p - 0.0001., laterality Ž F Ž1r26. s 30.1; p - 0.0001., and condition Ž F Ž2r52. s 31.7; p 0.0001.. Two two-way interactions welectrode-site = laterality Ž F Ž4r104. s 20.2; p - 0.0001. and electrodesite = condition ŽFŽ8r208. s 11.3; p - 0.0001.x, and the three-way interaction Ž F Ž8r208. s 2.7; p - 0.01. were also significant. In general, N400 peaks were more negative over the left than over the right hemisphere. At all electrode sites the N400 was most negative in the non-related condition, followed by the indirectly related and the directly related condition. These differences reached significance at most but not all electrode-sites.
3.2.2. Latencies N400 latencies increased in the order related — indirectly related — non-related condition. The two-way ANOVA yielded a main effect of condition Ž F Ž2r52. s 7.1; p s 0.002. and a significant interaction between electrode site and condition Ž F Ž4r104. s 3.1; p s 0.02.. Post hoc analysis revealed that at all electrode-sites N400 peaked earlier in the directly related condition than in the non-related condition. On Pz and Oz the difference between the directly related condition and the indirectly related condition reached also significance. Only on Oz N400 evoked by indirectly related targets peaked significantly earlier than N400 elicited by non-related targets. The Three-way ANOVA revealed significant main effects of electrode site Ž F Ž4r104. s 3.5; p s 0.01., laterality Ž F Ž1r26. s 4.9; p s 0.04., and condition Ž F Ž2r52. s 7.1; p s 0.002.. The interaction between electrode-site and condition Ž F Ž8r208. s 2.3; p s 0.02., and the three-way interaction reached significance Ž F Ž8r208. s 2.5; p s 0.01.. Post hoc analysis revealed that N400 peaked earlier over the left than over the right hemisphere. At all electrode sites N400 peaked first in the directly related, second in the indirectly related, and last in the non-related condition. However, not at all electrode sites differences between N400 peak latencies in the three conditions reached significance.
3.2.3. Component oÕerlap As it can be seen in Fig. 3, the N400 is followed and partially overlapped by a later positive going complex, probably a member of the P300 family. Since P300 latency may vary with semantic relationship between prime and target, its overlap with the N400 may differ depending on the prime target relationship and, therefore, may have confounded the described results especially with regard to N400 latencies. Since P300 is related to the RTs, we
293
splitted the averages in each condition according to the median RTs and compared N400 amplitudes and latencies for slow and fast responders calculating six additional ANOVAs Želectrode sites ŽCz, Pz, and Oz. as further factor.. With regard to N400 amplitudes groups were not different. N400 latency in the indirect condition peaked later in slow responders than in fast responders Ž F81r30. s 4.55; p s 0.05.. No differences for latencies were found in the other conditions. Hence, P300 overlap cannot explain our findings. Further studies, nevertheless, should use a design which allows to control for P300 overlap.
4. Discussion In this study a lexical decision task was used to elicit the N400 ERP component in healthy right-handed subjects. Primes and targets were either directly semantically related Že.g., black–white., indirectly semantically related Že.g., lemon– Žsour. –sweet. or not related Že.g., sofa–wing.. The main finding of this study consists of a significant semantic priming effect, a corresponding N400 amplitude reduction and a decrease in N400 latency for the indirectly related target words although prime and target word were only associated via an intermediate link. 4.1. BehaÕioral data The behavioral data gathered in this study reproduced the findings of earlier studies from our group, in which semantic content represented by the target word is activated by directly related primes, and — to some smaller extent — by an indirectly related prime word. Such an inverse relationship between semantic distance and the amount of activation has already been demonstrated by several authors w1,10,11,26x. 4.2. N400 The N400 component showed a pattern similar to the behavioral data: The most pronounced N400 amplitude and the latest peak latency was present in the non-related condition. Directly related targets evoked an N400 with a small amplitude and a comparatively short peak latency. The indirectly related condition elicited a smaller and earlier N400 amplitude than the non-related condition and a larger and later N400 than the related condition. Our findings bear on the question of what mechanisms are reflected by the N400 priming effects in this study. A popular functional model to date w30x posits three processes involved in semantic priming: automatic spread of activation, expectancy-induced priming, and semantic matching.
