Journal of Neurolinguistics 42 (2017) 12e22

Contents lists available at ScienceDirect

Journal of Neurolinguistics
journal homepage: www.elsevier.com/locate/jneuroling

Comprehension of scientific metaphors: Complementary

processes revealed by ERP
Xuemei Tang a, b, *, Senqing Qi b, Xiaojuan Jia b, Botao Wang b, Wei Ren b, **
a
Foreign Languages College, Anhui Polytechnic University, Wuhu, 241000, China
b
Key Laboratory of Modern Teaching Technology, Ministry of Education, Shaanxi Normal University, Xi'an, 710062, China

a r t i c l e i n f o a b s t r a c t

Article history: In order to complement current debates and open questions in the field of figurative
Received 7 December 2015 language comprehension, the current paper investigated how metaphors from different
Received in revised form 5 November 2016 kinds of contexts are electrophysiologically processed. For the first time, we compared
Accepted 7 November 2016
comprehension of scientific metaphors with that of conventional ones using event-related
Available online 15 November 2016
potentials (ERPs). Scientific metaphors have the unique semantic structure with two
different contexts and inference involvement for knowledge-understanding. By time-
Keywords:
locking the N400 and later LPC time windows, the present study shows the different
Scientific metaphors
ERP
stages of meaning integration when comprehending figurative language. The N400 am-
N400 plitudes to the last word of the sentence varied as a function of expression type in a graded
Late negativity manner increasing from literal sentences to conventional metaphors, and to scientific
Hemispheric asymmetry metaphors. N400s elicited by scientific metaphors showed central-parietal-right-biased
scalp distributions. Scientific metaphors also elicited a late negativity in the LPC window
simultaneously on the left and right hemispheres suggesting further attempts to integrate
meaning when scientific inference is involved. These findings of scientific metaphors
might test some related metaphor-processing models to a greater extent. The reported
results also demonstrate that the left and right hemispheres of the brain work together in a
complex dynamic pattern during literal and figurative language comprehension and that
the right hemisphere is necessarily involved, but not sufficient, for understanding meta-
phoric expressions.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction

The claim whether figurative and literal meanings are accessed concurrently or not is based largely on the comparison
between metaphors from the daily language and metaphors from the poetic language. Although metaphors are common in
the everyday and poetic language to talk about a wide range of subjects, metaphors are also pervasive in the scientific
language. Metaphors are popular in science and help scientists understand and communicate the intricacies, beauty and
strangeness of the natural world (Aubusson, Ritchie, & Harrison, 2006). Many of the significant advances in science utilized
metaphor as inference or reasoning tools. For example, Kekule , a German Chemist, derived his idea for a benzene ring from an
image of a snake biting its tail. Huygens, a Dutch physicist, used water waves to theorise that light is wavelike. However, in the

* Corresponding author. Foreign Languages College, Anhui Polytechnic University, Wuhu, 241000, China.
** Corresponding author. Key Laboratory of Modern Teaching Technology, Shaanxi Normal University, Xi'an, 710062. China.
E-mail addresses: tangxuemei@ahpu.edu.cn (X. Tang), renwei@snnu.edu.cn (W. Ren).

http://dx.doi.org/10.1016/j.jneuroling.2016.11.003
0911-6044/© 2016 Elsevier Ltd. All rights reserved.
 X. Tang et al. / Journal of Neurolinguistics 42 (2017) 12e22 13

current debate, the comprehension of scientific metaphors, such as those used in chemistry and physics, has been, for the
most part, overlooked. In fact, though both are creative and novel compared to conventional metaphors, scientific metaphors
and poetic metaphors have quite distinguishable properties. Firstly, scientific metaphors have a more complicated contextual
structure. For example, in the sentence “The circuit is a ladder”, the source domain (ladder) is from the daily context while the
target domain (circuit) is from the scientific context. In contrast, for conventional and poetic metaphors, mostly only daily
context is involved in the two semantic domains. This more complicated contextual structure of scientific metaphors implies
more difficulties in integrating the target and source domains. Secondly, compared to the emotion-arousing function of
conventional and poetic metaphors, scientific metaphors have its unique knowledge-understanding property. By triggering a
sort of inference from the source domain to the target domain, scientific metaphors could quite probably help readers quickly
understand certain new knowledge. Therefore, the aim of the present research is to enrich the debate of the neural mech-
anism of metaphor processing by using scientific metaphors with its unique properties instead of poetic metaphors as the
novel ones in the ERP experiment.

