Full Terms & Conditions of access and use can be found athttps://www.tandfonline.com/action/journalInformation?journalCode=pcem20

Cognition and Emotion

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/pcem20

The mental representation and social aspect ofexpressives

Stanley A. Donahoo & Vicky Tzuyin Lai

To cite this article: Stanley A. Donahoo & Vicky Tzuyin Lai (2020) The mental representationand social aspect of expressives, Cognition and Emotion, 34:7, 1423-1438, DOI:10.1080/02699931.2020.1764912

To link to this article: https://doi.org/10.1080/02699931.2020.1764912

View supplementary material

Published online: 18 May 2020.

Submit your article to this journal

Article views: 104

View related articles

View Crossmark data

The mental representation and social aspect of expressivesStanley A. Donahoo a,c and Vicky Tzuyin Laib,c

aDepartment of Linguistics, University of Arizona, Tucson, AZ, USA; bDepartment of Psychology, University of Arizona, Tucson, AZ,USA; cCognitive Science Program, University of Arizona, Tucson, AZ, USA

ABSTRACTDespite increased focus on emotional language, research lacks for the most emotionallanguage: Swearing. We used event-related potentials (ERPs) to investigate whetherswear words have content distinct from function words, and if so, whether thiscontent is emotional or social in nature. Stimuli included swear (e.g. shit, damn),negative but non-swear (e.g. kill, sick), open-class neutral (e.g. wood, lend), andclosed-class neutral words (e.g. while, whom). Behaviourally, swears were recognisedslower than valence- and arousal- matched negative words, meaning that there ismore to the expressive dimension than merely a heightened emotional state. InERPs, both swears and negative words elicited a larger positivity (250–550 ms) thanopen-class neutral words. Later, swears elicited a larger late positivity (550–750 ms)than negative words. We associate the earlier positivity effect with attention due tonegative valence, and the later positivity effect with pragmatics due to socialtabooness. Our findings suggest a view in which expressives are not merelyfunction words or emotional words. Rather, expressives are emotionally and sociallysignificant. Swears are more than what is indicated by valence ore arousal alone.

ARTICLE HISTORYReceived 15 January 2020Revised 12 April 2020Accepted 17 April 2020

KEYWORDSSwearing; ERP; emotion;language; taboo

Swear words are words such as damn, hell, and shit,and are often perceived as both negative and highlyarousing. How are swear words understood? Tra-ditionally, swears have been considered categoricallyas closed-class, function words, much like prepositions(Leech et al., 1982; Segalowitz & Lane, 2000). However,swear words differ from function words in that theygive rise to physical effects. For instance, swearingincreases tolerance for physical pain (Stephens et al.,2009), and elicits stronger skin conductance responsesthan neutral words (Bowers & Pleydell-Pearce, 2011).Recent theories consistent with a lack-of-contentview propose that swear words have no propositionalcontent themselves. Rather, they exert influence ontheir neighbouring content words (Gutzmann, 2019;Potts, 2005b, 2007). Further, this lack of propositionalcontent allows for great flexibility as to where theswear is interpreted with respect to the surroundinglexical context (Frazier et al., 2015, 2018). Anotherrational way to understand swear words is to treatthem as emotional words, as swear words are

oftentimes perceived as both negative and arousing.Following this line of reasoning, is the category ofswear words a special kind of negative word? In thisstudy, we propose that swears are neither just func-tion words nor emotional words and that there issome use-conditional, social content encoded inswear words. We use event-related potentials (ERPs)to investigate whether swear words have content dis-tinct from function words, and if so, whether thiscontent is emotional or social in nature.

According to linguistic theories that view swearwords as lacking propositional content, swearing isstudied as expressive content (Gutzmann, 2019; Potts,2007). Expressives are thus considered not-at-issuecontent, which is difficult to isolate from the utterancein which it occurs (Simons et al., 2010). For example, aninfelicitous follow-up to The damn dog is on the couchwould be something like, That’s not true, Fido is alovely dog, demonstrating the difficulty an interlocutorhas in trying to isolate the expressive content on itsown. Given that the expressive content of the swear

© 2020 Informa UK Limited, trading as Taylor & Francis Group

CONTACT Stanley A. Donahoo stanleydonahoo@email.arizona.eduSupplemental data for this article can be accessed https://doi.org/10.1080/02699931.2020.1764912

COGNITION AND EMOTION2020, VOL. 34, NO. 7, 1423–1438https://doi.org/10.1080/02699931.2020.1764912

word damn is not what is at-issue in an utterance, it hasbeen argued that the expressive content represents adimension of meaning which is separate from therest of the descriptive, truth-conditional content ofthe utterance (Gutzmann, 2013; Potts, 2005a, 2007;see also Anderbois et al., 2015; Barker et al., 2010). Evi-dence in support of an additional dimension ofmeaning are providedbyboth clinical andneurotypicalpopulations. Spontaneous usage of expressive contentis an infamous trait of Tourette syndrome (Robertson,2000; Van Lancker & Cummings, 1999). In healthypopulations, Dörre and colleagues (Dörre et al., 2018)examined not-at-issue content using German modalparticles and found processing differences comparedto their at-issue counterparts, suggesting differencesin meaning representations.

