ArticlePDF Available

The Interaction of Morphosyntactic and Semantic Processing in Japanese Sentence Comprehension: Evidence from Event-Related Brain Potentials

Authors:

Abstract and Figures

The present study examined event-related brain potential (ERP) responses to apparent Case-assignment violations to explore how morphosyntactic and semantic processing interact with each other during Japanese sentence comprehension. Consistent with previous studies on Case-assignment violation, the present results found that a Case-assignment violation elicited a left anterior negativity (LAN), followed by a posterior P600 compared with its grammatical counterpart. Crucially, the LAN-P600 effects were also elicited by a morphosyntactically well-formed sentence with a thematic implausible argument that may potentially force the sentence processor to perceive an apparent Case-assignment violation. Provided that the LAN effects can be interpreted as a morphosyntactic violation effect, these findings suggest that the morphosyntactic and semantic processing streams operate at least partially in parallel during sentence comprehension and that they begin to interact with one another at approximately 400 ms (at the latest) after encountering a verb.
Content may be subject to copyright.
言語研究(Gengo Kenkyu149: 43–592016
e Interaction of Morphosyntactic and Semantic Processing
in Japanese Sentence Comprehension: Evidence
from Event-Related Brain Potentials
M Y T S
Kyushu University/ Kyushu University
JSPS Research Fellow
Abstract: e present study examined event-related brain potential (ERP)
responses to apparent Case-assignment violations to explore how morphosyn-
tactic and semantic processing interact with each other during Japanese sentence
comprehension. Consistent with previous studies on Case-assignment violation,
the present results found that a Case-assignment violation elicited a left anterior
negativity (LAN), followed by a posterior P600 compared with its grammatical
counterpart. Crucially, the LAN-P600 eects were also elicited by a morphosyn-
tactically well-formed sentence with a thematic implausible argument that may
potentially force the sentence processor to perceive an apparent Case-assignment
violation. Provided that the LAN eects can be interpreted as a morphosyntactic
violation eect, these ndings suggest that the morphosyntactic and semantic
processing streams operate at least partially in parallel during sentence com-
prehension and that they begin to interact with one another at approximately
400 ms (at the latest) after encountering a verb.*
Key words: Japanese, Psycholinguistics, Sentence Comprehension, Event-
Related Brain Potentials (ERPs), Left Anterior Negativity (LAN)
1. Introduction
Sentence comprehension involves various informational processes, such as syntac-
tic, semantic, morphological, and pragmatic processes. Psycholinguists have been
researching how and when such information is processed and interacts in real-time
sentence comprehension. One of the core assumptions in traditional sentence
comprehension models is that syntactic processing guides sentence processing. For
example, in the syntax-rst model, the parser initially builds a syntactic structure
for the sequence based on (morpho)syntactic information (Ferreira and Clifton,
* We would like to thank two anonymous reviewers for their valuable comments. We are
also grateful to Dr. Hiroaki Oishi and Dr. Hajime Ono for their insightful comments and
discussion. Dr. Tsutomu Sakamoto passed away before this paper was completed. Due to his
signicant contribution to this work, we acknowledge him as a co-author. is study was
supported by a Grant-in-Aid for JSPS Research Fellows (#13J04854) from the Japan Soci-
ety for the Promotion of Science.
44 M Y  T S
1986; Frazier, 1979, 1987; Friederici, 1995, 2002). In the latter stage, the proces-
sor constructs a semantic representation based on the output of the syntactic
processing.
However, recent event-related potentials (ERP) evidence seemingly contradicts
the traditional core assumption that the semantic processor computes an analysis
that is consistent with syntactic phrase structures. A growing number of studies
have repeatedly observed that the thematic plausibility of arguments has an inu-
ence on syntactic analysis (i.e., the semantic P600 phenomenon). is observation
has led some researchers to propose the multi-stream processing model in which
independent semantic processing proceeds at least partially in parallel with syntac-
tic processing. For this model, the semantic processor can compute a semantically
plausible interpretation, even when it contradicts unambiguous syntactic infor-
mation and overwhelms the syntactic analysis in some circumstances. However,
there is also controversy with regard to the interpretation of the semantic P600
phenomenon. Other researchers have argued that the semantic P600 phenom-
enon does not indicate the existence of a syntax-independent semantic processing
stream (Brouwer, Fitz and Hoeks, 2012; Chow and Phillips, 2013). ese studies
suggest that the previous interpretation of the phenomenon is largely based on
misunderstanding the functional signicance of ERP components.
In this paper, rst, we review major ERPs and their (traditional) functional
interpretations to explain the semantic P600 phenomenon and its implication
for sentence comprehension mechanisms (sections 2 and 3). Second, we aim to
marshal the controversies on single- vs. multi-stream models to establish a founda-
tion for testing their predictions (section 4). ird, we evaluate these two types of
models by conducting an ERP experiment (sections 5 and 6). Section 7 concludes
our study.
2. Event-related brain potentials in sentence comprehension
e N400 is an ERP component that consists of a negative deection that peaks
approximately 400 ms after the onset of every content word. Semantically anoma-
lous words in a given context, such as in (1b), typically elicit an N400 eect com-
pared with (1a) (Kutas and Hillyard, 1980, 1983):
(1) a. He spread the warm bread with butter.
b. *He spread the warm bread with socks.
e N400 eect is traditionally interpreted to reect the semantic process-
ing costs of integrating an anomalous word in a given context into the preceding
semantic representation.
By contrast, (morpho)syntactic anomalies elicit left anterior negativity (LAN)
and/or the P600 eect. e LAN eect is often observed in response to a mor-
phosyntactic violation, such as a Case-assignment violation and a subject-verb
disagreement in (2) (Coulson et al., 1998; Friederici, Pfeifer and Hahne, 1993;
Friederici and Frisch, 2000; Osterhout and Mobley, 1995):
e Interaction of Morphosyntactic and Semantic Processing in Japanese Sentence Comprehension 45
(2) a. e elected ocials hope to succeed.
b. *e elected ocials hopes to succeed.
A P600 eect is an ERP component that consists of a positive deection that
peaks at approximately 600 ms. e P600 eect has been found for the syntacti-
cally anomalous sentence in (3) (Hagoort, Brown and Groothusen, 1993; Hagoort,
Brown and Osterhout, 1999; Kaan and Swaab, 2003a, 2003b; Osterhout and
Holcomb, 1992):
(3) a. e hungry guests helped themselves to the food.
b. *e hungr y guests helped himself to the food.
e traditional functional interpretation is that the LAN reects a detection of
morphosyntactic error, and the P600 reects the syntactic repair process.
3. Semantic P600 phenomenon
Now, let us return to the issue of the semantic P600 phenomenon. Recent ERP
evidence has shown that a syntactically well-formed but semantically anomalous
word embedded in a context, such as (4b), elicits a P600 eect yet not an N400
eect (Hoeks, Stowe and Doedens, 2004; Kim and Osterhout, 2005; Kim and
Sikos, 2011; Kolk et al., 2003; Kuperberg et al., 2003, 2006, 2007; van Herten,
Kolk and Chwilla, 2005; see Bornkessel-Schlesewsky and Schlesewsky, 2008;
Brouwer, Fitz and Hoeks, 2012; Kuperberg, 2007 for review):
(4) a. Passive Control: e hearty meal was devoured by the kids.
b. Violation: *e hearty meal was devouring the kids.
Given the traditional assumption of the syntax-rst model and the functional
interpretation of the N400 and P600, this nding is unexpected for two reasons:
First, according to the syntax-rst model, the English parser should assign an
AGENT role to ‘the hearty meal’ based on information related to its syntactic posi-
tion and the inection of the verb. Because the semantic processor should subse-
quently detect the thematic implausibility of ‘the hearty meal as an AGENT, the
processor should judge (4b) as a semantically anomalous sentence. However, (4b)
does not elicit an N400 eect compared with (4a) (but see Hoeks et al., 2004 for
the N400-P600 pattern). e second unexpected result is that (4b) elicited a P600
eect, although it is grammatically well-formed.
