BRAIN AND LANGUAGE 62, 311–341 (1998)
ARTICLE NO. BL971906

First-Pass versus Second-Pass Parsing Processes in a
Wernicke’s and a Broca’s Aphasic: Electrophysiological
Evidence for a Double Dissociation
Angela D. Friederici, Anja Hahne, and D. Yves von Cramon
Max-Planck-Institute of Cognitive Neuroscience
The present paper is a first attempt to integrate the classical brain lesion behavioral
impairment approach of functional neuroanatomy and the electrophysiological brain
mapping approach in the domain of syntactic processing. In a group of normal agematched controls we identified three electrophysiological components previously
observed in correlation with language comprehension processes: an early left anterior negativity normally seen in correlation with syntactic first-pass parsing processes (ELAN), a centroparietal negativity seen in correlation with processes of
lexical-semantic integration (N400), and a late centroparietal positivity observed in
correlation with secondary syntactic processes of reanalysis and repair (P600). The
early left anterior negativity was absent in a patient with an extended lesion in the
anterior part of the left hemisphere sparing the temporal lobe, although the late
centroparietal positivity and the centroparietal N400 were present. In a patient with
a left temporal-parietal lesion the early left anterior negativity was found to be
present, whereas the N400 component was absent. These findings suggest that firstpass parsing and secondary processes are subserved by distinct brain systems.
 1998 Academic Press

INTRODUCTION

The debate concerning the underlying deficit in Broca’s aphasia is still
ongoing. In the mid 1980s it became clear that most Broca patients although
agrammatic in their output as well as in their comprehension1 possessed a
The research reported here was supported by grants from the Deutsche Forschungsgemeinschaft (Fr 519/12-2) and the Alfried Krupp von Bohlen-und Halbach Prize awarded to
A.F. Portions of these data were presented at the 33. Academy of Aphasia Conference, 1995,
San Diego, CA. We thank Erdmut Pfeifer for his support in software and statistical matters
and Doug Saddy for his help in language matters.
Address correspondence and reprint requests to Angela D. Friederici, Max-Planck-Institute
of Cognitive Neuroscience, Inselstrasse 22-26, 04103 Leipzig, Germany. Fax: ⫹⫹49/(0)3 41/
99 40-113. E-mail: angelafr@cns.mpg.de.
1
There are, however, rare cases with agrammatic output but intact comprehension (Miceli,
Menn, & Goodglass, 1983).
311
0093-934X/98 $25.00
Copyright  1998 by Academic Press
All rights of reproduction in any form reserved.

312

FRIEDERICI, HAHNE, AND VON CRAMON

surprising amount of grammatical knowledge (Linebarger, Schwarz, & Saffran, 1983; Wulfeck, 1988; Friederici & Kilborn, 1989). Subsequent research
tried to specify the deficit in terms of a computational deficiency. While
some researchers assumed that reduced memory capacities were responsible
for the observed agrammatic comprehension (Kolk & van Grunsven, 1985)
others hypothesized that the deficit was a reflection of an alteration in the
temporal structure of either access or integration processes (Friederici, 1988;
Hagoort, 1990; Prather, Zurif, Stern, & Rosen, 1992).
In one of these proposals (Friederici, 1988, 1995) the temporal structure
of language processing is assumed to be altered due to a slower access to
specifically syntactic information in a two stage parsing model. Such a two
stage parsing mechanism had been proposed by Frazier (Frazier, 1978; Frazier & Fodor, 1978) to describe normal parsing. She assumed a first stage
during which the parser assigns an initial syntactic structure to the input and
a second stage during which thematic role assignment takes place. The initial
first-pass parsing stage supposedly operates quite fast and automatic mainly
on the basis of word category information building local phrase structure,
whereas the second stage during which syntactic and lexical-semantic information are mapped onto each other for interpretation may involve controlled
processes as well (Gorrell, 1995). Under these assumptions, slower access
to syntactic information results in an inefficient first-pass parse.
Although there is some debate as to whether the initial stage is purely
structure driven (Frazier, 1978; Ferreira & Clifton, 1986; Gorrell, 1995) or
may also consider lexically encoded information such as verb-argument
structure (McClelland, St. John, & Taraban, 1989; Shapiro, Nagel, & Levine,
1993) we will assume an initial stage of structure building that is primarily
structure driven and working in a highly automatic fashion. This stage may
guarantee the ease with which normals parse the incoming linguistic information. If this initial processing stage is slowed down efficient on-line parsing
may no longer be possible.
This assumption was supported by behavioral studies showing that Broca’s aphasics with comprehension problems are delayed in their on-line syntactic processes (Friederici, 1985; Friederici & Kilborn, 1989). It was demonstrated that Broca’s aphasics’ performance improved when more time was
given to process the relevant information, for example, by increasing the
interstimulus interval between the sentence prime and the target in a syntactic
priming task (Haarmann & Kolk, 1990). Additional support for the view that
these patients may have lost the automaticity of the syntactic processes came
from a syntactic priming study in which normal controls and Broca’s aphasics were required to ignore the sentence prime. Broca’s aphasics who did
demonstrate a syntactic priming effect with a considerable delay under normal priming conditions were able to attentionally ignore the prime whereas
normal controls were not (Kilborn & Friederici, 1994). This strongly suggests that Broca’s aphasics, in contrast to normals, process syntactic informa-

FIRST-PASS VERSUS SECOND-PASS PARSING

313

tion in a controlled way. On the basis of these findings the hypothesis was
formulated that the Broca’s area and adjacent areas support the fast and automatic syntactic processes necessary for efficient on-line parsing (Friederici,
1995).
A different way of exploring the assumed language-brain relationship is
to register brain activity in a normal brain while presenting language stimuli.
From the different methods available such as SPECT, PET, fMRI, MEG,
and EEG only the latter two allow monitoring of language processes on line,
millisecond by millisecond. This may be the reason why most studies on
syntactic processing have used event-related brain potential (ERP) measurements.2 From an overview of recent ERP studies focusing on syntactic processes in normals it is evident that there are two components which are observed in correlation with syntactic parsing, an early left anterior negativity
(ELAN) present around 200 ms or later after the critical stimulus and a late
positive component present around 600 ms (P600) after the critical stimulus
(Friederici, 1995).
The Early Negative ERP Component
The early negativity which has its localized maximum over the anterior
part of the left hemisphere was observed in correlation with the processing
of phrase structure errors (Neville, Nicol, Barss, Forster, & Garrett, 1991;
Friederici, Pfeifer, & Hahne, 1993; Münte, Heinze, & Mangun, 1993; Friederici, Hahne, & Mecklinger, 1996) as well as with the processing of function words in correct sentences (Neville, Mills, & Lawson, 1992; Nobre &
McCarthy, 1994). An ERP component sensitive to syntactic word class
around 200 ms was also reported by Samar and Berent (1986) in a principle
component analysis also suggesting early syntactic operations.3 A left anterior negativity that emerged around 400 ms was observed in correlation with
errors of inflectional morphology (Gunter, Vos, & Mulder, 1995; Coulson,
King, & Kutas, 1998; Penke, Weyers, Gross, Zander, Münte, & Clahsen,
1997). Thus the latency for this left anterior negativity seems to vary across
studies.
However, those studies which are comparable with respect to presentation
mode and stimulus material show consistent results. Both these factors may
influence the component’s latency. Let us first consider the presentation
mode. Note that the ERP is usually time locked to the word onset. Thus we
2
But see Mazoyer, Tzourio, Frak, Syrota, Murayama, Levrier, Salamon, Dehaene, Cohen,
and Mehler (1993) for a recent PET study including a syntactic condition in a variety of
conditions that were subtracted from each other and Stromswold, Caplan, Alpert, and Rauch
(1996) for a PET study which focuses on syntactic processing as such.
3
In this principle component analysis they identified one component at 140 ms that was
sensitive to syntactic phrase structure violations and one at 220 ms that was sensitive to word
category information.