In emotion research, many view swear words asbeing negatively valenced and highly arousing (Reillyet al., 2020). Emotion language researchers typicallyassume a two-dimensional model, comprised ofvalence and arousal (Citron, 2012; Citron et al., 2013;Lang et al., 1998; Russell, 1980). Valence represents avalue measure where higher scores represent a wordthat could be considered good or positive, whereaslower values represent a word considered bad or nega-tive. Arousal is an excitement measure, where highlyarousing words produce feelings of stimulation or exci-tement and low arousal words produce feelings ofrelaxation or calmness (Russell & Barrett, 1999; Scottet al., 2018). In what follows we briefly review some ofthe behavioural and electrophysiological literature onemotional words relevant to swears.

Behavioural studies have found faster and moreaccurate recognition for words with any emotionalvalence, be it negative or positive, in comparisonwith neutral words (Citron et al., 2013; Kousta et al.,2009). This processing advantage is theorised to rep-resent the relevance of emotional stimuli in both survi-val generally and goal attainment specifically. Whilethe directionality of the effects of word valence aremixed, (Algom et al., 2004; Kanske & Kotz, 2007; Knick-erbocker et al., 2015; Kuperman et al., 2014; Scott et al.,2014; Wentura et al., 2000), overall, researchers havetaken a theoretical stance which favours a processingadvantage for negative words as a reflection of survivaloften coinciding with fleeing from negative situations(Fox et al., 2001; Knickerbocker et al., 2015). Generally,emotional stimuli, regardless of valence directionality,capture attention because they are motivationally sig-nificant (Lang et al., 1990, 1997). More recently, it hasbeen argued that prior studies on emotional language

processing conflated one or more linguistic variables(e.g. word frequency), and that, when all known vari-ables are accounted for, we are left with a U-shapemodel, where more extreme valence values acrossthe spectrum speed up lexical decision times (Kuper-man et al., 2014). Further, while arousal has beenshown to facilitate lexical decision times whenvalence is held constant (Hofmann et al., 2009),valence, rather than arousal, has been demonstratedto be a more robust predictor in modelling factors oflexical access (Bradley & Lang, 2002; Kousta et al.,2009; Kuperman et al., 2014; Vinson et al., 2014).Specifically, valence is a significant predictor of lexicaldecision response latencies, even when other factors,including arousal, are included in regression models(Kousta et al., 2009). Finally, valence, unlike arousal,appears to remain a modulating factor in tasksbeyond the word level (Bayer et al., 2010).

Electrophysiological studies, with their superior tem-poral resolution, informus that the emotional content ofaword is activated rapidly. Event Related Potential (ERP)studies of emotional words have identified at least twoneural correlates (Citron, 2012). One is the early pos-terior negativity (EPN), which peaks roughly between200 and 300 ms after stimulus onset, and is larger forboth positive and negative valenced words than forneutral words (Herbert et al., 2008). It has beensuggested that the EPN reflects automatic, implicit pro-cessing of emotion (Palazova et al., 2011). However,studieswhich reported an EPN tend to employ tasks dis-tinct from natural reading, like multiple repetitions ofsingle words (Kissler et al., 2007) or counting thenumber of times a word from a grammatical classoccurs (Kissler et al., 2009). A second neural correlaterobustly associated with emotional word processing isthe late positive component (LPC), which peaksroughly between 500 and 750 ms (Bayer et al., 2010,2012; Bayer & Schacht, 2014; Yao et al., 2016). The LPCis an index of a more sustained, explicit, processing ofemotional content; or, put another way, the LPCindexes higher-order stimulus evaluation (Schacht &Sommer, 2009a). Negative words have been shown toincrease the amplitude of the LPC (Kanske & Kotz,2007; L. Wang & Bastiaansen, 2014). Collapsing overpositive and negative words, emotional words eliciteda larger LPC relative toneutralwords, afinding inparallelwith behavioural studies, where valence tends to facili-tate lexical access (Citron et al., 2013; Kousta et al., 2009).In general, more valenced words produce larger ERPamplitudes, which is associated with emotional proces-sing (Citron et al., 2013; Kanske & Kotz, 2007).

1424 S. A. DONAHOO AND V. T. LAI

If the not-at-issue content in swear words is primar-ily emotional, we would expect to find the ERP corre-lates discussed in the above-mentioned literature.However, here we suggest that the two dimensionsof valence and arousal may not fully capture theuse-conditional content of swear words. In additionto emotionality, there are at least three possibilities;the content is: (1) a pragmatic speech act, (2) taboon-ess, (3) a socially threatening function. Frazier et al.(2015, 2018) argue for pragmatically treating expres-sives as independent speech acts, independent ofthe speech act conveyed by the rest of the utterance.As a result, swears thus have flexibility as to wherethey are applied to the rest of the at-issue content.Frazier and colleagues found evidence for this claimusing offline sentence ratings. Participants read sen-tences such as The policeman gave me a damn ticketand rated whether the speaker felt negative towardsthe ticket (object reading), the policeman (subjectreading), or the entire situation (situation reading)itself. Even when the expressive was in an object pos-ition, participants still preferred the situation responsenearly 42% of the time. Similarly, damn presented insubject position yielded a situation interpretation in39% of ratings and damn in situation position (e.g.Damn. The policeman gave me a ticket.) yielded a situ-ation interpretation in only 63% of ratings, empiricallydemonstrating the flexibility of not-at-issue swears.

