Dear Author,

Please, note that changes made to the HTML content will be

added to the article before publication, but are not reflected
in this PDF.
Note also that this file should not be used for submitting
corrections.

Our reference: NSL 30962

P-authorquery-v9

AUTHOR QUERY FORM

Journal: NSL

Please e-mail or fax your responses and any corrections to:

E-mail: corrections.essd@elsevier.thomsondigital.com

Article Number: 30962

Fax: +353 6170 9272

Dear Author,
Please check your proof carefully and mark all corrections at the appropriate place in the proof (e.g., by using on-screen
annotation in the PDF file) or compile them in a separate list. Note: if you opt to annotate the file with software other than
Adobe Reader then please also highlight the appropriate place in the PDF file. To ensure fast publication of your paper please
return your corrections within 48 hours.
For correction or revision of any artwork, please consult http://www.elsevier.com/artworkinstructions.
Any queries or remarks that have arisen during the processing of your manuscript are listed below and highlighted by flags in
the proof. Click on the Q link to go to the location in the proof.
Location in
article
Q1
Q2
Q3
Q4
Q5

Query / Remark: click on the Q link to go

Please insert your reply or correction at the corresponding line in the proof
Please confirm that given names and surnames have been identified correctly.
Please check the presentation of affiliation "a" and the country name for affiliation "b", and correct if
necessary.
This sentence has been reworded for clarity. Please check that the meaning is still correct and amend if
necessary.
Please check the hierarchy of the section headings.
Please update Ref. [27].

Please check this box or indicate your approval if

you have no corrections to make to the PDF file

Thank you for your assistance.

G Model

ARTICLE IN PRESS

NSL 30962 15

Neuroscience Letters xxx (2014) xxxxxx

Contents lists available at ScienceDirect

Neuroscience Letters
journal homepage: www.elsevier.com/locate/neulet

Short communication

Early processing of auditory lexical predictions revealed by ERPs

Q1

4
5

Martijn Baart a, , Arthur G. Samuel a,b,c

Q2

BCBL Basque Center on Cognition, Brain and Language, Paseo Mikeletegi 69, 2nd oor, 20009 DonostiaSan Sebastin, Spain
IKERBASQUE, Basque Foundation for Science, Spain
c
Stony Brook University, Department of Psychology, Stony Brook, NY, USA
b

h i g h l i g h t s

8
9
10
11
12
13

Q3

Listeners heard (pseudo-) words in which the third syllable determined lexical status.
We measured ERPs that were time-locked to third syllable onset.
Items were either naturally timed or delayed third syllable.
Across conditions, words yielded more positive ERPs than pseudo-words at 200 ms.

14

a r t i c l e

15
28

i n f o

a b s t r a c t

16

Article history:
Received 3 September 2014
Received in revised form 6 November 2014
Accepted 26 November 2014
Available online xxx

17
18
19
20
21
22

27

Keywords:
Speech perception
N200 effect
Early lexicality effects
Robust lexical effects

29

1. Introduction

23
24
25
26

30
31
32
33
34
35
36
37
38
39
40
41
42
43
44

Q4

Auditory lexical processing starts within 200 ms after onset of the critical stimulus. Here, we used electroencephalography (EEG) to investigate whether (1) the so-called N200 effect can be triggered by
single-item lexical context, and (2) such effects are robust against temporal violations of the signal. We
presented items in which lexical status (i.e., is the stimulus a word or a pseudoword?) was determined at
third syllable onset. The critical syllable could be naturally timed or delayed (by 440 or 800 ms). Across
all conditions, we observed an effect of lexicality that started 200 ms after third syllable onset (i.e., an
N200 effect in naturally timed items and a similar effect superimposed on the P2 for the delayed items).
The results indicate that early lexical processes are robust against violations of temporal coherence.
2014 Published by Elsevier Ireland Ltd.

Despite the fact that speech input is variable, complex, and

often degraded, humans rapidly decode the signal into a meaningful linguistic percept. Initial processing of different features of
the speech signal (e.g., phonological, lexical, and semantic properties) takes place within 200 ms after onset of the stimulus [e.g.,14].
For instance, event-related potentials (ERPs) within 150250 ms
after onset of an incongruent target-word such as maze embedded in the sentence the painter colored the details with a small
maze, are more negative than for congruent target-words such
as paint brush. This so-called N200 effect has been argued to
reect a selection process during which initial phonological assessment of the stimuli interacts with predictive information derived
from the semantic/syntactic context [5,6]. Even though the N200
effect sometimes has been considered to be a part of the later auditory N400 effect that is triggered by similar stimulus manipulations

Corresponding author. Tel.: +34 943 309 300x228.

