- Email: [email protected]
Human brain potentials to reading syntactic errors in sentences of different complexity
Neuroscience Letters 235 (1997) 105–108
Human brain potentials to reading syntactic errors in sentences of different complexity Thomas F. Mu¨nte a , b ,*, Andras Szentkuti a, Bernardina M. Wieringa a, Mike Matzke a, So¨nke Johannes a b
a Department of Neurology, Med. Hochschule Hannover, Hannover, Germany Department of Cognitive Science, University of California San Diego, La Jolla, CA 92093-0515, USA
Received 18 June 1997; received in revised form 19 September 1997; accepted 19 September 1997
Abstract In order to determine if an event-related brain potential (ERP) effect described for syntactic violations (P600/SPS) varies with the amount of reprocessing entailed by a violation, number incongruencies were presented either within simple declarative or within subordinate clauses. ERPs were recorded while 12 German subjects read the stimulus materials presented word by word on a video monitor. The ERPs showed a P600/SPS effect for all sentence types, which was smallest in amplitude and earliest in latency for simple declarative sentences. This effect therefore qualifies as a metric for the amount and timing of syntactic reprocessing entailed by a syntactic error. In addition, a late frontal negativity (1000–1400 ms range) was found for the simple declarative sentences. 1997 Elsevier Science Ireland Ltd.
Keywords: Event-related potentials; Syntactic violation; N400; P600
One of the core areas of cognitive neuroscience is the investigation of the cognitive architecture and the neural structures subserving language processing. In particular, it is of interest if distinctions between different levels of processing (e.g. semantic, syntactic, pragmatic) made by linguists [3,13] are also reflected by the organization of the brain and its responses. One method that has been successfully applied to language research is the recording of event-related brain potentials (ERPs) [6]. For example, when semantically incongruous words are presented visually or auditorily, these words give rise to a negative potential with a peak latency of about 400 ms and a maximum over central scalp points relative to the semantically correct words (N400 component). The N400 has been extremely valuable for the investigation of semantic aspects of language [6] and has fueled the search for ERP effects to other aspects of language processing, such as morphology [12] and syntax [4,5,7–11]. With regard to the latter, two findings are most * Corresponding author. Fax: +1 619 5341128; e-mail: [email protected]
consistent: (1) a parietal positivity with an onset of about 500 ms that has been observed to a variety of syntactic errors such as agreement violations and phrase structure violations [1,4,5,7,9–11] and has been termed the P600 [10] or syntactic positive shift (SPS) [5] and (2) a left frontal negativity in an earlier latency range (250–500 ms) [1,4,8,9]. The left frontal negativity, present only in some of the published studies, has been proposed to reflect failure of the first pass of the parser [4]. For the positivity it has been suggested that it reflects aspects of reanalysis required by a syntactic incongruency or ambiguity [4,9,11] and that its amplitude is indicative of the cost of reprocessing [11] and its latency and duration a function of onset and duration of parsing processes [4]. Here we present evidence for a sensitivity of the P600/ SPS to syntactic complexity by presenting number incongruencies (for verbs) either in simple declarative clauses or in subordinate clauses. We hypothesized that the amplitude of the SPS effect should be greater for errors in subordinate clauses, as the interdependencies between the sentence constituents are more complex and syntactic reanalysis there-
0304-3940/97/$17.00 1997 Elsevier Science Ireland Ltd. All rights reserved PII S0304- 3940(97) 00719- 2
106
T.F. Mu¨nte et al. / Neuroscience Letters 235 (1997) 105–108