1.1. Metaphor processing models

Some psycholinguistic models may help illuminate those hypothesized conceptual mapping of metaphor processing
better. According to the structural-mapping model (Gentner, 1983) and the career of a metaphor model (Bowdle & Gentner,
2005; Gentner & Wolff, 1997), mapping is a process of analogical comparison of similarities between the source and the
target. When processing conventional metaphors, the analogical comparison might be quicker and easier because the two
domains usually become associated with each other. But, when processing unfamiliar metaphors, the analogical comparison
tends to be slower and more difficult for the lack of a well-defined association between the two domains. For example, in
“Energy flows through an ecosystem”, the target concept ENERGY and the source concept WATER are aligned by the predicate
“flows through”, and therefore, the invisible “energy” is understood as the visible “water”. Compared to the mapping between
BRIDGE and LANGUAGE in the conventional metaphor “The language is a bridge”, the mapping between WATER and ENERGY is
more difficult. These two models predict that both conventional and scientific metaphors should be somewhat cognitively
taxing due to an initial stage for structural alignment that is needed for mappings. But processing scientific metaphors should
be more difficult than processing conventional metaphors due to always having to compare concepts covering two different
contexts and generate scientific inference on-line. Therefore, compared to poetic metaphors used by most previous studies as
the novel ones, scientific metaphors should have its own advantages to be used as novel metaphors to study the possible
mapping process between different conceptual domains. Scientific metaphors cover the daily and scientific contexts, which
brings more difficulties in integrating the source and target domains thus probably showing better the differentiation be-
tween novel metaphors and conventional metaphors. Furthermore, the later knowledge-understanding inference in un-
derstanding scientific metaphors should probably distinguish the different processing stages of conceptual mapping more
clearly.
Moreover, some other models may help understand different hemispheric functions in the semantic processing of met-
aphors. According to the graded salience hypothesis (Giora, 2003), the degree of meaning salience determines the time-
course of meaning processing. Salience here refers to those meanings foremost in speakers’ minds at time of speaking,
which are characterized by conventionality, prototypicality, familiarity, and frequency. The figurative meaning in conven-
tional metaphors is commonly more salient than the literal one, and is processed first in the left hemisphere. In contrast,
when a novel or unfamiliar metaphor is encountered, the salient meaning is the literal one, and the figurative meaning is
inferred later by contextual mechanisms and is processed mainly in the right hemisphere. Another more specific psycho-
linguistic theory addressing hemispheric functions in semantic processing is the Fine-Coarse Semantic Coding Theory
(Beeman, 1998, pp. 255e284, Beeman, 2005), which has suggested that the two cerebral hemispheres process language
qualitatively differently. The right hemisphere loosely activates and maintains larger semantic fields containing more distant
associates and more unconventional meanings (coarse semantic coding) whereas the left hemisphere focuses on a single
dominant interpretation (fine semantic coding). These two models together have some important implications for using
scientific metaphors as novel ones in the comparison with conventional metaphors. These two models predict that literal
language is processed primarily in the dominant left hemisphere, while the right hemisphere should have faster access to the
figurative meaning of novel metaphors. As mentioned above, the two semantic domains of scientific metaphors covering two
different contexts are so distant from each other and the scientific inference is needed to achieve the integration of these two
domains to get some sort of new knowledge. The semantic structure of scientific metaphors could involve a selective pro-
cessing of non-salient meanings in the right hemisphere. We hypothesize that by comparing conventional metaphors with
those in a creative and scientific context, evidence from new perspectives could be gathered to complement the debate of the
actual cognitive processing of metaphors.

1.2. Event-related brain potentials (ERPs) and metaphor comprehension

ERPs can be utilized to measure the temporal dynamics of metaphor comprehension through two ERP waveforms: the
N400 and the late positive component (LPC). Most of the ERP studies on online metaphor processing have closely observed
the N400, a centro-parietal deflection peaking around 400 ms after stimulus onset, which is sensitive to violations of semantic
relatedness (Kutas & Hillyard, 1980). According to the conceptual blending view (Coulson & Petten, 2002; Yang, Bradley, Huq,
14 X. Tang et al. / Journal of Neurolinguistics 42 (2017) 12e22

Wu, & Krawczyk, 2013), the amplitude of the N400 has been regarded as an important index showing how easy or difficult the
stored information of a word could be retrieved for the meaning integration (Kutas & Federmeier, 2011). Compared to poetic
metaphors, scientific metaphors with more distant connection between the two semantic domains should show higher
degree of semantic unrelatedness and should be more difficult to retrieve information for meaning integration. Processing
scientific metaphors is expected to take more time for selection and reselection of appropriate inferences to understand a
type of new knowledge. Adopting scientific metaphors as the novel ones may better show the different N400 and LPC effects
between novel metaphors and conventional metaphors when there are more dramatic differences of relatedness between the
two semantic domains.
Much has been known about increased N400 amplitudes for novel metaphors (mostly poetic metaphors) compared to
conventional metaphors and literal expressions (Arzouan, Goldstein, & Faust, 2007a; Grauwe, Swain, Holcomb, Ditman, &
Kuperberg, 2010; Lai, Curran, & Menn, 2009; Rutter et al., 2012), suggesting higher cognitive cost for the understanding of
novel metaphors. Conventional metaphors evoke larger N400 amplitudes than literal expressions. The graded N400 differ-
ence may be interpreted in terms of increased mapping difficulties for metaphoric compared to literal meanings, whereby the
underlying language processing mechanisms are assumed to be the same (Goldstein, Arzouan, & Faust, 2012; Lai & Curran,
2013; Lai et al., 2009; Rutter et al., 2012). But other studies reported similar N400 amplitudes for literal expressions and
conventional metaphors (Arzouan et al., 2007a; Iakimova, Passerieux, Laurent, & Hardy-Bayle, 2005), which do not tax
conceptual mappings because of their high salience and familiarity.
Regarding later time windows of processing stages following the retrieval of the semantic information of words, findings
in the literature are less consistent. The LPC usually appears following the N400 and has been thought to reflect sentence-
level integration (Kaan, Harris, Gibson, & Holcomb, 2000) or reanalysis (Friederici, Steinhauer, & Frisch, 1999) and mem-
ory retrieval processes (Paller & Kutas, 1992). Larger LPCs for metaphoric sentences have been reported, compared to literal
and literal-mapping endings (Grauwe et al., 2010). This was interpreted as reflecting recovery and integration of additional
material from semantic memory. In contrast, smaller LPCs for metaphoric meanings, especially novel metaphoric meanings,
have also been reported implying a possible late negativity suggesting further attempts to integrate meaning in a figurative
context (Arzouan et al., 2007a, 2007b; Kazmerski, Blasko, & Dessalegn, 2003; Pynte, Besson, Robichon, & Poli, 1996; Rutter
et al., 2012). Other studies did not find significant LPC differences between literal expressions and conventional meta-
phors, or between conventional and unfamiliar metaphors. However, they do report an effect of context relevance. Metaphors
that were preceded by irrelevant context elicited larger N400s and smaller LPCs than those following relevant context,
regardless of familiarity (Iakimova et al., 2005).
Besides, the two hemispheres of the brain are also believed to play different roles at different stages during language
processing. In particular, the involvement of the right hemisphere in understanding figurative meanings has been studied
extensively in ERP, fMRI or PET researches. Although the right hemisphere has shown its importance in language processing,
the results remain rather inconsistent. For ERP studies, the right hemisphere activity was expected to be manifested mainly at
the N400 and LPC time windows, which reflect linguistic integration processes. Some studies report symmetric scalp dis-
tribution of ERP components (Coulson & Petten, 2002, 2007), and some report a small right topographic effect (Goldstein
et al., 2012; Sotillo et al., 2005). Besides, the empirical findings of neuroimaging studies are also controversial: several
experimental studies have found evidence for lateralization of figurative language processing (Diaz, Barrett, & Hogstrom,
2011; Mashal, Vishne, Laor, & Titone, 2013; Schmidt & Seger, 2009; Yang, 2014; Yang, Edens, Simpson, & Krawczyk, 2009)
while others did not find selective right hemisphere activations for non-literal language (Bohrn, Altmann, & Jacobs, 2012;
Chen, Widick, & Chatterjee, 2008; Kasparian, 2013; Rapp, Mutschler, & Erb, 2012; Uchiyama et al., 2012).
The controversial findings between studies may result from the influence of moderator variables on the lateralization of
figurative language processing, such as familiarity, difficulty, context or task (Schmidt & Seger, 2009). So, putting scientific
metaphors with a different contextual structure and unique knowledge-understanding inference into the current discussion
might throw a new light into the related issue.