314

FRIEDERICI, HAHNE, AND VON CRAMON

may not only expect general differences between the visual (holistic) and
auditory (left-to-right) presentation of the critical stimulus, but also between
auditorily presented words whose word category decision point (e.g., verb
vs. noun) appears in the word stem of the critical item (e.g., eat vs. meal )
and those whose word category decision point appears at the suffix (e.g.,
refined vs. refinement). It has been shown that in the auditory domain the
latency of the left anterior negativity observed for phrase structure violations
including the latter type of words is longer (Friederici, Hahne, & Mecklinger,
1996) than that for phrase structure violations including the former type of
words (Friederici, Pfeifer, & Hahne, 1993). Second, latency variations may
be influenced by the different types of syntactic violation, namely, phrase
structure violation on the one hand versus inflectional violation on the other,
as these differ functionally. According to Frazier (1978) first-pass parsing
processes consist of structure building processes on the basis of category
information, be it for lexical elements (e.g., N ⫽ noun, V ⫽ verb, etc.) or
for functional elements (e.g., INFL ⫽ inflection). This information is sufficient to build up a phrase structure tree. Lexical specification of these
categories [e.g., N → meal, refinement, etc.; INFL → -∅ (3. person
singular), -s (3. person plural)] is only required later during the process.
Therefore, violations of word category information should be detected earlier
than violations of agreement indicated by particular inflectional forms. Thus,
inflectional errors, in contrast to word category errors, lead to a left anterior
negativity around 400 ms. This negativity is followed by a late positivity in
some studies (in Dutch: Gunter, Vos, & Mulder, 1995; Gunter, Stowe,
& Mulder, 1997; in English: Coulson, King, & Kutas, 1998; in German:
Penke et al., 1997; but see Hagoort et al., 1993; and Osterhout & Mobley,
1995).
The Late Positive ERP Component
The late positivity was found in correlation with syntactic anomalies either
requiring a reanalysis as in garden-path sentences (Osterhout & Holcomb,
1992, 1993; Mecklinger, Schriefers, Steinhauer, & Friederici, 1995) or a repair as in syntactically incorrect sentences (Neville et al., 1991; Hagoort,
Brown, & Groothusen, 1993; Gunter, Vos, & Mulder, 1995; Coulson,
King, & Kutas, 1998; Osterhout & Mobley, 1995). This late positive component is widely distributed over centroparietal areas and is usually observed
with an onset around 500 ms and a duration of several hundred milliseconds
(labeled P600). It has been correlated with the cost of reprocessing during
sentence comprehension (Osterhout & Holcomb, 1992).
If one tries to characterize those syntactic anomalies that elicit a biphasic
ERP pattern (left anterior negativity followed by a late positivity) and those
that only elicit a late positivity it appears that syntactic errors realized as a
phrase structure violation or as an agreement violation are correlated with

FIRST-PASS VERSUS SECOND-PASS PARSING

315

a biphasic pattern.4 Whereas syntactic anomalies realized as garden-path sentences usually elicit a positivity only.
On the basis of the available literature it was hypothesized that the early
left anterior negativity may reflect first-pass parsing processes during which
an initial syntactic structure is built and that the late positivity might reflect
secondary syntactic processes including processes of structural reanalysis
and repair (Friederici, 1995).5
A recent paper by McKinnon and Osterhout (1996) suggests an interesting
specification of this view. In their study, they investigate structural violations
of long distance dependencies, i.e., structural violations that span across
clauses. As they found a late positivity as the only component correlated
with the violation, their study may suggest that the early left anterior negativity must be viewed as a component reflecting on-line processes of local
phrase structure building. Violations that span across clauses may not be
reflected in this component.
We take the two ERP components, i.e., the ELAN and the P600, observed
in correlation with syntactic processes to reflect different stages of syntactic
parsing (Friederici, 1995; Friederici & Mecklinger, 1996; Friederici et al.,
1996). The early negativity observed over left anterior sites (ELAN) is
viewed to reflect a first-pass parsing stage responsible for local phrase structure building. The late centroparietal positivity (P600) is assumed to reflect
a second stage responsible for thematic role assignment including processes
of syntactic reanalysis and repair.
The N400 Component
In addition to these two syntactic components a number of studies have
identified a particular component in correlation with lexical-semantic processes (Kutas & Hillyard, 1980; for review see Kutas & Van Petten, 1994).
This lexical-semantic component can be described as a negativity present
around 400 ms after the onset of the critical stimulus distributed over the
posterior parts of both hemispheres. Due to its polarity and temporal structure
this component is called N400.
The N400 component was investigated most recently in aphasic patients
with different lesion sites. Revonsuo and Laine (1996) present a case with
4

But see Hagoort, Brown, and Groothusen (1993) for an exception.
There is, however, some debate about the nature of the left anterior negativity. While most
researchers take it to reflect syntactic processes (Neville et al., 1991; Friederici et al., 1993;
Münte et al., 1993), Kluender and Kutas (1993) argue that it should be considered as a working
memory component. This argument is based on the finding that the left anterior negativity is
observed in correlation with sentence types including a filler-gap relation, i.e., sentences in
which a particular element has been moved out of its initial position as for example in whquestions. The argument, however, ignores at least some of the data in the literature, especially
those early left anterior negativities reported in correlation with the processing of syntactic
errors.
5

