Faculty of Psychology and Educational Sciences
Second Master Theoretical and Experimental
Psychology
Master thesis
PROCESSING NEUTRAL WORDS, NEGATIVE
WORDS AND TABOO WORDS: AN ERP
STUDY
Anneleen Mortier
Masterthesis
Supervisor: dr. Elisah D’Hooge
Promotor: Prof. dr. Robert Hartsuiker
Department of Experimental Psychology
2012-2013
Aknowledgments
First of all, I would give a
BIG
thanks to my supervisor, dr. Elisah D’Hooge. She
taught me to set up ERP experiments, administer and analyse them. Further, she was a big help
for my thesis, and also for other matters. Next, I would like to thank my promotor, Professor dr.
Robert Hartsuiker, for his guidance and support. I also would like to thank the other students
from 2nd master Experimental Psychology. Of course, thanks to my parents, for supporting me
those five years and for believing in me. As last, I want to thank Martijn for his support and for
keeping up with me when things were not going in the right direction.
PROCESSING NEUTRAL WORDS, NEGATIVE WORDS AND TABOO WORDS: AN ERP
STUDY
Abstract
Bilinguals have reported that they have a stronger emotional experience in their native language
than in their second language. Behavioural studies are inconclusive about this question. However,
studies that used skin conductance levels as a dependent variable are clearer about this effect:
participants elicit a stronger skin conductance response on emotional and taboo words in the
native language compared to the second language (e.g. Harris, Ayçiçegi and Gleason, 2003).
Moreover, this skin conductance response was higher for taboo words than for negative words,
indicating that taboo words are a special type of emotional words. This study tried to investigate
the difference in emotional experience between the native language and the second language with
a more direct and online measurement: event related potentials. It consisted of two experiments.
The first experiment used neutral and negative words in the first language and the second
language, the second experiment used neutral and taboo words in the first and second language.
Both experiments used Dutch-English bilinguals that performed a delayed naming task. The
results of the event related potentials showed that negative and taboo words elicited a P300
followed by a late positive potential. These components were attenuated when neutral words
were being presented. Moreover, the difference in amplitude between negative/taboo words and
neutral words was more pronounced in the native language. These findings are in line with our
hypothesis. The results are discussed with reference to previous findings on bilingual emotion
experience.
Keywords: Bilingualism, Emotion, Taboo, ERP, P300, LPP
Contents
Taboo and negative words . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
Qualitative Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2
Quantitative Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3
Bilinguals and physiological studies . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7
Present study
9
Experiment 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14
Experiment 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19
Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19
Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
General Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
26
Appendix A
37
Appendix B
38
Appendix C
39
Appendix D
40
Appendix E
49
Appendix F
50
Appendix G
51
1
Maria (L1: Spanish, L2: English):
“I never swear in Spanish. I simply cannot. The words are too heavy and are truly a
taboo for me.”
(Dewaele, 2004b:95).
The above quote illustrates how most of the bilinguals feel about their two (or more)
languages: they experience greater emotional intensity when using their native language (L1)
than when using their second language (L2) or other languages (LX). This intuition has been
investigated by several different paradigms and different stimuli. Various studies have shown
the importance of including different types of stimuli; since they can result in different outcomes
(e.g., Guillet & Arndt (2009) and Jay, Caldwell-Harris, & King, 2008). In this paper, we will
first discuss the different findings between taboo and negative words in bilingual research. Then,
we will give a short overview of the qualitative research in this domain; next we will discuss
previous quantitative research. Furthermore, we will present a short overview about the different
physiological studies that have been conducted in this field. Finally, we proceed to our own
investigation.
Taboo and negative words
Taboo words are a special category of negative words. They include terms referring to
bodily products, ethnic and racial insults, profanity, slang, vulgarity, body parts and sexual acts
(Jay, 1992, 2000). Several studies have found their influence on different types of tasks such
as memory tasks (e.g., Guillet, & Arndt, 2009; Jay, et al., 2008), rapid serial representation
task (e.g., Mathewson, Arnell & Mansfield, 2008), lexical decision (Zeelenberg, Bocanegra &
Pecher, 2011) and language production (Dhooge, De Baene, & Hartsuiker, submitted; Severens,
Janssens, Kûhn, Brass & Hartsuiker, 2011). Additionally, various studies have shown the effect
of taboo words in different domains: e.g., enhanced attention (e.g., Eilola, Havelka & Sharma,
2007; Mathewson et al., 2008), superior recall (e.g., Ayçiçegi & Harris, 2004) and heightened skin
conductance responses (e.g,. Harris, Ayçiçegi & Gleason, 2003). Furthermore, there is evidence
that taboo words may involve in a specialized neural mechanism: patients with aphasia are still
able to swear (Van Lancker & Cummings, 1999). On the contrary, inability to swear has also
been reported when the right hemispheric basal ganglia is impaired (Speedie, Wertman, Ta’ir &
Heilman, 1993).
It has been suggested that taboo words are a special type of emotional words. Several
2
studies have confirmed this suggestion. Jay et al. (2008) have administered several memory
tasks. They found that participants were better at recalling taboo words than negative words,
even when they had to report those taboo words as last. Guillet & Arndt (2009) also found that
participants had a better memory for taboo words than for negative words. However, they used
a different task than Jay et al. (2008). Participants had to remember certain words (central
information) in the presence of a distracter (peripheral information). This distracter could be
either taboo or negative. They found that the central information was remembered better when
the distracter was a taboo word, not when it was a negative word. In a study of Mathewson
et al., (2008), a single-target rapid serial visual representation task (RSVP) was used. Their
results showed that the taboo words received more attention than negative words.
Thus, it seems that taboo words evoke a different response than negative words. As
a result, when a study investigates the emotionality in different languages (or just in a single
language), it is not sufficient to only use emotional words, taboo words should be included as
well.
Qualitative Research
The difference in emotional intensity between L1 and L2 has already been reported in
early clinical studies. Buxbaum (1949) reported that L1 could help to remember lost memories,
in comparison to L2. If memories are retrieved in L2, they are colourless and lifeless. Additionally,
when patients were talking about negative memories, they used their L2 more often, but when
they recalled warm and happy memories, they would use their L1 (Javier, 1996. For an overview:
Schrauf, 2000). In contrast with this, some bilinguals can experience detachment from their L1
when they underwent negative experiences (e.g., German with World War II). Further, when
people are more attracted to their ‘new’ language, they can experience this language as more
emotional as their L1. This indicates that we cannot simply conclude that L1 is more emotional
than L2: other factors can still influence this process.
Dewaele (2004a, 2004b and 2008) tried to investigate the difference in emotional intensity
between the two languages in a more systematically way. In his series of studies, he administered
questionnaires from multilinguals, to investigate the perception of their expression of taboo words
and swearwords (2004a, 2004b) and their expression of love (2008). The studies in 2004 showed
that multilinguals found the emotional force of the taboo words and swearwords were stronger
in their L1. This effect diminished for languages that are learned at a later time. Additionally,
3
multilinguals tend to swear in their dominant language. This dominant language is determined
by age of acquisition, frequency of use and the learned setting (naturalistic, (e.g.,: parents), or
instrumentally, (e.g.,: classroom)). The dominant language is most of the time L1: participants
found that L1 has more force to express themselves. Also, they felt more emotionally involved
when using L1. Others said that they could not swear in their native language because the
emotional weight is too heavy or the words are too taboo. Some participants also mentioned that
they could not express their emotional feelings in another language than the native language (i.e.,
LX), because they cannot find the emotional load of that language. Additionally, multilinguals
might swear in their other languages when they want to soften the force that those words are
carrying or when they do not want to be associated with their expression.
In his study in 2008, Dewaele investigated whether the emotional intensity of the emotion
‘love’ depended on language. He again administered an online questionnaire about experiencing
love to a sample of multilinguals. They had to answer open and closed questions about the
perceived emotional weight of ‘I love you’ in the different languages the participants spoke. He
found that the majority of the respondents found this phrase to have the strongest emotional
value in their L1. However, this effect was dependent on certain other variables: age of acquisition,
self-perceived language dominance and context of the acquisition of the L2. Participants felt
that the emotional weight is the strongest in their L1 and that no other language could match
this feeling. Thus, multilinguals prefer to express one of the most universal emotions in their L1.
The findings by Dewaele (2004a, 2004b, 2008), Buxbaum (1949) and Javier (1996) seem
to suggest that L1 is a personal language with emotional affiliation and is used to express feelings.
In contrast with this, L2 is a more emotionally distant language and is used when we do not
want to be associated with that expression (Pavlenko, 2002). However, the results of Dewaele
(2004a, 2004b, 2008) were based upon self-report, which can be inaccurate. The emotional
processes that are associated to the different languages are not necessarily consciously accessible
to the speaker (Eilola, 2009). Also, clinical case studies are not representative for the entire
population. Hence, the next section discusses studies that explored the emotional difference
between L1 and L2 in an experimental way.
Quantitative Research
Studies with monolinguals have shown that we have a better and more detailed memory
for negative words (e.g. Kensinger & Corkin, 2003) and for taboo words (e.g. Grosser &
4
Walsch, 1966; Hadley & MacKay, 2006; Jay et al., 2008; MacKay et al., 2004, among others).
Furthermore, as the above studies suggest, the recall can be influenced by the kind of language
we use. The study of Anooshian and Hertel (1994) is one of the first who investigated this
effect in bilinguals. They asked late Spanish-English bilinguals to rate emotional and neutral
words presented in English or Spanish. The presentation of these words was blocked for each
language. The experiment was followed by a surprise recall task. Anooshian and Hertel found
that their participants were better at recalling emotional words, but only when those words were
presented in their L1. The authors suggested that when we learn an L2 at a later time, this
wouldn’t be associated with emotional experiences and because of this, will be less emotionally
intense. Ayçiçegi and Harris (2004) tried to replicate and extend the effect that Anooshian
and Hertel (1994) found. They included reprimands and taboo words in their stimulus set.
Additionally, they altered the design. The presentation of the words in each language was mixed
for each block. In contrast to the results of what they hypothesized (and thus in contrast with
the results of Anooshian and Hertel), they found the opposite result with late Turkish-English
participants (learned L2 at the age of 12 to 18 and migrated to the US at the average age of 22).
Participants were equally good at recalling negative English and Turkish words. Moreover, this
effect was stronger in the L2 for some stimuli. Concerning taboo words, participants had better
recall and recognition for these words in comparison with negative words. Ayçiçegi-Dinn and
Caldwell-Harris (2009) did a follow up study: perhaps their findings could be related to the level
of processing (shallow processing or deep processing). They had four conditions: word rating
on emotional intensity (replication of previous study), counting letters (shallow), translation
(deep) and word association (deep). Averaging over the four tasks, they found that recall was
equally strong for L1 and L2, except for the reprimands (more recall in L2). For the taboo
words, the same conclusions hold as their previous study (Ayçiçegi & Harris, 2004): participants
had a better recall and recognition for them compared to negative and neutral words. Those
two studies supports the suggestion that taboo words should be seen as a ‘special’ type of
emotional words. Ferré, García, Fraga, Sánchez-Casas and Molero (2010) investigated whether
this effect was dependent of age of acquisition or the setting where bilinguals learned their
second language. This last variable had two factors. Either the participants had learned their L2
outside a classroom, i.e., naturalistic context, or the participants had learned it in the classroom,
i.e. instrumental context. Early Spanish-Catalan bilinguals (naturalistic context – experiment 2)
and late Spanish-English bilinguals (instrumental context – experiment 3) had to rate emotional
or neutral words. The language in which the words were presented was blocked (one block
5
for each language). In both experiments, there was no difference in recall between the two
languages. Also, when they compared both groups, there was no difference in the magnitude of
the emotional effect in L2.
These four studies indicate that there is a great variability of results in the studies on
memory recall in bilinguals. All of them found different effects: Anooshian and Hertel (1994)
found better recall in L1, Ayçiçegi and Harris (2004) found better recall in L2 (for some stimuli),
Ayçiçegi-Dinn and Caldwell-Harris (2009) and Ferré et al. (2010) found no differences between
the two languages. Furthermore, Ferré et al. showed that language dominance, age of acquisition
or context of acquisition of L2 does not influence the emotional experience in L2. This is in
contrast with the studies of Dewaele (2004a, 2004b) and the results of Anooshian & Hertel
(1994). For taboo words, these studies show that participants have an advantage in recall for
these words, but on average, there is no difference between L1 and L2. The findings of those
four studies give support to the notion that the emotional intensity between different languages
is influenced by other variables. This could be due to characteristics of the participants and/or
of the stimuli.