fore presumably more difficult. Similarly, it was predicted that the onset and peak latencies of the SPS should be delayed for subordinate clause conditions. Twelve healthy, right-handed native speakers of German (subjects: five women, aged 23–29 years) gave informed consent to participate. A total of 480 stimulus units, each comprising 17 German words and two or three sentences, were constructed. Half of the units contained a number mismatch for a verb, while the other half was syntactically correct. The number mismatch occurred either in a simple declarative sentence (condition simple), in an embedded subordinate clause (condition embedded), or in a right branching relative clause being the terminal word of the sentence (condition terminal). Some examples are shown below. simple Der Opa hat zwei Maikaefer gefunden. Sie *brummt/ brummen beim Fliegen laut. Er zeigt die Tiere seinen Enkeln. The grandfather has found two june bugs. They *hums/ hum loudly when flying. He shows the animals to his grandchildren. embedded Zwei Maikaefer, die beim Fliegen laut *brummt/brummen, hat der Opa gefunden. Er zeigt die Tiere seinen Enkeln. The grandfather two june bugs, which *hums/hum loud when flying, has found. He shows the animals to his grandchildren. terminal Der Opa hat zwei Maikaefer gefunden, die beim Fliegen laut *brummt/brummen. Er zeigt die Tiere seinen Enkeln. The grandfather has found two june bugs, which *hums/hum loudly when flying. He shows the animals to his grandchildren. (English translations preserve word order, critical words underlined) Subjects were presented with a total of 480 units (80 in each condition, random order). Terminal and embedded conditions were distinguished in order to be able to differentiate effects of complexity and position on syntactic incongruencies. To ensure reading, subjects had to fill out a questionnaire after the recording session. Each unit was seen only once by a subject but was presented in all six conditions across subjects, as six scenarios were constructed. The stimuli were presented one word at a time (duration 300 ms, onset asynchrony 700 ms, height 0.8°, width 1.5–4.3° (visual angle) in yellow letters against a blue background. The interval between sentences was 2500 ms. The EEG was recorded with tin-electrodes mounted in an Electro-Cap from all standard sites of the 10/20 system plus 10 additional intermediate sites with reference electrodes placed on both mastoid processes (algebraic mean). The electrooculogram was recorded for
artifact rejection purposes. Biosignals were amplified (bandpass 0.01–100 Hz) and digitized with 4 ms resolution. After artifact correction for eye blinks using a spatial filter [2] averages were obtained for 2048 ms epochs beginning 200 ms prior to presentation of the stimuli. Waveforms were quantified by mean-amplitude measures. These measures were subjected to repeated analyses of variance (ANOVA) with the Greenhouse–Geisser correction applied as necessary. Reported P-values are corrected. In the grand average ERPs for the simple condition (Fig. 1) the response to each word is characterized by an initial negativity at 100 ms followed by a positivity with a latency of about 230 ms. ERPs are timelocked to the critical verb. Two effects are apparent for incorrect words: (1) a positivity with an onset latency of about 500 ms and a peak latency of about 800 ms that is maximal over parietal scalp regions (P600/SPS), and (2) an anterior negativity occurring actually during the response to the next word and having an onset of 1000 ms to the number incongruency. In keeping with previous experiments there was a slight left frontal negativity in the 300–500 ms range to the number incongruencies, which proved to be statistically unreliable. The ERPs from the three different conditions are contrasted in Fig. 2A. The effects to the number incongruencies in subordinate clauses clearly differed from the simple declarative case. Both, the terminal and the embedded condition showed a larger amplitude for the SPS compared to the simple condition (see also Fig. 2C). Also, both of these conditions lack the frontal negativity in the 1000–1400 ms
Fig. 1. Grand average ERPs (n = 12) to the critical words in the simple condition. Parietally a positivity (P600/SPS) emerges in response to the incongruent stimuli while over frontal scalp regions a negativity is observed.