316

FRIEDERICI, HAHNE, AND VON CRAMON

an ‘‘extensive infarction in the left frontotemporal areas . . . the lesion encompassed the Broca’s area, part of the Wernicke’s area, insula, and underlying white matter’’ (p. 31).6 One week postonset the patient was diagnosed
as a global aphasic. The patient, however, was only tested in the ERP experiment 21/2 months postonset. No independent neuropsychological or language
examination was carried out at that time. In this ERP experiment semantically congruous and incongruous sentences were presented auditorily to the
patient. The patient was required to listen to the sentences attentively. Similar
to normals he showed an N400 component for the incongruous words.7
Hagoort, Brown, and Swaab (1996) report ERP results from a group of
Broca’s (N ⫽ 13) and a group of Wernicke’s aphasics (N ⫽ 7). For semantically incongruous sentences they observed a significant N400 component for
Broca’s aphasics and a reduced N400 component for Wernicke’s aphasics.
Lesion data from these patients are presented averaged over Broca’s aphasics
and Wernicke’s aphasics, respectively. The figures indicate that Wernicke’s
aphasics’ lesions are mostly restricted to the left posterior regions, whereas
Broca’s aphasics’ lesions included anterior and posterior regions. Due to this
grouping based on clinical aphasia types independent of lesion site, conclusions about the functional brain topography, unfortunately, are not possible.
The Present Study
Based on the functional and neurotopographical specification of the language related ERP components we set out to investigate the patterns of brain
activity during sentence comprehension in a Broca patient with a left anterior
lesion and a Wernicke patient with a left posterior lesion. Based on earlier
reaction time studies we hypothesized that Broca’s aphasics have lost the
fast and highly automatic first-pass parsing mechanism, but that they may
be able to access grammatical knowledge by more controlled mechanisms.
This was based on the finding that Broca’s aphasics do show a syntactic
priming effect during sentence comprehension, however, with considerable
delay (Friederici & Kilborn, 1989; Haarmann & Kolk, 1991, 1994). The
priming effect must be taken as an indication that these patients do have
access to syntactic information during sentence processing. Moreover, these
patients’ syntactic ability in grammaticality judgment tasks was demonstrated by Linebarger, Schwartz, and Saffran (1983). The finding that Broca’s
aphasics, unlike normal controls, are able to ignore sentence fragments in a
syntactic priming experiment and perform a lexical decision task on the target uninfluenced by the prime (Kilborn & Friederici, 1994) is taken to sup6
From the displayed CAT scan of this patient 4 days postonset, however, extension of the
lesion to Wernicke’s area cannot be seen.
7
In addition for 10% of the experimental sentences the patient was required to give an
acceptability judgment. For these he showed chance level performance. The combined data
were taken to indicate semantic processing without conscious understanding.

FIRST-PASS VERSUS SECOND-PASS PARSING

317

port the notion that these patients’ syntactic processes are subject to the individual’s control. We predicted that if the Broca’s area supports this early
parsing mechanism we should not be able to observe the early syntactic component (ELAN) in a patient with a lesion including Broca’s area, but possibly
the late syntactic component (P600) and the semantic component (N400).
Wernicke patients who have been shown to demonstrate some on-line syntactic capacities (Friederici, 1985; Blumstein, Milberg, Dworetzky, Rosen, &
Gershberg, 1991) in the absence of lexical and sentential comprehension are
expected to differ in their brain activity from a Broca’s aphasic. We predicted
that a patient with a circumscribed lesion in the Wernicke’s area should show
the early syntactic component (ELAN), but no or a reduced component reflecting lexical-semantic integration or thematic role assignment (N400).
In the present experiment we examined one Broca patient and one Wernicke patient in addition to eight normal controls. We recorded their eventrelated brain potentials while they were listening to sentences including
lexical-semantic and syntactic violations presented as connected speech.
METHODS

Participants
In this study two aphasic patients and eight normal age-matched controls
were tested.
Aphasic Cases
The individual patient histories of the two aphasic cases are displayed in
Table 1.
Case W.S. The first case, W.S., is a 62-year-old right-handed male who had
suffered from a stroke in 1981. He was classified as a Broca’s aphasia on the
basis of the Aachen Aphasia Test. The test scores of the relevant subtests are
displayed in Table 1. He showed the typical agrammatic output seen in agrammatic Broca’s aphasia. Lesion site was confirmed by an MRI (see Fig. 1).
On August 17, 1995 (i.e., 14 years/5 months/3 weeks postonset), serial
T1- and T2-weighted 6-mm images were obtained parallel and perpendicular
to the CA-CP-line. The T1-weighted native horizontal images revealed an
extended lesion in the territory of both the left middle and the left anterior
cerebral artery.
The continuous lesion area encloses: (1) the superior frontal gyrus, from
the frontal pole to the paracentral lobule (not including its posterior frontomedian cortex); (2) the middle frontal gyrus, except its most rostral portion;
(3) the opercular and an adjacent band of the triangular part of the inferior
frontal gyrus; (4) the rostral and upper insular (including the external capsules); and (5) the frontodorsal white matter down to the moderately enlarged
lateral ventricle.
The following left-sided brain structures are largely spared: (1) the anterior

Age

62

38

Patient

W.S.

G.R.

f

m

Sex
CVA
(1981)
CVA
(1994)

Etiology
Broca
(100%)
Wernicke
(71,7%)

Classification
AAT

TABLE 1

114
(100%)
93
(71%)

Comp.
AAT
101
(82%)
111
(96%)

Naming
9
(83%)
36
(36%)

Token
test

Left temporoparietal

Left frontal and insula

Lesion

318
FRIEDERICI, HAHNE, AND VON CRAMON

FIRST-PASS VERSUS SECOND-PASS PARSING

FIG. 1.

319

MR image of patient W.S.

cingulate cortex and the cingular bundle; (2) the caudate nucleus and the
putamen; (3) the parietal lobe; and (4) the temporal lobe (and in particular
its mediobasal parties).
Taken together, the NMR images show large, pseudocystically transformed infarction in the supply area of the left middle and left anterior cerebral artery. The left middle cerebral artery infarction is centered on the
branches of the frontal flank, the left anterior cerebral artery infarction on
the internal frontal artery.
With respect to the question at issue the anterior language area (including
Broca’s area), the anterior arcuate fasciculus and (at least) the anterior supplementary motor area are most likely lesioned, whereas the posterior language area and the subcortical nuclei appear completely intact.
Case G.R. The second case G.R. is a 37-year-old right-handed female who
suffered from a stroke in 1994. She was classified as a Wernicke’s aphasia
on the Aachen Aphasia Test. Scores of the relevant subtests are displayed
in Table 1. Lesion site was confirmed by CT (see Fig. 2).

320

FRIEDERICI, HAHNE, AND VON CRAMON

FIG. 2.

CT scan of patient G.R.

On October 19, 1994 (20 months/2 weeks postonset), serial axial CT
8-mm images parallel to the orbitomeatal line were obtained.
The images revealed an area of significantly decreased density within the
territory of temporal and parietal branches of the left middle cerebral artery.
The well-marked lesion area is centered on the posterior third of the upper
temporal lobe (presumably on the superior temporal gyrus) and the adjacent
portion of the parietal lobe (presumably on the inferior angular and less possibly also on the supramarginal gyrus).
Except for a small strip of tissue between the involved parietal cortex and
the not significantly enlarged atrium of the lateral ventricles the temporalparietal white matter is largely intact. The posterior insula appears spared;
the insular cistern, however, is slightly enlarged.
Taken together, the CT images show a mainly cortical, left-sided middle
cerebral artery infarction restricted to the territories of posterior temporal

FIRST-PASS VERSUS SECOND-PASS PARSING

321

TABLE 2
Material
(1)

(2)

(3)

(4)

Correct sentence:
Der Finder wurde belohnt.
The finder was rewarded.
Semantic violation:
Die Wolke wurde begraben.
The cloud was buried.
Phrase structure violation:
Der Freund wurde im besucht.
The friend was in the visited.
Morphosyntactic violation:
Der Schatz wurde bewache (1. Pers. Sing.).
The treasure was guard.