Another line of research is investigating attention towards emotional words. A frequently
used paradigm is the rapid serial visual presentation (RSVP). This is a paradigm where stimuli
are presented very brief to participants. When distractor and target a presented within 500 ms
from each other, an attentional blink can occur, resulting in worse performance on the target.
Studies with monolinguals have already shown that when this distractor is an emotional or a
taboo word, people have a slower reaction on the target (e.g. Zeelenberg et al.). Moreover,
Mathewson et al., 2008 found that taboo words led to more interference than emotional words
(even negative words). Additionally, when the emotional word is a target, the attentional blink
is reduced (Anderson, 2005). This suggests that emotional words capture the attention more
than neutral words and that there are less attentional resources required to process emotional
words. Colbeck and Bowers (2012) investigated this effect with bilinguals. They showed a
stream of 18 words, with 1 critical distractor (English taboo or sexual word) and 1 color target
to Chinese-English bilinguals and native English speakers. The participants had to identify the
color-word. For the native English speakers, they found the classical finding: they were worse at
identifying the color word when it was preceded by an emotional or a sexual distractor. For the
bilinguals, this effect was reduced: they performed better on the task than the native speakers
when there was an emotional or a sexual distractor. Even when they only took the ‘best scoring’
bilinguals in account, the effect still persists.
6
Another paradigm where the attention for emotional words is investigated is the Stroop
task (Stroop, 1935). The emotional Stroop task is a modified version of the original Stroop task.
In this task, participants are shown emotional and neutral words in different colours. Their task
is to say the colour of the words aloud, and ignore the written word. The general finding is that,
when the target word is an emotional word, there is more interference when saying the colour of
the word (e.g. Pratto & John, 1991). This interference is caused by the emotional value the word
has, and not by incongruency (as in the classical Stroop task). Sutton, Altarriba, Gianico and
Basnight-Brown (2007) administered this emotional Stroop task from Spanish-English bilinguals.
However, note that the bilinguals were more proficient in their second language than in their
first. The results showed that there was an emotional Stroop effect: the participants were slower
when they had to pronounce the color of an emotional word. Nevertheless, this effect was not
moderated by language: there was no difference between L1 and L2. This could conclude that
this null-effect could be due to the proficiency of the L2: the participants used their L2 at
that time more than their first, which could have altered their emotional experience in their
L2. Eilola et al., (2007) did an emotional Stroop task as well, but their participants were late
Finnish-English bilinguals. However, they were still very proficient in their L2. The authors used
emotional and taboo words as distractors. Participants were slower on taboo words and negative
words compared to neutral words. Likewise, they found no difference in effect between the L1
and L2. Eilola et al. concluded that the emotional effect is dependent on the proficiency of the
language: the more proficient you are, the more emotionally attached and the less discrepancy
between L1 and L2 in the emotional Stroop task. The authors suggested that proficiency does
play a role in the emotional experience towards a certain language. However, these two studies
are again in contrast with what Ferré et al. (2010) and Colbeck and Bowers (2012) found and
concluded. This again shows that there are mixed results concerning the emotional intensity of
L1 and L2 in behavioural studies. Moreover, there is no clear evidence towards what actually
influences this effect.
In short, these behavioural studies are trustworthier than the self-reported studies from
Dewaele (2004a, 2004b, 2008). These studies also suggest that there is a difference in emotional
experience between L1 and L2, but this difference can be influenced by various factors. Some of
the above studies have proposed several variables that might influence this effect: e.g., age of
acquisition, proficiency,. . . However, other studies dismissed these variables. The results of the
behavioural studies can also be influenced by other processes (e.g. attention, memory,. . . ). Hence,
it is recommended to use a different dependent variable than reaction time. The brain-based
7
perspective suggests that all subjective experiences have a measurable physiological correlate
(Harris, 2004). Thus, physiological measurements can investigate the emotional difference
between L1 and L2 in a purer and clearer way (e.g. this measurement can pick up effects that are
not even conscious). Additionally, another advantage is that this can investigate the underlying
mechanisms. Thus, the following studies investigated if L1 and L2 differ from each other in
emotional intensity using physiological correlates.
Bilinguals and physiological studies
The study of Harris et al. (2003) is one of the first that investigated the emotional
experience of bilinguals with skin conductance levels (SCL’s, formerly known as Galvanic skin
response). They presented English and Turkish target words to Turkish-English bilinguals using
the auditory and visual modality. These target words consisted of neutral words, positive and
negative words, reprimands and taboo words. The reprimands were childhood reprimands such
as ‘Shame on you!’. The language of the words was mixed per block. Results showed that the
taboo words elicited the strongest autonomic response; these were even stronger when the taboo
words were presented in their L1. Another finding was that when the stimuli were presented in
the auditory modality, the skin conductance level was higher than when they were presented in
the visual modality. In contrast with this, when the words were presented in L2, there was no
difference between modalities.
Harris (2004) found that age of acquisition mediated the relationship between emotional
intensity in a language and the skin conductance level. She showed the same English target
words as in Harris et al. (2003) with their Spanish translations to Spanish-English bilinguals.
She had two groups of participants: early bilinguals (average age of acquisition: 3.7 years old)
and late bilinguals (average age of acquisition: 7.9 years old). The skin conductance level was
measured as well. Results showed that late bilinguals had a stronger skin conductance response
when they saw childhood reprimands in their first language (L1 – Spanish). In contrast with this,
early bilinguals had no difference in skin conductance response if the childhood reprimands were
presented in L1 or L2. Harris explained this by age of acquisition: the early Spanish-English
bilinguals had acquired their L2 at an early age. Also, they moved to the US at an early age
(average: 3.1 years old), thus they were very familiar with this culture and emotional value of it.
Harris and Ayçiçegi-Dinn (2009) also used skin conductance responses, but they let
Turkish-English bilinguals read Turkish and English true or false statements. They found a
8
stronger skin conductance response when the participants read false statements and this response
was even stronger for L2 (English). These findings supported the ‘double stressor’ account: for
most participants it was harder to lie in L2, because speaking a second language requires more
effort than speaking a native language. Moreover, lying also requires more effort and cognitive
resources than speaking the truth. Thus, it is not sure what causes these stronger responses: it
could be the result of the additional cognitive resources (‘I want to do my best, because it is in
English, and I do not want to look like a fool’), or it could be the result of a more emotional
experience to their L2.
Eilola and Havelka (2010) did a follow up on their previous emotional Stroop experiment
(Eilola et al., 2007). Native English speakers and Greek-English bilinguals had to react to the
colour of an English target word, and ignore the meaning of it. However, what was new to their
experiment was that they measured skin conductance levels. The behavioural results showed
no difference between monolinguals and bilinguals, but the SCLs clearly showed a difference:
it was significantly higher for the monolinguals in the condition of emotional and taboo words
than for the Greek-English bilinguals. This study is very crucial because it suggests that at
the behavioural level, there is no difference between the emotional perception of L1 and L2 (no
difference in RT), but the physiological measurements are suggesting the opposite. Thus, it is
very important to study the underlying mechanisms.
To summarize, the physiological studies show more clarity: overall, emotional and taboo
words in L1 elicited stronger skin conductance responses than in L2. Moreover, as Eilola et
al. (2010) showed, this is independent from behavioural results. However, skin conductance
levels cannot give information why these levels are higher in L1 than in L2 (as in Harris and
Ayçiçegi-Dinn (2009)). As the authors suggested, it could simply be explained by the emotional
strength of L1. Nevertheless, SCLs are not time-locked to the stimuli and is a more diffuse
measurement, so more general and non-task related factors could influence this response (e.g.
mood of the participant, external conditions,. . . ). Moreover, SCLs measure the reaction of
the sympathetic nervous system. Thus, it would be better to study the origin of this reaction
(the brain). A last remark is that SCLs habituate when stimuli are repeated (Codispoti, & De
Cesarei, 2007). This habituation does not happen with ERPs (LPP – Codispoti, & De Cesarei,
2007). Also, ERPs are a continuous, direct and online measurement. It makes it possible to
investigate every change between the stimulus presentation and the response. Another advantage
of ERPs is that this measurement is time-locked to the processing of the stimuli, so we know
with certainty that an event has elicited that response in the ERPs. Thus, when using ERPs, we
9
can be more conclusive about why these differences between L1 and L2 exist. As a conclusion,
this is a better measurement to investigate these differences.
Present study
Emotional stimuli have the property to automatically capture our attention more than
neutral stimuli, indicating differential processing of both types of stimuli (see previous studies:
e.g. Sutton et al., 2007; Zeelenberg, et al., 2011). This difference is also reflected in the ERPs:
when an emotional stimulus is presented, two components can be found. First, emotional stimuli
elicit a larger peak within the 200 – 300 ms range (e.g. Dhooge, et al., (submitted); for an
overview, see Hajcak, MacNamara & Olvet, 2010; Schupp et al., 2006). Additionally, the P300
might be followed by a Late Positive Potential (LPP), which also shows more positive potentials
for emotional stimuli. This is a broad superior-posterior positivity (Hajcak et al., 2010). The
difference between the P300 and the LPP is that the P300 is seen as a reflection of the increase
in attention towards task-relevant stimuli. The LPP on the other hand is a reflection of the
vigilance and processing of intrinsically motivating stimuli (Hajcak et al., 2010).
Previous studies already showed that ERPs are effective at measuring the emotional value
of a certain language. Wu and Thierry (2012) used ERPs in a translation-priming paradigm.
Chinese-English bilinguals were presented English and Chinese emotional and neutral words.
The first word served as a prime and they had to judge the the second one whether it was related
in meaning to the prime. The crucial part of their experiment was that for some words pairs,
prime and target shared the same sounds when they were translated into Chinese, but they were
unrelated in meaning and form in English (e.g. failure – poet : shi bai – shi ren). The results of
the ERPs showed that when prime and target are related in sounds in Chinese, the ERPs in the
negative valence condition were significantly different from the ERPs in the positive and neutral
conditions. Moreover, the negative valence failed to elicit the ERPs amplitude. These results
suggest that bilinguals use the L1 (Chinese) translation to judge whether the target was related
to the prime, but this only happened with neutral and positive words. This indicates that
negative valenced words have a ‘special status’ when compared to the positive valenced words.
The authors suggested that we are trying to protect ourselves against the negative connotation
of these words by not activating their meaning in our L1. This might indicate that this meaning
is too emotional or too offensive in our L1.
Opitz and Degner (2012) investigated the different sense of emotionality in L1 and
10
L2 with ERPs. They presented positive, negative and neutral nouns to German-French and
French-German bilinguals. The participants had to make a lexical decision on the target words.
Their dependent measure was the early posterior negativity (EPN). What they found was no
difference in amplitude for the EPN in L1 and L2, but it was delayed in L2. However, this
experiment has some shortcomings, so that the results might be distorted. The authors matched
their stimuli pairwise, but it has not been mentioned if this was also done for every condition.
Previous studies have shown that the frequency can influence the amplitude: words with a low
frequency can elicit a smaller P300 and high frequency words can elicit a smaller P1. For the
EPN, low frequency neutral words elicit a larger EPN than high frequency neutral words. High
frequent negative words elicit a larger EPN than low frequency negative words. Moreover, this
difference is not observed for positive words (Scott, O’Donnell, Leuthold & Sereno, 2009). So,
frequency can influence the amplitudes of the ERPs. Thus, the difference between the amplitude
for positive and negative words could be due to the frequency difference than to the emotionality
between the two languages. A second remark is that they used stimuli from a database according
to valence and arousal rated by monolinguals (German and French). When a study tries to
investigate the emotionality of words in L1 and L2, several words will be more emotional in L1
than the translation equivalent in L2. They tried to solve this problem by using a balanced
bilingual design. However, it would have been better if they have rated their stimuli by a similar
sample of the population on valence and arousal to control for this effect. Another remark is
that in their ERPs, they only looked at the effect on the early time course. However, there can
be some differences later in that time course (e.g. in the LPP). As a last remark, the authors
used a lexical decision task. However, the retrieval of the word can influence the effect the word
has when it was presented. Thus, the results of their study could be biased by factors they did
not control. The present study tried to control for these confounds.