To account for these two ndings, some researchers have challenged the tradi-
tional sentence processing models instead of challenging the functional interpreta-
tion of ERP components. For example, Kim and Osterhout (2005) propose that
the semantic combinatorial processor can compute a semantically attractive analy-
sis even when it contradicts simple unambiguous syntactic information. Under
this model, the semantic processor analyses ‘the hearty meal as a THEME in (4b)
because a meal is much more likely to be devoured than to devour. Accordingly,
the sentence processor fails to detect the semantic anomaly of the sentence. us,
the N400 is not elicited (cf. semantic illusion, Hoeks et al., 2004). In such a case
46 M Y  T S
wherein the semantic attraction to the THEME interpretation is so compelling,
the semantic processor overwhelms an analysis dictated by the syntactic informa-
tion and forces the syntactic processor to erringly perceive a grammatically well-
formed sentence as being ill-formed. is incorrect recognition of the syntactic
anomaly triggers a syntactic repair process, leading to the P600 eect in (4b).
e semantic P600 phenomenon, in the case where the syntactic reanalysis
of unambiguous syntactic information is triggered by thematic implausibility, is
seemingly unexpected in the syntax-rst model. is is because the model assumes
that the syntactic processor guides sentence processing. us, it has been argued
that this nding constitutes evidence against the model. According to Kim and
Osterhout (2005), the semantic P600 eect indicates the existence of a syntax-
independent semantic processing stream and that a semantic processor can chal-
lenge a syntactic analysis.
4. Counterarguments to syntax-independent semantic processing
e essential assumption of the multi-stream models is that the N400 is tied to
a semantic integration diculty and the P600 to a syntactic integration diculty.
However, accumulating ERP evidence challenges the traditional functional inter-
pretation of the N400 and P600. e N400 is not always sensitive to semantic/
pragmatic violations (e.g., Fischler et al., 1983; Urbach & Kutas, 2010). us, the
semantic integration view of the N400 fails to account for this observation.
A number of previous studies found that the N400 is sensitive to seman-
tic relatedness and word predictability (see Kutas and Federmeier, 2011, 2000
for review). is observation suggests that the N400 reects the ease of lexical
retrieval of a word from long-term memory. From this perspective, N400 reduc-
tion to semantically related words is attributable to the fact that lexical access to
relevant lexical representation is facilitated by the preceding semantically related
words that activate it to some extent via automatic spreading activation. e fact
that a more predicted word attenuates the N400 amplitudes can be explained by
suggesting that comprehenders pre-activate the meaning of upcoming words in
advance of their actual input. Accordingly, the lexical retrieval of more predicted
words is easier than that of less predicted words.
e lexical retrieval view of the N400 can account for the absence of an N400
eect in the thematically reversed sentence in (4). e retrieval cost for the criti-
cal verb should not dier between the conditions because the critical verb has an
identical semantic relatedness with the subject-noun phrase (NP) (i.e., the hearty
meal/devour) (Brouwer et al., 2012; Chow and Phillips, 2013). Once the N400
is interpreted to reect pre-integrative lexical access, the absence of an N400 in
role-reversed sentences should not be taken as evidence that the independent
semantic processor computes a semantically plausible analysis that is inconsistent
with the syntactic information and thus does not encounter a semantic integration
diculty.
With regard to P600, recent ERP experiments have demonstrated that the
P600 is not specic to syntactic (repair) processes. For example, a P600 is found
e Interaction of Morphosyntactic and Semantic Processing in Japanese Sentence Comprehension 47
for a word that establishes a new reference in a discourse, an ironic word in a given
situation or a grammatical sentence that incorrectly describes a picture (Burkhardt,
2005, 2006; Ragel, Gunter and Friederici, 2011; Vissers et al., 2008). Considering
the one-to-many relationship between a P600 eect and its antecedent conditions,
it is dicult to infer what factor triggers a P600 eect. In other words, the P600
eect in role-reversed sentences does not necessarily reect the syntactic repair
process triggered by a semantic-thematic implausibility.
We have reviewed two dierent views on the time course of syntactic and
semantic processing. e rst view assumes a syntax-rst single processing stream
model. Under this view, the N400 reects lexical access, and the P600 reects a
more general process than syntactic processing. e second view posits a multi-
stream model in which the independent semantic processor computes a seman-
tically plausible analysis and sometimes challenges an analysis dictated by the
syntactic information. e proponents of this view interpret the N400 as an index
of semantic integration diculty and the P600 as an index of syntactic integration
diculty (Kim and Osterhout, 2005).
Given that the functional interpretation of the N400 and P600 diers between
the two views, we cannot use these ERPs to reliably evaluate which model is more
adequate for the architecture of human sentence comprehension. It is appropri-
ate to use a morphosyntactically related component, or a phasic LAN. ere are
several advantages to using the LAN to investigate whether morphosyntactic pro-
cessing diculty arises due to thematic implausibility in morphosyntactically well-
formed but semantically anomalous sentences. e LAN reects more automatic
process than the P600. Crucially, the functional interpretation of a LAN eect is
relatively unambiguous and widely acknowledged: a LAN relates that there are
morphosyntactic processing diculties. For example, a morphosyntactic violation
elicits a greater LAN, whereas a semantic violation, such as in (1), does not elicit
a LAN. Assuming a LAN eect reects an automatic morphosyntactic processing
diculty but not a semantic process, it is more credible to infer that the sentence
processor detects a morphosyntactic anomaly if we observe the LAN eect under
certain conditions (but see the Discussion below).
Using the LAN as an index of the morphosyntactic process, we aim to re-
evaluate the single-stream model and the multi-stream model. e crucial dier-
ence between these models is that only the latter assumes that morphosyntactic
processing may be susceptible to thematic plausibility. e multi-stream model
assumes that the independent semantic processor computes a semantically coher-
ent analysis and interacts with a morphosyntactic processor. On the other hand, in
the single-stream model, the morphosyntactic processor initially begins to analyse
an input based on word category and agreement information, regardless of the
semantic and thematic plausibility of the sentence. e semantic processor sub-
sequently computes a semantic analysis based on the parsed structure. us, this
model predicts that the thematic plausibility of an argument does not have any
inuence on morphosyntactic processing when morphosyntactic information is
unambiguous.
48 M Y  T S
Considering these dierent assumptions, if the LAN is enhanced by the
thematic implausibility of an argument despite the absence of morphosyntactic
violations, then the implication is that the morphosyntactic analysis is challenged
by the semantic processor and that the sentence processor wrongly perceives a
morphosyntactically well-formed sentence as morphosyntactically anomalous (by
analogy to Kim and Osterhouts (2005) interpretation of the semantic P600 phe-
nomenon). In other words, a semantic LAN eect would be evidence in favour of
multi-stream models. By contrast, if the LAN is not modulated by the thematic
plausibility of an argument, then there is no essential need to assume independent
semantic processing, thus oering weak support for the single-stream model.
To test whether a morphosyntactic analysis respects the thematic plausibility
of an argument and to re-evaluate which processing model is more adequate, we
conducted an ERP experiment that examines apparent Case-assignment violations
in Japanese.
5. ERP experiment
e present study employed Japanese sentences with apparent Case-assignment
violations to elicit LAN and P600 eects. A Case-assignment violation has repeat-
edly been found to elicit biphasic LAN-P600 eects in English and German sen-
tences (Coulson et al. 1998, Friederici and Frisch, 2000). Taking into consideration
the well-established observation that Case-assignment anomalies elicit LAN
eects, it is appropriate to use such anomalies to elicit a LAN eect and to test
whether a LAN eect is enhanced by the thematic implausibility of an argument.
5.1. Stimuli
We used verbs with transitive/intransitive alternations in Japanese (e.g., agar-u
‘rise’ vs. age-ruraise’; ak-u ‘open vi’ v s . ake-ru ‘open vt.’). e experimental sen-
tences, which consisted of two phrases, were adopted from Sakamoto, Arao, and
Suwazono (2011). A total of 120 sets of stimuli were created by combining two
types of Case particles (nominative/accusative) and two types of verbs (unaccusa-
tive intransitive/transitive).