E-mail address: m.baart@bcbl.eu (M. Baart).

[e.g.,79], there is evidence that functionally differentiates the two

[5,6], with the N200 effect being related to lexical selection, and the
N400 effect to semantic integration [6].
Consistent with the time-frame of the N200 effect, oddball
paradigms in which repetitions of a standard stimulus are occasionally followed by a deviant, have revealed that the mismatch
negativity-response [i.e., MMN, [10]] generated by the deviant
is modulated by lexical properties of the signal [e.g.,1113]. For
instance, van Linden et al. [13] repeatedly presented listeners
with a stimulus in which the nal consonant was replaced by an
ambiguous sound in between t and p (i.e., ? in vloo? or hoo?)
and investigated the MMN generated by an occasional presentation of a canonical t sound (the deviant) embedded in vloot or
hoot, respectively. In line with Ganong [14], van Linden et al. [13]
observed that the lexicon perceptually changed the ? in vloo? to
a t (vloot is Dutch for eet, vloop is a non-word), and the ?
in hoo? to a p (hoop means hope, hoot is a non-word). The
authors observed a smaller MMN when vloo? was followed by
vloot than when hoo? was followed by hoot.
One way to interpret this result is that the standards had generated predictions about the phonological identity of the subsequent

http://dx.doi.org/10.1016/j.neulet.2014.11.044
0304-3940/ 2014 Published by Elsevier Ireland Ltd.

Please cite this article in press as: M. Baart, A.G. Samuel, Early processing of auditory lexical predictions revealed by ERPs, Neurosci.
Lett. (2014), http://dx.doi.org/10.1016/j.neulet.2014.11.044

45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

G Model
NSL 30962 15
2
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109

M. Baart, A.G. Samuel / Neuroscience Letters xxx (2014) xxxxxx

sound, and that the MMN increased whenever such a prediction

was violated. According to this view, the prediction about the
upcoming sound that is generated by repeated presentations of
? that is disambiguated toward p, would be violated upon hearing t (as one expects to hear the same p-like sound), generating
an MMN, assuming that the MMN reects a violation of a built-up
memory trace in echoic memory [15]. However, the standard sound
could also have generated a lexical prediction about the upcoming
sound. Since the hoot deviant was the only pseudoword in the set,
its larger MMN may therefore be related to the fact that the listener had expected to hear a lexically valid item, but instead heard
a pseudoword.
There is accumulating evidence favoring such a predictive coding account on a single item level [e.g.,16], and the brain indeed
responds differently to words and pseudowords in the timewindow of the N200 effect when measured from the point where
lexical status in the item is determined [1]. Here, we sought to
determine whether the N200 effect can be triggered by lexical predictions generated within the context of a single word (as opposed
to a multi-word context) previous behavioral research on phonemic restoration [17] and phonetic categorization [18,19] suggests
more direct perceptual effects of within-item lexical context than
sentential context. We presented Spanish listeners (N = 16) with
three-syllable auditory stimuli in which the third syllable determined the lexical status of the item (e.g., lechuga, meaning
lettuce, versus the pseudoword lechuda), while measuring ERPs
time-locked to the onset of the third syllable. If the N200 effect
is sensitive to lexical predictions generated on a single-item level,
we should observe a more negative ERP for pseudowords than for
words about 200 ms after onset of the third syllable.
The second aim of the study was to determine whether this N200
effect was robust against violations in temporal coherence within
the items, and thus, whether lexical predictions survive across time.
We therefore included conditions in which we inserted a 440 or
800 ms period of silence between the second and third syllables
(henceforth delay conditions). These conditions were expected to
generate ERPs with a different morphology than for the naturally
timed items, as the delayed onset of the third syllable should elicit
an N1/P2 complex. Lexical effects may shift in time based on stimulus specications [e.g.,20] and similar effects can be superimposed
on different ERP components [e.g.,21]. If the lexicality effect can
survive the delay, we should observe more negative activity for
pseudowords than words, but with the lexicality effect at around
200 ms now superimposed on the P2 component.