The goal of this study was to investigate with ERPs whether there is a difference in
emotional experience in L1 and in L2. Based on the previous studies, we hypothesize that there
is a difference in emotional intensity between L1 and L2. More specifically, L1 will elicit a
stronger emotional response in comparison to L2. (e.g., Harris, 2004; Harris et al., 2003). We
used a delayed-naming task to investigate this matter. In this task, participants are presented
with the to-be named target. They can prepare their response prior to the onset of the target
word. Since we also wanted to investigate the difference between negative and taboo words,
our critical stimuli were negative and neutral words (experiment 1) and taboo and neutral
words (experiment 2). The reason why we did not include any positive stimuli is because of the
11
difference in processing between positive and negative words: e.g. subjects are faster on positive
words compared to negative and neutral words (Kissler & Koessler, 2011; Palazova, Mantwill,
Sommer & Schacht, 2011). Also, we investigated the difference between taboo and emotional
words in L1 and L2. Thus, negative words are more in line with the taboo words. We examined
this effect with an objective, online and direct method – ERP signals. The dependent measures
were the P300 and the long positive potential (LPP). Our hypothesis was that negative and
taboo words would elicit a P300 and an LPP. The amplitude of the P300 and the LPP would be
larger in L1 than in L2.
Experiment 1
In the first experiment, we wanted to investigate if there was a difference in the emotional
intensity for negative and neutral words. Second, we wanted to establish whether this difference
differed between L1 and L2. A delayed naming paradigm with Dutch and English words was
used to study this difference. We chose to use ERPs as the dependent measure, since it is very
sensitive, online and direct method. More specifically, we focussed on the P300 and the LPP. We
hypothesized that the amplitudes would be larger for negative words in comparison to neutral
words. Moreover, the difference in amplitude between negative words and neutral words would
be larger in L1 than in L2.
Method
Participants.
Twenty psychology students from Ghent University participated in this
experiment (2 male and 18 female). They received a small compensation. Before they were
accepted to this study, they had to fill out the LEXTale English and Dutch (Lemhöfer &
Broersma, 2011) to make sure that all participants had a sufficient knowledge of English (L2).
Only participants that had a score higher than 85% on Dutch and 60% on English were allowed
to participate. Also, participants that scored higher on the English version than on the Dutch
version were excluded. The average score on the Dutch version was 93.62% and for the English
version 78.44%. None of the participants had a history of neurological or psychiatric disorders.
All participants were right handed (The Dutch handedness questionnaire, Van Strien,
2002). They reported that Dutch was their native language (L1), and the language they used
the most (table 1). All participants received English (L2) education in secondary school and
they lived in a Dutch dominant environment. All of them still came in contact on a regular basis
12
with English by internet, tv, films, radio, etc. For their L2, most of the participants learned this
at home (n = 16) and some of them at school (n = 2). The other 2 participants learned English
on holidays.
Table 1: Self-Assessed Ratings (7-point Likert Scale) of L1, L2 and L3 Proficiency (Experiment
1). Standard deviations are included in parentheses
Language
L1 (Dutch)
L2 (English)
L3 (French)
Stimuli.
Skill
Experiment 1
Writing
6.20 (.53)
Reading
6.70 (.47)
Understanding
6.60 (.60)
Speaking
6.50 (.69)
General proficiency
6.5 (.61)
Writing
4.89 (.88)
Reading
5.75 (.64)
Understanding
5.75 (.64)
Speaking
5.40 (.75)
General proficiency
5.16 (.51)
General proficiency
3.70 (1.22)
Before we administered the experiment, we chose to rate our stimuli. Volunteers,
who received a small monetary contribution, performed this rating. The volunteers were similar
to the subjects in the actual experiment in terms of education, age and L1 and L2 knowledge.
First, a huge pool of potential emotional nouns was gathered, using the ANEW and previous
studies (Ayçiçegi-Dinn & Caldwell-Harris, 2009; Belezza, Greenwald & Banaji 1986; Bertels,
Kolinsky & Morais, 2009; Bertels, Kolinsky & Morais, 2010; Eilola et al., 2007; Ferré et al.,
2010; Guillet & Arndt, 2009; Harris et al., 2003; Harris, 2004; Janschewitz, 2008; Pavlenko,
2002; Thomas, Johnstone & Gonsalvez, 2007 and Zeelenberg et al., 2011). English words were
being translated to Dutch and vice versa. Then, we omitted all cognates and homophones
from our stimulus set, because those words could enhance the bilingual lexical processing (e.g.
Costa, Caramazza & Sebastian-Galles, 2000). Further, using the SUBTLEX-Dutch (Keuleers,
Brysbaert & New, 2010) and SUBTLEX-English (Brysbaert & New, 2009), we searched for
13
control neutral words, which are matched on frequency, word category (nouns) and number of
letters (identical). Using a paired t-test, we found no significant difference between all categories
of words in terms of frequency (p > .1). Eventually, only 50 words each category remained (in
total: 200). Those stimuli were randomized for each participant in our rating study, so that
every participant received a different order. Participants had to judge the stimuli on a 7-point
likert scale on valence (1 = very negative, 7 = very positive) and arousal (1 = not arousing, 7 =
very arousing). To make sure we had negative stimuli, we selected all Dutch negative words
that had a rating lower than 2.5. For the neutral words, Dutch words were selected that had
a rating higher than 4 and lower than 5. We only selected on Dutch words, because else we
would have influenced the difference of perceived emotional intensity of both languages. If we
would have used the same criteria to select on the English words as well, then the perceived
difference between Dutch and English words would vanish. From our rated data pool, 26 stimuli
per category were selected. Those critical stimuli were being used in the delayed naming task.
They were still matched in terms of frequency, number of letters and word category (p > .1)
(see table 2). For the full list of critical items: see appendix A.
Table 2: The characteristics of the stimuli in experiment 1. Standard deviations are displayed in
parentheses. Reported p-values: paired t-test between different conditions.
# letters
p value letters
English
Dutch
Negative
7.27 (2.32)
7.00 (2.08)
Control
7.27 (2.32)
7.00 (2.08)
p-value
Identical
Identical
Procedure.
word freq
p-value freq
English
Dutch
>.5
2.63 (.54)
2.67 (.57)
>.4
>.5
2.72 (.55)
2.69 (.51)
>.7
>.25
>.6
Before the participants begun with the experiment, they were asked to fill
out the informed consent and the relevant questionnaires. Next, they were being prepared for
the ERP-experiment.
Participants were seated at 60cm in front of a computer screen. Responses were collected
by means of a voice key. We emphasized on the fact that they should not blink during the
trial (from fixation cross until after the response). At the beginning of the experiment, Dutch
instructions were presented on the screen. The subjects started with a practice block in one
language, followed by 3 experimental blocks in the same language. Then again a practice block
14
in the other language, followed by 3 experimental blocks in that language. The language blocks
were contra-balanced over participants
Every trial started with a fixation cross presented in the middle of the screen for 700
ms. Immediately the target word was presented for 1000 ms. Next, an exclamation mark was
shown in the centre of the screen, which was the cue for the participants to pronounce the word
they have just seen. This remained on the screen until the participants pronounced the word, or
until the maximum response time was expired (= 5000 ms). The intertrial interval was 1500
ms, which is a sufficient amount of time to blink. Between each block, subjects could take a
self-paced break.
Every practice block contained 20 trials (10 emotion words and 10 control words). Those
words did not occur in the actual experiment. The experimental blocks contained 52 trials (26
emotion words and 26 neutral words). Each block contained the same words, but they were
randomized every time. An additional restriction was included so that an emotional word could
not be presented on more than three consecutive trials.
After the experiment, the participants had to make backwards and forwards translations
on the critical stimuli. Also, they had to judge each word on emotionality on a 7-point likert
scale (1 = negative, 4 = neutral, 7 = positive). This was done so that unknown or wrongly
translated words could be removed from the analyses. Additionally, words that were rated into
a different category were also excluded from analyses.
EEG Recording.
EEG was recorded continuously at a sampling rate of 512Hz from 64
channels with the Biosemi Active Two system referenced online to the CMS-DRL ground. To
record vertical eye-movements, four unipolar electrodes were placed 1 cm below and above the
eyes. Two other unipolar electrodes were placed beyond the outer canthus of the eyes, to register
horizontal eye-movements. Additionally, two electrodes were placed on the mastoids.
Results and discussion
Behavioural Results.
Trials with a RT slower than 150 ms and faster than 3 standard
deviations above the mean RT for each condition of each subject were excluded from analysis.
Participants that were excluded from the EEG analyses, were also removed here (4 participants,
see below). Additionally, wrongly or not translated words were also excluded. In total, 24.05%
of data were not included in the analyses. A repeated measures ANOVA was performed with
emotion (2; negative vs neutral) and language (2; Dutch vs English) as crossed variables. There
15
was a marginal significant interaction between language and emotion, F1 (1, 15) = 3.84, p =
0.069; F2 (1, 24) = 10.66, p = .003.
1
Further, there was a main effect of language, F1 (1, 15)
= 7.15, p = .017, F2 (1, 24) = 33.50, p = .000 but no significant main effect of emotion, F1 (1,
15) = .24, p = .629, F2 (1, 24) = 0.01, p = .907.
Table 3: Average naming latency in experiment 1 (in ms) for negative and neutral stimuli in L1
and L2 in the delayed naming paradigm.
Neutral L1
Negative L1
Neutral L2
Negative L2
Mean
523.87
512.74
569.77
588.72
Standard Deviation
142.49
125.96
139.18
149.92
EEG Results and discussion.
All EEG data were analyzed using EEPROBE (ANT
inc.). The data were offline re-referenced to the average of the right and the left mastoids. The
EEG data were filtered offline with a bandpass filter of .01Hz – 30 Hz. Additionally, the data
were corrected for blinking, using PCA-transformed EOG components (Nowagk & Pfeifer, 1996).
The ERPs were time-locked to the onset of the target, with epochs being generated in a 1200ms
time window (baseline: -200ms – 0ms), with 0ms being the onset of the target word. Before
averaging, the epochs containing errors (e.g. EEG artefacts,...) and wrongly translated words
were removed. Outliers that were removed from the behavioural analyses were still include here,
apart from the trials were the voicekey did not trigger. Four participants were excluded due to
insufficient (i.e. fewer than 15% in one of the conditions) after artifact rejection. The total data
loss was 53.23%.
Visual inspection of the grand average waveforms showed a large difference between
negative and neutral words, which appeared to be larger for Dutch than English. The effect
appeared to be at its maximum between 300ms and 600 ms. To determine the moment in
time where the waves started to differ significantly, a repeated measures ANOVA with emotion
(negative vs neutral) x language (Dutch vs English) x frontality (posterior vs mid posterior vs
central vs mid frontal vs frontal electrodes) x laterality (left hemisphere vs mid left hemisphere
vs midline vs mid right hemisphere vs right hemisphere) was performed for 20ms bins. In a
series of 3 consecutive time bins, two had to reveal significant effects in order to be taken into
account when determining larger time windows. Given our hypotheses, all effects with emotion
1
We did not hypothesize to find any effect in our reaction time, so this was an unexpected finding. Even
though it was only marginal significant, we performed a paired samples t-test to investigate this interaction.
However, there were no significant results for the relevant pairs (negative Dutch – neutral Dutch: t(15) = 1.00, p
= .333 ; negative English – neutral English: t(15) = -1.73, p = .104.
16
(i.e. main effect and interactions) were taken into account when determining the larger time
window. The electrodes that were being used were, represented over frontality: F3/F1/Fz/F2/F4;
FC3/FC1/FCz/FC2/FC4; C3/C1/Cz/C2/C4; CP3/CP1/CPz/CP2/CP4; P3/P1/Pz/P2/P4.
In all analyses, the Greenhouse-Geisser correction was applied when the sphericity
assumption was violated. Here, the uncorrected degrees of freedom are reported. Analyses of
the bins revealed reliable effects between 400ms until 540ms and 600ms until 880ms. Therefore,
a repeated measures ANOVA with the same factors as before was performed in these time
windows.