(5) Intransitive constructions:
a. Intransitive verbs with a nominative NP (Ga-intransitive)
Nedan-ga agaru.
price- rise
‘Price rises.’
b. Intransitive verbs with an accusative NP (O-intransitive)
*Nedan-o agaru.
price- rise
Lit. ‘*rises price.’
e Interaction of Morphosyntactic and Semantic Processing in Japanese Sentence Comprehension 49
(6) Transitive constructions:
a. Transitive verbs with an accusative NP (O-transitive)
pro nedan-o ageru.
(Someone) price- raise
‘(Someone) raises price.’
b. Transitive verbs with a nominative NP (Ga-transitive)
*Nedan-ga pro ageru.
price- (something) raise
Lit. ‘*Price raises (something).’
e intransitive verb agaru ‘rise’ can assign a nominative Case to its argument
but cannot assign an accusative Case. us, (5b) involves a Case-assignment viola-
tion, whereas (5a) does not. e ungrammatical (5b) is supposed to elicit a LAN-
P600 eect compared with the grammatical (5a).
e transitive verb ageru ‘raise’ can assign both a nominative Case and an
accusative Case to its argument. However, (6b) is less acceptable than (6a) because
nedan ‘price’ is more likely to be a THEME than an AGENT of ageru ‘raise’ (i.e.,
role-reversed sentence).
5.2. Procedure for the ERP experiment
A total of 120 sets of stimuli in (5) and (6) were distributed into four lists such
that the participants did not see more than one item of the same set. e lists and
response buttons were counterbalanced among the participants. e stimuli were
presented randomly between the participants using Presentation 16.3 software
(Neurobehavioral Systems).
e participants were seated in a dimly lit soundproof room with a CRT
monitor positioned approximately 130 cm in front of them. e presentation of
the stimuli occurred in the centre of the screen in a non-cumulative manner. Each
trial started with the xation of 1,000 ms followed by a blank for 300 ms. After
the blank, the rst phrase (i.e., nedan-ga/o ‘price-nom/acc’) was presented for
700 ms, with an inter-stimulus interval of 600 ms. e second phrase (i.e., agaru/
ageru ‘rise/raise’) was presented for 800 ms followed by a blank for 500 ms. After
the blank, an acceptability judgement task was given in each trial to check how
acceptable the experimental sentences were to the participants.
5.3. Participants
e participants were 16 native speakers of Japanese at Kyushu University (14
females and two males, M = 21.5, SD = 2.5). All participants were classied as
right-handed based on the Edinburgh handedness inventory (Oldeld, 1971), and
all of them had normal or corrected-to-normal vision. None of the participants
had a history of reading disability or neurological disorders. Written informed
consent was obtained from all participants prior to the experiment, and the par-
ticipants were paid for their participation.
50 M Y  T S
5.4. Electrophysiological recording
EEGs were recorded from 19 Ag electrodes (Nihon Kohden, NE-113A) located
at Fp1, Fp2, F3, F4, C3, C4, P3, P4, O1, O2, F7, F8, T3, T4, T5, T6, Fz, Cz, and
Pz according to the international 10–20 system (Jasper, 1958). Additional elec-
trodes were placed on the left side of and beneath the left eye to monitor horizon-
tal and vertical eye movements. e C3 and C4 electrodes served as online refer-
ences, and EEGs were re-referenced to the average value of the earlobes oine.
e impedances of all electrodes were maintained at less than 5 kΩ throughout
the experiment. e EEGs were amplied with a bandpass of 0.03 to 60 Hz and
digitised at 200 Hz.
5.5. Electrophysiological data analysis
Trials with large artefacts (exceeding ±80 µV) were automatically removed from
the analyses (0.78%). All EEGs were ltered oine with a 30 Hz low-pass lter.
e baseline was set to 100 ms prior to the onset of the second phrase. e ERPs
were quantied by calculating the mean amplitude for each participant relative to
the baseline using the following three time windows: 350–500 ms (LAN), 500–
700 ms (early P600), and 700–1000 ms (late P600) (cf. Kaan & Swaab, 2003a,
2003b; Molinaro, Barber and Carreiras, 2011).
e analyses were conducted separately at the midline (Fz, Cz, and Pz), lateral
(F3, F4, C3, C4, P3, and P4), and temporal (Fp1, Fp2, F7, F8, T3, T4, T5, T6, O1,
and O2) arrays. e method, in which ERP data from all electrodes are divided
into sub-groups, is standard in ERP experiments on sentence comprehension (e.g.,
Chow & Phillip, 2014; Kim & Osterhout, 2005; Kuperberg et al., 2003, 2006,
2007; van Herten et al., 2005). e midline analysis consisted of repeated measures
of the analyses of variance (ANOVAs) with three within-group factors: C
(nominative/accusative) × V (intransitive/transitive) × A. e lat-
eral and temporal analyses consisted of four within-group factors: C × V ×
L × A. e Greenhouse-Geisser correction was applied for
all eects involving more than one degree of freedom (Greenhouse and Geisser,
1959). We report the original degrees of freedom and the corrected p-value.
5.6. Prediction
e O-intransitive sentence in (5b) involves a Case-assignment violation. us,
it should elicit a LAN-P600 eect. e focus of the experiment is on whether a
similar LAN eect is elicited in the Ga-transitive condition compared with the
O-transitive condition.
e single-stream processing model assumes that morphosyntactic processing
is completed before semantic combinatorial processing. Although nedan (‘price’)
is thematically plausible as a THEME, this model predicts that the morphosyn-
tactic processor would analyse nedan-gaprice-NOM’ as an AGENT based on
unambiguous morphosyntactic cues, such as a Case particle and the subcategorisa-
tion information of a verb. us, the processor cannot perceive a nominative Case
marking as a Case violation. As a result, this model predicts no LAN eect related
e Interaction of Morphosyntactic and Semantic Processing in Japanese Sentence Comprehension 51
to morphological error detection in the Ga-transitive condition. In other words,
only the ungrammatical O-intransitive condition may elicit a LAN eect relative
to the other three conditions. Note that this model may expect P600 eects in the
Ga-transitive condition, reecting diculty in updating/reorganising a mental
representation because the output of the subsequent semantic processing would
contradict world knowledge (cf. Brouwer et al., 2012).
e multi-stream model oers a dierent prediction. is model assumes that
thematic plausibility can aect morphosyntactic analysis even when morphosyn-
tactic information is unambiguous and less complex to process. us, the thematic
plausibility of nedan ‘price’ as a THEME may immediately aect morphosyntactic
analysis, leading the parser to wrongly perceive a Case-assignment violation in
the Ga-transitive condition and result in a LAN-P600 eect. In other words, the
two unacceptable conditions (i.e., the Ga-transitive and O-intransitive conditions)
would elicit larger LAN eects than the acceptable conditions.
5.7. Results
5.7.1. Behavioural data
At the end of each trial, the acceptability judgement task was conducted to assess
the extent to which Japanese speakers accept each condition of the experimental
sentences. Figure 1 shows the mean acceptability of each condition across the
participants.
Figure 1. Mean acceptability in the acceptability judgement
task. Error bars indicate standard errors.
e repeated measure ANOVA was conducted with two within-group factors:
C (nominative/accusative) × V (intransitive/transitive). e main eects of
C and V were not signicant (C: F(1, 15) = 0.12, p = 0.72, V: F(1,
15) = 1.57, p = 0.22). Because the interaction was signicant (F(1, 15) = 1531.48, p
< 0.001), we conducted a planned comparison. e result showed that both of the
52 M Y  T S
simple main eects were signicant (ps < 0.001). is result indicates that the par-
ticipants judged the O-intransitive and Ga-transitive conditions to be unaccept-
able and the Ga-intransitive and O-transitive conditions to be acceptable, which is
consistent with the ve-scale oine acceptability judgement test (Sakamoto, Arao
and Suwazono, 2011).
5.7.2. Electrophysiological data
Figure 2 shows the grand average ERPs in the second phrase. A visual inspec-
tion suggested that the ERPs of the unacceptable conditions (O-intransitive and
Ga-transitive) elicited a greater negativity with an anterior focus compared with
those of the acceptable conditions (Ga-intransitive and O-transitive) at 350 ms.