T.F. Mu¨nte et al. / Neuroscience Letters 235 (1997) 105–108
range observed in the simple condition. When quantified by a 800–1200 ms mean amplitude measure for the parietooccipital electrodes (Cp1/2, P3/4, O1/2, Pz), no main effect of condition was found (F(2,22) = 0.38, n.s.). The effect of number incongruency was highly significant (F(1,11) = 16.94, P , 0.002) as well the interaction between condition and number incongruency (F(2,22) = 8.12, P , 0.01) reflecting the smaller positivity in the simple condition. When each condition was tested individually, an incongruency effect was revealed in all cases (simple: F(1,11) = 6.30, P , 0.03; embedded: F(1,11) = 7.39, P , 0.02; terminal: F(1,11) = 11.51, P , 0.006). The frontal negativity was quantified by a 1000–1400 ms mean amplitude measure for the frontal electrode set (Fp1/2, F3/4, Fc1/2, Fz). No main effects of condition or number incongruency were attained (both F , 1, n.s.). There was an interaction between condition and number incongruency (F(2,22) = 5.41, P , 0.05). Posthoc testing of each of the conditions revealed that this interaction was due to the exclusive presence of the negativity in the simple condition (simple F(1,11) = 9.19, P , 0.012; embedded F(1,11) = 0.002, n.s.; terminal F(1,11) = 0.01, n.s.). While the difference waves (ERPs to correct minus ERPs to incorrect stimuli; Fig. 2B, Pz electrode) show a similar onset latency of the three conditions, the peak latency of the SPS was earliest in the simple condition and latest in the embedded condition (peak latency measure after 3.0 low-pass filter: simple: 824 ms; embedded: 939 ms; terminal: 909 ms, F(2,22) = 2.52, P = 0.1, fractional area latency, 50% area 700–1200 ms window, simple: 808 ms;
107
embedded: 896 ms; terminal: 884 ms; F(2,22) = 4.61, P , 0.05). The main goal of the present experiment was to test predictions derived from the syntactic reanalysis account of the P600/SPS [4,9,11]. These predictions are borne out: the easiest condition (simple) in which the number incongruencies were presented in simple declarative sentences had the smallest P600/SPS response with respect to the two conditions that contained errors within the subordinate clause. As the positions of the errors within the stimulus units as well as the content of the units had been carefully matched across conditions, the difficulty of the syntactic analysis appears the only reason for the amplitude effect on the P600/SPS. In a simple declarative sentence there is usually only one candidate word against which the verb has to be matched, whereas in subordinate clauses several noun phrases have to be checked. Likewise, the peak latency of the P600/SPS appeared to vary with the complexity of the sentences with the earliest latency found for the simple condition. ERP data obtained during the processing of relative clauses support the notion of a latency variability of the P600/SPS as a function of difficulty [7,8]. The second effect that has been often found in response to syntactic errors, the left anterior negativity in the 300–500 ms range, was virtually absent in the present set of data. Coulson et al. [1] showed that this effect varies with the salience of the incongruency. The rather unobtrusive nature of the number mismatches used here might therefore explain the lack of an early left anterior negativity.
Fig. 2. (A) Comparison of the three conditions for the midline electrodes. Only the simple condition is associated with a late frontal negativity. The parietal positivity is more pronounced in the ERPs from the embedded and terminal conditions. (B) Difference waveforms for the critical stimuli obtained by subtracting the ERPs to the incorrect words from the ERPs to the correct words, Pz electrode. The positivity in response to number incongruencies is largest in the terminal and smallest in the simple condition. (C) Mean difference amplitude (correct − incorrect, mean amplitude 900–1000 ms) for the midline sites. The effect of number incongruency is smallest in the simple condition.
108
T.F. Mu¨nte et al. / Neuroscience Letters 235 (1997) 105–108
An unexpected finding was an anterior negativity that occurred in the 1000–1400 ms range after the number incongruency only in the simple condition. Previous studies that used syntactic incongruencies have found negativities in the sense of an N400 for words following the error, which has been explained by the fact that these words were rendered semantically incongruous by the foregoing syntactic error. However, the N400 response in general has a very different scalp distribution, being maximal at central and parietal sites and more pronounced over the right hemisphere [6], from the negativity seen for the simple condition. A simple means to decide, whether the negativity actually occurs in an effort to integrate the word following the error with the preceding context, or whether the effect is still related to the processing of the actual error, would be to manipulate the intervals between the words. In the first case the effect should move with changing intervals, while in the second case it should not. Such experiments are underway in our laboratory. The technical assistance of J. Kilian is gratefully acknowledged. This work was supported by a grant from the Hermann and Lilly Schilling Foundation to T.F.M. [1] Coulson, S., King, J.W. and Kutas, M., Expect the unexpected: event-related brain response to morphosyntactic violations, Lang. Cogn. Proc., (1997) in press. [2] Dale, A.M., Source localization and spatial discriminant analysis of event-related potentials: linear approaches. Dissertation. University of California, San Diego, CA, 1994. [3] Frazier, L., Sentence processing: a tutorial review. In M. Coltheart (Ed.), Attention and Performance XII, Erlbaum, Hillsdale, NJ, 1987, pp. 559–586.