branches and the inferior angular artery. The lesion area encloses a portion
of the posterior language cortex and most likely Wernicke’s area.
Normal Controls. In addition to these two aphasic patients eight righthanded age-matched participants without neuropathological history served
as normal controls (NCs). Four served as controls for patient W.S. (BrocaNCs) and four for patient G.R. (Wernicke-NCs). Broca-NCs were chosen to
match W.S. in sex, age, and handedness. The four Broca-NCs were righthanded males at a mean age of 61.75 years ranging from 61 to 63 years.
Four NCs were chosen to match G.R. in sex, age, and handedness. WernickeNCs were right-handed females at a mean age of 36.25 years ranging from
34 to 39 years.
Material
The material used in the experiment consisted of sentences which were
either correct (Sentence Type 1) or incorrect in that the sentence final word
did not match the previous sentence context. Examples of each sentence type
are displayed in Table 2. Sentence Type 2 was incorrect with respect to
lexical-semantic aspects due to a selectional restriction violation. Sentence
Type 3 was incorrect with respect to the required phrase structure due to a
word category error. Note that German is a verb final language and that the
preposition requires a noun and not a verb as a correct continuation. Sentence
Type 4 contained a morphosyntactic error due to an incorrect verb inflection.8
8
In all sentences the critical word was at sentence final position. Although the argument
may be raised that specific effects may confound with so-called end-of-sentence wrap-up processes (e.g., Osterhout & Holcomb, 1992; Osterhout & Mobley, 1995), results from earlier
studies indicated that the specific effects under consideration were present when the critical
word was in mid-sentence position (Friederici et al., 1996) or in sentence final position (Friederici et al., 1993).

322

FRIEDERICI, HAHNE, AND VON CRAMON

There was a total of 240 sentences with 40 sentences in each experimental
category and 80 filler sentences to balance the amount of correct and incorrect sentences.
Procedure
Participants were seated in a comfortable chair and were requested to carefully listen to sentences which were presented over headphones and to indicate whether the sentence heard was correct or not. Nineteen tin scalp electrodes mounted in an elastic cap (Electro Cap) were used—13 were placed
according to the 10–20 system (Jaspers, 1958) and 6 were placed at additional sites over the Broca’s area and the Wernicke’s area left and the homologous areas at the right side as well as at a left anterior temporal and a right
anterior temporal side, i.e., Broca’s region (BL was defined in the frontotemporal region as crossing point between T3-Fz and F7-Cz) and the right hemisphere homologue (BR, defined as the crossing point between T4-Fz and
F8-Cz) and the Wernicke’s region (WL was defined in the posterior-temporal
region as crossing point between T3-P3 and C3-T5) and the right hemisphere
homologue (W.R., i.e., crossing point between T4-P4 and C4-T6). The left
anterior temporal electrode (ATL) was defined as half the distance between
F7 and T3 and the right anterior temporal electrode (ATR) corresponding
as half the distance between F8 and T4. The EEG was recorded monopolar
against linked mastoids.
Vertical electrooculogram (EOG) was recorded from electrodes placed
above and below the right eye; horizontal EOG was recorded from electrodes
placed beside the left and the right eye.
Electrode impedance was maintained below 5 kΩ. Biosignals were recorded with a DC-amplifier (low-pass filter settings 40 Hz) and digitized
with 4 ms resolution.
ERPs were calculated from the onset of the critical word until 1500 ms
hereafter. The onset of the critical word was marked within each sentence
to allow the time locking of the ERP. The first 100 ms of the critical word
were used as a baseline.
Sentences were presented auditorily. Prior to each sentence a fixation cross
was presented on a computer screen for 500 ms. Sentence duration was approximately 1800 ms. The fixation cross was present throughout the sentence
presentation and 3200 ms thereafter. Then two faces, a frowning face and
a smiling face, were presented to the left and right of the center of the screen.
Participants were instructed that a sentence judgment was required and that
the frowning face to the left of the center of the screen symbolized an unacceptable sentence and the smiling face to the right of the center of the screen
symbolized an acceptable sentence. Judgment response was required by a
button press. No special instruction concerning the speed of the response
was given; correctness of the response was emphasized. Participants were

323

FIRST-PASS VERSUS SECOND-PASS PARSING

TABLE 3
Grammaticality Judgment Performance: Number of Correct Responses of 40 Trials
Sentence type

Participants
Normal age-matched
controls
Broca-NCs
A
B
C
D
Wernicke-NCs
A
B
C
D
Aphasics
Case 1 (W.S.)a
Case 2 (G.R.)

Phrase
structure
violation

Morphosyntactic
violation

Age (years)

No. correct

Semantic
violation

62
61
63
61

39
40
40
40

39
39
40
39

40
40
40
37

40
40
39
39

37
34
35
39

35
40
40
38

39
39
38
40

39
39
39
40

40
40
40
39

62
37

77
21

77
39

80
35

80
38

a

Due to a high number of eye movement related artifacts case 1 took part in a second
session, resulting in an overall trial number of 80 trials.

holding a response panel on their laps with two buttons. The left button
(corresponding to the left of the screen) was the NO/unacceptable button;
the right button (corresponding to the right of the screen) was the YES/acceptable button. After 2500 ms the faces disappeared and the screen was
blank for 1500 ms before the next trial began.
A total of 240 sentences was presented in six blocks with 40 items each.
Experimental and filler items were systematically distributed over the six
blocks in a quasi-randomized order. Two different randomizations were constructed and distributed equally across subjects.
RESULTS

Behavioral Data
The number of correct responses for each condition and subject is displayed in Table 3.
Normal controls as well as W.S. showed a high level of judgment accuracy. G.R., in contrast, classified most of the incorrect sentences correctly
as being incorrect, but her performance was at chance level for the correct
sentences.

324

FRIEDERICI, HAHNE, AND VON CRAMON

ERP Data
As the ERP data for normal controls were group data while the aphasic
data were single case data we treated them differently in the statistical analyses.
ERPs for normal controls were analyzed using standard analyses of variance (ANOVAs). These analyses focused on the critical components observed in a prior experiment with younger adults using the same stimulus
materials and procedure (Friederici, Pfeifer, & Hahne, 1993): (1) a left anterior negativity around 180 ms for the phrasal violation condition, (2) a centroparietal negativity around 400 ms (N400) for the semantic condition, and
(3) a centroparietal positivity between 600 and 1500 ms.
For each violation type an ANOVA was calculated including the factors
Condition (correct/incorrect) and Group (Wernicke-control/Broca-control).
The analysis for the phrasal condition calculated for the time window 100–
300 ms over the electrodes in the left anterior quadrant revealed a significant
main effect of condition (F(1, 6) ⫽ 12.01, p ⬍ .01, MSe ⫽ 0.53), but neither
main effect of Group (F(1, 6) ⬍ 1) nor an interaction of the two variables
(F(1, 6) ⫽ 2.52, p ⬍ .16). The analysis for the semantic condition was calculated for the time window 300–500 ms over the electrodes in the two posterior quadrants. It revealed a significant main effect of condition (F(1, 6) ⫽
11.07, p ⬍ .02, MSe ⫽ 0.54), but no main effect of Group (F(1, 6) ⬍ 1)
and no interaction (F(1, 6) ⬍ 1). The analysis of the morphosyntactic condition for the time window 600–1500 ms over the electrodes of the two posterior quadrants, finally, revealed a main effect of condition (F(1, 6) ⫽ 13.46,
p ⬍ .01, MSe ⫽ 2.16), but again neither a main effect of Group (F(1, 6) ⬍
1) nor an interaction (F(1, 6) ⬍ 1). Given that no group-related differences
were found for the relevant components tested here, the two groups were
collapsed to constitute a single group of normal controls.
As we are dealing with single case data for the aphasic participants we
were not able to use standard statistical ERP analysis procedures. Therefore
we decided to use running t tests to test the reliability of the differences
between the correct and the incorrect condition for each of the two aphasic
patients. These t tests were calculated at each sampling point (i.e., every
4 ms) at each electrode site. In the result figures areas in which the conditions
differ significantly are plotted below each electrode with gray areas indicating a p ⬍ .05 and the black areas indicating a p ⬍ .01.
Normal Controls
The group of normal controls showed an ERP pattern similar to that observed in an earlier study using the same stimulus material (Friederici et al.,
1993). As can be seen from Figs. 3–5 no clear N1 and P2 components are
visible in the waveforms (this also holds for Figs. 6 to 11). The absence of
these components is due to the fact that the critical words were presented