In the 400ms – 540ms time window, there was a significant interaction between emotion
and frontality, F(4, 60) = 3.87, ε = .205, p = .036). This interaction showed a marginal
significant effect on the electrodes on the posterior part of the brain (P3/P1/Pz/P2/P4): F(1,
15) = 3.71, ε = .198, p = .073. There was no significant effect for the frontal part of the brain
F(1, 15) = .00, ε = .00, p = .985, the mid-frontal part of the brain, F(1, 15) = 1.22, ε = .075, p
= .287, the central part of the brain, F(1, 15) = 2.58, ε = .147, p = .129, and the mid – parietal
part of the brain, F(1, 15) = 2.92, ε = .163, p = .108. Important for our hypotheses, there was
also a marginal significant interaction between language and emotion, F(1, 15) = 4.40, ε = .227,
p = .053. Here, a significant difference for the contrast Dutch neutral words and Dutch negative
words, F(1, 15) = 8.05, ε = .349, p = .013, was found, but no significant difference for English
neutral words and English negative words, F(1, 15) = 1.13, ε = .07, p = .304.
There were no significant effects for the main effect of emotion, F(1, 15) = 2.08, ε =
.122, p = .169. Further, the other interactions were not significant (no interaction between
emotion and lateralization, F(4, 60) = 2.28, ε = .132, p = .110, no interaction between emotion,
frontality and lateralization, F(16, 240) = 1.01, ε = .063, p = .420, no interaction between
language, emotion and frontalization, F(4, 60) = .70, ε = .045, p = .528, no interaction between
language, emotion an lateralization, F(4, 60) = 1.60, ε = .096, p = .217 and no 4-way interaction
between language, emotion, frontality and lateralization, F(16, 240) = 1.27, ε = .078, p = .295).
In the second time window between 600ms and 880 ms, we found a significant interaction
between language and emotion (F(1, 15) = 5.03, ε = .251, p = .04). Looking into this interaction,
we found a significant difference for the contrast Dutch neutral words and Dutch negative words
(F(1, 15) = 6.88, ε = .314, p = .019), but no significant difference for English neutral words and
English negative words (F(1, 15) = .43, ε = .028, p = .521).
The analyses for the other effects showed no significant effects (no main effect of emotion,
F(1, 15) = 2.08, ε = .122, p = .170, no interaction between emotion and frontalization, F(4, 60)
17
= 1.24, ε = .076, p = .306, no interaction between emotion and lateralization, F(4, 60) = 1.47,
ε = .089, p = .241, no interaction between emotion, frontality and lateralization, F(16, 240) =
1.32, ε = .081, p = .4257, no interaction between language, emotion and frontalization, F(4, 60)
= 1.26, ε = .078, p = .299, no interaction between language, emotion an lateralization, F(4, 60)
= 1.17, ε = .072, p = .328 and no 4-way interaction between language, emotion, frontality and
lateralization, F(16, 240) = 1.08, ε = .067, p = .352).
See Appendix B and Appendix C for the average and standard deviation of the amplitude
of the waveform over participants per condition per electrode for each significant time window.
Additionally, see Appendix D for the grand averages for each electrode.
Figure 1. Grand averages for negative words (solid lines) and neutral words (dotted lines) for
L1 (green lines) and L2 (red lines) of electrode Fz.
Figure 2. Grand averages for negative words (solid lines) and neutral words (dotted lines) for
L1 (green lines) and L2 (red lines) of electrode Cz.
18
Figure 3. Grand averages for negative words (solid lines) and neutral words (dotted lines) for
L1 (green lines) and L2 (red lines) of electrode Pz.
We hypothesized that we would not find any significant results on the reaction time.
However, we did find significant results here. The main effect of language can be explained by
the task that the participants had to execute. This effect showed that participants were slower
for English words in comparison to Dutch words. Even though it was a delayed naming task, it
might be that participants found it hard to formulate the words in L2 and needed more time to
pronounce the words. This could explain why they were slower to name L2 words. There was
also a marginal significant interaction between language and emotion. However, further analyses
showed that the relevant contrasts were not significant.
Concerning the ERPs, we found a large positive amplitude emerging around 400ms until
540ms and another positive wave between 600ms and 800ms. Both components are distributed at
the posterior part of the brain, which was also reflected in the analyses: the interaction between
emotion and frontality revealed a marginal significant interaction between emotion and language
on the posterior part of the brain. According to the timing and the distribution of these peaks,
they can be interpreted as the P300 and the LPP, respectively. The P300 has been shown to
reflect the allocation of resources that have limited capacity for stimuli that are task-relevant
(Sutton, Braren, Zubin & John, 1965). In this experiment, we used negative words. Emotion
stimuli have an intrinsic motivational significance and thus can be automatically processed as
task-relevant (Hajcak et al., 2010), reflecting in a P300. Next, the LPP emerges when emotional
stimuli are being presented (e.g., Cuthbert, Schupp, Bradley, Birbaumer & Lang, 2000 and
Hajcak et al., 2010). Thus, the observation of those two components is valid in this study.
There was also a marginal significant interaction between language and emotion that
was significant over all the electrodes. Further analyses showed that the contrast for negative L1
19
words and neutral L1 words was significant. In contrast to this, the contrast for negative L2
words and neutral L2 words was not significant. In other words, negative Dutch words have a
larger positive amplitude in comparison to neutral Dutch words. This difference was significant
for L1, but not for L2. These results support our hypothesis that L1 is more emotional intense
in comparison to L2.
This experiment only investigated this effect with one type of emotional words. To
elaborate the results, the next experiment investigated if this effect also holds for taboo words.
As previously mentioned, studies that have included taboo words in their stimuli set found that
these stimuli differ from other emotion stimuli (e.g. Guillet & Arndt, 2009; Mathewson et al.,
2008). As a result, with experiment 2 we wanted to replicate the findings of experiment 1 with
taboo words.
Experiment 2
In this experiment, we wanted to investigate if there is any difference in the emotional
experience when participants see a taboo word in either Dutch or English. Also, with the two
experiments executed in this study, we could compare the differences in experience between
emotional and taboo words. We wanted to investigate whether there was a difference in brain
activity between taboo and neutral words and whether this was modulated by language. Jay et
al. (2008) already showed that there is a physiological difference between taboo and neutral
words: taboo words elicited a skin conductance response more frequently than other types of
words. However, ERPs are a more sensitive and online and direct method as skin conductance
responses, so we chose to investigate this difference with ERPs. Because of this, we could follow
the process that happened between the stimulus presentation and the response in real time.
Again, we set up a rating study for our stimuli set. We also used the same procedure and
method as in experiment 1.
Method
Participants.
Eighteen volunteers participated in this experiment (6 male and 12 female)
and received a small monetary compensation. The same criteria were used as in experiment
1. Additionally, they had not participated in the first experiment. Before they were accepted
to this study, they filled out the same questionnaires as in experiment 1. Participants were all
right handed (Van Strien, 2002). Their all reported that their L1 was Dutch, L2 was English
20
and L3 was French (table 4). All of them still came in contact on a regular basis with English
by internet, tv, films, radio, etc. For their L2, most of the participants learned English at school
(n = 8), followed by at home (n = 6). The other participants learned English on holidays (n =
3) and on language camp (n = 1).
Table 4: Self-Assessed Ratings (7-point Likert Scale) of L1, L2 and L3 Proficiency (Experiment
2). Standard deviations are included in parentheses
Language
L1 (Dutch)
L2 (English)
L3 (French)
Stimuli.
Skill
Experiment 1
Writing
6.20 (.51)
Reading
6.67 (.59)
Understanding
6.72 (.46)
Speaking
6.61 (.61)
General proficiency
6.67 (.59)
Writing
4.78 (.81)
Reading
5.72 (.46)
Understanding
5.72 (.46)
Speaking
5.17 (.71)
General proficiency
5.39 (.70)
General proficiency
4.00 (0.91)
Stimuli were also gathered from previous studies (Ayçiçegi-Dinn & Caldwell-
Harris, 2009; Bertels et al., 2009; Eilola, et al., 2007; Guillet & Arndt, 2009; Harris et al., 2003;
Harris, 2004; Janschewitz, 2008; MacKay et.al., 2004 and Zeelenberg et al., 2011) and from
websites that indicated some words as taboo and inappropriate. Control words were matched
the same was as in experiment 1. Here, also a rating study was used similar to experiment 1. In
contrast with experiment 1, participants had to to judge the stimuli on a 7-point likert scale for
different dimensions similar as in Janschewitz (2008): personal use (1 = never use this word, 7
= use this word all the time), offensiveness (1 = not offended, 7 = very offensive), familiarity (1
= never encountered, 7 = encounter this all the time), tabooness personal (1 = not taboo, 7 =
very taboo), tabooness interpersonal (1 = not taboo, 7 = very taboo), arousal (1 = not arousing,
7 = very arousing) and imaginability (1 = does not recall a mental picture, 7 = recall a very
vivid mental picture). Despite our efforts get a good taboo rating, participants tended to give
21
low scores on the taboo words when rating for tabooness (overall average rating taboo: 2.58,
overall average rating neutral: 1.09; on a likert scale ranging from 1 to 7). Thus, we decided to
reject the taboo words with the lowest ratings. Additionally, verbs were also included to have a
sufficient amount of stimuli. Eventually, 30 stimuli in each condition were constructed (taboo
Dutch, neutral Dutch, taboo English and neutral English). These critical stimuli were used in
the delayed naming task. They were still matched on the same criteria as in experiment 1 (see
table 5). For the full list of critical items: see appendix B.
Table 5: The characteristics of the stimuli in experiment 2. Standard deviations are displayed in
parentheses. Reported p-values: paired t-test between different conditions.
# letters
p value letters
English
Dutch
Taboo
5.40 (1.94)
6.10 (2.26)
Control
5.40 (1.94)
6.10 (2.26)
p-value
Identical
Identical
Procedure.
word freq
p-value freq
English
Dutch
>.1
2.54 (.52)
2.74 (.56)
>.1
>.1
2.72 (.57)
2.69 (.50)
>.5
>.1
>.6
The same procedure was used as in experiment 1 with the exception that
the participants saw 60 words per block.
EEG Recording.
The recording and analyses were identical to experiment 1.
Results and discussion
Behavioural Results.
The same criteria as in experiment 1 were used for data-reduction.
In total, 24.05% of data were not included in the analyses. A repeated measures ANOVA
was performed with emotion (negative vs neutral) and language (Dutch vs English) as crossed
variables. There were no significant interactions, F1 (1, 14) = .78, p = .393, F2 (1, 29) = .06, p
= .808), a marginal main effect of emotion for participants, F1 (1, 14) = .3.166, p = .097, F2 (1,
29) = .825, p = .371, and no significant main effect of language, F1 (1, 14) = .064, p = .804, F2
(1, 29) = 1.461, p = .237.
22
Table 6: Average naming latency in experiment 2 (in ms) for taboo and neutral stimuli in L1
and L2 in the delayed naming paradigm.
Neutral L1
Taboo L1
Neutral L2
Taboo L2
Mean
607.92
604.56
613.89
603.94
Standard Deviation
123.07
127.11
122.32
115.68
EEG Results and discussion.
Before the server crashed, we only had 11 participants
to perform the analyses on, which is not sufficient to draw any conclusions on. However, to give
an indication, analyses and discussion are still reported below.
The same pre-processing was used as in experiment 1. Visual inspection of the grand
average waveforms showed a large difference between taboo Dutch and neutral Dutch words.
This difference was smaller for taboo English and neutral English words. The effect appeared to
be at its maximum between 400ms and 800 ms. The same method was used as in experiment 1
to determine the moment in time where the waves started to differ significantly. Also, the same
effect were considered and the same electrodes were used.
In all analyses, the Greenhouse-Geisser correction was applied when the sphericity
assumption was violated. Here, the uncorrected degrees of freedom are reported. Analyses of
the bins revealed reliable effects between -200ms until -140ms, -60ms until 40ms and 380ms until
820ms. Therefore, a repeated measures ANOVA with the same factors as before was performed
in these time windows.