[4] Friederici, A.D., Hahne, A. and Mecklinger, A., The temporal structure of syntactic parsing: early and late effects elicited by syntactic anomalies, J. Exp. Psychol. Learn. Mem. Cogn., 22 (1996) 1219–1248. [5] Hagoort, P., Brown, C. and Groothusen, J., The syntactic positive shift (SPS) as an ERP measure of syntactic processing, Lang. Cogn. Proc., 8 (1993) 439–483. [6] Kutas, M. and Van Petten, C., Event related brain potential studies of language. In P.K. Ackles, J.R. Jennings and M.G.H. Coles (Eds.), Advances in Psychophysiology 3, JAI Press, Greenwich, CT, 1988, pp. 129–187. [7] Mecklinger, A., Schriefers, H., Steinhauer, K. and Friederici, A.D., Processing relative clauses varying on syntactic and semantic dimensions: an analysis with event-related brain potentials, Mem. Cogn., 23 (1995) 477–494. [8] Mu¨nte, T.F. and Heinze, H.J., ERP negativities to syntactic violations in written text. In H.J. Heinze, T.F. Mu¨nte and G.R. Mangun (Eds.), Cognitive Electrophysiology, Birkha¨user, Boston, 1994, pp. 211–238. [9] Mu¨nte, T.F., Matzke, M. and Johannes, S., Brain activity associated with syntactic incongruencies in words and pseudowords, J. Cogn. Neurosci., 9 (1997) 300–311. [10] Osterhout, L. and Holcomb, P.J., Event-related brain potentials elicited by syntactic anomaly, J. Mem. Lang., 31 (1992) 785– 806. [11] Osterhout, L., Holcomb, P.J. and Swinney, D.A., Brain potentials elicited by garden path sentences: evidence of the application of verb information during parsing, J. Exp. Psychol. Learn. Mem. Cogn., 20 (1994) 768–803. [12] Penke, M., Weyerts, H., Gross, M., Zander, E., Mu¨nte, T.F. and Clahsen, H., How the brain processes complex words: an ERP study of German verb inflection, Cogn. Brain Res., (1997) in press. [13] Taraban, R. and McClelland, J.L., Constituent attachment and thematic role assignment in sentence processing: influences of content-based expectations, J. Mem. Lang., 27 (1988) 597– 632.
Human brain potentials to reading syntactic errors in sentences of different complexity Thomas F. Mu¨nte a , b ,*, Andras Szentkuti a, Bernardina M. Wieringa a, Mike Matzke a, So¨nke Johannes a b
a Department of Neurology, Med. Hochschule Hannover, Hannover, Germany Department of Cognitive Science, University of California San Diego, La Jolla, CA 92093-0515, USA
Received 18 June 1997; received in revised form 19 September 1997; accepted 19 September 1997
Abstract In order to determine if an event-related brain potential (ERP) effect described for syntactic violations (P600/SPS) varies with the amount of reprocessing entailed by a violation, number incongruencies were presented either within simple declarative or within subordinate clauses. ERPs were recorded while 12 German subjects read the stimulus materials presented word by word on a video monitor. The ERPs showed a P600/SPS effect for all sentence types, which was smallest in amplitude and earliest in latency for simple declarative sentences. This effect therefore qualifies as a metric for the amount and timing of syntactic reprocessing entailed by a syntactic error. In addition, a late frontal negativity (1000–1400 ms range) was found for the simple declarative sentences. 1997 Elsevier Science Ireland Ltd.