1.3. The present study

So far, ERP studies on metaphors have given inconsistent results. The present study intends to provide a new perspective
to observe the perception, comprehension, and neural representation of metaphoric language as a special form of pragmatic
language. In the present study, for the first time, scientific metaphors have been employed as the unfamiliar metaphors. This
may help optimize the experimental design by widening the scope of metaphor stimuli and improving the comparability
between metaphors and literal expressions as well as conventional metaphors and unfamiliar metaphors. In our experiment,
the three-element structure ‘X 是 (is) Y’ was adopted. Based on some previous studies, we collected conventional metaphors
from dictionaries, such as “恋爱是咖啡 (Love is coffee)” created from the concept “LOVE IS AN ENTITY”. More importantly, we
created scientific metaphors with the same syntactic structure from technical terms covering major disciplines of natural
science, such as mathematics, physics, chemistry, biology, etc. For example, “电路是阶梯 (The circuit is a ladder)” was created
from the physical term (ladder circuit), “离子是碎片 (An ion is a fragment)” from the chemical term “fragment ion”. Literal
sentences were also created like “教授是学者 (A professor is a scholar)” for comparison. To ensure that conventional meta-
phorical expressions are familiar and interpretable, and that scientific ones are unfamiliar but interpretable, pretests of
meaningfulness, figurativeness and familiarity were conducted. Pretest details are shown in experimental procedures.
 X. Tang et al. / Journal of Neurolinguistics 42 (2017) 12e22 15

The goal of the present study is to test the validity of those models of metaphor processing previously discussed by
adopting scientific metaphors as the novel ones. Our predictions are as follows. Compared to poetic metaphors used in most
previous studies, scientific metaphors covering scientific and daily contexts and involving knowledge-understanding inter-
ference mechanisms are expected to show the different stages of cognitive processing more clearly. Firstly, in the N400
window, due to the complicated contextual structure, scientific metaphors might elicit more negative N400 resulting from
more difficulties in retrieving the stored the information for meaning integration. Secondly, in the LPC window, due to the
later knowledge-understanding inference, scientific metaphors might show either more positive LPCs for greater sentence-
level integration or more negative late negativity for further integration of meaning. Thirdly, scientific metaphors with looser
semantic relations might elicit greater right hemisphere activity relative to conventional metaphoric and literal expressions.

2. Experimental procedures

2.1. Participants

Twenty-five (11 males, 14 females, average age 21.7) right-handed, native Chinese-speaking students at the Shaanxi
Normal University were enrolled in the study for monetary compensation. All the participants were from science disciplines
considering the possible difficulties in understanding the academic knowledge related to scientific metaphors. Exclusion
criteria were sinistrality, past or present psychiatric illness or neurological disorder or major head injury. Handedness was
tested by means of the Edinburgh Handedness Inventory (Oldfield, 1971). The experimental standards of the study were
approved by the local Review Board for Human Participant Research. Each subject provided written informed consent before
participating. Data from four subjects were not included in the analyses due to an insufficient number of usable trials leading
to a final sample size of 21 subjects (8 males, 13 females).