Tabooness is another potential factor that maydrive swear word recognition. Taboo language gen-erally produces slower reaction times across abroad variety of tasks, including modified Stroop(Eilola & Havelka, 2011; Guillet & Arndt, 2009;Mackay et al., 2004; Siegrist, 1995), attentional blink(Mathewson et al., 2008), and picture-word interfer-ence (Dhooge & Hartsuiker, 2011). This slowdown isconversely associated with the taboo-superiorityeffect: the phenomenon that memory is better fortaboo words over neutral words in a variety of exper-imental paradigms, for example in spontaneousrecall tasks (Hadley & MacKay, 2006; Mathewsonet al., 2008). Better recall has also been demon-strated for taboo words in word lists (Buchananet al., 2006; Grosser & Walsh, 1966), and in repetitionpriming (Thomas & LaBar, 2005). Madan et al. (2017)reported that factors other than valence, such asword frequency and tabooness, are the better pre-dictors of lexical decision latencies. It should be men-tioned however, that stimuli in Madan et al. (2017),while including words that could be consideredoffensive (e.g. skank), do not directly feature any

words that are commonly considered swears (e.g.damn, shit, hell).

Lastly, social threat has been associated with swearwords. One study explored the idea that social threatand physical threat are distinct (Wabnitz et al., 2012).Participantspassively read socially threateningnegativewords (swear words), physically threatening negativewords (e.g. bomb, explosion, wound), neutral, and posi-tive words, while monitoring for a dot presented in15% of trials. The socially threatening negative words(swear words) elicited an enhanced P1 (100–140 ms)relative to physically threatening negative words andpositive words. The authors interpreted the P1 as areflection of a fast-forward mechanism for evolutiona-rily relevant cues. In their so-called N400 time window(450–580 ms), negative and positive words elicited alarger negativity effect relative to neutral words, withno difference between the positive and negativewords. The authors interpreted the N400-like effect asevidence that swear words receive no special semanticconsideration relative to other emotionally valencedwords, at least not in a typical population. In peoplewith social anxiety disorder, swears did entertainsemantic consideration in the N400 time window(Wabnitz et al., 2016). A related study (Klein et al.,2019) reported LPC differences for socially threateningand physically threatening words in adolescents withPost-traumatic stress disorder (PTSD), but not inhealthy controls.

The present study uses ERPs to investigate thenature of the content, if any, in swear words. Partici-pants engaged in a lexical decision task while EEGwas simultaneously recorded on the scalp. We com-pared the response times and ERPs for swear words,negatively valenced but non-swear words, neutralopen-class words, and neutral closed-class functionwords. If swears are not as content heavy in theirbeing not-at-issue, they should behave similarly tofunction words. If swears have content and if thecontent is emotional in nature, then swears shouldbehave more like negative words. If swears havecontent and if the content is something more thanemotional in nature, then swears should be distinctfrom negative words.

1. Material and methods

1.1. Participants

Thirty-two right-handed native English speakers (15female; age range 18–23) participated in the

COGNITION AND EMOTION 1425

experiment for course or extra credit. According topower analyses using G*Power software (Faul et al.,2009), this was a sufficient sample size using a conven-tional threshold of α = .05 for Type I error, a power of.80, and a medium effect size (partial η2 = .06). All par-ticipants had normal or corrected-to-normal vision,and had no known language or neurological disorder.All gave informed consent before participation. Forinstitutional review board purposes, participantswere verbally informed that stimuli presented duringthe experiment included words that could be con-sidered offensive. But, in order to avoid any sort ofpriming effects, no examples of such words were pro-vided before the experiment began.

1.2. Stimuli

The stimuli consisted of 28 swear words (e.g. shit,damn), 28 negatively valenced but non-swear words(kill, sick), 28 open-class neutral words (e.g. wood,lend), 28 closed-class neutral words (e.g. while,whom), and 112 pseudowords. The number of wordsis largely constrained by the number of words con-sidered swears. The words were generated bynative-speaking research assistants with reference tothe literature on swears (Jay, 1992; Wang et al., 2014)and closed-class words (Craig & Kinney, 2009).

Twoweb-based norming surveys verified themanip-ulations of valence and tabooness. Thirty-nine native-speaking participants rated the words one word at atime on a 9-point Likert scale. In the valence norming(“how positive or negative a word is”), 1 was negativeand 9 was positive. In the norming for tabooness(“how offensive the word is to people in general”), 1was not at all offensive and 9 was very offensive.Arousal values were taken from existing corpus work(Warriner et al., 2013), supplemented by the web-based norming survey for the closed-class words.Basedon the results of thenorming survey, one stimulusitem initially coded as a swearword, christ, was removedfrom the analysis, as participants consistently rated theword as relatively positive (6.19) and not taboo (4.75).Table 1 summarises the word properties. Both swearsand negative words were rated more negative thanneutral words, (all t values > 12, all p values < .001).Swears and negative words were similarly negative(n.s.), and similarly arousing (n.s.). Swears were ratedmore taboo than negative words (t(43) = 6.74 p < .001).In addition, the conditions were matched in terms oflog frequency (Brysbaert & New, 2009), word length,and number of morphemes, phonemes, and syllables

(Balota et al., 2007). There was no significant differencebetween these conditions.

The pseudowords were constructed using the ARCPseudoword Database (Rastle et al., 2002), selectingonly strings with orthographically existing onsets,and using an average range of letter length (min: 4,max: 6), as well as orthographic (min: 4, max: 6) andphonological (min: 12, max: 14) neighbourhood sizesthat matched the target words in the task.