Keywords: Event-related potentials; Syntactic violation; N400; P600
One of the core areas of cognitive neuroscience is the investigation of the cognitive architecture and the neural structures subserving language processing. In particular, it is of interest if distinctions between different levels of processing (e.g. semantic, syntactic, pragmatic) made by linguists [3,13] are also reflected by the organization of the brain and its responses. One method that has been successfully applied to language research is the recording of event-related brain potentials (ERPs) [6]. For example, when semantically incongruous words are presented visually or auditorily, these words give rise to a negative potential with a peak latency of about 400 ms and a maximum over central scalp points relative to the semantically correct words (N400 component). The N400 has been extremely valuable for the investigation of semantic aspects of language [6] and has fueled the search for ERP effects to other aspects of language processing, such as morphology [12] and syntax [4,5,7–11]. With regard to the latter, two findings are most * Corresponding author. Fax: +1 619 5341128; e-mail: [email protected]
consistent: (1) a parietal positivity with an onset of about 500 ms that has been observed to a variety of syntactic errors such as agreement violations and phrase structure violations [1,4,5,7,9–11] and has been termed the P600 [10] or syntactic positive shift (SPS) [5] and (2) a left frontal negativity in an earlier latency range (250–500 ms) [1,4,8,9]. The left frontal negativity, present only in some of the published studies, has been proposed to reflect failure of the first pass of the parser [4]. For the positivity it has been suggested that it reflects aspects of reanalysis required by a syntactic incongruency or ambiguity [4,9,11] and that its amplitude is indicative of the cost of reprocessing [11] and its latency and duration a function of onset and duration of parsing processes [4]. Here we present evidence for a sensitivity of the P600/ SPS to syntactic complexity by presenting number incongruencies (for verbs) either in simple declarative clauses or in subordinate clauses. We hypothesized that the amplitude of the SPS effect should be greater for errors in subordinate clauses, as the interdependencies between the sentence constituents are more complex and syntactic reanalysis there-
0304-3940/97/$17.00 1997 Elsevier Science Ireland Ltd. All rights reserved PII S0304- 3940(97) 00719- 2
106
T.F. Mu¨nte et al. / Neuroscience Letters 235 (1997) 105–108
fore presumably more difficult. Similarly, it was predicted that the onset and peak latencies of the SPS should be delayed for subordinate clause conditions. Twelve healthy, right-handed native speakers of German (subjects: five women, aged 23–29 years) gave informed consent to participate. A total of 480 stimulus units, each comprising 17 German words and two or three sentences, were constructed. Half of the units contained a number mismatch for a verb, while the other half was syntactically correct. The number mismatch occurred either in a simple declarative sentence (condition simple), in an embedded subordinate clause (condition embedded), or in a right branching relative clause being the terminal word of the sentence (condition terminal). Some examples are shown below. simple Der Opa hat zwei Maikaefer gefunden. Sie *brummt/ brummen beim Fliegen laut. Er zeigt die Tiere seinen Enkeln. The grandfather has found two june bugs. They *hums/ hum loudly when flying. He shows the animals to his grandchildren. embedded Zwei Maikaefer, die beim Fliegen laut *brummt/brummen, hat der Opa gefunden. Er zeigt die Tiere seinen Enkeln. The grandfather two june bugs, which *hums/hum loud when flying, has found. He shows the animals to his grandchildren. terminal Der Opa hat zwei Maikaefer gefunden, die beim Fliegen laut *brummt/brummen. Er zeigt die Tiere seinen Enkeln. The grandfather has found two june bugs, which *hums/hum loudly when flying. He shows the animals to his grandchildren. (English translations preserve word order, critical words underlined) Subjects were presented with a total of 480 units (80 in each condition, random order). Terminal and embedded conditions were distinguished in order to be able to differentiate effects of complexity and position on syntactic incongruencies. To ensure reading, subjects had to fill out a questionnaire after the recording session. Each unit was seen only once by a subject but was presented in all six conditions across subjects, as six scenarios were constructed. The stimuli were presented one word at a time (duration 300 ms, onset asynchrony 700 ms, height 0.8°, width 1.5–4.3° (visual angle) in yellow letters against a blue background. The interval between sentences was 2500 ms. The EEG was recorded with tin-electrodes mounted in an Electro-Cap from all standard sites of the 10/20 system plus 10 additional intermediate sites with reference electrodes placed on both mastoid processes (algebraic mean). The electrooculogram was recorded for
artifact rejection purposes. Biosignals were amplified (bandpass 0.01–100 Hz) and digitized with 4 ms resolution. After artifact correction for eye blinks using a spatial filter [2] averages were obtained for 2048 ms epochs beginning 200 ms prior to presentation of the stimuli. Waveforms were quantified by mean-amplitude measures. These measures were subjected to repeated analyses of variance (ANOVA) with the Greenhouse–Geisser correction applied as necessary. Reported P-values are corrected. In the grand average ERPs for the simple condition (Fig. 1) the response to each word is characterized by an initial negativity at 100 ms followed by a positivity with a latency of about 230 ms. ERPs are timelocked to the critical verb. Two effects are apparent for incorrect words: (1) a positivity with an onset latency of about 500 ms and a peak latency of about 800 ms that is maximal over parietal scalp regions (P600/SPS), and (2) an anterior negativity occurring actually during the response to the next word and having an onset of 1000 ms to the number incongruency. In keeping with previous experiments there was a slight left frontal negativity in the 300–500 ms range to the number incongruencies, which proved to be statistically unreliable. The ERPs from the three different conditions are contrasted in Fig. 2A. The effects to the number incongruencies in subordinate clauses clearly differed from the simple declarative case. Both, the terminal and the embedded condition showed a larger amplitude for the SPS compared to the simple condition (see also Fig. 2C). Also, both of these conditions lack the frontal negativity in the 1000–1400 ms
Fig. 1. Grand average ERPs (n = 12) to the critical words in the simple condition. Parietally a positivity (P600/SPS) emerges in response to the incongruent stimuli while over frontal scalp regions a negativity is observed.