2.2. Stimuli

A list of 144 Chinese sentences was formed consisting of three categories: scientific metaphoric, conventional metaphoric
and literal, with 48 sentences in each sentence type (see Table 1 for examples). Categories were matched for grammar and
syntactic structure. All sentences were of the “X 是 (is) Y” format to exclude possible confounding factors such as sentence
length or complex syntactic processing. As described in the Introduction, conventional metaphors were created from dic-
tionaries, and scientific metaphors from scientific terms or concepts covering major disciplines of natural science, such as
mathematics, physics, chemistry, biology, etc. Literal sentences were also created for comparison. Examples of stimuli sen-
tences are shown in Table 1.
Prior to the neurophysiological study, several pilot surveys were conducted by raters who did not participate in the ERP
experiment. All participants were from science disciplines in order to avoid the possible difficulties in understanding the
scientific terms related to the scientific metaphoric category.
Firstly, 40 raters were presented with a list of sentences and asked to decide whether each expression is meaningful or not
on a 1e3 scale (1 ¼ not meaningful, 2 ¼ somewhat meaningful, 3 ¼ highly meaningful). According to the results of the
meaningfulness judgment, sentences rated by at least 80% of the judges as metaphorically/literally plausible (>2.5) were
selected as expressions with either a metaphoric or a literal meaning. Then those selected sentences were rated by another
group of 40 raters on a 1e3 scale regarding their figurativeness (1 ¼ not figurative, 2 ¼ somewhat figurative, 3 ¼ highly
figurative). On the final list of stimuli, 48 expressions with an average of less than 1.5 were chosen as literal expressions
whereas 96 expressions with an average of more than 2.5 were chosen as either scientific metaphors (48) or conventional
metaphors (48). Both scientific metaphors and conventional metaphors were much more figurative than literal expressions
(Fs > 200). Finally, another group of 40 raters was asked to rate all the 144 sentences on a 1e7 scale ranging from 1 (highly

Table 1
Sample stimuli.

Scientific metaphors 电路是阶梯。 The circuit is a ladder.

离子是碎片。 An ion is a fragment.
函数是斜坡。 The function is a ramp.
螺栓是鱼尾。 The bolt is a fishtail.
声音是波浪。 Sound is wave.
Conventional metaphors 恋爱是咖啡。 Love is coffee.
杭州是天堂。 Hangzhou is heaven.
家庭是港湾。 Home is a harbor.
手机是伙伴。 The cellphone is my partner.
语言是桥梁。 The language is a bridge.
Literal expressions 教授是学者。 A professor is a scholar.
汉语是语言。 Chinese is a language.
北京是首都。 Beijing is the capital.
蚂蚁是昆虫。 An ant is an insect.
小狗是宠物。 The dog is a pet.
16 X. Tang et al. / Journal of Neurolinguistics 42 (2017) 12e22

nonfamiliar) to 7 (highly familiar). Both literal expressions and conventional metaphors were more familiar than scientific
metaphors (Fs > 600). In the present study, compared to conventional metaphors, scientific metaphors were novel ones. The
results of the pretests of stimuli sentences are shown in Table 2.

2.3. Procedure

The experiment took place in a sound-attenuated, electrically shielded room. The sentences were presented in white
colour on a black background word by word in a quasi-random order. Stimuli on each trial were presented in the following
time sequence: fixation cross (800 ms), blank (200e500 ms), subject (1000 ms), verb (600 ms), blank (200e500 ms), object
(1000 ms) and question mark (3000 ms). At the sight of the question mark, participants gave their judgments about whether
the sentence was metaphoric or not by pressing a corresponding key with the right and left index fingers. Response period
was limited to 3s and was followed by a 1s intertrial interval. The overall sequence of events for a trial is illustrated in Fig. 1.
Before the ERP experiment, in order to make sure that the academic knowledge involved in scientific metaphors could be
understood during the experiment, participants firstly were asked to read a list of scientific terms, from which scientific
metaphors were created, together with their brief explanations. Before the main session of the experiment, there was a brief
practice session to familiarize the participants with the experimental procedure.

2.4. Electrophysiological recording

Scalp voltages were collected with the CURRY 7 system (Compumedics Neuroscan, Texas, USA) with 64 Ag/AgCl electrodes,
monitored by the CURRY recorder software and connected to a SynAmp amplifier (Compumedics Neuroscan, Texas, USA).
Amplified analog voltages were digitized at 1000 Hz. Impedances of individual sensors were kept below 5 kU. Eye movements
were monitored through bipolar electrodes which were placed above and below the right eye, as well as at the left and right
canthi. EEG was measured online with reference to the left mastoid, with a ground electrode on the medial frontal aspect, and
later was analyzed offline with re-reference to an average of the left and right mastoids.
EEG was analyzed with the SCAN 4.5 software (Compumedics Neuroscan, Texas, USA) and Matlab using the ERPLAB
toolbox (Lopez-Calderon & Luck, 2014). The EEG was digitally filtered at 0.1e30 Hz bandpass. Eye movements were corrected
with an ocular artifact correction algorithm (Gratton, Coles, & Donchin, 1983). Artifacts with amplitudes exceeding ±75 mV
were removed from analyses. ERPs were time-locked to the onset of the last word of the sentence and were obtained by
stimulus-locked averaging of the EEG recorded in each condition. Epochs were 1000 ms long with a 200 ms pre-stimulus
baseline. The resulting amplitudes of N400 or LPC were entered into repeated measures analyses of variance (ANOVA). All
ANOVA results were GreenhouseeGeisser corrected if assumption of sphericity was violated, and post-hoc multiple com-
parisons were carried out using Bonferroni-adjusted corrections.