FIG. 3. Average of ERPs of normal controls for the semantic violation condition. The waveforms are superimposed for the correct and the violation
condition. The vertical lines indicate the onset of the critical word. Below each electrode the significant difference obtained by a running t test calculated
over subjects is displayed.

FIRST-PASS VERSUS SECOND-PASS PARSING

325

326

FRIEDERICI, HAHNE, AND VON CRAMON

in a continuous speech stream and not as single words in a word-by-word
fashion. Due to this presentation mode no clear offset N1/P2 components
are visible.
For the lexical-semantic condition normals showed a negative going wave
about 400 ms postonset of the critical word (Fig. 3). This negativity is distributed over centroparietal areas in the left and the right hemisphere as the
running t test displayed below each electrode indicates. This pattern resembles that of the N400 component observed in correlation with lexicalsemantic integration processes (e.g., Kutas & Hillyard, 1983).
For the phrase structure condition the normal participants displayed (Fig.
4) a left anterior negativity between 100 and 300 ms after the onset of the
critical word. Moreover, a centroparietal positive going wave was observed
between 300 and 700 ms after the word indicating the word category violation. Early negativities (ELAN) similar to the one observed here have been
registered in correlation with the processing of word category errors (Neville
et al., 1991; Friederici et al., 1993). Centroparietal positivities have been
found to correlate with the processing of syntactic errors (Hagoort et al.,
1993) as well as with the processing of garden-path sentences (Osterhout,
Holcomb, & Swinney, 1994).
For the morphosyntactic condition the normal participants displayed a
centroparietal positive going wave between 600 and 1500 ms (P600) (see
Fig. 5). This is in agreement with studies investigating inflectional violations
(in Dutch: Hagoort et al., 1993; in English: Osterhout & Mobley, 1995).
However, a biphasic pattern consisting of a negative going wave preceding
a late positivity has been reported in correlation with the processing of inflectional errors (Friederici et al., 1993; Gunter et al., 1995). An explanation
for the observed differences between studies is not available and must await
an exploration in future experiments.
The discussion whether the observed positivity reflects primary syntactic
processes or processes of reanalysis and repair is still ongoing (Friederici &
Mecklinger, 1996; Mecklinger et al., 1995; Osterhout et al., 1994). We take
the two components observed in correlation with syntactic processes to reflect the two stages of syntactic parsing (Frazier, 1978), namely, the early
stage of initial parsing being reflected in the left anterior negativity and a
later secondary stage involving thematic mapping including processes of reanalysis and repair reflected in the late centroparietal positivity (Friederici,
1995).
Aphasic Cases
The number of artifact-free trials gathered from the two aphasic cases
participating in this study is sufficient to allow a calculation of averages
of the correctly answered trial in each condition, even for G.R. who only
participated in one session.

FIG. 4. Average of ERPs of normal controls for the syntactic, i.e., phrase structure violation condition. The waveforms are superimposed for the
correct and the violation condition. The vertical lines indicate the onset of the critical word. Below each electrode the significant difference obtained
by a running t test calculated over subjects is displayed.

FIRST-PASS VERSUS SECOND-PASS PARSING

327

FIG. 5. Average of ERPs of normal controls for the morphosyntactic violation condition. The waveforms are superimposed for the correct and
the violation condition. The vertical lines indicate the onset of the critical word. Below each electrode the significant difference obtained by a running
t test calculated over subjects is displayed.

328
FRIEDERICI, HAHNE, AND VON CRAMON

FIRST-PASS VERSUS SECOND-PASS PARSING

329

Case G.R. Wernicke patient G.R. similar to normals displayed an early
left anterior negativity for the phrase structure violation condition. This negativity present between 200 and 350 ms was significant at the left anterior
electrode F7 and ATL (Fig. 6). There was also a late positivity with a maximum at Pz which was significant between 1200 and 1500 ms. In contrast to
normal controls the onset of this positivity was considerably delayed.
For the semantic violation condition no N400 component was observed
(Fig. 7). There was, however, a late posterior positivity starting as late as
1200 ms and which was similar in latency and distribution to the one observed for the phrase structure violation condition. This is in clear contrast to
the pattern demonstrated by normal controls in this study and that of normal
participants in previous studies. In these studies the processing of sentences
with a semantic violation was always correlated with an N400 component,
but never with a late positivity.
For the morphosyntactic violation condition (Fig. 8) G.R. again displayed
a very late centroparietal positivity present between 1000 and 1500 ms.
Given the finding that G.R. showed a similar positivity for the two other
conditions it appears that it is not specific to the processing of the morphosyntactic violation.
Case W.S. Broca patient W.S. unlike both normal controls and unlike G.R.
demonstrated no early left anterior negativity (Fig. 9). There was, however,
a late positivity between 400 and 1100 ms in correlation with the processing
of phrase structure violations. This positivity was not only present over
centroparietal sites, but also at frontal sites of the right hemisphere.
Similar to normal controls W.S. showed a negative component between
500 and 950 ms for the semantically incorrect sentence (Fig. 10). Although
somewhat delayed the pattern of this negativity resembles that of an N400
component usually seen in correlation with lexical-semantic processes in
normals.
For the morphosyntactic violation condition (Fig. 11) W.S. displayed a
positive going wave between 600 and 1200 ms most pronounced over centroparietal areas but also present over frontal sites of the right hemisphere.
DISCUSSION

The study investigated lexical-semantic and syntactic processes in two
aphasic cases and eight normal age-matched controls using event-related
brain potential measures. The results of the normal controls are in agreement
with earlier findings, in particular, with those from a study using similar
language material in an auditory language comprehension experiment (Friederici et al., 1993). All three components observed in this earlier study in
correlation with language processes were present in the normal controls:
(1) an early left anterior negativity indicating the disruption of first-pass parsing processes was observed in the condition including a phrase structure

FIG. 6. Average of ERPs of Wernicke-patient G.R. for the syntactic, i.e., phrase structure violation condition. The waveforms are superimposed
for the correct and the violation condition. The number of trials that were included in the analysis is indicated in the lower left of the figure. These are
only those to which a correct judgment was given. The vertical lines indicate the onset of the critical word. Below each electrode levels of significance
for a running t test calculated over items are displayed.