110

2. Material and methods

111

2.1. Participants

112
113
114
115
116

117

118
119
120
121
122
123
124
125

ARTICLE IN PRESS

Sixteen native Spanish adults (11 females, mean age = 21 years,

S.D. = 1.5) with normal hearing and normal- or corrected-to-normal
vision participated in the experiment in return for payment
(D 10/h). All provided written informed consent. The experiment
was conducted in accordance with the Declaration of Helsinki.
2.2. Stimuli
Three-syllable items (with a CVCVCV, CCVCVCV, or CVCCVCV
structure) were selected from the EsPal subtitle database [22].
We excluded items that (1) were not (only) nouns, (2) had word
stress on the rst or third syllable, (3) were diminutives or had
lexically valid items embedded in them (according the on-line dic
tionary of the Real Academia Espanola),
(4) had a frequency <2
or >15 per million or (5) had a phonological neighbor (dened
by the addition, deletion, or substitution of one letter) with a

higher frequency than the item itself. These criteria yielded a set
of 6 words: brigada (brigade), lechuga (lettuce), granuja (rascal), laguna (lagoon), pellejo (hide/skin), and boleto ((lottery)
ticket). Next, pseudowords were created by rotating the nal syllables (1) without creating new embedded lexical items, (2) such that
all nal syllables occurred once in a word and once in a pseudoword,
and (3) so that the nal consonant in the pseudowords never
occurred after the rst two syllables in any existing Spanish words.
The resulting pseudowords were brigaja, lechuda, laguga, granuna, pelleto, and bolejo. The predictability of speech segments
is known to modulate early EEG activity [e.g.,23] and was therefore controlled in our stimuli, with very similar predictability from
syllable two to three for words and pseudowords. More precisely,
transition probabilities (derived from the CLEARPOND database
[24]) for syllables were .037 for words and .003 for pseudowords,
p = .21, and transition probabilities for biphones were .051 for
words and .036 pseudowords, p = .52.
A male native speaker of Spanish recorded the words and a set
of 12 pseudowords in which the nal consonant was replaced by
ch or sh. To control for acoustic properties and co-articulation,
all items were created from the ch or sh items through splicing.
For example, auditory da from brigada was spliced onto briga
from brigacha and lechu from lechusha to create brigada (a
real word) and lechuda (a pseudoword). All of the naturally-timed
stimuli sounded natural, with no audible clicks or irregularities.
3. Procedure
Participants were seated in a sound-attenuated, dimly lit, and
electrically shielded booth about 80 cm from a computer monitor. Stimuli were presented through a regular computer speaker (at
65 dB(A) at ear-level and a 48 kHz sampling rate) placed directly
above the monitor. There were 6 experimental blocks (8 min in
duration), and each block contained 108 experimental trials. Of
these, 54 were lexical items and 54 were non-lexical items (9 presentations of each of the 6 words and 6 pseudowords). In each block,
one-third of the items were naturally timed, one-third came from
the 440 ms delay condition (the silence in between syllables 2
and 3 was 400, 440 or 480 ms, 1 presentation per item per delay),
and one-third were from the 800 ms delay condition (760, 800,
or 840 ms). On an additional 18 trials per block (14% of the total
of 756 trials), a small white dot appeared on the screen (120 ms
in duration) at auditory onset of the third syllable. These trials
were included to keep participants oriented toward the monitor
(and the speaker above it) and minimize head movement during
testing. Participants were instructed to press a button upon detecting a dot. Each trial started with a 400 ms xation cross followed
by a 600, 800, or 1000 ms interval before the auditory stimulus
was delivered. Onset of the critical third syllable thus ranged from
1240 ms (no delay condition, 600 ms break) to 2280 ms (840 ms
delay, 1000 ms break) after the xation had disappeared. The intertrial interval between sound offset and xation onset was 1800 ms.
Trials were pseudo-randomly distributed across the six experimental blocks. Before the experiment started, participants were
instructed about the three delay conditions and completed a 12trial practice session to acquaint them with the procedures.
3.1. EEG recording and analyses
The EEG was recorded with a 32-channel BrainAmp system
(Brain Products GmbH) at a 500 Hz sampling rate. Twenty-seven
Ag/AgCl electrodes were placed in an EasyCap recording cap at
positions Fp1, Fp2, F7, F3, Fz, F4, F8, FC5, FC1, FC2, FC6, T7, C3,
Cz, C4, T8, CP5, CP1, CP2, CP6, P7, P3, Pz, P4, P8, O1, and O2. An
additional electrode at FCz served as ground, and 4 electrodes (2