1.3. Procedure

Preparation for the EEG recording lasted 30 min, inwhich participants were measured and fitted withthe electrode cap (ActiCap, Brain Products, GmbH).During the setup, participants provided generallanguage and demographic information and werebriefed as to the nature of the EEG technique. Partici-pants then entered a dimly lit and sound attenuatedroom. They were seated in a comfortable chair at adesk which featured a computer screen 70–80 cmaway from their eyes. The stimuli were presented visu-ally in white font (Font: Lucida Console; Font size: 20)against a black background via the E-prime 3.0 soft-ware (Psychology Software Tools, Inc.).

Figure 1 shows an example of a trial. In each trial, acentral fixation cross was presented for 500 ms, fol-lowed by a blank screen with a random duration of500–1500 ms. The stimulus (either a legal letterstring or illegal letter string) then appeared for a dur-ation determined by the number of letters in thestring, ranging from 248 ms to 400 ms (37 ms/letter+ 100 ms, max duration of 400 ms). Participants wereinstructed to decide whether the letter string was aword or not as quickly and as accurately as possibleby pressing one of two buttons on a response pad.The order of the button press was counterbalanced

Table 1. Mean ratings/values with standard deviations in parenthesesof word properties for each condition.

Wordproperties Swear Negative

Open classneutral

Closedclass

Valence 2.46 (.92) 2.04 (.65) 5.9 (1.01) 4.95 (.90)Arousal 5.57 (.94) 5.44 (.78) 3.81 (.81) 3.43 (.69)Tabooness 6.82 (.84) 5.03 (.90) 2.95 (.37) 3.21 (.31)Log subtlexword freq

3.31 (.68) 3.58 (.4) 3.25 (.52) 3.08 (.57)

Letter length 4.85 (1.32) 4.85 (.80) 5 (.85) 4.93 (.98)Nummorphemes

1.27 (.44) 1.22 (.42) 1.11 (.31) 1.14 (.34)

Numphonemes

3.81(1.04) 4.11 (.87) 4.04 (.78) 3.83 (1.15)

Num syllables 1.27 (.44) 1.52 (.5) 1.5 (.5) 1.48 (.5)

1426 S. A. DONAHOO AND V. T. LAI

with participant number. A trial timed out after2000 ms, but this duration proved adequate togarner a participant response.

There were 6 practice trials to familiarize partici-pants with the task. Participants were advised torefrain from moving or blinking during the trials, butencouraged to rest their eyes during break intervals,occurring after every 8 trials. Each item was presentedonce, in one block with a randomised presentationorder. We decided against repeating items, as thereis evidence that emotional words in particular are sus-ceptible to repetition priming effects, both immediateand delayed (D. Zhang et al., 2019).

1.4. EEG acquisition and pre-processing

Continuous EEG was recorded from 64 surface activeelectrodes using the actiCHamp system (Brain Pro-ducts, GmbH). Data were measured with respect to avertex reference, with the sampling rate of 500 Hz.Electrode impedances were below 10 kΩ throughout.

Data was analysed with Brain Vision Analyzer 2.0.Data was low-pass filtered at 30 Hz with a notch filterat 60 Hz to filter out line noise. Blinks were detectedand corrected offline via ocular artefact correction inde-pendent component analysis (ICA) using the InfomaxRestricted ICA algorithm. Artefacts were detectedsemi-automatically with Brain Vision. Trials contami-nated with artefacts due to body movements or peakdeflection exceeding ±100 µV were rejected. Individualbad channels were replaced through interpolation ofthe average of the surrounding channels (Brain VisionUser Manual) on a participant-by-participant basis.Remaining artefacts were reduced by individual

channel-interpolation or the epochs were rejected.1.5% of the channels were replaced across the exper-iment. Data were re-referenced to the average of theleft and right mastoids, and segmented from 200 msbefore the stimulus onset to 750 ms after, with baselinecorrection from −200 ms to 0 ms preceding the onsetof the target letter string. Finally, the ERP data wereaveraged across participants for each condition.

2. Results

2.1. Behavioural

Mean accuracy and reaction times results for trials withcorrect responses are displayed in Figures 2 and 3. Sep-arate Repeated-Measures ANOVAs were conducted tocompare the effect of word condition (Swear, Negative,Neutral, Closed, Pseudoword) on accuracy and on reac-tion times, respectively. Greenhouse-Geisser correctedp values are reported when appropriate for all testsinvolving factors with more than two levels. In addition,False Discovery Rate corrections were applied (Benja-mini & Hochberg, 1995). In terms of accuracy, therewas a significant main effect of condition, F(4,112) =20.38, p < .001, h2

G = .259. Crucially, there were no differ-ences for swear and neutral, F(1,28) = 1.70, p = .303;negative and neutral, F(1,28) = .15, p = .697; nor swearand negative, F(1,28) = 1.29, p = .319. There were signifi-cant differences between pseudoword and neutral, F(1,28) = 21.83, p < .002, h2

G = .199; closed and neutral, F(1,28) = 53.65, p < .002, h2

G = .312; and closed andswear, F(1,28) = 26.48, p < .002, h2

G = .236.In terms of reaction time, therewas a significantmain

effect of condition, F(4,112) = 47.58, p < .001, h2G = .139.

Figure 1. An example of a trial of the lexical decision task.

COGNITION AND EMOTION 1427

Figure 2. Mean accuracy and standard deviation for all conditions.

Figure 3. Mean reaction times and standard deviation for all conditions.