In the -200ms until -140ms time window, there was a significant main effect of emotion,
F(1, 11) = 8.58, ε = .438, p = .014). There was no interaction between language and emotion,
F(1, 11) = 1.76, ε = .138, p = .212, a marginal significant interaction between emotion
and frontalization, F(4, 44) = 2.97, ε = .213, p = .068 no interaction between emotion and
lateralization, F(4, 44) = .75, ε = .064, p = .481, no interaction between emotion, frontality and
lateralization, F(16, 176) = 1.14, ε = .094, p = .348, no interaction between language, emotion
and frontalization, F(4, 44) = .90, ε = .076, p = .414, no interaction between language, emotion
an lateralization, F(4, 44) = .52, ε = .045, p = .633 and no 4-way interaction between language,
emotion, frontality and lateralization, F(16, 176) = 1.19, ε = .098, p = .328.
In the -60ms until 40ms window, there was a significant main effect of emotion F(1, 11)
= 9.19, ε = .455, p = .011). There was no interaction between language and emotion, F(1, 11)
= .01, ε = .001, p = .931, no significant interaction between emotion and frontalization, F(4,
23
44) = 1.22, ε = .100, p = .315 no interaction between emotion and lateralization, F(4, 44) =
.59, ε = .051, p = .532, no interaction between emotion, frontality and lateralization, F(16, 176)
= 1.14, ε = .094, p = .348, no interaction between language, emotion and frontalization, F(4,
44) = .24, ε = .021, p = .725, no interaction between language, emotion an lateralization, F(4,
44) = 1.57, ε = .125, p = .236 and no 4-way interaction between language, emotion, frontality
and lateralization, F(16, 176) = 1.32, ε = .107, p = .281.
In the crucial time window between 380ms and 820ms, we found a significant main effect
of emotion, F(1, 11) = 8.98, ε = .450, p = .012. There was also a significant interaction between
emotion and lateralization, F(4, 44) = 3.50, ε = .241, p = .046. Further analyses showed that
there was an effect on the left hemisphere (electrodes F3, FC4, C3, CP3, P3), F(1, 11) = 7.65, ε
= .410, p = .018, on the mid left hemisphere (electrodes F1, FC1, C1, CP1, P1), F(1, 11) =
11.52, ε = .511, p = .006, on the midline (electrodes Fz, FCz, Cz, CPz, Pz), F(1, 11) = 12.93, ε
= .540, p = .004, and on the mid right hemisphere (electrodes F2, FC2, C2, P2), F(1, 11) =
6.37, ε = .367, p = .028. There was only a marginal significant effect on the right hemisphere
(electrodes F4, FC4, C4, CP4, P4), F(1, 11) = 3.70, ε = .252, p = .081. The main analyses
showed also a significant interaction between language, emotion and frontalization, F(4, 44) =
3.84, ε = .259, p = .009. Further analyses showed an interaction of emotion and language on the
posterior part of the brain (electrodes P3, P1, Pz, P2, P4), F(1, 11) = 5.173, ε = .320, p = .044.
There was no interaction of emotion and language on the anterior part of the brain (electrodes
F3, F1, Fz, F2, F4), F(1, 11) = 1.90, ε = .147, p = .196, the mid anterior part (electrodes FC3,
FC1, FCz, FC2, FC4), F(1, 11) = .741, ε = .063, p = .408, the central part (electrodes C3, C1,
Cz, C2, C4), F(1, 11) = 2.11, ε = .161, p = .174 and the mid posterior part (electrodes CP3,
CP1, CPz, CP2, CP4), F(1, 11) = 1.87, ε = .131, p = .223. We further analyzed the interaction
between language and emotion at the posterior part of the brain. This showed a significant
difference for the contrast Dutch neutral words and Dutch taboo words, F(1, 15) = 14.03, ε =
.561, p = .003, but no significant difference for English neutral words and English taboo words,
F(1, 15) = 1.31, ε = .107, p = .276.
The main analyses of this window revealed no significant interaction between emotion
and frontalization, F(4, 44) = 2.29, ε = .172, p = .112, no interaction between emotion, frontality
and lateralization, F(16, 176) = 1.31, ε = .107, p = .277, no interaction between language,
emotion an lateralization, F(4, 44) = .63, ε = .054, p = .519 and no 4-way interaction between
language, emotion, frontality and lateralization, F(16, 176) = 1.66, ε = .131, p = .186.
To see the averages and standard deviations of each electrode in the significant time
24
window, see Appendix F. To see the figures of the grand averages, see Appendix G.
Figure 4. Grand averages for taboo words (solid lines) and neutral words (dotted lines) for L1
(green lines) and L2 (red lines) of electrode Fz.
Figure 5. Grand averages for taboo words (solid lines) and neutral words (dotted lines) for L1
(green lines) and L2 (red lines) of electrode Cz.
25
Figure 6. Grand averages for taboo words (solid lines) and neutral words (dotted lines) for L1
(green lines) and L2 (red lines) of electrode Pz.
In contrast with the previous experiment, we did not find any significant results in the
reaction time. This is what we hypothesized in the beginning of this study: there would be no
significant differences between reaction times, whereas there would be a significant effect in the
ERPs. This dissociation between RTs and ERPs might seem arbitrary, however previous studies
have found the same pattern of results. For example, Carretié et al. (2008) did not always find
a difference between neutral and negative words in a lexical decision task in the behavioural
results. The ERPs, on the other hand, showed that negative words are processed differently
from neutral words. Comparing these findings with the previous experiment, we did not find an
interaction between language and emotion for taboo words. This could mean that the marginal
interaction in the first experiment was indeed not significant, as the further analyses already
showed. On the other hand, it could also mean that the effect of emotional difference between
the two languages is not present when taboo words are used. Future studies should investigate
this matter further.
Concerning the ERPs, we found some effects in our baseline. This is probably due to
the lack of participants in this experiment. Due to the server crash, it was impossible to do the
analyses with more participants. This should be resolved when more participants were added in
the analyses.
For our main effect, we found a large positive peak that emerged around 380ms. The
significant effect lasted until 820ms after target onset. Even though the timing is not at 300ms
precise, the first component can be interpreted as the P300. The timing of the P300 can shift to
a later moment in time (e.g. McCarthy & Dunchin, 1981 and Polich, 1987), which could have
26
happened in this experiment as well. The grand averages also showed that the P300 blends into
the LPP, which would explain why such a large window revealed a significant effect. Also, the
lack of participants can explain this large window as well. The P300 and the LPP are related
to the processing of emotional words (Hajcak et al., 2010), so the observation of those two
components is valid in this experiment. We found an interaction between language, emotion
and frontality, which revealed an interaction between emotion and language in the posterior
part of the brain. This last interaction showed that the difference in amplitude of the P300 is
larger for taboo Dutch words in comparison to neutral Dutch words. This was not the case for
taboo English words and neutral English words. This is a similar finding to the first experiment
with negative words. This larger difference between taboo words and neutral words in L1 is
again evidence for a more emotional experience in L1 in comparison to L2. There was also an
interaction between emotion and lateralization, which showed an effect of emotion on the entire
electrodes, except for the most right row of electrodes (marginal significant). This shows that
the difference between taboo and neutral words is picked up. This is independent from language:
just seeing a taboo word elicited a very strong response in the brain in comparison to seeing a
neutral word. This shows us that having taboo words have a very strong impact in the brain.
General Discussion
Research on bilingualism has revealed that bilinguals differ from monolinguals in several
aspects, such as language processing (even in their native language) and cognitive control
(Anooshian & Hertel, 1994). However, language is not only used to convey propositional
information from a speaker to a listener, but of course also allows us to express emotions
(Leventhal & Scherer, 1987). An intuition shared by many bilinguals is that the emotional
experience of a word in a second language (L2) is considerably less intense than in a first language
(L1) (Dewaele, 2004a, 2004b, 2008). The goal of this paper was to investigate whether L1 and
L2 differ in emotional intensity. This was done by conducting two experiments: in the first
experiment, negative and neutral words were used in L1 and L2. In the second experiment, the
results were replicated with taboo and neutral words in L1 and L2.
In the first experiment, there was a significant effect of language in the reaction times,
namely there was a difference between the two languages. Participants were slower to react on
their L2. This effect was no longer present in experiment 2. We believe that this is due to our
stimuli. In the first experiment we used negative and neutral words in both language. However,
27
in every day life, L2 emotional words are not often used or encountered. Thus, it can be harder
to name those words for bilinguals. On the other hand, in experiment 2 we used taboo and
neutral words in L1 and L2. Here, participants encountered and used the taboo words more
often than the negative words. It is quite common that Flemish people swear in English (“shit”
or “fuck”). Hence, it could be that the participants found those words easier to pronounce. This
then resulted in a non significant effect.
Considering the ERPs, in both experiments we found a positive wave emerging around
300ms and a late positive complex at 600ms. These can be interpreted as the P300 and the
late positive complex (LPP) respectively. Additionally, the topography of the P300 and LPP
is distributed in the posterior part of the brain, which is in accordance to our results. Both
have already been found in emotion literature (e.g. Opitz & Degner, 2012; Wu & Thierry,
2012): emotional words draw more attention towards them in comparison with neutral words
(Cuthbert, et al., 2000 and Hajcak et al., 2010). The finding of these two components is in line
with our predictions. Participants were presented with neutral and negative or taboo words.
The larger positive amplitude for taboo and negative words demonstrated that our manipulation
was successful: taboo and negative words elicited a larger emotional response in comparison to
neutral words. In addition to this, the difference in amplitude between negative and taboo words
with neutral words was larger in L1 (Dutch) when compared to the L2 (English). These findings
support our hypothesis: L1 has a stronger emotional experience than L2. This is supported as
well with negative words as with taboo words. Experiment 1 is in accordance with previous
behavioural findings (e.g., Anooshian & Hertel, 1994) and clinical studies (e.g., Javier, 1996).
Here, we found that negative words in L1 elicited a stronger emotional response in comparison
to negative words in L2. Experiment 2 is in accordance with previous physiological findings (e.g.,
Harris et al., 2003 and Harris, 2004) and self report (Dewaele, 2004a, 2004b). In experiment 2
we found that taboo words elicited a stronger emotional response in L1 in comparison to L2.
Moreover, when we asked the participants afterwards about their experience with the experiment,
most of them said that they found it weird to say the taboo words out loud in their L1. They
also reported to say them less loud in comparison to the L2 taboo words. The findings of these
two studies support the notion that L1 is considered as a personal language, the language you
use to express emotions with, while L2 is seen as more emotional distant, and more convenient
to say taboo words in (Pavlenko, 2002).
In experiment 2, we also found an interaction between emotion and lateralization. This
was not present in experiment 1. In the second experiment, the presentation of a taboo word
28
elicited a strong response in the brain and this was independent from language. This was not
the case with negative words. This could indicate that taboo words have a stronger emotional
intensity in comparison to negative words.
In the first experiment, we had a lot of data loss. This was mainly due to EEG artefacts.
The second reason for this data loss was because participants did not know the word, or
incorrectly translated them. We tried to prevent this by rating our stimuli in advance, so that
commonly unknown words would not be included in the stimuli set. Additionally, we screened
our participants on proficiency in L1 and L2, so that participants would have sufficient knowledge
of L2. Nevertheless, some trials were lost due to wrong or incorrect translations.
When we compare our results to the previous studies, they are in agreement with
the studies of Dewaele (2004a, 2004b, 2008), Anooshian & Hertel (1994) and Harris (2004).
Unfortunately, other studies did not find any difference between L1 and L2 when using emotional
stimuli. We believe that one of the main reason for this lack of results is the characteristics of
the participants. Previous research did not systematically consider these person variables. For
example, in memory studies, Anooshian & Hertel (1994) found better recall in L1, Ayçiçegi &
Harris (2004) found better recall in L2 for some stimuli and Ayçiçegi-Dinn & Caldwell-Harris
(2009) found no differences between the two languages. However, they did not control person
variables: Anooshian & Hertel and Ayçiçegi-Dinn & Caldwell-Harris used late L2 bilinguals,
while Ayçiçegi & Harris used L2 proficient bilinguals. It could be the case that person variables
influence the results (age of acquisition, learning context, proficiency and language environment).
Hence, future research should investigate if these variables indeed have an influence on the
results. If this is the case, then these variables should be taken into account.