294
M. Weisbrod et al.r CognitiÕe Brain Research 8 (1999) 289–298
4.3. Expectancy induced priming Expectancy-induced priming asserts that subjects when confronted with a prime word generate an expectancy set consisting of words which are potential targets. Expectancy-induced priming is thought to be a slowly acting postlexical process. Since expectancy-induced priming needs sufficient time to build up, the short SOA of 250 ms renders it unlikely that it can account for the N400 and behavioral priming obtained in the present study. In addition, expectancy-induced priming is thought to be a controlled postlexical process. The relatively small amount of obviously related targets in our study Ž16,7% of the targets were directly-related. did not encourage controlled semantic analysis. Hence, it seems to be very unlikely that indirect associated targets were included in the expectancy set created by the prime word.
al. w4x instructed subjects to perform a graphemic judgement. Both groups found N400 being modulated by the semantic relation between prime and target indicating influential automatic processes. In contrast, Chwilla and coworkers w7x found an N400 priming effect only with a lexical decision instruction and not with an instruction requiring a pure physical discrimination. Brown and Hagoort w6x compared in a lexical decision task the N400 elicited in a masked and in an unmasked condition. Whereas reaction times showed significant priming effects in both conditions, an N400 priming effect was not obtained in the masked condition which did not allow postlexical processing. Holcomb w14x used normal and degraded targets and found no difference in the size of the ERP priming effects indicating that the process underlying the N400 priming effect operates on a postlexical level. In sum, there is an ongoing controversy, if automatic spread of activation influences N400 or not.
4.4. Automatic spread of actiÕation
4.5. Semantic matching
Automatic spread of activation is thought to be a fast operating intralexical mechanism w9,28x which does not depend on attentional resources. Semantic features of words are assumed to be represented in the semantic memory as nodes in a network. During speech production and reception, these nodes become activated for a short period of time until passive decay or active inhibition occurs w8,28,29x. Automatic spread of activation further asserts that activation spreads to related nodes, thereby lowering their thresholds of activation. The results obtained in this study are parsimoniously explained within the framework of spreading activation within semantic networks. Fast automatic activation may spread to adjacent nodes. The activation of these nodes further activates adjacent nodes. Hence, the recognition of target words is facilitated. As the amount of activation decreases with the semantic distance between prime and target N400 in the indirectly related condition is less effected than N400 in the directly related condition. Findings of several studies suggest that automatic spread of activation influences N400. Holcomb w13x manipulated both the instructions given to the subjects and the proportion of related prime–target pairs in a way that one block was ideally suited for producing only automatic priming and another block for producing attentional priming. In the ‘‘automatic’’ block target words that followed a non-related word and target words that followed a neutral prime elicited pronounced N400s. However, the N400s were larger in the ‘‘attentional’’ block than in the ‘‘automatic’’ block suggesting that N400 may be sensitive to both automatic spread of activation and the allocation of attentional resources. Some authors used instructions which imply shallow processing and do not encourage deep semantic analysis of targets. Kutas and Hillyard w20x asked subjects to decide if a target letter is present either in the first or the second member of a word pair, and Besson et
According to the model of semantic matching, primes are matched with targets and the result of this matching process biases the wordrnon-word decision. Semantic matching is supposed to be a postlexical mechanism. Whether it is fast acting or whether it needs substantial time to proceed is currently a matter of debate. The indirect association between prime and target in the indirect condition enabled matching. The successful matching process may have biased the wordrnon-word decision and, hence, resulted in a reduced N400 amplitude. A further, recently very popular model was introduced by Rugg and Doyle Žfor a comprehensive overview see Ref. w36x.. In this view, the amplitude of N400 reflects the ease with which words are integrated