1428 S. A. DONAHOO AND V. T. LAI

Pairwise comparisons revealed that there were signifi-cant differences between pseudoword and neutral,F(1,28) = 132.59, p < .001, h2

G = .198; swear and neutral,F(1,28) = 18.03, p < .001, h2

G = .030; swear and negative,F(1,28) = 18.5, p < .001, h2

G = .056; closed and neutral,F(1,28) = 30.68, p < .001, h2

G = .054. Negative comparedto neutral was not significant, F(1,28) = 1.69, p = .204.Closed compared to swear was not significant, F(1,28)= 1.90, p = .204.

2.2. ERP

The grand averaged ERPs are displayed in Figure 4,and the scalp distributions of the effects, in Figure 5.Average trial counts for each condition were swear:26.78; negative: 25.75; neutral: 25.97; closed: 26.38;

pseudoword: 101.53. Descriptively, the overall ampli-tude for swear was greater than the other conditionsat early (250–550) and later (550–750) time windows:swear, M = 2.74 µV; negative, M = 2.12 µV; neutral, M= 1.17 µV; closed, M = 1.33 µV; pseudoword, M = -.55 µV. 550–750: swear, M = 5.64 µV; negative, M =3.41 µV; neutral, M = 2.85 µV; closed, M = 3.87 µV;pseudoword, M = 1.95 µV.

Visual inspection indicated that all waveformsshowed early components of N1 and P2, whichreflect visual processing and reassure data quality tosome extent. The conditions diverged as early as∼250 ms. Negative words elicited a larger positivitythan neutral words from 250 ms to 550 ms, centrallyand parietally distributed. Swears also elicited alarger positivity than neutral content words from

Figure 4. Grand averaged ERP waveforms for Swear (Red), Negative (Blue), Neutral (Black), Closed (Green), Pseudoword (Pink) conditions at thefrontal, central, and posterior locations.

COGNITION AND EMOTION 1429

250 ms to 550 ms, widespread across the scalp, butshowed more positive amplitudes than negativewords from 550 ms to 750 ms. Pseudowords elicited

a larger negativity than neutral open-class words start-ing at 250 ms, identified as N400. There was no EPN(See supplementary materials).

Figure 5. Scalp topographies at for earlier (250–550) (top panel) and LPC (550–750) (bottom panel) time windows. All voltage differences aresignificant unless otherwise noted.

1430 S. A. DONAHOO AND V. T. LAI

Mean amplitudes were exported from two timewindows of 250–550 ms and 550–750 ms at thefrontal group (Afz, F1, F2, Fz), central group (FCz, C1,C2, Cz), and the posterior group (CPz, P1, P2, Pz)along the midline. The omnibus Repeated-MeasuresANOVA model of 5 condition (Swear, Negative,Neutral, Closed, Pseudoword) × 3 location (Frontal,Central, Posterior) × 2 time (Early, Late) showed aninteraction of condition × time × location, F(8,248) =5.19, p < .001, h2

G = .001, and an interaction of con-dition × time F(4,124) = 5.70, p < .001, h2

G = .004.These interactions suggest that the patterns ofresults in the two time windows differed. Thus separ-ate Repeated-Measures ANOVAs of condition ×location were conducted within each time window.Greenhouse-Geisser corrected p values are reportedwhen appropriate for all tests involving factors withmore than two levels. False Discovery Rate correctionswere also applied.

2.2.1. 250–550 msThere was a condition × location interaction, F(8,248)= 3.36, p = .010, h2

G = .002, a main effect of condition,

F(4,124) = 14.41, p < .001, h2G = .067, and a main effect

of location, F(2,62) = 40.48, p < .001, h2G = .200. Pseudo-

word elicited a larger and widespread N400than neutral words (Frontal: F(1,31) = 18.40, p < .004,h2G = .060; Central: F(1,31) = 23.08, p < .004, h2

G = .086;Posterior: F(1,31) = 9.57, p = .013, h2

G = .028), consistentwith past ERP findings for pseudowords (Palazovaet al., 2011; Schacht & Sommer, 2009a). Figure 6 dis-plays relevant ERPs and scalp topographies for thesetwo conditions.

Comparing negative and neutral words, negativewords elicited a larger positivity than neutralwords, centrally and posteriorly distributed (Frontal:F(1,31) = 3.43, p = .074, h2

G = .009; Central: F(1,31) =5.35, p = .033, h2

G = .018; Posterior: F(1,31) = 7.66,p = .019, h2

G = .020).Comparing swear and neutral words, swears eli-

cited a larger positivity than neutral words, at alllocations (Frontal: F(1,31) = 9.04, p = .013, h2

G = .037;Central: F(1,31) = 5.89, p = .027, h2

G = .030; Posterior: F(1,31) = 6.78, p = .020, h2

G = .025).Comparing swear and negative words, the two

behaved similarly in this time frame. That is, swears

Figure 6. Grand averaged ERP waveforms and scalp topographies for pseudowords and open-class neutral words.

COGNITION AND EMOTION 1431

elicited a larger positivity than negative words in thefrontal group only (Frontal: F(1,31) = 4.28, p = .050,h2G = .011; Central: F < 1; Posterior: F < 1).Closed-class words behaved similarly to neutral

(content) words in this window (all Fs < 1). Butclosed-class words behaved differently compared toswears, with swears eliciting a larger positivity(Frontal: F(1,31) = 12.46, p = .004, h2

G = .032; Central:F(1,31) = 7.31, p = .019, h2

G = .020; Posterior: F(1,31) =7.1, p = .019, h2

G = .014).

2.2.2. 550–750 msThere was a main effect of condition, F(4,124) = 11.41,p < .001, h2

G = .053, but condition did not interact withlocation, F < 1. Thus in the following analyses we com-bined all electrode locations. Pseudoword elicited amarginally larger negativity than neutral words(F(1,31) = 4.17, p = .07, h2

G = .013), likely a continuationof the N400.