In the following time window, a positive-moving shift with a posterior focus was
observed in the O-intransitive and Ga-transitive conditions compared with the
Ga-intransitive and O-transitive conditions.
Figure 2. Grand average ERPs in the second phrase. e solid black line indicates the Ga-
intransitive condition, the dotted black line indicates the O-intransitive condi-
tion, the dotted grey line indicates the Ga-transitive condition, and the solid grey
line indicates the O-transitive condition. e X-axis represents the time dura-
tion, and each hash mark represents 100 ms. e Y-axis represents the voltage,
which ranged from -3 to 7 μV. Negativity is plotted upward.
e Interaction of Morphosyntactic and Semantic Processing in Japanese Sentence Comprehension 53
5.8.2.1. e LAN time window (350–500 ms)
Figure 3 shows the amplitudes of the LAN eects at F4, F3, and Fz. Figure 4
shows their topographical distributions at the intransitive (left) and transitive
(right) conditions.
Figure 3. e amplitude dierences in 350–500 ms in the intransitive (left) and transi-
tive (right) conditions. Negativity (µV ) is plotted upward. Error bars indicate
standard errors.
Figure 4. e topographical isovoltage map at 350–500 ms. e intransi-
tive condition represents the mean dierence calculated as the
O-intransitive condition minus the Ga-intransitive condition. e
transitive condition represents the mean dierence calculated as the
Ga-transitive condition minus the O-transitive condition. Negative
deection is indicated in grey.
e interaction of V × C was signicant in all arrays (M: F(1,
15) = 5.148, p < 0.05, L: F(1, 15) = 5.673, p < 0.05, T: F(1, 15) =
4.883, p < 0.05). e planned comparison revealed that the O-intransitive condi-
tion elicited a negative ERP component compared with the Ga-intransitive condi-
tion (M: F(1, 30) = 4.203, p < 0.05, L: F(1, 30) = 4.228, p < 0.05,
T: F(1, 30) = 4.229, p < 0.05). e Ga-transitive condition also elicited
a negativity compared with the O-transitive condition in the midline and lateral
54 M Y  T S
arrays, although only marginal eects were observed (M: F(1, 30) = 2.961,
p = 0.09, L: F(1, 15) = 3.406, p = 0.07). e interaction of V × C ×
A was marginally signicant in the midline because the negativity was
pronounced more anteriorly (F(2, 30) = 2.758, p = 0.07).
One may think that the LAN eect for the O-intransitive condition is
greater than that for the Ga-transitive condition. However, the primary purpose
of this experiment is to examine the presence/absence of the LAN eect in the
Ga-transitive condition. e amplitude dierence between the Ga-transitive and
O-intransitive conditions or the duration/timing of the LAN eect is of no inter-
est in the present study. us, we do not address the issue any further.
5.8.2.2. e Early P600 time window (500–700 ms)
In the temporal array, ANOVA revealed a signicant interaction of V × C
× A and V × C × L (F(4, 56) = 5.553, p < 0.01,
F(1, 14) = 8.163, p < 0.01). e planned comparison showed that the anterior
negativities in the O-transitive and Ga-transitive conditions continued into this
time-window in the anterior electrodes (O-transitive: Fp1/2: F(1, 14) = 19.334,
p < 0.01, F7/F8, Ga-transitive: Fp1/2: F(1, 14) = 4,012, p = 0.06). In other arrays,
the signicant eect of interest was not observed.¹
5.8.2.3. e Late P600 time window (700–1000 ms)
Figure 5 shows the amplitudes of the P600 eects at P3, Pz, and P4.
Figure 5. e amplitude dierences in 700–1000 ms in the intransitive (left) and
transitive (right) conditions. Positivity (µV) is plotted upward. Error bars
indicate standard errors.
In the midline and lateral arrays, the interaction of V × C × A
was signicant (M: F(2, 30) = 18.010, p < 0.01, L: F(2, 30) =
15.215, p < 0.01). In the temporal array, a four-way interaction of V × C ×
A × L was observed.
e planned comparison revealed that the O-intransitive condition elicited
a posterior positivity and a frontal negativity compared with the Ga-intransitive
¹is analysis was conducted in accordance with the reviewer’s comment.
e Interaction of Morphosyntactic and Semantic Processing in Japanese Sentence Comprehension 55
condition. e Ga-transitive condition also elicited a posterior positivity and a
frontal negativity relative to the O-transitive condition.
Repeated measures ANOVAs conrmed that there was no signicant main
eect of Case or interaction with Laterality/Anteriority in any time window at
the rst phrase (all ps > 0.10). us, the baseline of the second phrase was not con-
taminated by the potential ERP dierence of the preceding region.
Overall, these analyses conrmed our visual inspection. Both the O-intransitive
condition and the Ga-transitive condition elicited left anterior negativities at
350 ms, followed by a posterior positivity from 700 to 1000 ms.²
6. General discussion
is study employed Japanese sentences with (apparent) Case-assignment viola-
tions to examine the interaction of morphosyntactic and semantic processing.
Importantly, the results of the ERP experiment showed biphasic LAN-P600
eects for unacceptable intransitive and transitive sentences.
Previous studies suggest that the LAN eect is a manifestation of processing
diculty caused by a morphosyntactic mismatch (Coulson et al., 1998; Friederici
and Frisch, 2000; Münte, Matzke and Johannes, 1997). In the case of the ungram-
matical O-intransitive condition, the parser encounters an intransitive verb follow-
ing the accusative NP. us, the parser recognises the Case mismatch between a
nominative-taking verb and an accusative Case, which is reected in a LAN eect.
e posterior positivity has been repeatedly observed for syntactic constraint viola-
tions. According to Kaan and Swaab (2003a, 2003b), the posterior P600 (at least
partially) reects a syntactic repair process, although there is controversy concern-
ing the functional interpretation of the P600. In our experiment, the posterior
P600 in the O-intransitive condition may indicate syntactic repair diculties due
to a Case mismatch.
e major nding in our experiment is that the Ga-transitive condition also
elicited similar LAN eects compared with the O-transitive condition. If seman-
tic processing operates only after morphosyntactic processing, in accordance
with the single-stream model, then it would not detect a Case mismatch in the
²One may consider that ERP eects in transitive constructions reect the dierences of
pro-drop positions. In Japanese, object pro-drop is less frequent than subject pro-drop
(Ueno and Polinsky, 2008). However, a previous ERP study showed no ERP eect in any
time window between a subject pro-drop and object pro-drop construction when there was
no semantic attraction between a verb and its argument (proper name) (Yano, Tateyama and
Sakamoto, 2014). us, the LAN and P600 eects are not solely due to processing diculty
induced by object pro-drop.
(i) a. Kinoo Ichiro-ga pro hinanshita
Yesterday Ichiro- criticised
‘Ichiro criticised (someone) yesterday’ (object pro-drop)
b. Kinoo pro Ichiro-o hinanshita
Yesterday Ichiro- criticised
‘(someone) criticised Ichiro yesterday’ (subject pro-drop)
56 M Y  T S
Ga-transitive condition. us, under the single-stream model, it is less clear why
the LAN eect was observed in the Ga-transitive condition.
On the contrary, the LAN eect can be explained in the multi-stream model:
In this model, semantic information and morphosyntactic information may inter-
act with one another. When semantic information contradicts an analysis dictated
by morphosyntactic information, the semantic processor may challenge a morpho-
syntactic analysis. In the transitive condition, nedan (‘price’) is highly plausible as
a THEME but not as an AGENT. us, the sentence processor misperceives the
Case marking in the Ga-transitive condition as morphosyntactically anomalous.
is may lead to a LAN eect in the Ga-transitive condition.