Please cite this article in press as: M. Baart, A.G. Samuel, Early processing of auditory lexical predictions revealed by ERPs, Neurosci.
Lett. (2014), http://dx.doi.org/10.1016/j.neulet.2014.11.044

126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150

151

152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179

180

181
182
183
184
185
186

G Model
NSL 30962 15

ARTICLE IN PRESS
M. Baart, A.G. Samuel / Neuroscience Letters xxx (2014) xxxxxx

Fig. 1. (a) ERPs for words and pseudowords per delay condition, (b) results of point-wise t-tests in a 0300 ms window (left panels labeled as I) and six 50 ms windows
(right panels labeled as II), (c) the pseudowords minus words difference wave, (d) the results of corresponding point-wise ANOVAs, and (e) the scalp topographies of the
time-window in which the difference between conditions was most prominent. Asterisks indicate electrodes at which signicant differences between delay conditions (i.e.,
differences in lexicality effects) were observed.
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211

on the orbital ridge above and below the right eye and 2 on the lateral junctions of both eyes) recorded the vertical- and horizontal
electro-oculogram (EOG). Two electrodes were placed on the mastoids, of which the left was used to reference the signal on-line.
Impedance was kept below 5 k for mastoid and scalp electrodes,
and below 10 k for EOG electrodes. The EEG signal was analyzed
using Brain Vision Analyzer 2.0. The signal was referenced off-line
to an average of the two mastoid electrodes and band-pass ltered
(Butterworth Zero Phase Filter, 0.130 Hz, 24 dB/octave). ERPs were
time-locked to auditory onset of the third syllable and the raw
data were segmented into 1100 ms epochs (i.e., 200 ms before, and
900 ms after third syllable onset). After EOG correction [25], segments that included artifacts were rejected (we allowed a maximal
voltage step of 50 V/ms, a 100 V maximal difference in a 200 ms
interval, a minimal/maximal amplitude of 50 V in a segment,
and the lowest allowed activity in a 100 ms interval was .5 V). The
average proportion of trials that survived artifact rejection was 95%
(with a range between 91% and 98% per electrode, and 94% and 95%
per condition).
ERPs for the dot-trials, during which a response was required
(97% of the dots were detected), were excluded from analyses
and the remaining ERPs were averaged separately for words and
pseudowords for all three delay conditions (no delay, 440 ms,
800 ms) and base-line corrected (200 ms before third syllable
onset).

In each delay condition, the difference between ERPs obtained

with words versus pseudowords was investigated through pointwise t-tests (see Fig. 1) that were corrected via temporal
signicance thresholds [26]. In this procedure, the autocorrelation
and length of the time-window are critical parameters. However,
since it is unclear how the autocorrelation should be estimated [27],
we applied the corrections four different ways: assuming low (.3)
or high (.9) autocorrelation, and in either one 0300 ms epoch or six
consecutive 50 ms windows. The four approaches yielded comparable results (see Fig. 1). The N200 effect (i.e., the wordpseudoword
difference) was compared across delay conditions in an overall
ANOVA on 50 ms time-windows in between 100 and 300 ms. We
also conducted one-way within-subject ANOVAs on the differencewaves at each electrode and time-point (in the 0300 ms epoch)
with delay condition (naturally timed, 440 ms, 800 ms) as the independent variable.
4. Results
As can be seen in Fig. 1a, there was a clear effect of stimulus lexicality in all conditions: Independent of whether the third syllable
was naturally timed or delayed relative to the second syllable, items
in which the third syllable produced a word yielded more positive
(yet similarly shaped) ERPs than items in which the third syllable produced a pseudoword. Early (300 ms) lexicality effects were

Please cite this article in press as: M. Baart, A.G. Samuel, Early processing of auditory lexical predictions revealed by ERPs, Neurosci.
Lett. (2014), http://dx.doi.org/10.1016/j.neulet.2014.11.044

212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227

228

229
230
231
232
233
234

G Model
NSL 30962 15
4

ARTICLE IN PRESS
M. Baart, A.G. Samuel / Neuroscience Letters xxx (2014) xxxxxx

266

explored by point-wise t-tests in each delay condition (see Fig. 1b).