Swears elicited a larger positivity than neutralwords, F(1,31) = 17.54, p < .003, h2

G = .078. Swear alsoelicited a larger positivity than negative words,F(1,31) = 12.75, p = .003, h2

G = .049. Negative andneutral words do not differ here (F(1,31) = 1.74,p = .197).

Closed-class words behaved similarly to neutral(content) words in this time window, F(1,31) = 2.76,p = .126. But closed-class words behaved differentlycompared to swear words, with swear words elicitinga larger positivity than the closed (F(1,31) = 9.96,p = .008, h2

G = .026).A potential confounding factor is task difficulty,

where the judgement of swears might be more orless difficult than negative words. We verified thatthis is not the case by conducting an ANCOVAusing mean reaction times for hits in the respectiveconditions as a within-subject covariate. TheANCOVA indicated that in the late positivitywindow (550–750 ms), the effect of swear wordsstill differs from the effect of negative words (F(1,62) = 7.97, p = .006). This suggests that aftertaking task difficulty (approximated by reactiontimes) into consideration, swear words are processeddifferently from the negative words in the late timewindow. In the early positivity window (250–550 ms), the effect of swear words does not differfrom the effect of negative words (F(1,62) = .817,p = .369), consistent with the analysis withouttaking task difficulty into consideration.

3. Discussion

The present study investigated whether swear wordshave an additional dimension of meaning, and if so,whether such meaning is functional, emotional, orsocial in nature. Participants read swear words, nega-tively valenced but non-swear words, neutral open-class words, neutral closed-class words, and pseudo-words. They made word/non-word decisions whiletheir EEG was recorded. The behavioural resultsshowed that, while participants were equally as accu-rate to respond to swears, negative, and neutralwords, swear words were recognised more slowlythan neutral open-class words, and neutral open-class words were recognised more slowly than nega-tive words. Swears were recognised as quickly as theclosed-class neutral words. For ERPs, both swearsand negative words elicited larger positivity effectsin the 250–550 ms time window relative to neutralwords. In the 550–750 ms time window, the positivityeffect in swears continued but the positivity effect inthe negative words condition converged with theneutral. Swears were also distinct from closed-classwords in both time windows, in that swears elicitedlarger positivity effects than closed-class words.

3.1. Behavioural results

That swear words were recognised more slowly thanneutral open-class words could be interpreted as ataboo superiority effect, and that neutral open-classwords were recognised more slowly than negativewords could be viewed as an example of well-attestednegative bias (Ito et al., 1998; L. Wang & Bastiaansen,2014). Recent behavioural work in speech productionhas argued that taboo effects are pre-lexical (Hansenet al., 2017). However, our results suggest that tabooeffects emerge later, around 250 ms – at whichlexical access already occurred (Dien, 2009; Kim &Lai, 2012). Looking at the issue of when emotionalmeaning affects word processing specifically, Palazovaand colleagues used lexical decision simultaneouslywith EEG to conclude that emotionality has a post-lexical focus (Palazova et al., 2011). While there werebehavioural differences for open-class and closed-class neutral words, with the closed-class wordsbeing identified more slowly and less accurately,there were no ERP differences in the two, suggestingthat the reaction time differences are attributable tooverall familiarity with the two conditions.

1432 S. A. DONAHOO AND V. T. LAI

In our opinion, ERPs and behaviours are measuringdifferent aspects and timings of the same process, andtherefore behaviours might not always explain ERPs.Considering reaction times and ERPs for swears andnegative words together, a relatively consistentpicture emerges: The negative valence (present inboth swears and negative words), or rather the atten-tion allocation to negative valence is processed rela-tively earlier, whereas the social taboo aspect(present in swears and not in negative words) is pro-cessed later, after 550 ms.

3.2. Swears and early effects

Our results indicate that swears in isolation were per-ceived similarly to negative words initially until550 ms, and then continued to be processed fortheir pragmatic content. Both swears and negativewords elicited larger ERP positivity effects relative toneutral words in the 250–550 ms window. We areconfident in associating the positivity effect herewith negative valence, based on past literature (Hino-josa et al., 2014; Kanske & Kotz, 2007; L. Wang & Bas-tiaansen, 2014; Zhang et al., 2018; but see Bayer &Schacht, 2014). Note that compared to those studies,our positivity effect time window 250–550 ms seemsearlier, e.g. 400–650 ms in Hinojosa et al. (2014). Weincluded an earlier portion because we consideredeffects in all conditions, including the pseudowordeffect which started at 250 ms. Such inclusion math-ematically counteracted statistical significance, yetwe still found significance, pointing to the robustnessof the positivity effect. An additional post-hoc analysisusing the 400–600 ms window verified the signifi-cance of the larger positivity effect for negative thanfor neutral words (F(1,31) = 7.88, p = .009, h2

G = .017).We attribute the positivity effect in the 250–550 ms

window to attention allocation (Hinojosa et al., 2014).According to Global Resource Theory (Mackay et al.,2004), attentional capacities are finite. Emotionalwords demand more attentional resources and incura processing disadvantage, which possibly accountsfor the delay we observed in lexical decisions timesfor participants responding to the swear condition. Arelated approach, automatic vigilance (Erdelyi, 1974;Pratto & John, 1991; Wentura et al., 2000), suggeststhat negative stimuli engage attention longer thanneutral stimuli because humans preferentially attendto negative stimuli, though it remains unclear atwhat stage of processing the effect arises (Kupermanet al., 2014). Task type and relevance also appear to

play a role in the processing of negative comparedto neutral stimuli (Trauer et al., 2012, 2015).