with what went before. In the directly related condition the target word is preceded by a directly associated prime word. Hence, the integration of the target word is easy and, therefore, the N400 becomes small. In the indirectly related condition the integration of the target word is more difficult than in the directly condition, but facilitated in comparison to the non related condition, resulting in an N400 of medium size. With regard to the short SOA used in this study an additional assumption is inevitable concerning semantic matching to explain the obtained N400 priming effects: it has to be assumed that semantic matching is fast acting and does not demand substantial time to proceed even in the indirect condition in which an intermediate link between prime and target has to be mentally generated. SOA may be a critical variable which determines the processes influencing N400 priming effects. Van Petten and Kutas w43x recorded ERPs while presenting subjects a series of sentences, each ending with a polysemous word. At 200 ms SOA, the ERP elicited by targets related to the inappropriate meaning was similar to the ERP for the target related to the appropriate meaning during the first 500 ms
M. Weisbrod et al.r CognitiÕe Brain Research 8 (1999) 289–298
epoch. In contrast, at 700 ms SOA, non-related targets and targets related to the inappropriate meaning elicited similar N400s, both larger than targets related to the appropriate meaning. These findings suggest that intra- and postlexical processes may influence N400: first, a fast acting automatic spread of activation within semantic memory networks may immediately arise and decay w28x and, later, processes which address contextual constraints may take over. To conclude, our findings demonstrate that indirect associations modulate N400 in a lexical decision task, even when SOA is short Ž250 ms. and no sentential constraints are present. Expectancy induced priming clearly does not contribute to the obtained N400 priming effect. Nevertheless, our data do not unequivocally allow to decide whether automatic spread of activation Žintralexically., semantic matching Žpostlexically. or both influenced the N400. However, if one assumes that semantic matching contributed to the N400 effect in the indirectly related condition semantic matching must be a fast acting process in order to integrate the meaning of words which are only associated via an intermediate link. Therefore, we believe that automatic spread of activation is the most parsimonous mechanism to account for the present results. Further studies, i.e., experiments crossing different prime target relations with several SOAs, are clearly needed to further characterize the mechanisms underlying N400 priming effects.
Acknowledgements This work was supported by an established investigator’s award by the National Alliance for Research in Schizophrenia and Depression ŽNARSAD., a grant from the Deutsche Forschungsgemeinschaft, DFG ŽGerman Research Community. to Manfred Spitzer, and a grant from the German Research Community ŽDeutsche Forschungsgemeinschaft, DFG. to M. Weisbrod. We thank Kerstin Herwig and Sabine Meidner for her help during data acquisition.
Indirectly related pairs ŽEnglish translation in parentheses. Prime Mediating Link Žnot presented. Amboß Žanvil. Hammer Žhammer. Auge Žeye. Nase Žnose. Auster Žoyster. Perle Žpearl. Bier Žbeer. Wein Žwine. Blume Žflower. Rose Žrose. Decke Žceiling. Boden Žfloor. Gabel Žfork. Messer Žknife. Hand Žhand. Fuß Žfoot. Hart Žhard. Weich Žsoft. Hemd Žshirt. Hose Žtrousers. Heu Žhay. Gras Žgrass.
295
Appendix A Directly related pairs ŽEnglish translation in parentheses. Prime Target Ast Žbough. Außen Žoutside. Bauer Žfarmer. Bein Žleg. Blitz Žlightning. Blut Žblood. Dach Žroof. Dunst Žmist. Essen Žto eat. Ferne Ždistance. Fischer Žfisherman. Fluß Žriver. Frage Žquestion. Gast Žguest. Žprison. Gefangnis ¨ Gold Žgold. Groß Žbig. Haß Žhate. Kleid Ždress. Kopf Žhead. Kuchen Žcake. Luft Žair. Morgen Žmorning. Mutter Žmother. Nonne Žnun. Ort Žplace. Pfeil Žarrow. Salz Žsalt. Schuler ¨ Žpupil. Schwarte Žrind. Tante Žaunt. Tuch Žcloth. ¨ Žsick. Ubel Vieh Žcattle. Wurst Žsausage. Zimmer Žroom.
Zweig Žbranch. Innen Žinside. Hof Žfarm. Arm Žarm. Donner Žthunder. Rot Žred. Ziegel Žtile. Nebel Žfog. Trinken Žto drink. Nahe ¨ Žnearness. Boot Žboat. Bach Žbrook. Antwort Žanswer. Wirt Žhost. Polizei Žpolice. Silber Žsilver. Klein Žsmall. Liebe Žlove. Rock Žskirt. Hals Žthroat. Teig Ždough. Atmen Žto breathe. Abend Ževening. Vater Žfather. Žmonk. Monch ¨ Dorf Žvillage. Bogen Žbow. Zucker Žsugar. Lehrer Žteacher. Speck Žfat. Onkel Žuncle. Stoff Žmaterial. Schlecht Žbad. Stall Žbarn. Kase ¨ Žcheese. Raum Žchamber.