Effects for naturally timed items started at around 60 ms in centroparietal electrodes over the right hemisphere and lasted 40 ms.
A more fronto-centrally oriented difference between words and
pseudowords was observed in an 80 ms time-window starting at
around 170 ms. At around 280 ms, differences became more widespread. Lexicality effects in both delay conditions started somewhat
later than in the naturally timed condition (at around 200 ms) and
were more wide-spread and prominent for the 800 ms delay condition than the 440 ms condition.
Next, we calculated the lexicality effect for each delay condition (i.e., the pseudowordword difference waves, see Fig. 1c.) and
averaged the difference in four 50 ms time-bins from 100 to 300 ms
(which is the time-frame that captures the N200 effect and is consistent with the positive and negative peaks in the ERPs visible in
Fig. 1a). We submitted the data to a 3 (delay condition: naturally
timed, 440, 800) 4 (time window: 100150 ms, 150200 ms,
200250 ms, 250300 ms) 9 (electrode: Fz, FC1, FC2, C3, Cz, C4,
CP1, CP2, Pz1 ) ANOVA. There was a main effect of time window, F(3,45) = 3.08, p = .04, p 2 = .17, because the lexicality effect
increased from .34 V in the rst time-window to .91 V in
the second time-window, t(15) = 2.22, p = .04. However, the N200
effect was quite consistent as none of the other comparisons across
time-windows reached signicance (ps > .06). The ANOVA showed
no other main or interaction effects (ps > .09) indicating that the
effect of stimulus lexicality was alike across delay conditions and
electrodes. One-way point-wise within-subject ANOVAs on the
pseudowordword difference scores (see Fig. 1d), showed prominent main effects from 164 to 214 ms, because the lexicality effect
for the 440 ms delay was smaller than in the other two conditions
(see Fig. 1e for the pair-wise follow-up tests on the average activity
across the electrodes that showed a signicant effect).

267

5. Discussion

235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265

268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294

Previous studies have reported that activation of lexical representations based on semantic/syntactic context is associated with
the N200 effect. In the current study, we addressed two questions:
(1) can we nd evidence that single word level lexical predictions
are reected in an N200 effect? And if so, (2) is the pattern of lexical
activation robust against a violation of temporal coherence? Our
naturally timed items yielded an unambiguous N200 effect that
was superimposed on a negative peak, as observed before [5,6].
Furthermore, we observed similarly sized and shaped difference
waves across all three conditions. Our ndings are consistent with
reports of single word lexical effects in oddball paradigms [1113],
and demonstrations of similarly timed and sized N200 effects
(occurring at 150250 ms and ranging between .71 and .94 V)
for lexical predictions generated by sentence contexts [5,6]. Similarities across contexts presumably arise because the N200 effect
reects phonological violation of lexical predictions [e.g.,28], which
is essentially the case in both sentential and single word contexts.
As mentioned, the N200 is likely to be functionally distinct from the
subsequent N400 (that is also related to semantic/syntactic context) [e.g.,5,6,9,29]; Hagoort [29] has proposed that the N200 effect
reects lexical selection at the interface of lexical form and content
(meaning), whereas the N400 is driven by content only.
Critically, the effect of lexicality remained intact when we
delayed the third syllable by 440 or 800 ms, even though the
ERPs for naturally timed items were clearly different from those in
both delay conditions. This may seem surprising because the system usually does not need to retain within-word lexical predictions