3. Results

3.1. Behavioral results

A repeated-measures ANOVA performed on the accuracy rates yielded significant effects of expression type, F
(2,40) ¼ 29.09, p < 0.001, h2p ¼ 0.59. Pairwise comparisons showed differences between scientific condition and the other two
conditions (scientific vs. Literal, p ¼ 0.001, scientific vs. Conventional, p < 0.001), but no difference between conventional and
literal conditions (p ¼ 0.63). These results indicated that participants responded more correctly to both literal and conven-
tional metaphoric expressions compared to scientific metaphoric sentences. Besides, the false alarm rate was quite low
(0.065) implying good discrimination of participants and very few possibility of using strategies during the experiment.
For reaction time, a main effect between conditions was found [F (2, 40) ¼ 10.23, p < 0.001, h2p ¼ 0.34]. Pairwise com-
parisons indicated differences between the scientific (M ¼ 485.27, SD ¼ 212.83) and literal (M ¼ 439.79, SD ¼ 178.26) con-
ditions (p < 0.05), as well as the scientific and conventional (M ¼ 409.31, SD ¼ 157.41) conditions (p < 0.01), but no difference
between the conventional and literal conditions (p ¼ 0.19). These results indicated participants responded more slowly to
scientific metaphoric expressions compared to both literal sentences and conventional metaphoric expressions.

Table 2
The results of pretests.

Meaningfulness Figurativeness Familiarity

M SD M SD M SD
Scientific metaphors 2.36 0.27 2.5 0.24 3.19 0.41
Conventional metaphors 2.63 0.22 2.6 0.3 5.45 0.38
Literal expressions 2.7 0.19 1.44 0.28 5.59 0.39
 X. Tang et al. / Journal of Neurolinguistics 42 (2017) 12e22 17

Fig. 1. Experimental paradigm.

Fig. 2. Grand average ERP waveforms recorded at the 9 chosen electrodes. Topographic plots of scientificeliteral and conventionaleliteral for the N400
(350e450ms) window and the LPC (600e900 ms) window are presented at the bottom.
18 X. Tang et al. / Journal of Neurolinguistics 42 (2017) 12e22

3.2. Event related potentials

ERPs were derived by averaging correctly classified trials on each condition for each participant. The waveforms in the
present study were characterized by the early perceptual components of N100 and P200 observed clearly at the frontal and
central sites, a sizable negative deflection around 400 ms at all scalp sites identified as the N400, and a later positive
deflection, which appeared less positive for the scientific metaphor condition. Grand averaged waveforms for the sentence-
final words are shown in Fig. 2. Analyses involved mean amplitude measurements of the following time windows: (a)
350e450 ms, which captured the N400, and (b) 600e900 ms, covering the activity of the late positive complex (LPC). The
selected windows captured most of the activity in each component with a minimal overlap between them.
In order to analyse ERP components, 3  3  3 ANOVAs for repeated measures were conducted for the mean amplitude of
both time windows. Due to the distinct central-parietal topographic distribution (see also Fig. 2), within-subjects factors were
condition, region (frontal, medial/central, parietal) and hemisphere (left, midline, right). Accordingly, nine electrode positions
were assigned with respect to region (frontal: F3, Fz, F4; central: C3, Cz, C4; parietal: P3, Pz, P4) and hemisphere (left: F3, C3,
P3; midline: Fz, Cz, Pz; right: F4, C4, P4), respectively. Because effects and interactions involving the condition factor are of
interest, only these are reported here.

3.3. 350e450 ms

Analyses of ERPs recorded over the nine chosen sites indicated that scientific metaphors did show unique topography of
the waveforms compared to conventional metaphors. The condition  region  hemisphere ANOVA revealed significant main
effects of condition [F (2,40) ¼ 10.77, p < 0.001, h2p ¼ 0.35]. Scientific metaphors elicited the most negative N400 (M ¼ - 0.15,
SD ¼ 2.91), followed by conventional metaphors (M ¼ 1.26, SD ¼ 3.23) and literal sentences (M ¼ 4.36, SD ¼ 4.25) (See Fig. 2).
There were significant condition  region interactions [F (4,80) ¼ 5.8, p ¼ 0.014, h2p ¼ 0.23]. Post-hoc analysis showed
significant differences between scientific metaphors and literal sentences in all the three regions [Fs(1,20) > 7, ps < 0.05], and
those between conventional metaphors and literal sentences in the central and parietal regions [Fs(1,20) > 7, ps < 0.05] but
not in the frontal region (p ¼ 0.3).
In order to make a clear comparison between the two kinds of metaphoric expressions, separate pairwise ANOVAs for all
the six electrodes in the central and parietal regions were performed. The differences between scientific metaphors and
conventional metaphors were significant at the right group [C4: F (1,20) ¼ 8, p ¼ 0.01, h2p ¼ 0.29; P4: F (1,20) ¼ 7.43, p ¼ 0.013,
h2p ¼ 0.27], marginally significant at the central group [Cz: F (1,20) ¼ 4.27, p ¼ 0.052, h2p ¼ 0.18; Pz: F (1,20) ¼ 3.89, p ¼ 0.063,
h2p ¼ 0.17], and not significant at the left group (ps > 1) (See Fig. 3).