330
FRIEDERICI, HAHNE, AND VON CRAMON

FIG. 7. Average of ERPs of Wernicke-patient G.R. for the semantic violation condition. The waveforms are superimposed for the correct and the
violation condition. The number of trials that were included in the analysis is indicated in the lower left of the figure. These are only those to which
a correct judgment was given. The vertical lines indicate the onset of the critical word. Below each electrode levels of significance for a running t test
calculated over items are displayed.

FIRST-PASS VERSUS SECOND-PASS PARSING

331

FIG. 8. Average of ERPs of Wernicke-patient G.R. for the morphosyntactic violation condition. The waveforms are superimposed for the correct
and the violation condition. The number of trials that were included in the analysis is indicated in the lower left of the figure. These are only those to
which a correct judgment was given. The vertical lines indicate the onset of the critical word. Below each electrode levels of significance for a running
t test calculated over items are displayed.

332
FRIEDERICI, HAHNE, AND VON CRAMON

FIG. 9. Average of ERPs of Broca-patient W.S. for the syntactic violation condition. The waveforms are superimposed for the correct and the
violation condition. The number of trials that were included in the analysis is indicated in the lower left of the figure. These are only those to which
a correct judgment was given. The vertical lines indicate the onset of the critical word. Below each electrode levels of significance for a running t test
calculated over items are displayed.

FIRST-PASS VERSUS SECOND-PASS PARSING

333

FIG. 10. Average of ERPs of Broca-patient W.S. for the semantic violation condition. The waveforms are superimposed for the correct and the
violation condition. The number of trials that were included in the analysis is indicated in the lower left of the figure. These are only those to which
a correct judgment was given. The vertical lines indicate the onset of the critical word. Below each electrode levels of significance for a running t test
calculated over items are displayed.

334
FRIEDERICI, HAHNE, AND VON CRAMON

FIG. 11. Average of ERPs of Broca-patient W.S. for the morphosyntactic violation condition. The waveforms are superimposed for the correct and
the violation condition. The number of trials that were included in the analysis is indicated in the lower left of the figure. These are only those to which
a correct judgment was given. The vertical lines indicate the onset of the critical word. Below each electrode levels of significance for a running t test
calculated over items are displayed.

FIRST-PASS VERSUS SECOND-PASS PARSING

335

336

FRIEDERICI, HAHNE, AND VON CRAMON

error, (2) an N400 component indicating the disruption of lexical-semantic
integration processes was present in the condition including a lexical-selection error, and (3) a late positivity taken to reflect secondary processes of
syntactic repair was present both for the phrase structure violation and the
morphosyntactic violation condition.
Although it is widely accepted that the N400 component reflects lexicalsemantic (e.g., Kutas & Hillyard, 1983; Kutas & Van Petten, 1994) and in
particular lexical integration processes (Brown & Hagoort, 1993; Chwilla,
Brown, & Hagoort, 1995), it is less clear which subprocesses the two other
components observed in correlation with syntactic processing reflect. While
Hagoort et al. (1993) take the late positivity to reflect syntactic processes,
Osterhout and Holcomb (1992, 1993) view the late positivity to index the
cost of reprocessing when confronted with syntactic anomalies, not assigning
a functional role to the left anterior negativity observed in some studies. We
(Friederici, 1995; Friederici et al., 1996) have tried to specify the functional
role of the two ERP components found in correlation with different types
of syntactic anomalies by proposing that the early left anterior negativity
might reflect primary syntactic processes and the late positivity to reflect
secondary processes of syntactic reanalysis and repair. Within this approach
primary syntactic processes are defined as first-pass parsing processes responsible for building up the initial syntactic structure on line as it is usually
observed in correlation with the processing of local phrase structure errors
(Neville et al., 1991; Friederici, Pfeifer, & Hahne, 1993). Results from a
recent study investigating syntactic violations of long-distance dependencies
that showed no left anterior negativity (McKinnon & Osterhout, 1996) might
suggest that the early left anterior negativity (ELAN) may only be elicited
by within-clause phrase structure violations, but not by across-clause phrase
structure violations. The secondary processes which we assume to be reflected in the late positivity (P600) may include processes of syntactic reanalysis as in the case of syntactically correct garden-path sentences and processes of syntactic repair as in the case of syntactically incorrect sentences.
The present finding that different ERP components identified in normal
listeners can be selectively impaired as a function of lesion site and aphasic
breakdown supports the view that the different components are generated
by different brain systems. In detail, the ERP data from the two aphasic
patients suggest the following conclusions: W.S., a patient with the clinical
classification of an agrammatic Broca’s aphasia and with a large left anterior
lesion including Broca’s area leaving the posterior language area and subcortical nuclei intact, demonstrated a selective absence of the early left anterior
negativity. This component normally seen in correlation with first-pass parsing processes responsible for building up local phrase structure seems to
depend upon the intactness of the left anterior cortical regions. It is not clear,
however, whether the left anterior cortical areas themselves generate this
component or whether these areas are part of a network generating this com-

FIRST-PASS VERSUS SECOND-PASS PARSING

337

ponent. Interestingly, this patient who has lost the ability to produce grammatically correct sentences in an on-line fashion and to process sentences
during language comprehension on line9 showed a late positive component
indicating that he detected the syntactic violation although presumably on
the basis of secondary processing mechanisms. Processing mechanisms that
have been discussed involve those of controlled syntactic processes which
appear to be separable from automatic first-pass parsing processes (Kilborn & Friederici, 1994). W.S., moreover, demonstrated an N400 component
normally seen in correlation with lexical-semantic processes, although with
some delay compared to the normal controls. This latter finding is compatible
with a recent study by Hagoort, Brown, and Swaab (1996) reporting that
Broca’s aphasics although somewhat delayed when integrating this type of
information into preceding context showed an N400 pattern. A recent ERP
study with a case of global aphasia whose CT scan suggest a primarily anterior lesion also displayed a negativity between 300 and 800 ms in response
to a semantic anomaly (Revonsuo & Laine, 1996). Thus, the ERP result for
the syntactic condition observed for W.S. is compatible with the view that
Broca patients have lost the automatic first-pass parsing processes, but may
have retained controlled second-pass parsing processes (Friederici & Kilborn, 1989; Haarmann & Kolk, 1994).10 Furthermore, it is compatible with
the view that left anterior brain areas are involved in the cortical network
supporting language comprehension (see Neville et al., 1992, for a similar
argument on the basis of an ERP study with deaf signers). The role this
anterior brain system plays in the comprehension process may be restricted
to procedures supporting automatic first-pass parsing processes.
G.R., a patient classified as a Wernicke’s aphasic and with a circumscribed
lesion centered on the posterior third of the superior temporal gyrus and the
adjacent portion of the inferior angular/(supramarginal) gyrus, demonstrated
the selective presence of the early left anterior negativity indicating intact
first-pass parsing processes. This finding was predicted on the basis of earlier
behavioral results indicating the intactness of some aspects of automatic syntactic processes (Friederici, 1985; Blumstein et al., 1991). The absence of
9
W.S. participated on an on-line sentence comprehension experiment examining patient’s
on-line sensitivity to inflectional elements as word category markers (noun vs. verb) (Friederici, Wessels, Emmorey, & Bellugi, 1992). In this study W.S. showed no on-line grammatically
effect.
10
Tyler, Ostrin, Cooke, and Moss (1995) have tried to argue against this view showing
that four of four agrammatic patients display a sensitivity for subcategorization violation and
morphological substitutions. These data, however, do not concern the argument put forward
here that first-pass parsing processes are defined to involve structure building on word category
information alone. The only relevant test is a patient’s sensitivity to word category violations.
Only one of four agrammatic patients showed such a sensitivity on line. This patient, however,
was a special case as his aphasia resulted from a motorcycle accident which he had when he
was 16 years old. It is not clear to what extent brain plasticity at this young age may have
allowed functional reorganization.