T.F. Mu¨nte et al. / Neuroscience Letters 235 (1997) 105–108
range observed in the simple condition. When quantified by a 800–1200 ms mean amplitude measure for the parietooccipital electrodes (Cp1/2, P3/4, O1/2, Pz), no main effect of condition was found (F(2,22) = 0.38, n.s.). The effect of number incongruency was highly significant (F(1,11) = 16.94, P , 0.002) as well the interaction between condition and number incongruency (F(2,22) = 8.12, P , 0.01) reflecting the smaller positivity in the simple condition. When each condition was tested individually, an incongruency effect was revealed in all cases (simple: F(1,11) = 6.30, P , 0.03; embedded: F(1,11) = 7.39, P , 0.02; terminal: F(1,11) = 11.51, P , 0.006). The frontal negativity was quantified by a 1000–1400 ms mean amplitude measure for the frontal electrode set (Fp1/2, F3/4, Fc1/2, Fz). No main effects of condition or number incongruency were attained (both F , 1, n.s.). There was an interaction between condition and number incongruency (F(2,22) = 5.41, P , 0.05). Posthoc testing of each of the conditions revealed that this interaction was due to the exclusive presence of the negativity in the simple condition (simple F(1,11) = 9.19, P , 0.012; embedded F(1,11) = 0.002, n.s.; terminal F(1,11) = 0.01, n.s.). While the difference waves (ERPs to correct minus ERPs to incorrect stimuli; Fig. 2B, Pz electrode) show a similar onset latency of the three conditions, the peak latency of the SPS was earliest in the simple condition and latest in the embedded condition (peak latency measure after 3.0 low-pass filter: simple: 824 ms; embedded: 939 ms; terminal: 909 ms, F(2,22) = 2.52, P = 0.1, fractional area latency, 50% area 700–1200 ms window, simple: 808 ms;
107
embedded: 896 ms; terminal: 884 ms; F(2,22) = 4.61, P , 0.05). The main goal of the present experiment was to test predictions derived from the syntactic reanalysis account of the P600/SPS [4,9,11]. These predictions are borne out: the easiest condition (simple) in which the number incongruencies were presented in simple declarative sentences had the smallest P600/SPS response with respect to the two conditions that contained errors within the subordinate clause. As the positions of the errors within the stimulus units as well as the content of the units had been carefully matched across conditions, the difficulty of the syntactic analysis appears the only reason for the amplitude effect on the P600/SPS. In a simple declarative sentence there is usually only one candidate word against which the verb has to be matched, whereas in subordinate clauses several noun phrases have to be checked. Likewise, the peak latency of the P600/SPS appeared to vary with the complexity of the sentences with the earliest latency found for the simple condition. ERP data obtained during the processing of relative clauses support the notion of a latency variability of the P600/SPS as a function of difficulty [7,8]. The second effect that has been often found in response to syntactic errors, the left anterior negativity in the 300–500 ms range, was virtually absent in the present set of data. Coulson et al. [1] showed that this effect varies with the salience of the incongruency. The rather unobtrusive nature of the number mismatches used here might therefore explain the lack of an early left anterior negativity.