Target Nagel Žnail. Duft Žsmell. Kette Žnecklace. Traube Žgrape. Dorn Žthorn. Teppich Žcarpet. Schneiden Žto cut. Tritt Žkick. Wolle Žcotton. Gurtel ¨ Žbelt. Grun ¨ Žgreen.
296 Hochzeit Žwedding. Holle ¨ Žhell. Hose Žpants. Kreide Žchalk. Kunde Žcustomer. Lowe ¨ Žlion. Nest Žnest. Netz Žweb. Priester Žpriest. Rasch Žfast. Rauh Žrough. Schnur Žstring. Schwer Žheavy. Sommer Žsummer. Sport Žsport. Stier Žbull. Tag Žday. Tal Žvalley. Taube Ždove. Tee Žtea. Woche Žweek. Wolke Žcloud. Wuste ¨ Ždesert. Zahn Žtooth. Zitrone Žlemon.
M. Weisbrod et al.r CognitiÕe Brain Research 8 (1999) 289–298 Ring Žring. Himmel Žheaven. Hemd Žshirt. Weiß Žwhite. Konig ¨ Žking. Tiger Žtiger. Vogel Žbird. Spinne Žspider. Kirche Žchurch. Langsam Žslow. Glatt Žsmooth. Faden Žthread. Leicht Žlight. Winter Žwinter. Fußball Žsoccer. Kuh Žcow. Nacht Žnight. Berg Žmountain. Frieden Žpeace. Kaffee Žcoffee. Tag Žday. Regen Žrain. Sand Žsand. Žbrush. Burste ¨ Sauer Žsour.
Non-related pairs ŽEnglish translation in parentheses. Prime Target Anker Žanchor. ¨ Žanger. Arger Bahn Žtrack. Bitten Žto beg. Bleistift Žpencil. Dame Žlady. Durst Žthirst. Esel Ždonkey. Feld Žfield. Fels Žrock. Fenster Žwindow. Form Žshape. Frack Žtuxedo. Geschichte Žhistory. Gespann Žteam. Glas Žglas. Kamin Žfireplace. Kampf Žfight. Kasten Žbox. Klopfen Žto knock. Kuh Žcow. Leib Žbody. Neid Ženvy. Rast Žleisure. Rechnung Žinvoice. Reich Žempire. Rolle Žrole. Schach Žchess. Seele Žsoul. Spielen Žto play. Stein Žstone.
Herr Žmister. Perle Žpearl. Obst Žfruit. Hahn Žcock. Uhr Žclock. Bau Žbuilding. Kessel Žcettle. Rosa Žpink. Horen ¨ Žto hear. Kabel Žcable. Tugend Žvirtue. Henne Žhen. Reis Žrice. Moor Žmoor. Frosch Žfrog. Flagge Žflag. Waage Žscale. Most Žmust. Spott Žmockery. Strecke Ždistance. Lehrling Žapprentice. Zoll Žcustoms. Deckel Žlid. Getreide Žgrain. Stille Žsilence. Blick Žlook. Sahne Žcream. Kamm Žcomb. Žspoon. Loffel ¨ Ž Faß barrel. Lage Žposition.
Finger Žfinger. Erde Žearth. Kragen Žcollar. Schwarz Žblack. Kaiser Žemperor. Streifen Žstripes. Fliegen Žto fly. Insekt Žinsect. Turm Žtower. Schnecke Žsnail. Seide Žsilk. Nadel Žneedle. Feder Žfeather. Schnee Žsnow. Tor Žgoal. Milch Žmilk. Dunkel Ždark. Hoch Žhigh. Krieg Žwar. Bohne Žbean. Nacht Žnight. Naß Žwet. Strand Žbeach. Haar Žhair. Suß ¨ Žsweet.
Stricken Žto knit. Strom Žstream. Truhe Žvault. Unkraut Žweed. Vase Žvase.
Hopfen Žhop. Treffer Žhit. Sau Žboar. Sieger Žwinner. Mantel Žcoat.