It is noteworthy, however, that recent ERP studies suggest that (E)LAN may
reect the violation of expectation for upcoming functional morphology or phrase
structures (Molinaro et al., 2011; Lau et al., 2006). For instance, Lau et al. (2006)
report that the ELAN is enhanced by the violation of expectation for the subse-
quent word category (e.g., *Although Erica kissed Mary’s mother, she did not kiss
the Dana’s of the bride (NP ellipsis possible) vs. *Although the bridesmaid kissed
Mary, she did not kiss Dina’s of the bride (NP ellipsis impossible)). In transitive
conditions, the nominative Case attached to the inanimate noun may trigger an
expectation for an unaccusative verb, whereas the accusative Case may lead to the
anticipation of a transitive verb. In the Ga-transitive condition, the appearance of
transitive verbs may violate such an expectation. Accordingly, the LAN eect in
the Ga-transitive condition may reect such an expectation violation rather than
morphosyntactic processing diculty. If this is the case, then the LAN eect does
not falsify the single-stream processing models. erefore, to examine the interac-
tion mechanism of morphosyntactic and semantic processing, future research must
take the eect of expectation on the ERPs of verbs into consideration.³
7. Conclusion
e present study explored the interaction of morphosyntactic and semantic pro-
cessing during Japanese sentence comprehension. e result of our ERP experi-
ment showed that violations of Case assignment elicited left anterior negativity
and posterior P600 eects relative to their grammatical counterpart. Similar elec-
trophysiological patterns were observed in an apparent Case assignment violation
induced by the thematic implausibility of an NP. Provided that the LAN eects
can be interpreted as morphosyntactic violation eects, these results oer support
for the immediate interaction of syntactic and semantic processing in Japanese.
References
Bornkessel-Schlesewsky, Ina and Matthias Schlesewsky (2008) An alternative perspec-
³Yano (2015) conducted two ERP experiments that manipulated SOAs and verb types to as-
sess the extent to which the expectation aected the ERPs of verbs and what type of expecta-
tion is generated by the rst phrase. e results of these ERP experiments will be reported in
another paper currently in preparation. We appreciate the reviewer’s insightful comment.
e Interaction of Morphosyntactic and Semantic Processing in Japanese Sentence Comprehension 57
tive on “semantic P600” eects in language comprehension. Brain Research Review 59:
55–73.
Brouwer, Harm, Hartmut Fitz and John Hoeks (2012) Getting real about semantic illu-
sions: Rethinking the functional role of the P600 in language comprehension. Brain
Research 1446: 127–143.
Burkhardt, Petra (2005) e syntax-discourse interface: Representing and interpreting depen-
dency. Philadelphia, Pennsylvania: John Benjamins.
Burkhardt, Petra (2006) Inferential bridging relations reveal distinct neural mechanisms:
Evidence from event-related brain potentials. Brain and Language 98: 159–168.
Chow, Wing-Yee and Colin Phillips (2013) No semantic illusions in the “semantic P600”
phenomenon: ERP evidence from Mandarin Chinese. Brain Research 1506: 76–93.
Coulson, Seana, Jonathan W. King and Marta Kutas (1998) Expect the unexpected: Event-
related brain response to morphosyntactic violations. Language and Cognitive Processes
13: 21–58.
Ferreira, Fernanda and Charles Clifton Jr. (1986) e independence of syntactic processing.
Journal of Memory and Language 25: 348–368.
Fischler, Ira, Paul A. Bloom, Donald G. Childers, Salim E. Roucos and Nathan W. Perry
Jr. (1983) Brain potentials related to stages of sentence verication. Psychophysiology 20:
400–409.
Frazier, Lyn (1979) On comprehending sentence: Syntactic parsing strategies. Doctoral dis-
sertation, University of Connecticut.
Frazier, Lyn (1987) Sentence processing: A tutorial review. In: Max Coltheart (ed.) Attention
and performance 12: Psychology of reading, 559–586. Hillsdale, New Jersey: Lawrence
Erlbaum Associates.
Friederici, Angela D. (1995) e time-course of syntactic activation during language pro-
cessing: A model based on neuropsychological and neurophysiological data. Brain and
Language 50: 259–281.
Friederici, Angela D. (2002) Towards a neural basis of auditory sentence processing. Trends
in Cognitive Sciences 6: 78–84.
Friederici, Angela D. and Stefan Frisch (2000) Verb argument structure processing: e role
of verb-specic and argument-specic information. Journal of Memory and Language 43:
476–507.
Friederici, Angela D., Erdmut Pfeifer and Anja Hahne (1993) Event-related brain poten-
tials during natural speech processing: Eects of semantic, morphological and syntactic
violations. Cognitive Brain Research 1: 183–192.
Greenhouse, Samuel W. and Seymour Geisser (1959) On methods in the analyses of prole
data. Psychometrika 24: 95–112.
Hagoort, Peter, Colin Brown and Jolanda Groothusen (1993) e syntactic positive shift
(SPS) as an ERP measure of syntactic processing. Language and Cognitive Processes 8:
439–483.
Hagoort, Peter, Colin Brown and Lee Osterhout (1999) e neurocognition of syntactic
processing. In: Peter Hagoort and Colin Brown (eds.) e neurocognition of language
273–316. Oxford, United Kingdom: Oxford University Press.
Hoeks, John, Laurie A. Stowe and Gina Doedens (2004) Seeing words in context: e
interaction of lexical and sentence level information during reading. Cognitive Brain
Research 19: 59–73.
Jasper, Herbert H. (1958) e ten-twenty electrode system of the international federation.
Electroencephalography and Clinical Neurophysiology 10: 371–375.
58 M Y  T S
Kaan, Edith and Tamara Y. Swaab (2003a) Electrophysiological evidence for serial sentence
processing: A comparison between non-preferred and ungrammatical continuations.
Cognitive Brain Research 17: 621–635.
Kaan, Edith and Tamara Y. Swaab (2003b) Repair, revision, and complexity in syntactic
analysis: An electrophysiological dierentiation. Journal of Cognitive Neuroscience 15:
98–110.
Kim, Albert and Lee Osterhout (2005) e independence of combinatory semantic pro-
cessing: Evidence from event-related potentials. Journal of Memory and Language 52:
205–225.
Kim, Albert and Les Sikos (2011) Conict and surrender during sentence processing: An
ERP study of syntax-semantics interaction. Brain and Language 118: 15–22.
Kolk, Herman. H. J., Dorothee J. Chwilla, Marieke van Herten and Patrick J. W. Oor
(2003) Structure and limited capacity in verbal working memory: A study with event-
related potentials. Brain and Language 85: 1–36.
Kuperberg, Gina R. (2007) Neural mechanisms of language comprehension: Challenges to
syntax. Brain Research 1146: 23–49.
Kuperberg, Gina R., David Caplan, Tatiana Sitnikova, Marianna Eddy and Phillip J. Hol-
comb (2006) Neural correlates of processing syntactic, semantic and thematic relation-
ships in sentences. Language and Cognitive Processes 21: 489–530.
Kuperberg, Gina R., Donna A. Kreher, Tatiana Sitnikova, David Caplan and Phillip J.
Holcomb (2007) e role of animacy and thematic relationships in processing active
English sentences: Evidence from event-related potentials. Brain and Language 100:
223–237.
Kuperberg, Gina. R., Tatiana Sitnikova, David Caplan and Phillip J. Holcomb (2003)
Electrophysiological distinctions in processing conceptual relationships within simple
sentences. Cognitive Brain Research 17: 117–129.
Kutas, Marta and Kara D. Federmeier (2000) Electrophysiology reveals semantic memory
use in language comprehension. Trends in Cognitive Sciences 4: 463–470.
Kutas, Marta and Kara D. Federmeier (2011) irty years and counting: Finding meaning
in the N400 component of the event-related brain potential (ERP). Annual Review of
Psychology 62: 621–647.
Kutas, Marta and Steven A. Hillyard (1980) Reading senseless sentences: Brain potentials
reect semantic incongruity. Science 207(4427): 203–205.
Kutas, Marta and Steven A. Hillyard (1983) Event-relate brain potentials to grammatical
errors and semantic anomalies. Memory and Cognition 11: 539–550.
Lau, Ellen, Clare Stroud, Silke Plesch and Colin Phillips (2006) e role of structural pre-
diction in rapid syntactic analysis. Brain and Language 98: 74–88.
Münte, omas F., Mike Matzke and Sönke Johannes (1997) Brain activity associated with
syntactic incongruencies in words and pseudo-words. Journal of Cognitive Neuroscience
9: 318–329.
Oldeld, Richard C. (1971) e assessment and analysis of handedness: e Edinburgh
inventory. Neuropsychologia 9: 97–113.