Fig. 3. Mean values of N400 amplitudes for the three sentence conditions over the three electrode groups (left, midline, right) in the central and parietal regions.
Error bars indicate the standard errors of the mean. (* indexes p < 0.05, þ indexes p < 0.08, ns indicates no significant difference).
 X. Tang et al. / Journal of Neurolinguistics 42 (2017) 12e22 19

3.4. 600e900 ms

Similar to the former window, the condition  region  hemisphere ANOVA of the later window (600e900 ms) also
revealed significant main effects of condition [F (2,40) ¼ 35.27, p < 0.001, h2p ¼ 0.64]. As can be seen in Fig. 2, a sustained
activity was observed for scientific metaphors, whose ERPs were least positive over all the nine chosen electrodes. The linear
increase of amplitudes from scientific metaphors (M ¼ 1.01, SD ¼ 2.72) to conventional metaphors (M ¼ 3.34, SD ¼ 3.1) and to
literal sentences (M ¼ 4.56, SD ¼ 2.53) led to interactions between condition and region [F (4,80) ¼ 6.73, p ¼ 0.002, h2p ¼ 0.25].
There were no significant interaction between condition and hemisphere (p ¼ 0.341).
Post-hoc analysis showed significant differences between scientific metaphors and literal sentences in all the three regions
[Fs(1,20) > 18, ps < 0.001], and those between conventional metaphors and literal sentences in the central and parietal regions
[Fs(1,20) > 5, ps < 0.05] but not in the frontal region (p ¼ 0.812). Most importantly, the differences between scientific
metaphors and conventional metaphors were dramatic in all the three regions [Fs(1,20) > 17, ps < 0.001] (See Fig. 4).
Moreover, the ERP patterns in the later window (600e900 ms) were further illustrated by subtracting the amplitude of the
literal condition from that of each of the two metaphoric conditions, and the resulting difference curves are shown in Fig. 5.
Following their prominent N400 effects, scientific metaphors and conventional metaphors displayed a second negativity
peaking around 760 ms. There were significant main effects of condition [F (1,20) ¼ 30.48, p < 0.001, h2p ¼ 0.6]. The interaction
effects between condition and hemisphere [F (2,40) ¼ 6.6, p ¼ 0.005, h2p ¼ 0.25] indicated that the differences were more
prominent at the right hemisphere.

4. Discussion

The present study chose scientific metaphors as the novel ones to provide a complementary access to the debate about the
cognitive mechanism of metaphor comprehension, specifically about the following two aspects: (1) to what degree mapping
processes are extant during the comprehension of scientific and conventional metaphorical expressions, and (2) the inter-
action over time between the left and right hemispheres to further define the specific contribution of the right hemisphere to
the comprehension of metaphors.
There are two main findings in the present study, one regarding the N400 for the contextual factor and the other regarding
the late negativity for the knowledge-understanding inference factor. Consistent with our predictions, the first finding of this
study was a grading in N400 amplitudes, with the lowest for literal expressions, increasing slightly in negativity for con-
ventional metaphors, the most negative for scientific metaphors. According to the conceptual blending view (Coulson &
Petten, 2002; Yang et al., 2013), the amplitudes of N400 indicate the degree of difficulties in retrieving the stored concep-
tual knowledge during the semantic processing to integrate these elements in metaphor comprehension. Our findings firstly
replicate previous studies (Lai & Curran, 2013; Rutter et al., 2012; Schneider et al., 2014), which found that metaphors, either
novel or conventional, elicited more negative N400s relative to literal expressions. The graded N400 amplitudes found here
suggests literal mapping occurs during metaphor comprehension. However, the contextual factor affecting the blending of
the two domains remains poorly understood in previous studies. The more important contribution of the present study is to
put the contextual factor into consideration. The target and source domains in scientific metaphors (The circuit is a ladder) are
from different contexts, namely, the scientific target (circuit) and the daily source (ladder). Greater N400 amplitudes for
scientific metaphors might result not only from their unfamiliarity but also from their more complicated contextual structure,
which further enhanced the difficulty of meaning integration. It should be more difficult to retrieve background knowledge to
integrate meaning when the two semantic domains are from different contexts. Therefore, retrieving the stored conceptual
knowledge associated with scientific metaphors is more demanding than retrieving knowledge of conventional metaphors or
literal expressions.

Fig. 4. Mean values of LPC amplitudes for the three sentence conditions over the three regions (frontal, central, parietal). Error bars indicate the standard errors of
the mean. (*** indexes p < 0.001, * indexes p < 0.05, ns indicates no significant difference).
20 X. Tang et al. / Journal of Neurolinguistics 42 (2017) 12e22

Fig. 5. Difference curves of the grand average ERP s recorded at the 9 chosen electrodes between the literal condition and each of the two metaphoric conditions.

Secondly, in the LPC window, like several previous studies (Arzouan et al., 2007a, 2007b; Kazmerski et al., 2003; Rutter
et al., 2012), a late negativity reflecting secondary semantic integration processes was elicited by novel metaphors, scienti-
fic metaphors here. There appeared a late negativity peaking around 760 ms in the late part of the LPC window. This late
negative component was elicited partly overlapping in space and time with the LPC. Thus, although a more positive LPC was
expected for scientific metaphors, the late negativity masked it, appearing as a rather low LPC. This late negativity has been
interpreted as the manifestation of a further attempt to select an appropriate meaning (Goldstein et al., 2012). According to
the structural-mapping model (Gentner, 1983) and the career of a metaphor model (Bowdle & Gentner, 2005; Gentner &
Wolff, 1997), when processing scientific metaphors, a secondary integration of meaning might be required. During the
secondary integration, some sort of analogical comparison might be triggered in order to understand the related scientific
knowledge. This kind of inference from the daily source domain to the scientific target domain could attenuate the amplitude
of the late negativity (Berkum, Hagoort, & Brown, 1999). However, for processing either conventional metaphors or poetic
metaphors in previous studies, there was no such kind of knowledge-understanding inference because the purpose for them
is to get a sort of emotional echo instead of new knowledge. Such knowledge-understanding inference further enhances the
cognitive taxing for processing scientific metaphors. Besides, the sequential view might also be supported by the longer
reaction time for scientific metaphoric expressions.
Furthermore, in the present study, the ERP topography for scientific metaphors in the two time windows was different
from the one for conventional metaphors, implying different relative contributions of hemispheres. Compared to conven-
tional metaphors, scientific metaphors displayed a central-parietal-right-biased N400 topography. Comparatively, the N400
distribution of scientific metaphors covered much larger area extending from the right hemisphere to the left one and from
the parietal region to the frontal one. For the LPC, conventional metaphors did not show any right-biased distribution, and
scientific metaphors also symmetrically activated the left and right hemispheres. However, more regions were involved in
understanding scientific metaphors. More interestingly, for the late negativity, compared to conventional metaphors, sci-
entific metaphors showed a right-biased distribution probably indicating the special role of right hemisphere in under-
standing metaphors from scientific context. As reflected in some studies, the right hemisphere was involved in novel
metaphor processing (Gold, Ben-Artzi, & Elisheva, 2012; Yang, 2014). In the present study, the right hemisphere seems
possibly show its special role in understanding scientific metaphors. However, like some other studies reporting symmetric
scalp distribution of ERP components (Arzouan et al., 2007a; Coulson & Petten, 2002; Coulson & Petten, 2007), the present
study shows that the left hemisphere is also important in understanding scientific metaphors. Therefore, lateralized distri-
butions may not definitely reflect lateralized activation, and in fact, the commonly reported right-bias of N400 has been
 X. Tang et al. / Journal of Neurolinguistics 42 (2017) 12e22 21