The N200 effect was largest at Cz (i.e., .76 V) and we therefore included the
9 mid-central electrodes in the ANOVA.
1

about the upcoming sounds for hundreds of milliseconds. However

this does not necessarily imply that lexical information is transient
by nature, and our ndings suggest that lexically generated predictions about upcoming sounds remain effective until overridden by
new input.
Even though we observed a lexicality effect at around 200 ms
across all delay conditions, there were differences between
conditions. As indicated in Fig. 1a, the lexicality effect for
the naturally-timed items reached signicance very early, in a
50100 ms window. MacGregor et al. [1] used single-syllable
items in which lexicality of the stimulus was dened by the nal
plosive and observed similarly early effects of lexicality (5080 ms
after the point where lexical status was determined). Given the
absence of such effects in our delay conditions, the natural context
possibly not only generated predictions about what to expect, but
also about when to expect it. For example, hearing p after shee
in a naturally-timed sheep token conrms the lexical prediction
as well as the temporal prediction about when it should occur.
The same is true for the natural condition in the current study in
which lexical status of the stimulus was always determined at the
naturally timed third syllable onset. In contrast, in our delay conditions, participants knew that lexical information would become
available at onset of the third syllable, but their normal predictions
about when this would happen were violated. Under these conditions, the lexical effect remained robust in the 200 ms range, but
the early part of it disappeared. A Bayesian perspective may provide an interesting hypothetical frame-work on this issue: The prior
belief about encountering pseudowords in the world (e.g., hearing a speaker in a language unknown to the listener) is likely to
be higher than encountering the unnatural delay of third syllables
as used here. In contrast, the probability of encountering a pseudoword in the experiment on any given trial was lower (.50) than
encountering a delay (.67). Thus, Bayesian inference (i.e., updating
of the prior based on evidence) may have been relatively large for
delayed items, possibly reducing early lexical processes.
However, the 3-way comparison across conditions (Fig. 1d) did
not show prominent differences among conditions in the <100 ms
window, going against the Bayesian hypothesis as sketched above.
In fact, the effect in the naturally-timed and 800 ms delay conditions were quite similar in shape (see Fig. 1c). In contrast, the
440 ms delay condition showed a positive trend in the data
(already apparent before third-syllable onset, see Fig. 1a) and a
100 ms delay in the difference waves relative to both other conditions, despite the similar shape of the difference waves across
conditions (see the 0200 ms window in Fig. 1c and Fig. 1e). Given
that the onset of the lexicality effect for the items with a 800 ms
delay was similar to the naturally timed items, the delayed onset of
the effect in the 440 ms condition is unlikely to be caused by the
delay in the stimulus per se. Instead, it is possible that the delayed
effects for the 440 ms condition are related to an increased temporal uncertainty in this condition. That is, when the default state (i.e.,
the naturally-timed syllable) did not occur, listeners did not know
whether the third syllable would occur at 440 ms or at 800 ms.
However, as soon as the 440 ms window had passed, only one
alternative remained.
Clearly however, the between condition differences are less
prominent than the consistent lexicality effects: we observed evidence for lexical processing starting at 200 ms, and these effects
were to a large extent similar in shape and size across all three
delay conditions. Note that our words (and pseudowords) were
drawn from a small stimulus set and presented many times, which
could have caused saturation (which holds for auditory as well as
printed items, see [e.g.,30,31]). Saturation could work both ways:
the lexical status of words, as well as the non-lexical status of pseudowords, may become less clear with repetition. However, since we

Please cite this article in press as: M. Baart, A.G. Samuel, Early processing of auditory lexical predictions revealed by ERPs, Neurosci.
Lett. (2014), http://dx.doi.org/10.1016/j.neulet.2014.11.044

295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359

G Model
NSL 30962 15

ARTICLE IN PRESS
M. Baart, A.G. Samuel / Neuroscience Letters xxx (2014) xxxxxx

374

observed clear effects of lexicality even with saturation possibly

working against us, it seems likely that our effects are genuine.
There are some parallels between the current results and recent
ERP investigations of audiovisual speech integration. In those studies, visual context that provided temporal predictions modulated
auditory processing about 100 ms post-stimulus [32], much as the
naturally-timed stimuli here did. And speech-specic audiovisual
integration was observed at the P2 [33], with effects of phonetic
stimulus congruency (i.e., whether the speech sound phonetically matched the visual speech context) observed approximately
200 ms after the congruency becomes apparent [3337]. The current experiment shows that lexicality effects are sufciently robust
in time to be investigated as a superimposed effect at the P2
peak, which opens up future possibilities to investigate interactions
between lip-read and lexical speech context.