338

FRIEDERICI, HAHNE, AND VON CRAMON

the N400 component in the semantic condition suggests that lexical-semantic
integration processes are impaired. These findings although compatible with
earlier behavioral studies reporting lexical-semantic deficits in Wernicke’s
aphasics seem only partly compatible with a recent ERP study in which a
group of Wernicke’s patients had shown a reduced N400 effect (Hagoort
et al., 1996). However, a direct comparison with this study may not be valid
for two reasons. First, in the Hagoort et al. study ERPs were averaged across
all correct vs. incorrect sentences, respectively, independent of a control of
whether they were processed correctly or not. As in this study no reaction
indicating the patient’s judgment of the sentence was required; a statement
about the reduced presence or absence of the N400 with respect to the correctly answered trials is not possible. A reduced N400 pattern may be the
result of averaging over correctly and incorrectly understood sentences. Second, ERPs were averaged over patients with different lesions. On the basis
of group data, however, it is difficult to describe whether a single patient
with a particular lesion in the temporal-parietal language area may or may
not have revealed an N400 component. In addition, it is not unlikely that
the presence or absence of the N400 depends not only on the lesion site but
on other tissue variables such as type and stage of necrosis or chronicity of
the lesion. In the present study G.R. also showed a very late positive component starting around 1200 ms which, however, was present in the semantic
as well as in the two syntactic conditions. Given that this component appeared in all three conditions we are hesitant to attribute it to a specific
linguistic process. It may rather be taken to reflect a process of a general
kind, for example, some nonspecific surprise component (Donchin, 1981)
reflecting the patient’s ‘‘feeling’’ that there is something wrong with the
sentence without being able to specify it.11
CONCLUSION

The ERP pattern for the two patients provides a picture of a double dissociation: Patient G.R. showed a selective presence of the early left anterior
negativity in the absence of an N400 component, whereas patient W.S.
showed a selective absence of the early left anterior negativity in the presence
of an N400 component. When considering the presence of the early left anterior negativity and the absence of the N400 in G.R. in light of the patient’s
lesion site, we may conclude that the intactness of some part of the posterior
language area seems a necessary condition for the generation of an N400
component. The finding of an absence of the left anterior negativity and the
presence of an N400 in W.S. in whom anterior brain areas are lesioned suggests that posterior brain areas must be supported by an additional anterior
11
G.R. reported that she often felt that something was wrong with the sentence presented,
but that she was not sure.

FIRST-PASS VERSUS SECOND-PASS PARSING

339

cortical regions of the left hemisphere to guarantee the efficiency with which
the normal brain processes language on line.
REFERENCES
Blumstein, S. E., Milberg, W. P., Dworetzky, B., Rosen, A., & Gershberg, F. 1991. Syntactic
priming effects in aphasia: An investigation of local syntactic dependencies. Brain and
Language, 40, 393–421.
Brown, C. M., & Hagoort, P. 1993. The processing nature of the N400: Evidence from masked
priming. Journal of Cognitive Neuroscience, 5, 34–44.
Chwilla, D. J., Brown, C. M., & Hagoort, P. 1995. The N400 as a function of the level of
processing. Psychophysiology, 32, 274–285.
Coulson, S., King, J., & Kutas, M. 1998. Expect the unexpected: Event-related brain response
to morphosyntactic violations. Language and Cognitive Processes, 13, 21–58.
Donchin, E. 1981. Surprise! . . . Surprise! Psychophysiology, 18, 493–513.
Ferreira, F., & Clifton, C. 1986. The independence of syntactic processing. Journal of Memory
and Language, 25, 348–368.
Frazier, L. 1978. On comprehending sentences: Syntactic parsing strategies. Doctoral dissertation. University of Connecticut.
Frazier, L., & Fodor, J. D. 1978. The sausage machine: A new two-stage model of the parser.
Cognition, 6, 291–325.
Friederici, A. D. 1985. Levels of processing and vocabulary types: Evidence from on-line
comprehension in normals and agrammatics. Cognition, 19, 133–166.
Friederici, A. D. 1988. Agrammatic comprehension: Picture of a computation mismatch. Aphasiology, 2, 279–282.
Friederici, A. D. 1994. Auditory on-line language processing: Evidence from event-related
potentials. European Journal of Neuroscience, Suppl. 7, 7.
Friederici, A. D. 1995. The time course of syntactic activation during language processing:
A model based on neuropsychological and neurophysiological data. Brain and Language,
50, 259–281.
Friederici, A. D., Hahne, A., & Mecklinger, A. 1996. The temporal structure of syntactic
parsing: Early and late ERP effects elicited by syntactic anomalies. Journal of Experimental Psychology: Learning, Memory and Cognition, 5, 1219–1248.
Friederici, A. D., & Kilborn, K. 1989. Temporal constraints on language processing: Syntactic
priming in Broca’s aphasia. Journal of Cognitive Neuroscience, 1, 262–272.
Friederici, A. D., & Mecklinger, A. 1996. Syntactic parsing as revealed by brain responses:
First-pass and second-pass parsing processes. Journal of Psycholinguistic Research, 25,
1, 157–176.
Friederici, A. D., Pfeifer, E., & Hahne, A. 1993. Event-related brain potentials during natural
speech processing: Effects of semantic morphological and syntactic violations. Cognitive
Brain Research, 1, 183–192.
Friederici, A. D., Wessels, J., Emmorey, K., & Bellugi, U. 1992. Sensitivity of inflectional
morphology in aphasia: A real-time processing perspective. Brain and Language, 43,
747–763.
Gorrell, P. 1995. Syntax and Parsing. Cambridge, MA: Cambridge Univ. Press.
Gunter, T. C., Jackson, J. L., & Mulder, G. 1992. An electrophysiological study of semantic
processing in young and middle aged academics. Psychophysiology, 29, 38–54.
Gunter, T. C., Vos, S. H., & Mulder, G. 1995. Syntactic violations and ERPs: P600 or P3b?