Another explanation of the mixed results in the literature is the stimuli set that was
used in the previous experiments. Either the amount of words per condition was low (e.g.
Ayçiçegi & Harris, 2004; Harris, 2004), the stimuli set contained cognates (e.g. Harris, 2004),
the stimuli set was not been rated by other participants (e.g. Wu & Thierry, 2012), or the
stimuli set was not controlled for different variables (Ayçiçegi & Harris, 2004; Ayçiçegi-Dinn, &
Caldwell-Harris, 2009; Harris et al., 2003). These biases could have influenced the results of
the previous studies. Additionally, some emotion words can have different meanings in their
translation equivalents (Altarriba & Bauer, 2004). Moreover, there are terms of emotions that
only exists in one language (e.g., ‘schadenfreude’ in German, ‘saudade’ in Portugese – “Emotions
and Culture”, 2012). In this experiment, we tried to control our stimuli as best as possible.
There were a sufficient amount of words per category (ca. 27 in each category). Cognates and
29
homographs were left out the stimuli set, since they could activate the other language (Costa
et al., 2000). Additionally, our stimuli set in both experiments is matched over categories on
frequency, amount of letters and word category. Moreover, we did not make use of a database
like ANEW, since it was not been constructed for Dutch words when the study started (now
Moors et al., 2013, have constructed a database for Dutch words). Nevertheless, when using
a database, several words can be more emotional in L1 than the translation equivalent in L2.
This can be solved by rating the stimuli. Here, all our stimuli were rated before the experiment
begun, so that we would have a stimuli set with the desired characteristics, rated by a sample
that is similar to the participants in the ERP experiments. With these caretakings, we hope
that we ruled out any (or most) biases that could have influenced the results.
To summarize, the mixed results found in the literature could be due to the person
variables and/or due to the characteristics of the stimuli set. This experiment tried to control
these variables: we recruited late L2 bilinguals, who lived in an L1 dominant environment and
were proficient in their L2. For our stimuli sets, we controlled for different variables (see above)
and rated the sets by other participants. Unfortunately, there are also some flaws in these
experiments. First of all, it is difficult to literally translate taboo words, and even if there were
literal translation available, those translations can differ in emotional experience in the two
languages. However, in this experiment we made sure that the two targets had the equal amount
of emotional experience in L1 and L2. Additionally, because literally translated taboo words
are rare, this experiment made use of targets with different synonyms (e.g. ‘screw’ and ‘fuck),
which could be exchanged with other targets. This has been done to make sure that there were
sufficient amount of targets in each condition. Also, in the taboo experiment we made use of
verbs as well. This was done because otherwise there would be an insufficient amount of targets
per condition. Nonetheless, the experiment would have been cleaner if the stimuli set would
only consists of nouns, similar to the first experiment.
Second, when we constructed the stimuli set, we used the frequency of the words from
the SUBTLEX-Dutch database (Keuleers et al., 2010) and the SUBTLEX-English database
(Brysbaert & New, 2009). Yet, the construction of the SUBTLEX-English database relies on
the frequency of native English speakers. The frequency of these words can differ from the
frequency of L2 English speakers. We tried to accommodate to this shortcoming by rating the
stimuli. Nevertheless, it would also be preferred to have frequency ratings from an L2 database.
However, no such databases exist up to now.
For follow-up studies, we first of all suggest to try to replicate these results with positive
30
words. Palazova et al. (2011) have demonstrated that positive words are being processed faster in
comparison to negative and neutral words. Although several studies did not find any significant
effects for positive words (e.g. Harris, et al., 2003 and Scott et al., 2009), the effect should also
be present with positive words if bilinguals indeed have different emotional experience in L1
and L2. Additionally, the stimuli can also be presented in the auditory modality. Harris et
al. (2003) already showed that when they presented stimuli in auditory, the skin conductance
levels were higher than in the visual stimuli. Thus, it could be stimuli presented auditory in
L1 and L2 have an even stronger effect. Another follow-up experiment could be to replicate
these findings with pictorial stimuli. Previous studies have shown that using pictures elicited
a stronger emotional response in comparison with emotional words (e.g. Hinojosa, Carretié,
Valcárcel, Méndez-Bértolo,& Pozo, 2009). Thus, using pictures and/or presenting stimuli in an
auditory modality could provide us more more information about this difference in emotional
intensity between L1 and L2.
A second line of research is to investigate which person variables influence the emotional
experience in L1 and L2. The four most important ones are age of acquisition, L2 proficiency,
learning context of L2 and the language environment. Harris et al., (2006) have already proposed
an emotional context of learning theory. When L2 is learned after a certain age, this will
influence the emotional responsiveness in L2. Also, when bilinguals are less proficient in their
L2, this also results into a general decline in emotional responsiveness. The same can be stated
for language environment. When bilinguals are immersed into their L2 environment, it can have
a beneficial effect on the emotional experience in the second language. However, the learning
context of L2 is not straightforward. One could say that learning L2 in a naturalistic way would
benefit the emotional experience in L2. Yet, in this experiment taboo words were used – words
that are often learned in a naturalistic environment – . Thus, following this reasoning, one would
expect to find no difference in experiment 2 between L1 and L2 in the difference in amplitudes
for taboo words and neutral words. But this is not what we have found. Hence, it is not so clear
how strong this learning context have to be. Perhaps it has more influence on the emotional
experience if the entire language is learned in a naturalistic setting.
To conclude these two studies, we have shown that there is a difference in emotional
experience between L1 and L2. This has been demonstrated by negative and taboo words. We
have compared our results to previous findings and have proposed some follow-up studies to
elaborate this effect.
31
References
Altarriba, J., & Bauer, L.M. (2004). The distinctiveness of emotion concepts: a comparison
between emotion, abstract and concrete words. American Journal of Psychology, 117,
389 – 410. http://www.jstor.org/stable/4149007
Anderson, A. K. (2005). Affective influences on the attentional dynamics supporting awareness.
Journal of Experimental Psychology: General, 134, 258 – 281. DOI: 10.1037/00963445.134.2.258
Anooshian, L., & Hertel, P. (1994). Emotionality in free recall: Language specificity in bilingual
memory. Cognition and Emotion, 8 (6), 503 – 514. DOI: 10.1080/02699939408408956
Ayçiçegi, A. & Harris, C.L. (2004). Bilinguals’ recall and recognition of emotion words. Cognition
and Emotion, 18 (7), 977 – 987. DOI: 10.1080/02699930341000301
Ayçiçegi-Dinn, A. & Caldwell-Harris, C.L. (2009). Emotion-memory effects in bilingual speakers:
a levels-of-processing approach. Bilingualism: Language and Cognition 12 (3), 291 – 303
DOI: 10.1017/S1366728909990125
Belezza, F.S., Greenwald, A.G., & Banaji, M.R. (1986). Word High and Low in Pleasantness as
Rated by Male and Female College Students. Behaviour Research Methods, Instruments,
& Computers, 18 (3), 299 – 303. DOI: 10.3758/BF03204403
Bertels, J., Kolinsky, R. & Morais, J. (2009). Norms of Emotional Valence, Arousal, Threat
Value and Shock Value for 80 Spoken French Words: Comparison between Neutral
and Emotional Tones of Voice. Psychologica Belgica, 49 (1), 19 – 40. http://www
.ingentaconnect.com/content/acad/psyb/2009/00000049/00000001/art00002
Bertels, J., Kolinsky, R. & Morais, J. (2010). Emotional Valence of Spoken Words Influences the Spatial Orienting of Attention. Acta Psychologica 134, 264 – 278. DOI:
10.1016/j.actpsy.2010.02.008.
Brysbaert, M. & New, B. (2009). Moving beyond Kucera and Francis: a critical evaluation of
current word frequency norms and introduction of a new and improved word frequency
measure for American English. Behaviour Research Methods, 41 (4), 997 – 990. DOI:
10.3758/BRM.41.4.977.
Buxbaum, E. (1949). The Role of Second Language in the Formation of Ego and Superego.
Psychoanalytic Quarterly, 18, 279 – 289.
Carretié, L., Hinojosa, J.A., Albert, J., López-Martin, S., de la Gándara, B.S., Igoa, J.M., &
Sotillo, M. (2008). Modulation of ongoing cognitive processes by emotionally intense
32
words. Psychophysiology, 45 (2), 188 – 196. DOI: 10.1111/j.1469-8986.2007.00617.x
Codispoti, M., & De Cesarei, A. (2007). Arousal and attention: Picture size and emotional
reactions. Psychophysiology, 44, 680–686. DOI: 10.1111/j.1469-8986.2007.00545.x
Colbeck, K.L. & Bowers, J.S. (2012). Blinded by taboo words in L1 but not in L2. Emotion, 12
(2), 217 – 222. DOI: 10.1037/a0026387
Costa, A., Caramazza, A., & Sebastian-Galles, N. (2000). The cognate facilitation effect:
Implications for models of lexical access. Journal of Experimental Psychology Learning
Memory and Cognition, 26, 1283 – 1296. DOI: 10.1037/0278-7393.26.5.1283
Cuthbert, B.N., Schupp, H.T., Bradley, M.M., Birbaumer, N., & Lang, P.J. (2000). Brain
potentials in affective picture processing: covariation with autonomic arousal and affective
report. Biological Psychology, 52, 95 – 111. DOI: 10.1016/S0301-0511(99)00044-7,
Dewaele, J.-M. (2004a). The Emotional Force of Swearwords and Taboo Words in the Speech of
Multilinguals. Journal of multilingual and multicultural development, 25 (2&3), 204 –
222. DOI: 10.1080/01434630408666529
Dewaele, J.-M. (2004b). Blistering Barnacles! What Language do Multilinguals Swear in?!
Estudios de Sociolingüística 5 (1), 83 – 105. DOI: 10.1558/sols.v5i1.83
Dewaele, J.-M. (2008). The Emotional Weight of I love you in Multilinguals’ Languages. Journal
of Pragmatics, 40, 1753 – 1780. DOI: 10.1016/j.pragma.2008.03.002
Dhooge, E. , De Baene, W., Severens, E., & Hartsuiker, R. J. (submitted). Lifting the ban: How
social context affects the brain’s response to taboo words.
Eilola, T.M. (2009). Processing emotionally charged words in the first and second language.
Unpublished doctoral dissertation, University of Kent, Canterbury, United Kingdom.
Eilola, T.M., & Havelka, J. (2010). Behavioural and physiological responses to the emotional
and taboo Stroop tasks in native and non-native speakers of English. International
Journal of Bilingualism, 15 (3), 353 – 369. DOI: 10.1177/1367006910379263
Eilola, T.M., Havelka, J. & Sharma, D. (2007). Emotional Activation in the First and Second
Language. Cognition & Emotion, 21(5), 1064 – 1076. DOI: 10.1080/02699930601054109
Emotions and Cultures (2012). In Wikipedia. Retrieved May 11, 2012, from Wikipedia
Ferré, P., Garcia, T., Fraga, I., Sanchez-Casus, R. & Molero, M. (2010). Memory for Emotional
Words in Bilinguals: Do Words Have the Same Emotional Intensity in the First and
in the Second Language? Cognition and Emotion 24 (5), 760 – 785. Full accDOI:
10.1080/02699930902985779
Grosser, G.S., Walsh, A.A. (1966). Sex differences in the differential recall of taboo and neutral
33
words. Journal of Psychology, 63 (2), 219 – 227. DOI: 10.1080/00223980.1966.10543035
Guillet, R. & Arndt, J. (2009). Taboo Words: The Effect of Emotion of Memory for Peripheral
Information. Memory & Cognition, 37 (6), 866 – 879. DOI: 10.3758/MC.37.6.866.
Hadley, C.B. & MacKay, D.G. (2006). Does emotion help or hinder immediate memory? Arousal
versus priority-binding mechanisms. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 32 (1), 79 – 88. DOI: 10.1037/0278-7393.32.1.79
Hajcak, G., MacNamara, A., & Olvet, D.M. (2010). Event-related potentials, emotion and
emotion regulation: an integrative review. Developmental neuropsychology, 35 (2), 129 –
155. DOI: 10.1080/87565640903526504.
Harris, C.L. (2004). Bilingual Speakers in the Lab: Psychophysiological Measures of Emotional
Reactivity. Journal of Multilingual and Multicultural Development, 25 (2 & 3), 223 –
247. DOI: 10.1080/01434630408666530
Harris, C.L. & Ayçiçegi-Dinn, A. (2009). Emotion and lying in a non-native language. International Journal of Psychophysiology, 71, 193 – 204. DOI: 10.1016/j.ijpsycho.2008.09.006.