375

Acknowledgements

360
361
362
363
364
365
366
367
368
369
370
371
372
373

379

This work was funded by Rubicon grant 446-11-014 by the

Netherlands Organization for Scientic Research (NWO)to MB, and
MINECO grant PSI2010-17781 from the Spanish Ministry of Economics and Competitiveness to AGS.

380

References

376
377
378

[13]

[14]
[15]

[16]
[17]
[18]
[19]

[20]
[21]

[22]

[23]
[24]

381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414

[1] L.J. MacGregor, F. Pulvermller, M. van Casteren, Y. Shtyrov, Ultra-rapid

access to words in the brain, Nat. Commun. 3 (2012) 711.
[2] N.Y. Egorova, Y. Shtyrov, F. Pulvermller, Early and parallel processing of
pragmatic and semantic information in speech acts: neurophysiological
evidence, Front. Hum. Neurosci. 7 (2013) 86.
[3] F. Pulvermller, Y. Shtyrov, O. Hauk, Understanding in an instant:
neurophysiological evidence for mechanistic language circuits in the brain,
Brain Lang. 110 (2009) 8194.
[4] Y. Shtyrov, F. Pulvermller, Early MEG activation dynamics in the left
temporal and inferior frontal cortex reect semantic context integration, J.
Cognit. Neurosci. 19 (2007) 163316642.
[5] D. van den Brink, P. Hagoort, The inuence of semantic and syntactic context
constraints on lexical selection and integration in spoken-word
comprehension as revealed by ERPs, J. Cognit. Neurosci. 16 (2004)
106810684.
[6] D. van den Brink, C.M. Brown, P. Hagoort, Electrophysiological evidence for
early contextual inuences during spoken-word recognition: N200 versus
N400 effects, J. Cognit. Neurosci. 13 (2001) 967985.
[7] C. Van Petten, S. Coulson, S. Rubin, E. Plante, M. Parks, Time course of word
identication and semantic integration in spoken language, J. Exp. Psychol.
Learn. 25 (1999) 394417.
[8] J.F. Connolly, N.A. Phillips, Event-related potential components reect
phonological and semantic processing of the terminal word of spoken
sentences, J. Cognit. Neurosci. 6 (1994) 256266.
[9] J.F. Connolly, N.A. Phillips, S.H. Stewart, W.G. Brake, Event-related potential
sensitivity to acoustic and semantic properties of terminal words in
sentences, Brain Lang. 43 (1992) 118.
[10] R. Ntnen, A.W.K. Gaillard, S. Mntysalo, Early selective-attention effect in
evoked potential reinterpreted, Acta Psychol. 42 (1978) 313329.
[11] C.M. Pettigrew, B.E. Murdoch, C.W. Ponton, S. Finnigan, P. Alku, J. Kei, R.
Sockalingam, H.J. Chenery, Automatic auditory processing of English words as
indexed by the mismatch negativity, using a multiple deviant paradigm, Ear
Hearing 25 (2004) 284301.
[12] F. Pulvermller, T. Kujala, Y. Shtyrov, J. Simola, H. Tiitinen, P. Alku, K. Alho, S.
Martinkauppi, R.J. Ilmoniemi, R. Ntnen, Memory traces for words as

[25]
[26]
[27]

[28]

[29]

[30]

[31]
[32]

[33]

[34]

[35]

[36]

[37]

revealed by the mismatch negativity, Neuroimage 14 (2001)