Word non-word pairs ŽEnglish translation in parentheses. Prime Target Abt Žabbot. Adel Žnobility. Amt Žoffice. Apfel Žapple. April Žapril. Arzt Žphysician. Atom Žatom. Ausfall Žloss. Backen Žto bake. Žbaker. Backer ¨ Backstein Žbrick. Balg Žpelt. Balsam Žbalsam. Bar ¨ Žbear. Bart Žbeard. Base Žbase. Bastler Žmodeller. Blech Žtin. Brei Žbroth. Bremse Žbrake. Brezel Žbrezel. Brust Žchest. Bus Žbus. Chor Žchoire.
Rinte Bacht Fobbe ¨ Tacker ¨ Nogel Meich Schech Fiza Tenge Kapfel Wuse Lahrt Wemm Falsis Zerv Tek Keltan Gabist Schimp Winn Waler Zarr Mulse ¨ Zunste
M. Weisbrod et al.r CognitiÕe Brain Research 8 (1999) 289–298
Denken Žto think. Drama Ždrama. Dung Žmenure. Durre ¨ Ždrought. Eifrig Žeager. Elefant Želephant. Erbse Žpea. Ernte Žharvest. Erz Žore. Farn Žfern. Fell Žskin. Flut Žflood. Freund Žfriend. Fruh ¨ Žearly. Futter Žfeed. Gerust ¨ Žscaffolding. Gesetz Žlaw. Glanz Žshine. Glied Žjoint. Grad Ždegree. Gut Žgood. Haufen Žheap. Hecke Žfence. Hexe Žwitch. Himmel Žsky. Horn Žhorn. Jammer Žmisery. Jubeln Žto cheer. Kafer ¨ Žbeetle. Kammer Žchamber. Kanal Žcanal. Kappe Žcap. Kerl Žguy. Kern Žcore. Kino Žcinema. Kirche Žchurch. Kirsche Žcherry. Klee Žclover. Korb Žbasket. Kraut Žweed. Kreuz Žcross. Kur Žcure. Lang Žlong. Last Žburden. Macht Žpower. Mark Žmark. Menge Žquantity. Metall Žmetal. Mittag Žmidday. Montag ŽMonday. Mull ¨ Žgarbish. Nerv Žnerve. Note Žnote. Papier Žpaper. Pfarrer Žpriest. Plage Žplague.
Hajel Bergel Mehte Zammer Ronster Aket Wirst Zarrer Trachsar Ocktas Altim Tirm Sorb Afnet Astier Birmul Dant Stoge Trogge Neuz Ellas Nelle Agurt Stee Auner Famto Fegal Bantse Dirten Ogten Fame ¨ Gimmer Nuser Dollum Nahmel Beiner Paut Waxe ¨ Saunt ¨ Lirsche Prat Gambi Enster Schor Badel Tor ¨ Trast Onsten Pegle Gimber Asto Sart Derse Sutter Gernte Minge
Plakat Žsignpost. Radio Žradio. Rahmen Žframe. Rede Žspeech. Regel Žrule. Roß Žhorse. Ruder Žrudder. Ruine Žruin. Scherz Žjoke. Schicht Žlayer. Schimmer Žshine. Schreiner Žjoiner. See Žsea. Soldat Žsoldier. Stimme Žvoice. Suchen Žto seek. Theater Žtheater. Ton Žsound. Torf Žturf. Tropen Žtropics. Verkehr Žtraffic. Wach Žawake. Wanne Žtub. Weich Žsoft. Wein Žwine. Wert Žvalue. Wurzel Žroot. Zwei Žtwo.