340

FRIEDERICI, HAHNE, AND VON CRAMON

Paper Presented at the Eighth Annual CUNY Conference on Human Sentence Processing,
Tucson, AZ, March 1995.
Gunter, T. C., Stowe, L. A., & Mulder, G. 1997. When syntax meets semantics. Psychophysiology, 34, 660–676.
Haarmann, H., & Kolk, H. 1990. A computer model of the temporal course of agrammatic
sentence understanding: The effects of variation in severity and sentence complexity.
Cognitive Science, 15, 49–87.
Haarmann, H., & Kolk, H. 1991. Syntactic priming in Broca’s aphasics: Evidence for slow
activation. Aphasiology, 5(3), 247–263.
Haarmann, H. J., & Kolk, H. 1994. On-line sensitivity to subject-verb agreement violations
in Broca’s aphasics: The role of syntactic complexity and time. Brain and Language,
46, 493–516.
Hagoort, P. 1990. Tracking the time course of language understanding in aphasia. Doctoral
dissertation. Catholic University of Nijmegen.
Hagoort, P., Brown, C., & Groothusen, J. 1993. The syntactic positive shift as an ERP-measure
of syntactic processing. Language and Cognitive Processes, 8, 439–483.
Hagoort, P., Brown, C., & Swaab, T. Y. 1996. Lexical-semantic event-related potential effects
in patients with left hemisphere lesions and aphasia, and patients with right hemisphere
lesions without aphasia. Brain, 119, 627–649.
Harbin, T. J., Marsh, G. R., & Harvey, M. T. 1984. Differences in the late components of
the event-related potential due to age and to semantic and non-semantic tasks. Electroencephalography and Clinical Neurophysiology, 59, 489–496.
Holcomb, P. J., & Neville, H. J. 1991. Natural speech processing: An analysis using eventrelated brain potentials. Psychobiology, 19, 286–300.
Jasper, H. Hv. 1958. Report of the committee of methods of clinical examination in electroencephalography. Electroencephalography and Clinical Neurophysiology, 10, 370–375.
Kilborn, K. W., & Friederici, A. D. 1994. Cognitive penetrability of syntactic priming in
Broca’s aphasia. Neuropsychology, 8, 83–90.
Kluender, R., & Kutas, M. 1993. Bridging the gap: Evidence from ERP’s on the processing
of unbounded dependencies. Journal of Cognitive Neuroscience, 2, 196–214.
Kolk, H. H. J., & Van Grunsven, M. J. F. 1985. Agrammatism as a variable phenomenon.
Cognitive Neuropsychology, 2, 347–384.
Kutas, M., & Hillyard, S. A. 1980. Reading senseless sentences: Brain potentials reflect semantic incongruity. Science, 207, 203–205.
Kutas, M., & Hillyard, S. A. 1983. Event-related brain potentials to grammatical errors and
semantic anomalies. Memory and Cognition, 11, 539–550.
Kutas, M., & Van Petten, C. 1994. Psycholinguistics electrified. Event-related brain potential
investigations. In M. A. Gernsbacher (Ed.), Handbook of psycholinguistics (pp. 83–143).
San Diego: Academic Press.
Linebarger, M. C., Schwarz, M., & Saffran, E. M. 1983. Sensitivity to grammatical structure
in so-called agrammatic aphasia. Cognition, 13, 361–392.
Mazoyer, B. M., Tzourio, N., Frak, V., Syrota, A., Murayama, N., Levrier, O., Salamon, G.,
Dehaene, S., Cohen, L., & Mehler, J. 1993. The cortical representation of speech. Journal
of Cognitive Neuroscience, 5, 467–479.
McClelland, J. L., St. John, M., & Taraban, R. 1989. Sentence comprehension: A parallel
distributed processing approach. Language and Cognitive Processes, 4, 287–336.
McKinnon, R., & Osterhout, L. 1996. Constraints on movement phenomena in sentence pro-

FIRST-PASS VERSUS SECOND-PASS PARSING

341

cessing: Evidence from event-related brain potentials. Language and Cognitive Processes, 11, 495–523.
Mecklinger, A., Schriefers, H., Steinhauer, K., & Friederici, A. D. 1995. Processing relative
clauses varying on syntactic and semantic dimensions: An analysis with event-related
potentials. Memory and Cognition, 23, 477–494.
Miceli, G., Mazzucchi, A., Menn, L., & Goodglass, H. 1983. Contrasting cases of Italian
agrammatic aphasia without comprehension disorder. Brain and Language, 19, 65–97.
Münte, T. F., Heinze, H.-J., & Mangun, G. R. 1993. Dissociation of brain activity related to
syntactic and semantic aspects of language. Journal of Cognitive Neuroscience, 5, 335–
344.
Neville, H.-J., Mills, D. L., & Lawson, D. L. 1992. Fractionating language: Different neural
subsystems with different sensitive periods. Cerebral Cortex, 2, 244–258.
Neville, H.-J., Nicol, J. L., Barss, A., Forster, K. I., & Garrett, M. F. 1991. Syntactically based
sentence processing classes: Evidence from event-related brain potentials. Journal of
Cognitive Neuroscience, 3, 155–170.
Osterhout, L., & Holcomb, P. J. 1992. Event-related brain potentials elicited by syntactic
anomaly. Journal of Memory and Language, 31, 785–804.
Osterhout, L., & Holcomb, P. J. 1993. Event-related potentials and syntactic anomaly: Evidence of anomaly detection during the perception of continuous speech. Language and
Cognitive Processes, 8, 413–437.
Osterhout, L., Holcomb, P. J., & Swinney, D. A. 1994. Brain potentials elicited by gardenpath sentences: Evidence of the application of verb information during parsing. Journal
of Experimental Psychology: Learning, Memory and Cognition, 20, 786–803.
Osterhout, L., & Mobley, L. A. 1995. Event-related brain potentials elicited by failure to agree.
Journal of Memory and Language, 34, 739–773.
Penke, M., Weyerts, H., Gross, M., Zander, E., Münte, T. F., & Clahsen, A. 1997. How the
brain processes complex words: An ERP-study of German verb inflections. Cognitive
Brain Research, 6, 37–52.
Prather, P., Zurif, E., Stem, C., & Rosen, T. J. 1992. Slowed lexical access in non-fluent
aphasia. Brain and Language, 43, 336–348.
Revonsuo, A., & Laine, M. 1996. Semantic processing without conscious understanding in a
global aphasic: Evidence from auditory event-related brain potentials. Cortex, 32, 29–
48.
Samar, V. J., & Berent, G. P. 1986. The syntactic priming effect: Evoked response evidence
for a prelexical locus. Brain and Language, 28, 250–272.
Shapiro, L. P., Nagel, H. N., & Levine, B. A. 1993. Preferences for a verb’s complement and
their use in sentence processing. Journal of Memory and Language, 32, 96–114.
Stromswold, K., Caplan, D., Alpert, N., & Rauch, S. 1996. Localization of syntactic comprehension by positron emission tomography. Brain and Language, 52, 452–473.
Swaab, T. Y., Brown, C. M., & Hagoort, P. 1997. Spoken sentence comprehension in aphasia:
Event-related potential evidence for a lexical integration deficit. Journal of Cognitive
Neuroscience, 9, 39–66.
Tyler, L. K., Ostrin, R. K., Cooke, M., & Moss, H. E. 1995. Automatic access of lexical
information in Broca’s aphasics: Against the automaticity hypothesis. Brain and Language, 48, 131–162.
Wulfeck, B. 1988. Grammatically judgments and sentence comprehension in agrammatic
aphasia. Journal of Speech and Hearing Research, 31, 72–81.