415
607616.
416
S. van Linden, J.J. Stekelenburg, J. Tuomainen, J. Vroomen, Lexical effects on
417
auditory speech perception: an electrophysiological study, Neurosci. Lett. 420
418
(2007) 4952.
419
W.F. Ganong, Phonetic categorization in auditory word perception, J. Exp.
420
Psychol. Human 6 (1980) 110125.
421
R. Ntnen, P. Paavilainen, K. Alho, K. Reinikainen, M. Sams, Do event-related
422
potentials reveal the mechanism of the auditory sensory memory in the
423
human brain? Neurosci. Lett. 98 (1989) 217221.
424
P. Gagnepain, R.N. Henson, M.H. Davis, Temporal predictive codes for spoken
425
words in auditory cortex, Curr. Biol. 22 (2012) 615621.
426
A.G. Samuel, Phonemic restoration: insights from a new methodology, J. Exp.
427
Psychol. Gen. 110 (1981) 474494.
428
C.M. Connine, C. Clifton, Interactive use of lexical information in speech
429
perception, J. Exp. Psychol. Human 13 (1987) 291299.
430
C.M. Connine, Constraints on interactive processes in auditory word
431
recognition: the role of sentence context, J. Mem. Lang. 26 (1987)
432
527538.
433
T.B. ORourke, Holcomb, electrophysiological evidence for the efciency of
434
spoken word processing, Biol. Psychol. 60 (2002) 121150.
435
P. Praamstra, A.S. Meyer, W.J.M. Levelt, Neurophysiological manifestations of
436
phonological processing: latency variation of a negative ERP component
437
timelocked to phonological mismatch, J. Cognit. Neurosci. 6 (1994) 204219.
438
A. Duchon, M. Perea, N. Sebastin-Galls, M. Carreiras, EsPal: one-stop
439
shopping for Spanish word properties, Behav. Res. Methods 45 (2013)
440
12461258.
441
L.B. Astheimer, L.D. Sanders, Predictability affects early perceptual processing
442
of word onsets in continuous speech, Neuropsychologia 49 (2011) 35123516.
443
V. Marian, J. Bartolotti, S. Chabal, A. Shook, CLEARPOND: cross-Linguistic
444
easy-access resource for phonological and orthographic neighborhood
445
densities, PLoS ONE 7 (2012) e43230.
446
G. Gratton, M.G. Coles, E. Donchin, A new method for off-line removal of
447
ocular artifact, Electroencephalogr. Clin. Neurophysiol. 55 (1983) 468484.
448
D. Guthrie, J.S. Buchwald, Signicance testing of difference potentials,
449
Psychophysiology 28 (1991) 240244.
450
Q5451
V. Piai, K. Dahlsltt, E. Maris, Statistically comparing EEG/MEG waveforms
through successive signicant univariate tests: how bad can it be?
452
Psychophysiology (2014) (in press).
453
J.F. Connolly, S.H. Stewart, N.A. Phillips, The effects of processing
454
requirements on neurophysiological responses to spoken sentences, Brain
455
Lang. 39 (1990) 302318.
456
P. Hagoort, The fractionation of spoken language understanding by measuring
457
electrical and magnetic brain signals, Philos. Trans. R. Soc. B: Biol. Sci. 363
458
(2008) 10551069.
459
J. Kounios, S.A. Kotz, P.J. Holcomb, On the locus of the semantic satiation
460
effect: evidence from event-related brain potentials, Mem. Cognit. 28 (2000)
461
13661377.
462
X. Tian, D.E. Huber, Playing duck duck goose with neurons: change detection
463
through connectivity reduction, Psychol. Sci. 24 (2013) 819827.
464
J. Vroomen, J.J. Stekelenburg, Visual anticipatory information modulates
465
multisensory interactions of articial audiovisual stimuli, J. Cognit. Neurosci.
466
22 (2010) 15831596.
467
M. Baart, J.J. Stekelenburg, J. Vroomen, Electrophysiological evidence for
468
speech-specic audiovisual integration, Neuropsychologia 53 (2014)
469
115121.
470
J.J. Stekelenburg, J. Vroomen, Neural correlates of multisensory integration of
471
ecologically valid audiovisual events, J. Cognit. Neurosci. 19 (2007)
472
19641973.
473
V. Klucharev, R. Mttnen, M. Sams, Electrophysiological indicators of
474
phonetic and non-phonetic multisensory interactions during audiovisual
475
speech perception, Cognit. Brain Res. 18 (2003) 6575.
476
V. van Wassenhove, K.W. Grant, D. Poeppel, Visual speech speeds up the
477
neural processing of auditory speech, Proc. Natl. Acad. Sci. U. S. A. 102 (2005)
478
11811186.
479
J. Besle, A. Fort, C. Delpuech, M.H. Giard, Bimodal speech: early suppressive
480
visual effects in human auditory cortex, Eur. J. Neurosci. 20 (2004) 22252234.
481

Please cite this article in press as: M. Baart, A.G. Samuel, Early processing of auditory lexical predictions revealed by ERPs, Neurosci.
Lett. (2014), http://dx.doi.org/10.1016/j.neulet.2014.11.044