297
Ettes Jehr Fersti Uht Wansker Wente Sammer Begelt Vise Gerbst Ater Mirsche Hurz Werm Pfanst Frame Linge Relt Dast Dels Daget Penti Lund Barzt Barn Kauler Anjum Faumle
References w1x D.A. Balota, R.F. Lorch Jr., Depth of automatic spreading activation: Mediated priming effects in pronunciation but not in lexical decision, Journal of Experimental Psychology: Learning, Memory, and Cognition 12 Ž1986. 336–345. w2x S. Bentin, Event-related potentials, semantic processes, and expectancy factors in word recognition, Brain Lang. 31 Ž1987. 308– 327. w3x S. Bentin, G. McCarthy, C.C. Wood, Event-related potentials, lexical decision, and semantic priming, Electroencephalography and Clinical Neurophysiology 60 Ž1985. 343–355. w4x M. Besson, I. Fischler, T. Boaz, G. Raney, Effects of automatic associative activation on explicit and implicit memory tests, Journal of Experimental Psychology: Learning, Memory, and Cognition 18 Ž1992. 89–105. w5x M. Besson, F. Macar, An event-related potential analysis of incongruity in music and other non-linguistic contexts., Psychopysiology 24 Ž1987. 14–25. w6x C.M. Brown, P. Hagoort, The processing nature of the N400: Evidence from masked priming, J. Cogn. Neurosci. 5 Ž1993. 34–44. w7x D.J. Chwilla, C.M. Brown, P. Hagoort, The N400 as a function of the level of processing, Psychophysiology 32 Ž1995. 274–285. w8x A.M. Collins, E.F. Loftus, A spreading-activation theory of semantic processing, Psychol. Rev. 82 Ž1975. 407–428. w9x A.M.B. De Groot, Primed lexical decision: combined effects of the proportion of related prime–target pairs and stimulus-onset asynchrony of prime and target, The Quarterly Journal of Experimental Psychology Ž1984. 281–297.
298
M. Weisbrod et al.r CognitiÕe Brain Research 8 (1999) 289–298
w10x A.M.B. De Groot, The range of automatic spreading activation in word priming, Journal of Verbal Learning and Verbal Behavior 22 Ž1983. 417–436. w11x K. Den Heyer, K. Briand, Priming single digit numbers: automatic spread of activation dissipates as a function of semantic distance, Am. J. Psychol. 99 Ž1986. 315–339. w12x I. Fischler, P.A. Bloom, D.G. Childers, S.E. Roucos, N.W. Perry, Brain potentials related to stages of sentence verification, Psychophysiology 20 Ž1983. 400–409. w13x P.J. Holcomb, Automatic and attentional processing: An event-related brain potential analysis of semantic priming, Brain Lang. 35 Ž1988. 66–85. w14x P.J. Holcomb, Semantic priming an stimulus degradation: implications for the role of the N400 in language processing, Psychophysiology 30 Ž1993. 47–61. w15x P.J. Holcomb, H.J. Neville, Auditory and visual semantic priming in lexical decision: a comparison using event-related brain potentials, Language and Cognitive Processes 5 Ž1990. 281–312. w16x P.J. Holcomb, H.J. Neville, Natural speech processing: An analysis using event-related brain potentials, Psychobiology 19 Ž1991. 286– 300. w17x M. Kiefer, M. Weisbrod, I. Kern, S. Maier, M. Spitzer, Right hemisphere activation during indirect semantic priming: evidence from event-related potentials, Brain Lang. 64 Ž1998. 377–408. w18x S.A. Kounios, P.J. Holcomb, Structure and process in semantic memory: evidence from event-related potentials and reaction times, J. Exp. Psychol. Gen. 121 Ž1992. 460–480. w19x M. Kutas, S.A. Hillyard, Brain potentials during reading reflect word expectancy and semantic association, Nature 307 Ž1984. 1161–1163. w20x M. Kutas, S.A. Hillyard, An electrophysiological probe of incidential semantic association, J. Cogn. Neurosci. 1 Ž1989. 38–49. w21x M. Kutas, S.A. Hillyard, Event-related potentials to grammatical errors and semantic anomalies, Mem. Cogn. 11 Ž1983. 539–550. w22x M. Kutas, S.A. Hillyard, Reading senseless sentences: brain potentials reflect semantic incongruity, Science 207 Ž1980. 203–205. w23x M. Kutas, H.J. Neville, P.J. Holcomb, A preliminary comparison of the N400 response to semantic anomalies during reading, listening, and signing, Electroencephalography and Clinical Neurophysiology 39 Ž1987. 325–330, Supplement. w24x M. Kutas, C.K. Van Petten, Psycholinguistics electrified. Event-related brain potential investigations, in: M.A. Gernsbacher ŽEd.., Handbook of Psycholinguistics, Academic Press, New York, 1994, pp. 83–143. w25x W.C. McCallum, S.F. Farmer, P.V. Pocock, The effects of physical and semantic incongruities on auditory event-related potentials, Electroencephalography and Clinical Neurophysiology 59 Ž1984. 477– 488. w26x T.P. McNamara, J. Altaribba, Depth of spreading activation revisited: semantic mediated priming occurs in lexical decisions, Journal of Memory and Language 27 Ž1988. 545–559. w27x D.E. Meyer, R.W. Schvaneveldt, Facilitation in recognizing pairs of words: evidence of a dependence between retrieval operations, J. Exp. Psychol. Gen. 90 Ž1971. 227–234.