Harris, C.L., Ayçiçegi, A. & Gleason, J.B. (2003). Taboo Words and Reprimands Elicit
Greater Autonomic Reactivity in a First Language than in a Second Language. Applied
Psycholinguistics, 24 (4), 561 – 579. DOI: 10.1017/S0142716403000286
Hinojosa, J.A., Carretié, L., Valcárcel, M.A., Méndez-Bértolo, C., & Pozo, M.A. (2009).
Electrophysiological differences in the processing of affective information in words
and pictures. Cognitive, Affective& Behavioural Neuroscience, 9, 173 – 189. DOI:
10.3758/CABN.9.2.173
Janschewitz, K. (2008). Taboo, Emotionally Valenced, and Emotionally Neutral Word Norms.
Behaviour Research Methods, 40 (4), 1065 – 1074. DOI: 10.3758/BRM.40.4.1065
Javier, R.A. (1996). In Search of Repressed Memories in Bilingual Individuals. In R.P. Foster,
M., Moskovitz & R.A. Javier (Eds.), Reaching across boundaries of culture and class:
Widening the scope of psychotherapy (225 – 241). Northvale, NJ: Aronson.
Jay, T.B. (1992). Cursing in America. Philadelphia: John Benjamins.
Jay, T.B. (2000). Why We Curse: A Neuro-Psycho-Social Theory of Speech. Philadelphia: John
Benjamins Pub Co.
Jay, T., Caldwell-Harris, C. & King, K. (2008). Recalling Taboo and Nontaboo Words. American
Journal of Psychology, 121 (1), 83 – 103. https://www.mcla.edu/Undergraduate/
uploads/textWidget/1457.00016/documents/RecallTabooNonTabooWords.pdf
Kensinger, E.A. & Corkin, S. (2003). Memory enhancement for emotional words: Are emotional
34
words more vividly remembered than neutral words? Memory & Cognition, 31 (8), 1169
– 1180. DOI: 10.3758/BF03195800
Keuleers, E., Brysbaert, M. & New, B. (2010). SUBTLEX-NL: A new frequency measure for
Dutch words based on film subtitles. Behavior Research Methods, 42(3), 643-650. DOI:
10.3758/BRM.42.3.643.
Kissler, J., & Koessler, S. (2011). Emotionally positive stimuli facilitate lexical decisions – An
ERP study. Biological Psychology, 86, 254 – 264. DOI: 10.1016/j.biopsycho.2010.12.006.
Lemhöfer, K., & Broersma, M. (2011). Introducing LexTale, a quick and valid lexical test for
advanced learners of English. Behaviour Research Methods DOI: 10.3758/s13428-0110146-0
Leventhal, H. & Scherer, K. (1987). The relationship of emotion and cognition: a functional approach to a semantic controversy. Cognition & Emotion, 1 (1), 3 – 28 DOI:
10.1080/02699938708408361
MacKay, D.G., & Ahmetzanov, M.V. (2005). Emotion, Memory and Attention in the Taboo
Stroop Paradigm – An Experimental Analogue of Flashbulb Memories. Psychological
science, 16 (1), 25 – 32. DOI: 10.1111/j.0956-7976.2005.00776.x
MacKay, D.G., Shafto, M., Taylor, J.K., Marian, D.E., Abrams, L. & Dyer, J.R. (2004).
Relations between emotion, memory and attention: Evidence from taboo Stroop, lexical
decision, and immediate memory tasks. Memory & Cognition, 32 (3), 474 – 488. DOI:
10.3758/BF03195840
Mathewson, K.J., Arnell, K.M. & Mansfield, C.A. (2008). Capturing and holding attention: The
impact of emotional words in rapid serial visual presentation. Memory and Cognition,
36 (1), 182 – 200. DOI: 10.3758/MC.36.1.182
McCarthy, G., & Donchin, E. (1981). A metric for thought: a comparison of P300 latency and
response time. Science, 211, 77 – 80. DOI: 10.1126/science.7444452
Moors, A., De Houwer, J., Hermans, D., Wanmaker, S., van Schie, K., Van Harmelen, A.L.,
De Schryver, M., De Winne, J., & Brysbart, M. (2013). Norms of valence, arousal,
dominance and age of acquisition for 4,300 Dutch words. Behaviour Research Methods,
45 (1), 169 - 177. DOI: 10.3758/s13428-012-0243-8.
Nowagk, R., Pfeifer, E. (1996). UNIX implementation of the ERP evaluation package (EEP 3.0).
In A.D. Friederici & D.Y. Von Cramon (Eds.), Annual report of Max-Planck Institute of
Cognitive Neuroscience. Liepzig, Germany: Max-Planck-Institute.
Opitz, B., & Degner, J. (2012). Emotionality in a second language: It’s a matter of time.
35
Neuropsychologia, 50 (8), 191 - 1967. DOI: 10.1016/j.neuropsychologia.2012.04.021
Palazova, M., Mantwill, K., Sommer, W., & Schacht, A. (2011). Are effects of emotion single
words non-lexical? Evidence from event-related brain potentials. Neuropsychologia, 49,
2766 – 2775. DOI: 10.1016/j.neuropsychologia.2011.06.005.
Pavlenko, A. (2002). Bilingualism and Emotions. Multilingua, 21. 45 – 78. DOI: 10.1515/mult.2002.004
Polich, J. (1987). Task difficulty, probability, and interstimulus interval as determinants of P300
from auditory stimuli. Electroencephalography and Clinical Neurophysiology, 68, 311 –
320. DOI: 10.1016/0168-5597(87)90052-9
Pratto, F. & John, O.P. (1991). Automatic vigilance: The attention-grabbing power of negative
social information. Journal of Personality and Social Psychology, 61, 380 – 391. DOI:
10.1037/0022-3514.61.3.380
Scott, G.C., O’Donnell, P.J., Leuthold, H., & Sereno, S.C. (2009). Early emotion word
processing: evidence from event-related potentials. Biological Psychology, 80, 95 – 104.
DOI: 10.1016/j.biopsycho.2008.03.010.
Schrauf, R.W. (2000). Bilingual Autobiographical Memory: Experimental Studies and Clinical
Cases. Culture & Psychology, 6 (4), 387 – 417. DOI: 10.1177/1354067X0064001
Schupp, H. T., Cuthbert, B. N., Bradley, M. M., Hillman, C. H., Hamm, A. O., & Lang, P.
J. (2004). Brain processes in emotional perception: Motivated attention.Cognition and
Emotion, 18, 593–611. DOI:10.1080/02699930341000239
Severens, E., Janssens, I., Kühn, S., Brass, M., Hartsuiker, R.J. (2011). When the brain tames
the tongue: Covert editing of inappropriate language. Psychophysiology, 86, 1252 – 1257.
DOI: 10.1111/j.1469-8986.2011.01190.x.
Speedie, L.J., Wertman, E., Ta’ir, J., & Heilman, K.M. (1993). Disruption of automatic
speech following a right basal ganglia lesion. Neurology, 43 (9), 1768 – 1774. DOI:
10.1212/WNL.43.9.1768
Stroop, J.R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental
Psychology, 18, 643 – 662. DOI: 10.1037/0096-3445.121.1.15
Sutton, T.M., Altarriba J., Bianico, J.L. & Basnight-Brown, D.M. (2007). The automatic access
of emotion: Emotional Stroop effects in Spanish-English bilingual speakers. Cognition &
Emotion, 21 (5), 1077 – 1090. DOI: 10.1080/02699930601054133
Sutton, S., Braren, M., Zubin, J., & John, E.R. (1965). Evoked-potential correlates of stimulus
uncertainty. Science, 150, 1187 – 1188. DOI: 10.1126/science.150.3700.1187
Thomas, S.J., Johnstone, S.J. & Gonsalvez, C.J. (2007). Event-related Potentials During an
36
Emotional Stroop Task. International Journal of Psychophysiology, 63, 221 – 231. DOI:
10.1016/j.ijpsycho.2006.10.002
Van Lancker, D., & Cummings, J.L. (1999). Expletives: neurolinguistic and neurobehavioral
perspectives on swearing. Brain Research Reviews, 31, 83 – 104. DOI: 10.1016/S01650173(99)00060-0
Van Strien, J.W. (2002). The Dutch handedness questionnaire. Based on Van Strien, J.W.
(1988). Handedness and hemispheric laterality. Thesis, Vrije universiteit Amsterdam, The Netherlands. http://1011.psyweb.nl/homepage/jan_van_strien_files/
hquestionnaire_article.pdf
Wu, Y.J., & Thierry, G. (2012). How reading in a second language protect your heart. The
Journal of Neuroscience, 32 (19), 6485 – 6489. DOI: 10.1523/ JNEUROSCI.6119-11.2012
Zeelenberg, R., Bocanegra, B.R. & Pecher, D. (2011). Emotion-Induced Impairments in Speeded
Word Recognition Tasks. Experimental Psychology, 58 (5), 400 – 411. DOI: 10.1027/16183169/a000109.
37
Appendix A
Table 7: Critical items in experiment 1
English
Negative
Abuse
Addict
Avalanche
Bullet
Crime
Damage
Danger
Despair
Disease
Disgust
Dungeon
Fear
Grief
Humiliation
Injury
Manslaughter
Massacre
Misery
Mutilation
Paralysis
Plague
Poison
Rage
Resignation
Suffocation
Violence
Control
Pulse
Switch
Broadcast
Border
Entry
Weight
Amount
Descent
Current
Bracket
Lobster
Meat
Speed
Expectation
Player
Refrigerator
Corridor
Mirror
Replacement
Publisher
Button
Cattle
Soil
Inheritance
Hibernation
Evidence
Dutch
Negative
Misbruik
Verslaafde
Lawine
Kogel
Misdaad
Schade
Gevaar
Wanhoop
Ziekte
Afkeer
Kerker
Angst
Verdriet
Vernedering
Letsel
Doodslag
Bloedbad
Ellende
Verminking
Verlamming
Pest
Gif
Woede
Ontslag
Suffocation
Geweld
Control
Hartslag
Schakelaar
Omroep
Grens
Toegang
Belang
Aantal
Afkomst
Stroom
Beugel
Kreeft
Vlees
Snelheid
Verwachting
Speler
Koelkast
Doorgang
Spiegel
Vervanging
Uitgeverij
Knop
Vee
Bodem
Erfenis
Winterslaap
Bewijs
38
Appendix B
Table 8: Averages and standard deviation of the amplitude of the waveform in the critical time
window 400ms - 540ms in experiment 1.