We did not find any sort of EPN for the swear wordsor the negative valence words, replicating some pre-vious work (M. Zhang et al., 2018; M. Zhang & Guo,2014). Those studies reporting an EPN typicallyfound it for positive words or pictures (Hinojosaet al., 2009; Schacht & Sommer, 2009b), and swearstend to be used in negative contexts. Future workshould examine whether swears used in a positivecontext (e.g. it is fucking awesome) would give rise toan EPN.

3.3. Swears and the late positivity

Our primary finding is that there is more to swearingthan negative affect, as reflected by the late positivityeffect in the 550–750 ms window which was moreenhanced for swears than for negative words. Thereare a few possible interpretations for this. The first,and perhaps most parsimonious, interpretation isthat the late positivity was merely a continuation ofthe early positivity effect (or a more extreme versionof the effects). The effect was prolonged in swearsbecause some function in swears acted to amplifytheir emotional properties and made the perceivedemotionality more robust than what those propertiesactually were. This interpretation is partially consistentwith the not-at-issue view of swears. We say only par-tially because the not-at-issue view suggests thatswears impact words surrounding them, and here asa single word study, there were no words neighbour-ing the swears.

A more likely interpretation is that the late positiv-ity effect is a real effect in its own right. Under thisscenario, swears contribute their own dimension ofmeaning and the late positivity could reflect variablesinherent to swears, such as: (1) a pragmatic speech act,(2) social threat, (3) tabooness. Ultimately, we con-clude that the tabooness element unique to swearsis the best candidate, but we take each of these inturn.

First, the late positivity could reflect an extra stepassociated with processing a speech act. This wouldsupport the view in which expressives representtheir own distinct speech act (Frazier et al., 2015,2018). Theoretically, there are several factors to con-sider regarding any one speech act (Bach, 1998).These extra steps come with processing costs, someof which have been borne out experimentally forboth direct and indirect speech acts (Coulson &

COGNITION AND EMOTION 1433

Lovett, 2010; Gisladottir et al., 2015; Gisladottir et al.,2018). It is difficult to compare these ERP findings formulti-word speech acts to our lexical decision task.Gísladóttir and colleagues reported a late negativityfor speech acts in which an interlocuter makes anindirect offer (by saying I have a credit card as aresponse to I don’t have any money to pay for theticket) as opposed to a direct answer (I have a creditcard as a response to How are you going to pay forthe ticket?). Coulson and Lovett (2010) reported LPCeffects for differences in literal statements and indirectrequests (My soup is too cold to eat, said to either ahusband or a restaurant waiter), with a larger LPC forindirect requests. However, this positivity onlyemerged for words later in the request, much laterthan we see for words presented in isolation. Still,negative and neutral words alone in isolation do notrepresent a speech act in the way that a swear wordmight; for example, uttering shit after spilling coffeeon a computer desk can invoke a hearer to movethe keyboard. However, it remains an open questionwhether a swear word presented in a lexical decisiontask can be construed as conveying communicativemeaning in the way that a speech act typically does.

Second, the late positivity could reflect the proces-sing of social threat, an idea postulated for differencesobserved for taboo words in the two extant ERPstudies directly involving swearing (Wabnitz et al.,2012, 2016). Wabnitz and colleagues argued thatswear words receive no special semantic consider-ation relative to other emotionally valenced wordswhile our findings suggest otherwise. One likely expla-nation for the discrepancy is task difference. Wabnitzand colleagues relied on passive reading while ourstudy required that participants made an activedecision about each letter string. Such task differencesare known to affect lexical processing for emotionalwords (Kissler et al., 2009). Additionally, Wabnitz andcolleagues employed a block design; participantssaw the words in each condition six times. Such rep-etition can significantly alter ERPs (Van Petten et al.,1991). Finally, there is a potential confound in thedesign of both of these studies regarding the labelof swear word towards the stimuli used. BecauseWabnitz and colleagues were interested in socialthreat, the swear stimuli used were person-oriented,thus functioning more like an insult than a swearword specifically. Categorically, there is a differencein fuck or fucking compared to motherfucker (Blake-more, 2015). The former are swears, whereas thelatter is also an insult. All of the stimuli used by

Wabnitz and colleagues are person-oriented and inter-preted as a noun. It remains an open question as towhether these related but distinct conceptual cat-egories are processed and represented in similarfashion. In this vein, Klein et al. (2019) used the samestimuli as Wabnitz and colleagues and found no differ-ence in insults and physically threatening words inhealthy adolescents, whereas they did find differencesfor adolescents with PTSD. In general, all of thesewords bear some social offense, but there are linguis-tic distinctions. It is clear however, that expressives doreceive special linguistic consideration relative toother words that are equally as emotionally valenced.

Third, the late positivity reflects the processing ofan additional dimension, tabooness. That is, eventhough damn and death are both negatively valenced,only damn has the additional use-conditional taboodimension. Swearing is not merely the sum ofextreme values of valence and arousal. Instead, thesewords offer their own element of affect, namelytabooness, and contribute their own dimension ofmeaning, both of which are distinct from otherwords that are just as extreme in their valencevalues. This interpretation is consistent with recentexperimental work were tabooness demands moreattention in sentential contexts (Christianson et al.,2017). In our lexical decision task, relative to bothnegative and neutral words, swears elicited longerreaction times, a finding consistent with prior studiesinvolving a variety of tasks (Dhooge & Hartsuiker,2011; Eilola & Havelka, 2011; Guillet & Arndt, 2009;Mathewson et al., 2008). Regarding our ERP findings,it is a challenge to situate the positivity effect forswears relative to negative and neutral words, giventhat so few studies have investigated these specificaspects of emotional language.