Osterhout, Lee and Linda A. Mobley (1995) Event-related brain potentials elicited by fail-
ure to agree. Journal of Memory and Language 34: 739–773.
Osterhout, Lee and Phillip J. Holcomb (1992) Event-related brain potentials elicited by
syntactic anomaly. Journal of Memory and Language 31: 785–806.
Regel, Stefanie, omas C. Gunter and Angela D. Friederici. (2011). Isn’t it ironic? An
electrophysiological exploration of gurative language processing. Journal of Cognitive
e Interaction of Morphosyntactic and Semantic Processing in Japanese Sentence Comprehension 59
Neuroscience 23: 277–293.
Sakamoto, Tsutomu, Hiroshi Arao and Shugo Suwazono (2011) Zita doshi-to kakujoshi-
no kumiawase-nitaisuru bogo-washa-no yooninsei handan: kotonaru shudan-kan-no
hikaku [Native Japanese speakers’ acceptability judgment on combination of verbs and case
markers: A comparison between two dierent groups of participants]. Bulletin of the Faculty
of Letters Kyushu University 108: 31–48.
Ueno, Mieko and Maria Polinsky (2009) Does headedness aect processing? A new look at
the VO-OV contrast. Journal of Linguistics 45: 675–710.
Urbach, omas, P. & Marta Kutas (2010) Quantiers more or less quantify on-line: ERP
evidence for partial incremental interpretation. Journal of Memory and Language 63:
158–179.
van Herten, Marieke, Herman H. J. Kolk and Dorothee J. Chwilla (2005) An ERP study of
P600 eects elicited by semantic anomalies. Cognitive Brain Research 22: 241–255.
Vissers, Constance, . W. M., Herman H. J. Kolk, Nan van de Meerendonk and Doro-
thee J. Chwilla (2008) Monitoring in language perception: Evidence from ERPs in a
picture-sentence matching task. Neuropsychologia 46: 967–982.
Yano, Masataka (2015) Is semantic LAN eect elicited by thematic anomaly or expectation
violation? Evidence from Japanese sentence processing. e 28th Annual CUNY Confer-
ence on Human Sentence Processing, p. 268. University of Southern California.
Yano, Masataka, Yuki Tateyama, and Tsutomu Sakamoto (2014) Processing of Japanese
cleft constructions in context: Evidence from event-related brain potentials. Journal of
Psycholinguistic Research 43: 277–286.
Author’s contact information: [Received 9 September, 2014;
Graduate School of Humanities Accepted 15 January, 2016]
Kyushu University
6-19-1 Hakozaki, Higashi-ku, Fukuoka 812-0053, Japan
e-mail: masayano@kyudai.jp
要 旨
日本語文理解におる形態統語的意味的処理の相互作用
――事象関連脳電位を指標て――
矢野 雅貴        坂本 勉
九州大学/本学術振興会特別研究員 九州大学
本研究は,事象関連電位を指標として,形態統語的処理と意味的処理がどのように相互作
用しているのかを検討した。実験の結果,格違反文(値段をあがる)に対して左前頭部陰性
波と P600 が観察された。また,格違反は含まないが意味役割が逆転した文(値段があげる)
において,同様の左前頭部陰性波と P600 が観察された。もしこの左前頭部陰性波が形態統
語的な違反効果を反映しているとすれば,この結果は,形態統語的処理と意味的処理が,動
詞の呈示開始語 400 ms 辺りで相互作用していることを示唆している。
... Nashiwa et al. (2006) found that semantic processing can disturb syntactic processing. Yano and Sakamoto (2016) found a possibility that syntactic and semantic processing are processed at least partially in parallel. Despite many obvious advantages of EEG (e.g., temporal resolution of EEG is high), it is not without disadvantage. ...
... Our results are in harmony with Nashiwa et al. (2006) and Yano and Sakamoto (2016) who used EEG to study the relationship between syntactic and semantic processing in Japanese. Nashiwa et al. (2006) found that semantic violation can block syntactic processing. ...
... Nashiwa et al. (2006) found that semantic violation can block syntactic processing. Yano and Sakamoto (2016) found that syntactic processing and semantic processing operate at least partially in parallel. Although utilizing different tools and materials, none of the three studies show evidence supporting the primacy of syntactic processing in Japanese reading. ...
Article
Full-text available
In order to examine whether syntactic processing is a necessary prerequisite for semantic integration in Japanese, cortical activation was monitored while participants engaged in silent reading task. Congruous sentences (CON), semantic violation sentences (V-SEM), and syntactic violation sentences (V-SYN) were presented in the experiment. The participants’ oxygenated hemoglobin concentration changes during the reading task were measured using functional near-infrared spectroscopy. The results suggest that the CON sentences did not require additional cognitive load on syntactic processing or semantic processing. The V-SEM sentences demanded great cognitive load on semantic processing. Besides, it also elicited great cognitive load on syntactic processing. The V-SYN sentences induced great cognitive load on syntactic processing, but it did not induce additional load on semantic processing. These evidence demonstrates that, in Japanese language processing, the difficultness of semantic processing could influence the difficultness of syntactic processing, while the difficultness of syntactic processing would not influence the difficultness of semantic processing. Our findings are suggestive of the possibility that in Japanese language reading, semantic processing precedes syntactic processing, or semantic processing and syntactic processing are in parallel.
... The dependent variables included mean amplitudes calculated for each participant, trial, and electrode. 6 Although P600 effects were expected to appear in response to morphosyntactic violations in the time windows of 700-900 ms on the basis of previous studies in Japanese (Mueller, Hirotani, & Friederici, 2007;Nashiwa, Nakao, & Miyatani, 2007;Yano, 2018;Yano & Sakamoto, 2016;Yano, Suzuki, & Koizumi, 2018), the mean amplitudes of 300-500 ms and 500-700 ms were also analyzed to examine at which time-windows the ERPs started to change over the course of the experiment. The division of a P600 time-window into early and late time-windows (i.e., early P600 time-window of 500-700 ms and late P600 time-window of 700-900 ms) is motivated by previous studies showing that different types of sentences show a P600 effect with different peak latencies and topographical distributions (Barber & Carreiras, 2005;Carreiras, Salillas, & Barber, 2004;Hagoort & Brown, 2000;Kaan & Swaab, 2003b, 2003aMolinaro, Barber, & Carreiras, 2011, see also footenote 2). ...
... Since the peak latency of P600 is response-aligned (Sassenhagen & Bornkessel-Schlesewsky, 2015;Sassenhagen, Schlesewsky, & Bornkessel-Schlesewsky, 2014), it is plausible to expect a late P600 effect rather than an early P600 effect. Furthermore, previous experiments using similar materials and the same instruction have observed a P600 at the late time-window (Yano, 2018;Yano & Sakamoto, 2016;Yano et al., 2019. ...
... The morphosyntactic violations were expected to elicit a P600 effect, according to similar experiments in Japanese (Yano, 2018;Yano & Sakamoto, 2016; and the literature of other languages. A left anterior negativity (LAN) effect has also been observed for morphosyntactic violations in Indo-European languages with verb agreement system (see Molinaro et al., 2011 for review), but less often in East Asian languages including as Japanese, except for a special case such as a verb is distant from its subject, when a reader makes a strong prediction for a verb morphology (Yano, 2018). ...
Article
Full-text available
The present study conducted two event-related potential experiments to investigate whether readers adapt their expectations to morphosyntactically (Experiment 1) or semantically (Experiment 2) anomalous sentences when they are repeatedly exposed to them. To address this issue, we experimentally manipulated the probability of occurrence of grammatical sentences and syntactically and semantically anomalous sentences through experiments. For the low probability block, anomalous sentences were presented less frequently than grammatical sentences (with a ratio of 1 to 4), while they were presented as frequently as grammatical sentences in the equal probability block. Experiment 1 revealed a smaller P600 effect for morphosyntactic violations in the equal probability block than in the low probability block. Linear mixed-effect models were used to examine how the size of the P600 effect changed as the experiment went along. The results showed that the smaller P600 effect of the equal probability block resulted from an amplitude's decline in morphosyntactically violated sentences over the course of the experiment, suggesting an adaptation to morphosyntactic violations. In Experiment 2, semantically anomalous sentences elicited a larger N400 effect than their semantically natural counterparts regardless of probability manipulation. Little evidence was found in favor of adaptation to semantic violations in that the processing cost associated with the N400 did not decrease over the course of the experiment. Therefore, these results demonstrated a dynamic aspect of language-processing system. We will discuss why the language-processing system shows a selective adaptation to morphosyntactic violations. 2
... The present study used P600 data from Yano & Sakamoto (2016) and Yano (2018), and the direct replication of Yano & Sakamoto (2016), and P300 data from Yano et al. (2019). Although these studies tested native Japanese speakers to obtain a large dataset of P600, the present result of the sample size analysis can be generalised to other language speakers since the P600 effect of previous studies does not vary substantially across different languages. ...