w28x J.H. Neely, Semantic priming and retrieval from lexical memory: roles of inhibitionless spreading activation and limited-capacity attention, J. Exp. Psychol. Gen. 106 Ž1977. 226–254. w29x J.H. Neely, Semantic priming effects in visual word recognition: A selective review of current findings and theories, in: D. Besner, G.W. Humphreys ŽEds.., Basic Progresses in Reading — Visual Word Recognition, Lawrence Erlbaum Associates, Hillsdale, NJ, 1991, pp. 264–333. w30x J.H. Neely, D.E. Keefe, Semantic context effects on visual word processing: A hybrid prospective–retrospective processing theory, in: G.H. Bower ŽEd.., The Psychology of Learning and Motivation: Advances in Research and Theory Vol. 24, Academic, New York, 1989, pp. 207–248. w31x H.J. Neville, M. Kutas, G. Chesney, A. Schmidt, Event-related brain potentials during the initial encoding and subsequent recognition memory of congruous and incongruous words, Journal of Memory and Language 25 Ž1986. 75–92. w32x A. Nigam, J.E. Hoffmann, R.F. Simons, N400 to semantically autonomous pictures and words, J. Cogn. Neurosci. 4 Ž1992. 15–22. w33x A.C. Nobre, G. McCarthy, Language-related ERPs: scalp distributions and modulation by word type and semantic priming, J. Cogn. Neurosci. 6 Ž1994. . w34x J.P. O’Halloran, R. Isenhart, C.A. Sandmann, L.S. Larkey, Brain responses to semantic anomaly in natural continuous speech, Int. J. Psychophysiol. 6 Ž1988. 243–254. w35x R. Oldfield, The assessment and analysis of handedness: The Edinburgh Inventory, Neuropsychologia 9 Ž1971. 97–113. w36x L. Osterhout, P.J. Holcomb, Event related potentials and language comprehension, in: M.D. Rugg, M.G.H. Coles ŽEds.., Electrophysiology of Mind, Oxford Psychology Series, Oxford Univ. Press, Oxford, 1997, pp. 171–215. w37x M.I. Posner, C.R.R. Snyder, Attention and cognitive control, in: R.L. Solso ŽEd.., Information Processing and Cognition: The Loyota Symposium, Lawrence Erlbaum Associates, Hillsdale, 1975, pp. w38x M.D. Rugg, The effects of semantic priming and word repetition on event-related potentials, Psychophysiology 22 Ž1985. 642–647. w39x H.V. Semlitsch, P. Anderer, P. Schuster, O. Presslich, A solution for reliable and valid reduction of ocular artefacts applied to the P300 ERP, Psychophysiology 23 Ž1986. 695–703. w40x M. Spitzer, U. Braun, L. Hermle, S. Maier, Associative semantic network dysfunction in thought-disordered schizophrenic patients: Direct evidence from indirect semantic priming, Biol. Psychiatry 34 Ž1993. 864–877. w41x M. Spitzer, U. Braun, S. Maier, L. Hermle, B.A. Maher, Indirect semantic priming in schizophrenic patients, Schizophr. Res. 11 Ž1993. 71–80. w42x M. Spitzer, S. Winkler, S. Maier, M. Weisbrod, Ereigniskorrelierte Potentiale bei semantischen Sprachverarbeitungsprozessen schizophrener Patienten, Nervenarzt 68 Ž1997. 212–225. w43x C. Van Petten, M. Kutas, Ambiguous words in context: An event-related potential analysis of the time course of meaning activation, Journal of Memory and Language 26 Ž1987. 188–208. w44x C. Van Petten, M. Kutas, Influences of semantic and context on open- and closed-class words, Mem. Cogn. 19 Ž1991. 95–112.