shown to paradoxically involve mostly the left hemisphere (Petten & Luka, 2006). We conclude here that language
comprehension involves widespread bilateral distribution with the right hemisphere having a special role in the appreciation
of metaphors.
In sum, this study adopted scientific metaphors with its complicated contextual structure and unique knowledge-
understanding inference as the novel ones to provide a new perspective to observe the debate of metaphor processing
mechanism. The semantic structure of scientific metaphors makes them more difficult to be processed, as reflected by their
more negative N400 amplitudes. In addition, scientific metaphors elicited a more negative late negativity suggesting further
attempts to integrate meaning through knowledge-understanding inference. This is consistent with the sequential view of
metaphor comprehension. Nevertheless, the N400 and late negativity scalp distributions elicited by scientific vs. Conven-
tional metaphors suggest different relative contributions of the two hemispheres in accessing semantic information of target
words. Our findings differentiating scientific and conventional metaphors support an indirect processing model compatible
with the structure mapping/the career of metaphor model. Further studies and elaborations of processing models will be
needed in order to determine the best interpretation of those conceptual mappings involved in scientific metaphor
comprehension.

Acknowledgments

The study was performed in partial fulfillment of the requirements for the first author's doctoral dissertation at Shaanxi
Normal University. The experiments were carried out at the Key Laboratory of Modern Teaching Technology of Ministry of
Education, Shaanxi Normal University. John Quelch, an Australian teacher and writer, has kindly polished this paper several
times.