F3
F1
Fz
F2
F4
L1 neutral
Mean
0,99278
-0,26930
-0,27258
0,18616
-2,71765
Stdev
1,37943
2,11429
1,57795
1,91881
11,92100
L1 negative
Mean
1,09213
0,38673
0,37596
0,55266
-1,50621
Stdev
1,75569
1,99819
1,58566
1,99978
8,17608
L2 neutral
Mean
0,29932
-0,36802
0,25965
0,37243
-1,77550
FC3
FC1
FCz
FC2
FC4
Stdev
2,16610
3,52334
3,25674
2,75042
10,01782
L2 negative
Mean
0,38841
-0,47379
0,39067
0,14134
-0,62635
Stdev
2,14759
2,41585
1,81831
1,73593
2,56415
0,66627
0,22438
-0,13217
-0,32029
-0,39369
1,16127
1,42856
2,05770
2,16530
2,75201
1,36716
1,25182
0,53903
0,81383
0,55413
1,80479
1,93083
2,05608
2,27608
2,77868
1,00455
0,86452
0,76614
0,57338
0,36553
2,35944
2,77838
2,68919
3,10241
2,41620
0,95057
0,66656
0,82415
0,16241
0,04337
1,80915
2,03556
2,34313
2,35515
2,12826
C3
C1
Cz
C2
C4
1,80853
-0,26131
0,09137
-0,28010
0,47788
1,87405
2,54506
2,19231
2,20576
1,91876
2,94998
0,90166
1,07862
1,13203
1,52720
2,18175
2,67411
3,05945
2,72414
2,12243
2,37495
0,62780
0,58143
0,80783
0,24932
2,67510
3,24280
2,92091
2,87179
2,75661
2,30742
0,55301
0,77959
0,87430
0,55818
2,14304
2,54596
3,02018
2,22104
2,66235
CP3
CP1
CPz
CP2
CP4
1,77403
0,81620
0,43943
0,78572
1,64188
2,58534
2,81475
2,16491
2,06729
2,31414
3,07727
2,14542
1,81276
1,98782
2,30776
3,20901
2,88055
2,88965
2,59301
2,65516
1,81126
1,76326
1,50917
1,62433
1,38814
3,83035
2,46802
3,43296
2,96418
3,23413
2,38795
2,03441
1,28484
1,71138
1,36736
2,78664
2,30494
3,16935
2,65119
2,62942
P3
P1
Pz
P2
P4
1,54686
2,11391
1,55957
1,15094
1,84461
2,98707
1,95784
2,01366
2,65422
3,34149
3,02241
3,27787
3,21988
2,41034
2,95479
3,72484
2,72912
2,98329
2,74582
3,17788
2,11155
2,66562
1,73545
1,51620
1,94117
3,47351
3,27591
3,31021
3,21477
2,54075
2,59748
2,97223
2,10690
1,88220
2,17973
3,11896
2,92815
3,19391
2,91173
4,13656
39
Appendix C
Table 9: Averages and standard deviation of the amplitude of the waveform in the critical time
window 600ms - 880ms in experiment 1.
F3
F1
Fz
F2
F4
L1 neutral
Mean
-0,24151
-1,51777
-1,22519
-0,40744
-3,76361
Stdev
2,50877
2,52406
2,35666
2,12173
14,43403
L1 negative
Mean
0,14285
-1,09979
-0,79006
-0,41384
-2,21507
Stdev
2,91734
2,84958
3,61421
3,14340
9,26976
L2 neutral
Mean
-0,28833
-1,06424
-0,10124
0,38287
-2,41401
FC3
FC1
FCz
FC2
FC4
Stdev
2,44251
4,26945
3,55663
2,86952
13,31273
L2 negative
Mean
-0,10233
-1,31538
-0,03236
0,30843
-0,09176
Stdev
2,82969
3,81686
2,38268
2,14758
2,43001
-1,00620
-1,25869
-1,09078
-1,13435
-1,30545
2,48349
2,42150
2,83157
3,31278
3,29075
-0,36019
-0,22167
-0,95416
-0,36124
-0,79821
3,16826
3,84772
3,67631
3,98512
3,80934
0,63092
0,75908
0,51364
0,76560
1,00096
2,69856
3,27646
2,88926
3,41783
2,54552
0,68909
0,52710
0,77067
0,41092
1,00280
2,00531
2,57287
3,26268
3,10931
2,29505
C3
C1
Cz
C2
C4
-0,18579
-2,08415
-0,88581
-1,32391
-0,77997
3,20756
3,31670
2,72994
2,86092
2,48934
0,54556
-1,47273
-0,51475
-0,59945
-0,25479
4,00206
4,26005
4,75378
3,99250
2,51596
2,18937
0,38205
0,76917
1,06387
0,54381
3,19354
4,39163
3,57425
3,50125
3,02785
1,95551
0,28724
1,09562
1,42351
0,88826
2,53948
3,22052
3,69328
2,66997
2,90898
CP3
CP1
CPz
CP2
CP4
-0,73061
-1,30646
-1,07675
-0,81100
-0,04177
3,67812
3,52017
2,74536
2,35905
2,83846
0,21773
-0,37472
-0,43975
-0,32678
0,27645
4,90126
3,85175
4,27595
3,69473
4,29503
1,50923
1,60834
1,76142
1,77383
1,40493
4,04210
2,95759
4,22230
3,43636
3,68027
1,83424
1,82565
1,40662
1,96746
1,30432
3,17893
2,08496
3,74979
2,82976
2,81214
P3
P1
Pz
P2
P4
-1,62776
-0,43240
-0,90815
-1,19913
-0,28532
4,18591
2,79894
2,71175
3,10985
4,06261
-0,79973
0,03048
0,22284
-0,70783
-0,39594
5,18372
4,03504
3,96271
3,85340
3,39667
1,24234
2,03956
1,15495
1,05482
1,99409
3,76014
3,65705
4,09625
4,17755
3,69446
1,59026
2,34867
1,64044
1,54743
1,81299
3,29705
3,08576
3,64373
3,09586
4,58834
40
Appendix D
Figure 7. Grand averages for emotion words (solid lines) and neutral words (dotted lines)
for L1 (green lines) and L2 (red lines) of electrodes F3, F1 and Fz.
41
Figure 8. Grand averages for emotion words (solid lines) and neutral words (dotted lines)
for L1 (green lines) and L2 (red lines) of electrodes F2, F4 and FC3.
42
Figure 9. Grand averages for emotion words (solid lines) and neutral words (dotted lines)
for L1 (green lines) and L2 (red lines) of electrodes FC1, FCz and FC2.
43
Figure 10. Grand averages for emotion words (solid lines) and neutral words (dotted
lines) for L1 (green lines) and L2 (red lines) of electrodes FC4, C3 and C1.
44
Figure 11. Grand averages for emotion words (solid lines) and neutral words (dotted
lines) for L1 (green lines) and L2 (red lines) of electrodes Cz, C2 and C4.
45
Figure 12. Grand averages for emotion words (solid lines) and neutral words (dotted
lines) for L1 (green lines) and L2 (red lines) of electrodes CP3, CP1 and CPz.
46
Figure 13. Grand averages for emotion words (solid lines) and neutral words (dotted
lines) for L1 (green lines) and L2 (red lines) of electrodes CP2, CP4 and P3.
47
Figure 14. Grand averages for emotion words (solid lines) and neutral words (dotted
lines) for L1 (green lines) and L2 (red lines) of electrodes C1, Pz and P2.
48
Figure 15. Grand averages for emotion words (solid lines) and neutral words (dotted
lines) for L1 (green lines) and L2 (red lines) of electrodee P4.
49
Appendix E
Table 10: Critical items in experiment 2
English
Taboo
Ass
Bitch
Blowjob
Boner
Boobs
Cocksucker
Crap
Cunt
Dickhead
Fart
Fatso
Sissy
Fuck
Fucker
Jerk
Loser
Moron
Motherfucker
Nigger
Puke
Queer
Rape
Retard
Screw
Shit
Shithead
Tits
Wanker
Whore
Control
Reet
Teef
Pijpbeurt
Stijve
Borsten
Rotzak
Kak
Kut
Eikel
Scheet
Vetzak
Mietje
Neuken
Klootzak
Smeerlap
Sukkel
Debiel
Hoerenjong
Neger
Kots
Janet
Verkrachting
Achterlijke
Poepen
Schijten
Lul
Tieten
Rukker
Hoer
Dutch
Taboo
Ant
Paint
Artwork
Peach
Watch
Profession
Poke
Push
Customer
Frog
Diver
Value
Send
Answer
Twin
Cover
Whale
Satisfaction
Shovel
Glue
Candy
Cure
Change
Erase
Jump
Granddad
Hill
Carpet
Guide
Control
Mier
Verf
Kunstwerk
Perzik
Horloge
Beroep
Por
Duw
Klant
Kikker
Duiker
Waarde
Sturen
Antwoord
Tweeling
Omslag
Walvis
Voldoening
Schop
Lijm
Snoep
Geneesmiddel
Verandering
Wissen
Springen
Opa
Heuvel
Tapijt
Gids
50
Appendix F
Table 11: Averages and standard deviation of the amplitude of the waveform in the critical time
window 380ms - 820ms in experiment 2.
F3
F1
Fz
F2
F4
L1 neutral
Mean
1,60014
2,16024
1,93882
1,76736
0,67580
Stdev
3,14775
4,14551
3,37358
4,56258
2,86920
L1 negative
Mean
2,33117
3,09324
3,27483
3,39822
1,30237
Stdev
3,22654
4,14443
4,03340
5,51658
2,42457
L2 neutral
Mean
1,63414
1,82506
1,94311
1,27487
0,41760
FC3
FC1
FCz
FC2
FC4
Stdev
2,68101
2,21441
2,39488
2,17824
3,40613
L2 negative
Mean
1,83958
2,39583
3,06417
1,61930
1,02912
Stdev
3,07845
2,21083
2,19999
2,13258
3,61294
2,69668
1,87671
1,62029
2,52633
2,64178
4,10560
4,38361
3,70422
3,89061
4,20885
3,10994
3,58435
3,04763
3,07410
3,83410
4,65398
5,33704
4,79089
4,58183
5,22411
2,61743
2,07417
1,09905
1,90575
1,98833
2,65975
2,99077
2,39983
2,33038
2,91540
2,95962
3,11657
2,07534
2,75983
2,05681
3,00411
2,88856
2,34462
2,34214
3,15018
C3
C1
Cz
C2
C4
2,71836
2,06443
1,77621
1,83536
2,68924
3,86746
4,05321
4,29959
4,13679
4,14092
3,80990
3,67644
3,28497
3,62526
3,97603
5,06650
4,94041
5,61486
5,08198
4,86204
2,71556
1,95137
1,63669
1,87680
2,71592
3,04742
2,50684
2,49290
2,66170
3,13307
3,56581
2,77942
2,48318
2,63188
3,05407
3,62755
2,45898
2,71233
2,80040
3,35882
CP3
CP1
CPz
CP2
CP4
2,74269
2,56472
2,13010
3,03039
3,63792
4,12760
4,54244
4,74765
4,09203
5,64263
4,32158
4,40484
4,29847
4,55287
4,84378
5,29785
5,14201
5,65638
5,13839
6,45185
3,51245
2,45080
2,49835
2,51319
3,26101
4,88377
2,80993
2,92720
3,19800
3,61060
3,50621
3,53938
3,75404
3,22433
4,62735
3,02503
3,27443
3,29099
3,63860
3,85284
P3
P1
Pz
P2
P4
2,47852
2,87847
2,30730
2,96265
2,48219
4,05894
5,91812
4,86513
4,48866
3,48167
4,67772
5,02280
4,50934
4,99206
5,04609
5,40456
6,57932
5,98991
5,43712
5,18523
2,30631
2,72942
2,69882
2,69735
4,06102
3,48684
3,08792
2,99706
3,53262
5,77600
3,85779
3,64779
3,74797
3,60059
2,52528
3,92068
3,61525
3,67632
3,97194
4,40752
51
Appendix G
Figure 16. Grand averages for taboo words (solid lines) and neutral words (dotted lines)
for L1 (green lines) and L2 (red lines) of electrodes F3, F1 and Fz.
52
Figure 17. Grand averages for taboo words (solid lines) and neutral words (dotted lines)
for L1 (green lines) and L2 (red lines) of electrodes F2, F4 and FC3.
53
Figure 18. Grand averages for taboo words (solid lines) and neutral words (dotted lines)
for L1 (green lines) and L2 (red lines) of electrodes FC1, FCz and FC2.
54
Figure 19. Grand averages for taboo words (solid lines) and neutral words (dotted lines)
for L1 (green lines) and L2 (red lines) of electrodes FC4, C3 and C1.
55
Figure 20. Grand averages for taboo words (solid lines) and neutral words (dotted lines)
for L1 (green lines) and L2 (red lines) of electrodes Cz, C2 and C4.
56
Figure 21. Grand averages for taboo words (solid lines) and neutral words (dotted lines)
for L1 (green lines) and L2 (red lines) of electrodes CP3, CP1 and CPz.
57
Figure 22. Grand averages for taboo words (solid lines) and neutral words (dotted lines)
for L1 (green lines) and L2 (red lines) of electrodes CP2, CP4 and P3.
58
Figure 23. Grand averages for taboo words (solid lines) and neutral words (dotted lines)
for L1 (green lines) and L2 (red lines) of electrodes P1, Pz and P2.
59
Figure 24. Grand averages for taboo words (solid lines) and neutral words (dotted lines)
for L1 (green lines) and L2 (red lines) of electrodes P4.