... The present study used P600 data from Yano & Sakamoto (2016) and Yano (2018), and the direct replication of Yano & Sakamoto (2016), and P300 data from Yano et al. (2019). Although these studies tested native Japanese speakers to obtain a large dataset of P600, the present result of the sample size analysis can be generalised to other language speakers since the P600 effect of previous studies does not vary substantially across different languages. ...
... In the experiment, the participants were instructed to read a sentence silently and press either YES (acceptable) or NO (unacceptable) button when seeing a response cue presented 500 ms after the offset of the sentence. The only difference between Yano & Sakamoto (2016) and Yano (2018) is the manipulation of the SOA; the first phrase was presented for 800 ms followed by a 500 ms blank in Yano & Sakamoto (2016), whereas it was presented for 500 ms with a 100 ms blank in Yano (2018). The duration of the second phrase 2 Japanese marks the subject of intransitive and transitive sentences with the nominative case "-ga" (abbreviated as NOM in (1) and (2)) and an object with the accusative case "-o" (abbreviated as ACC) (i.e., nominative-accusative type language). ...
Article
Full-text available
Reliably and efficiently detecting physiological differences between conditions of interest is of importance in psychophysiology. In particular, when it comes to the observation of relatively small differences, such as a P600 effect, a language-related brain potential elicited by ungrammatical sentences compared to grammatical sen- tences, inter-participant variability is a critical factor since a larger inter-participant variability decreases sta- tistical significance, and therefore increases the necessary sample size. The present study investigated how stable individual P600s are, at which sample sizes the P600 becomes stable, and how many participants are necessary to observe a P600 effect. P600s were recorded from 48 participants, as well as P300 (P3b) from 40 participants for comparison. Unlike the P3b effect, which had an approximately 10 μV difference between the target and standard stimuli, P600 increased in amplitude by only 1.4–1.7 μV at Pz during the processing of ungrammatical sentences relative to the grammatical counterparts. The sample size analysis suggests that 20 to 30 participants are needed to detect a P600 effect at Pz, and the distribution of variances does not change significantly with a larger sample size.
... More recently, a more controversial observation has been made. In Yano and Sakamoto's (2016) experiment, case particles (nominative/accusative) rather than voices (active/passive) were manipulated to create a Japanese version of a role-reversed sentence. The transitive verb "simeru" (close vt) in (2) can assign both a nominative case and an accusative case to its argument. ...
... (3b) involves a case-assignment violation because the unaccusative intransitive verb "simaru" (close vi) marks a single argument with a nominative case rather than an accusative case. Yano and Sakamoto (2016) found that the role-reversed sentence in (2b) elicited both a L(A)N effect and a P600 effect compared with its control sentence, as did the case-assignment anomalies in (3b). 1 A similar LAN elicitation for a role-reversed sentence has also been reported in English (Kim and Sikos, 2011, see Section 2.2.4. for discussion). ...
... 1 An increasing number of recent ERP studies have suggested that a biphasic response of the grand mean ERP can sometime reflect a result of a spatiotemporal component overlapping of N400-like negativities that some participants show and P600-like positivities that other participants show (Grey eta l., 2017;Kos et al., 2012;Nieuwland & Van Berkum, 2008;Osterhout, 1997;Tanner, 2015;Tanner & Van Hell, 2014;Tanner et al., 2013;. The biphasic response was a representative pattern of Yano & Sakamoto's (2016) experiment. Thus, the bi-phasic pattern was not a result of grand mean averaging with some participants showing either a negativity or a positivity. ...
Article
Full-text available
Several recent event-related potential (ERP) studies have observed a left (anterior) negativity (L(A)N) in (morpho)syntactically well-formed but semantically anomalous sentences, which are often referred to as semantic reversal anomalies (e.g., The window closes someone). Such a L(A)N elicitation for semantic reversal anomalies is not expected under the widely held assumption that LANs are associated with morphosyntactic processing loads and are insensitive to semantic anomalies. This raises the empirical question of why semantic reversal anomalies elicit a L(A)N effect. One possible explanation for this observation can be presented in relation the multi-stream processing model, according to which the independent semantic processing stream challenges an analysis outputted by the morphosyntactic processing stream, leading to a misperception of morphosyntactic ill-formedness. Alternatively, the L(A)N effect may reflect a mismatch between expected and actual syntactic structures. As an inanimate subject can trigger the expectation of an intransitive structure rather than an active transitive structure, the input of a transitive verb should violate such an expectation. The present study tests these two hypotheses by manipulating the temporal predictabilities of verb types to examine the underlying cognitive processes of L(A)N effects in semantic reversal anomalies. The results reveal a L(A)N effect for semantic reversal anomalies only when prediction of a verb type was possible, in favour of the latter prediction-based view. The implications of this finding are discussed with respect to predictive processing mechanisms and the architecture of processing systems.
... Federmeier, 2007;Federmeier and Kutas, 1999;Federmeier, 2000, 2011;Kutas and Hillyard, 1984;Lau et al., 2013). Furthermore, the LAN effect is modulated by a structural prediction mismatch (Lau et al., 2006;Molinaro et al. 2011;Yano and Sakamoto, 2016b). These observations suggest that linguistic processes performed at 300-500 ms involve access to information associated with actual input to compensate for a less predicted input rather than the reanalysis of a representation to incorporate an input word in the already-built representation. ...
... only P600) (see, e.g. Ito e al., 2016;Wlotko and Federmeier, 2015;Yano and Sakamoto, 2016b; see also Chow et al., 2015;, for a similar manipulation). ...
... From a broader perspective, our finding is consistent with the observation that other early negative effects in the 300-500 time window, such as N400 and LAN, are also sensitive to a prediction error and temporal predictabilities (Brouwer et al., 2012;Chow et al., 2015;Yano and Sakamoto, 2016b). N400 ...
Article
Full-text available
We conducted an event-related potential experiment using aspectual coercion to examine whether aspectual information is predictively processed prior to verb input in Japanese. In the experiment, experimental sentences were presented with two presentation durations (500 ms vs. 800 ms) to manipulate the temporal predictability of a verb. Aspectually coerced sentences elicited an early anterior negativity (AN) only when participants had a sufficient time to generate an aspectual prediction for a verb. A late AN was observed in aspectual coercion irrespective of the manipulation of the duration. This result indicates that early AN is elicited by a prediction error pertaining to aspectual information, suggesting that aspectual information is computed predictively based on adverbials without waiting for verb information.
... Also, there exists a clear cut boundary between each morpheme, and the agglutination processes are expressed through affixes, which do not affect the individual morpheme forms [46][47][48]. The processing of other agglutinative languages such as Japanese [49], Korean [50,51], Arabic [52], Finnish [53], Hebrew [54], Hindi [55,56], Urdu [57] and South Indian languages such as Kannada LAN is interpreted as a morphosyntactic violation effect, the morphosyntactic/syntactic and semantic processing operates at least in parallel during initial sentence comprehension Column headings: Authors-Name of the authors, Year-year of publication, Language-language in which the study was run, N-number of participants. [58,59] have been investigated using ERPs. ...