References

Arzouan, Y., Goldstein, A., & Faust, M. (2007a). Dynamics of hemispheric activity during metaphor comprehension: Electrophysiological measures. Neu-
roimage, 36(1), 222e231.
Arzouan, Y., Goldstein, A., & Faust, M. (2007b). Brainwaves are stethoscopes: Erp correlates of novel metaphor comprehension. Brain Research, 1160, 69e81.
Aubusson, P. J., Ritchie, S. M., & Harrison, A. G. (2006). Metaphor and analogy in science education. Springer.
Beeman, M. (1998). Coarse semantic coding and discourse comprehension. In Beeman, & Chiarello (Eds.), Right hemisphere language comprehension: Per-
spectives from cognitive neuroscience. Erlbaum.
Beeman, M. J. (2005). Bilateral brain processes for comprehending natural language. Trends in Cognitive Sciences, 9, 512e518.
Berkum, J. J., Hagoort, P., & Brown, C. M. (1999). Semantic integration in sentences and discourse: Evidence from the N400. Journal of Cognitive Neuroscience,
11, 657e671.
Bohrn, I. C., Altmann, U., & Jacobs, A. M. (2012). Looking at the brains behind figurative languageda quantitative meta-analysis of neuroimaging studies on
metaphor, idiom, and irony processing. Neuropsychologia, 50, 2669e2683.
Bowdle, B. F., & Gentner, D. (2005). The career of metaphor. Psychological review, 112(1), 193.
Chen, E., Widick, P., & Chatterjee, A. (2008). Functional-anatomical organization of predicate metaphor processing. Brain & Language, 107, 194e202.
Coulson, S., & Petten, C. (2002). Conceptual integration and metaphor: An event-related potential study. Memory and Cognition, 30(6), 958e968.
Coulson, S., & Petten, C. (2007). A special role for the right hemisphere in metaphor comprehension? ERP evidence from hemifield presentation. Brain
Research, 1146, 128e145.
Diaz, M. T., Barrett, K. T., & Hogstrom, L. J. (2011). The influence of sentence novelty and figurativeness on brain activity. Neuropsychologia, 49, 320e330.
Friederici, A. D., Steinhauer, K., & Frisch, S. (1999). Lexical integration: Sequential effects of syntactic and semantic information. Memory and Cognition, 27(3),
438e453.
Gentner, D. (1983). Structure-mapping: A theoretical framework for analogy. Cognitive Science, 7(2), 155e170.
Gentner, D., & Wolff, P. (1997). Alignment in the processing of metaphor. Journal of Memory and Language, 37(3), 331e355.
Giora, R. (2003). On our Mind: Salience, context, and figurative language. Language in Society, 34(2), 307e310.
Gold, R., Ben-Artzi, M. F., & Elisheva. (2012). Metaphors and verbal creativity: The role of the right hemisphere. Laterality Asymmetries of Body Brain &
Cognition, 17(5), 602e614.
Goldstein, A., Arzouan, Y., & Faust, M. (2012). Killing a novel metaphor and reviving a dead one: Erp correlates of metaphor conventionalization. Brain and
Language, 123(2), 137e142.
Gratton, G., Coles, M. G., & Donchin, E. (1983). A new method for off-line removal of ocular artifact. Electroencephalography and Clinical Neurophysiology,
55(4), 468e484.
Grauwe, S. D., Swain, A., Holcomb, P. J., Ditman, T., & Kuperberg, G. R. (2010). Electrophysiological insights into the processing of nominal metaphors.
Neuropsychologia, 48(7), 1965e1984.
Iakimova, G., Passerieux, C., Laurent, J. P., & Hardy-Bayle, M. C. (2005). ERPs of metaphoric, literal, and incongruous semantic processing in schizophrenia.
Psychophysiology, 42(4), 380e390.
Kaan, E., Harris, A., Gibson, E., & Holcomb, P. (2000). The P600 as an index of syntactic integration difficulty. Language and Cognitive Processes, 15(2),
159e201.
Kasparian, K. (2013). Hemispheric differences in figurative language processing: Contributions of neuroimaging methods and challenges in reconciling
current empirical findings. Journal of Neurolinguistics, 26, 1e21.
Kazmerski, V. A., Blasko, D. G., & Dessalegn, B. G. (2003). ERP and behavioral evidence of individual differences in metaphor comprehension. Memory &
Cognition, 31, 673e689.
Kutas, M., & Federmeier, K. D. (2011). Thirty years and counting: Finding meaning in the N400 component of the event related brain potential (ERP). Annual
Review of Psychology, 62, 621e647.
Kutas, M., & Hillyard, S. A. (1980). Reading senseless sentences: Brain potentials reflect semantic incongruity. Science, 207(4427), 203e205.
Lai, V. T., & Curran, T. (2013). ERP evidence for conceptual mappings and comparison processes during the comprehension of conventional and novel
metaphors. Brain & Language, 127(3), 484e496.
Lai, V. T., Curran, T., & Menn, L. (2009). Comprehending conventional and novel metaphors: An ERP study. Brain Research, 1284, 145e155.
Lopez-Calderon, J., & Luck, S. J. (2014). ERPLAB: An open-source toolbox for the analysis of event-related potentials. Frontiers in Human Neuroscience, 8, 213.
Mashal, N., Vishne, T., Laor, N., & Titone, D. (2013). Enhanced left frontal involvement during novel metaphor comprehension in schizophrenia: Evidence
from functional neuroimaging. Brain & Language, 124, 66e74.
Oldfield, R. C. (1971). The assessment and analysis of handedness: The Edinburgh inventory. Neuropsychologia, 9(1), 97e113.
22 X. Tang et al. / Journal of Neurolinguistics 42 (2017) 12e22

Paller, K. A., & Kutas, M. (1992). Brain potentials during memory retrieval provide neurophysiological support for the distinction between conscious
recollection and priming. Journal of Cognitive Neuroscience, 4(4), 375e392.
Petten, C. V., & Luka, B. J. (2006). Neural localization of semantic context effects in electromagnetic and hemodynamic studies. Brain & Language, 97,
279e293.
Pynte, J., Besson, M., Robichon, F. H., & Poli, J. (1996). The time-course of metaphor comprehension: An event-related potential study. Brain & Language, 55,
293e316.
Rapp, A. M., Mutschler, D. E., & Erb, M. (2012). Where in the brain is nonliteral language? A coordinate-based meta-analysis of functional magnetic
resonance imaging studies. Neuroimage, 63, 600e610.
Rutter, B., et al. (2012). Can clouds dance? Part 2: an ERP investigation of passive conceptual expansion. Brain and Cognition, 80(3), 301e310.
Schmidt, G. L., & Seger, C. A. (2009). Neural correlates of metaphor processing: The roles of figurativeness, familiarity and difficulty. Brain and Cognition, 71,
375e386.
Schneider, S., et al. (2014). Beyond the N400: Complementary access to early neural correlates of novel metaphor comprehension using combined elec-
trophysiological and haemodynamic measurements. Cortex, 53(4), 45e59.
Sotillo, M., Carretie, L., Hinojosa, J. A., Tapia, M., Mercado, F., Lopez-Martin, S., et al. (2005). Neural activity associated with metaphor comprehension: Spatial
analysis. Neuroscience Letters, 373, 5e9.
Uchiyama, H. T., Saito, D. N., Tanabe, H. C., Harada, T., Seki, A., Ohno, K., et al. (2012). Distinction between the literal and intended meanings of sentences: A
functional magnetic resonance imaging study of metaphor and sarcasm. Cortex, 48, 563e583.
Yang, J. (2014). The role of the right hemisphere in metaphor comprehension: A meta-analysis of functional magnetic resonance imaging studies. Human
Brain Mapping, 35, 107e122.
Yang, F. G., Bradley, K., Huq, M., Wu, D. L., & Krawczyk, D. C. (2013). Contextual effects on conceptual blending in metaphors: An event-related potential
study. Journal of Neurolinguistics, 26, 312e326.
Yang, F. G., Edens, J., Simpson, C., & Krawczyk, D. C. (2009). Differences in task demands influence the hemispheric lateralization and neural correlates of
metaphor. Brain & Language, 111, 114e124.