Fig. 2. (A) Comparison of the three conditions for the midline electrodes. Only the simple condition is associated with a late frontal negativity. The parietal positivity is more pronounced in the ERPs from the embedded and terminal conditions. (B) Difference waveforms for the critical stimuli obtained by subtracting the ERPs to the incorrect words from the ERPs to the correct words, Pz electrode. The positivity in response to number incongruencies is largest in the terminal and smallest in the simple condition. (C) Mean difference amplitude (correct − incorrect, mean amplitude 900–1000 ms) for the midline sites. The effect of number incongruency is smallest in the simple condition.
108
T.F. Mu¨nte et al. / Neuroscience Letters 235 (1997) 105–108
An unexpected finding was an anterior negativity that occurred in the 1000–1400 ms range after the number incongruency only in the simple condition. Previous studies that used syntactic incongruencies have found negativities in the sense of an N400 for words following the error, which has been explained by the fact that these words were rendered semantically incongruous by the foregoing syntactic error. However, the N400 response in general has a very different scalp distribution, being maximal at central and parietal sites and more pronounced over the right hemisphere [6], from the negativity seen for the simple condition. A simple means to decide, whether the negativity actually occurs in an effort to integrate the word following the error with the preceding context, or whether the effect is still related to the processing of the actual error, would be to manipulate the intervals between the words. In the first case the effect should move with changing intervals, while in the second case it should not. Such experiments are underway in our laboratory. The technical assistance of J. Kilian is gratefully acknowledged. This work was supported by a grant from the Hermann and Lilly Schilling Foundation to T.F.M. [1] Coulson, S., King, J.W. and Kutas, M., Expect the unexpected: event-related brain response to morphosyntactic violations, Lang. Cogn. Proc., (1997) in press. [2] Dale, A.M., Source localization and spatial discriminant analysis of event-related potentials: linear approaches. Dissertation. University of California, San Diego, CA, 1994. [3] Frazier, L., Sentence processing: a tutorial review. In M. Coltheart (Ed.), Attention and Performance XII, Erlbaum, Hillsdale, NJ, 1987, pp. 559–586.
[4] Friederici, A.D., Hahne, A. and Mecklinger, A., The temporal structure of syntactic parsing: early and late effects elicited by syntactic anomalies, J. Exp. Psychol. Learn. Mem. Cogn., 22 (1996) 1219–1248. [5] Hagoort, P., Brown, C. and Groothusen, J., The syntactic positive shift (SPS) as an ERP measure of syntactic processing, Lang. Cogn. Proc., 8 (1993) 439–483. [6] Kutas, M. and Van Petten, C., Event related brain potential studies of language. In P.K. Ackles, J.R. Jennings and M.G.H. Coles (Eds.), Advances in Psychophysiology 3, JAI Press, Greenwich, CT, 1988, pp. 129–187. [7] Mecklinger, A., Schriefers, H., Steinhauer, K. and Friederici, A.D., Processing relative clauses varying on syntactic and semantic dimensions: an analysis with event-related brain potentials, Mem. Cogn., 23 (1995) 477–494. [8] Mu¨nte, T.F. and Heinze, H.J., ERP negativities to syntactic violations in written text. In H.J. Heinze, T.F. Mu¨nte and G.R. Mangun (Eds.), Cognitive Electrophysiology, Birkha¨user, Boston, 1994, pp. 211–238. [9] Mu¨nte, T.F., Matzke, M. and Johannes, S., Brain activity associated with syntactic incongruencies in words and pseudowords, J. Cogn. Neurosci., 9 (1997) 300–311. [10] Osterhout, L. and Holcomb, P.J., Event-related brain potentials elicited by syntactic anomaly, J. Mem. Lang., 31 (1992) 785– 806. [11] Osterhout, L., Holcomb, P.J. and Swinney, D.A., Brain potentials elicited by garden path sentences: evidence of the application of verb information during parsing, J. Exp. Psychol. Learn. Mem. Cogn., 20 (1994) 768–803. [12] Penke, M., Weyerts, H., Gross, M., Zander, E., Mu¨nte, T.F. and Clahsen, H., How the brain processes complex words: an ERP study of German verb inflection, Cogn. Brain Res., (1997) in press. [13] Taraban, R. and McClelland, J.L., Constituent attachment and thematic role assignment in sentence processing: influences of content-based expectations, J. Mem. Lang., 27 (1988) 597– 632.