Article
The past few decades of research in language processing provides empirical data on the dimensions of the brain-language relationship. The processing of different language components is determined through various methods. The methodologies used to study language processing have evidenced an immense advancement over the years, tracking real-time processing events with millisecond precision. ERP (event-related potentials) is one such method which assists to visualize the neural mechanisms that underlie language processing. Different electrophysiological components mark different components of language depending on their structural and functional aspects. Since research on language processing is expanding its boundaries, the neural mechanisms for processing syntax components have been the focus of recent investigations across the languages of the world. The present review article aims to discuss the findings of studies on syntax processing besides highlighting the functional significance of P600, the electrophysiological marker of syntax processing. The results of the review elucidate the need for evaluating the finer details of syntax, including morpho-syntax concerning specific language structures. Studies across the languages of the world exemplify the uniqueness in the structure of different languages that may provide varied perspectives on the universality in syntax processing. The present review contributes a new dimension towards understanding the nature of syntax processing with respect to language specificity.
... According to Yano and Sakamoto (2016), the case-assignment and semantic reversal anomaly in (1) and (2) elicit LAN and late P600 effects as compared to their control sentences. Note that the ERP effects for semantic reversal anomaly are not attributable to the difference in pro-drop position as lack of ERP effects between a subject pro-drop and object pro-drop sentence was reported for cases with no semantic attraction between the verb and its argument (proper name) (Yano et al. 2014). ...
Article
Full-text available
The present study examined the locus responsible for the effect of emotional state on sentence processing in healthy native speakers of Japanese, using event-related brain potentials. The participants were induced into a happy, neutral, or sad mood and then subjected to electroencephalogram recording during which emotionally neutral sentences, including grammatical sentences (e.g. window-NOM close vi, ‘The window closes.’), morphosyntactically-violated sentences (e.g. window-ACC close vi, Lit. ‘Close the window.’), and semantically-reversed sentences (e.g. window-NOM close vt, ‘The window closes pro.’) were presented. The results of the ERP experiment demonstrated that while the P600 effect elicited by morphosyntactic violation was not modulated by mood, the P600 effect elicited by semantic reversal anomaly was observed only in participants previously induced into a happy mood. The LAN and N400 were not sensitive to the participants’ transient emotional state. These results suggest intact memory access and impaired integration of syntactic and semantic information in individuals in a sad mood.
Article
Full-text available
Numerous studies have found "subject gap preference" in relative clauses and cleft constructions in English, French, and other languages. In contrast, previous studies have reported "object gap preference" in cleft constructions in Japanese. However, the effect of integrating a filler and its gap may be influenced by the effect of transitional probabilities, so previous studies confounded these two factors. This study explores processing asymmetries in Japanese cleft constructions by conducting an event-related brain potential experiment by controlling transitional probabilities. The results demonstrate that the subject gap preference in Japanese is well aligned with that observed in other languages, suggesting that subject gap preference is a universal aspect of language comprehension.
Article
Full-text available
Event-related potentials were measured as subjects read sentences presented word by word. A small N400 and a robust P600 effect were elicited by verbs that assigned the thematic role of Agent to their preceding noun-phrase argument when this argument was inanimate in nature. The amplitude of the P600, but not the N400, was modulated by the transitivity of the critical verbs and by plausibility ratings of passivised versions of these sentences (reflecting the fit between the critical verb and the inanimate noun-phrase as the verb's Theme). The P600 was similar in scalp distribution although smaller in amplitude, than that elicited by verbs with morphosyntactic violations. Pragmatically unlikely verbs that did not violate thematic constraints elicited a larger N400 but no P600 effect. These findings support the theory that the cost of syntactic processing on a verb is influenced by the precise thematic relationships between that verb and its preceding arguments.
Article
Full-text available
This paper examines the relationship between headedness and language processing and considers two strategies that potentially ease language comprehension and production. Both strategies allow a language to minimize the number of arguments in a given clause, either by reducing the number of overtly expressed arguments or by reducing the number of structurally required arguments. The first strategy consists of minimizing the number of overtly expressed arguments by using more pro-drop for two-place predicates (Pro-drop bias). According to the second strategy, a language gives preference to one-place predicates over two-place predicates, thus minimizing the number of structural arguments (Intransitive bias). In order to investigate these strategies, we conducted a series of comparative corpus studies of SVO and SOV languages. Study 1 examined written texts of various genres and children's utterances in English and Japanese, while Study 2 examined narrative stories in English, Spanish, Japanese, and Turkish. The results for these studies showed that pro-drop was uniformly more common with two-place predicates than with one-place predicates, regardless of the OV/VO distinction. Thus the Pro-drop bias emerges as a universal economy principle for making utterances shorter. On the other hand, SOV languages showed a much stronger Intransitive bias than SVO languages. This finding suggests that SOV word order with all the constituents explicitly expressed is potentially harder to process; the dominance of one-place predicates is therefore a compensatory strategy in order to reduce the number of preverbal arguments. The overall pattern of results suggests that human languages utilize both general (Pro-drop bias) and headedness-order-specific (Intransitive bias) strategies to facilitate processing. The results on headedness-order-specific strategies are consistent with other researchers' findings on differential processing in head-final and non-head-final languages, for example, Yamashita & Chang's (2001) ‘long-before-short’ parameterization.
Article
Event-related brain potentials (EMS) were recorded while normal German subjects read either simple declarative sen- tences made up from real German words, or sentences that contained German pseudo-words instead of nouns and verbs. The verb (pseudo-verb) of the sentences disagreed in number with the subject noun (pseudo-noun) in 50% of the sentences. The subjects had the task either to read the sentences for an interspersed memory test (memory condition, pseudeword sentences only) or to make a syntactic judgment after each real-word/pseudo-word sentence. While in the real-word condition a late and widespread positivity resembling the previously described syntactic positive shift was found for the disagreeing verbs, a negativity with an onset latency of about 300 msec was seen for the disagreeing pseudo-verbs. In the pseudo-word conditions no positivity followed the initial negativity. This dissociation of negative and positive waves occurring in response to morphosyntactic mismatches by the pseudo/real-word manipulation suggests that the positive shift is a concomitant of a recomputation routine initiated to account for the number incongruency. This routine is based upon the semantics of the sentence and therefore is not observed in the pseudo-word conditions. The earlier negativity, on the other hand, appears to be a more direct index of morphosyntactic incongruency.
Article
Recent observations of unexpected ERP responses to grammatically well-formed role-reversed sentences (the "Semantic P600" phenomenon) have been taken to bear directly on questions about the architecture of the language processing system. This paper evaluates two central pieces of evidence for accounts that propose a syntax-independent semantic composition mechanism, namely the presence of P600 effects and the absence of N400 effects in role-reversed sentences. Experiment 1 examined the relative contribution of the presence of an animacy violation and the semantic relations between words ('combinability') to the ERP responses to role-reversed sentences. Experiment 2 examined the ERP responses to role-reversed sentences that are fully animacy-congruous. Results from the two experiments showed that animacy-violated sentences with no plausible non-surface interpretation elicited the same P600 effect as both types of role-reversed sentences; additionally, semantically anomalous target words elicited no N400 effects when they were strongly semantically related to the preceding words, regardless of the presence of animacy violations. Taken together, these findings suggest that the presence of P600s to role-reversed sentences can be attributed to the implausibility of the sentence meaning, and the absence of N400 effects is due to a combination of weak contextual constraints and strong lexical association. The presence of a plausible non-surface interpretation and animacy violations made no unique contribution to the ERP response profiles. Hence, existing ERP findings are compatible with the long-held assumption that online semantic composition is dependent on surface syntax and do not constitute evidence for a syntax-independent semantic composition mechanism.
Article
Event-related brain potentials (ERPs) were recorded from 13 scalp electrodes while subjects read sentences containing syntactic ambiguities. Words which were inconsitent with the “preferred” sentence structure elicited a brain potential (P600) quite distinct from the potential previously observed following contextually inappropriate words (N400). Furthermore, final words in sentences typically judged to be unacceptable elicited an N400-like effect, relative to final words in sentences typically judged to be acceptable. These findings suggest that ERPs are sensitive to syntactic anomaly, including anomaly engendered by disambiguating material following erroneous analysis of a syntactically ambiguous string (the “garden path” effect). We evaluate the speculation that the P600 and N400 effects are elicited as a function of anomaly type (syntactic and semantic, respectively).