The limitations of this study are as follows: (1) Thetrial counts are comparatively lower (∼26 each con-dition), which is due to the fact that there are notthat many swear words to begin with and that wehad to reject trials contaminated by artefacts. Otherstudies have used comparable trial counts: 20 in Hino-josa et al. (2009), 32 in Kanske et al. (2011), and 20 inYao et al. (2016). Note that these studies repeatedthese items several times in blocks, however. Weopted against using repetition in our design, giventhe evidence that emotional words in particular aresusceptible to immediate and delayed repetitionpriming effects (D. Zhang et al., 2019). In addition,our reported power analysis indicates sufficientsample size even with these lower item counts.

1434 S. A. DONAHOO AND V. T. LAI

Future studies should employ more trials. (2) Wematched multiple word properties (Table 1), but wedid not match part of speech, concreteness, or famili-arity. What part of speech a swear word is cannoteasily be determined. The majority of swear wordscan be nouns (What is this shit?), verbs (He shithimself.), adjectives (That was a shit film), or virtuallyany other part of speech. In this study, we comparedswears with closed-class function words to test theclaim in the literature that swears act as closed-classwords. In some sense, it is part of the research ques-tion, as opposed to being a variable that must becontrolled.

Concreteness of swears is also hard to pin downand likely context dependent. For example, in Thissong is my shit, shit is very abstract, but in My dogtook a shit, shit is concrete and reflects literal faecalmatter. In Kanske and Kotz (2007), concreteness inter-acted with emotion in that only concrete, negativewords (e.g. bomb) showed the enhanced LPC, butnot abstract negative words (e.g. violence).

With respect to familiarity, we matched the wordfrequency, which according to some (Tanaka-Ishii &Terada, 2011), can be viewed as an approximant offamiliarity. This is especially so given that our stimulihave a relatively high frequency, which typically alsohave high familiarity. The familiarity effect mattersmost for low frequency items. Finally, in our data, par-ticipants responded with over 90% accuracy to thecritical comparison conditions swear, negative, andneutral, suggesting that participants were familiarwith these stimuli. The reduced accuracy for theclosed-class words condition might possibly suggestthat familiarity played a role here, but these wordswere just as frequent as the other conditions,suggesting that participants have encountered theseitems on a regular basis.

4. Conclusion

The current study found that swear words havecontent distinct from function words and from nega-tive words, and that this content is emotional andsocial. Specifically, swears and negative words eli-cited a larger positivity (250–550 ms) than open-class neutral words, associated with attention dueto negative valence. Later on, swears elicited alarger late positivity (550–750 ms) than negativewords, associated with social tabooness. Ourfindings suggest that a two-dimensional model ofemotional language processing requires expansion

in order to fully capture all emotional languageincluding swears. With these results, we contributea new data set for probing our understanding ofone of the central properties of language: conveyingemotion.

Disclosure statement

No potential conflict of interest was reported by the author(s).

ORCID

Stanley A. Donahoo http://orcid.org/0000-0002-8235-4296

References

Algom, D., Chajut, E., & Lev, S. (2004). A rational look at theemotional Stroop phenomenon: A generic slowdown, not aStroop effect. Journal of Experimental Psychology: General,133(3), 323–338. https://doi.org/10.1037/0096-3445.133.3.323

Anderbois, S., Brasoveanu, A., & Henderson, R. (2015). At-issueproposals and appositive impositions in discourse. Journal ofSemantics, 32(1), 93–138. https://doi.org/10.1093/jos/fft014

Bach, K. (1998). Speech acts. In Routledge encylopedia of philos-ophy. Taylor and Francis. https://www.rep.routledge.com/articles/thematic/speech-acts/v-1

Balota, D. A., Yap, M. J., Hutchison, K. A., Cortese, M. J., Kessler, B.,Loftis, B., Neely, J. H., Nelson, D. L., Simpson, G. B., & Treiman, R.(2007). The English lexicon project. Behavior ResearchMethods, 39(3), 445–459. https://doi.org/10.3758/BF03193014

Barker, C., Bernardi, R., & Shan, C. (2010). Principles of interdimen-sional meaning interaction. Semantics and Linguistic Theory,20, 109–127. https://doi.org/10.3765/salt.v20i0.2569

Bayer, M., & Schacht, A. (2014). Event-related brain responses toemotional words, pictures, and faces—A cross-domain com-parison. Frontiers in Psychology, 5. https://doi.org/10.3389/fpsyg.2014.01106

Bayer, M., Sommer, W., & Schacht, A. (2010). Reading emotionalwords within sentences: The impact of arousal and valenceon event-related potentials. International Journal ofPsychophysiology, 78(3), 299–307. https://doi.org/10.1016/j.ijpsycho.2010.09.004

Bayer, M., Sommer, W., & Schacht, A. (2012). P1 and beyond:Functional separation of multiple emotion effects in word rec-ognition. Psychophysiology, 49(7), 959–969. https://doi.org/10.1111/j.1469-8986.2012.01381.x

Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discov-ery rate: A practical and powerful approach to multipletesting. Journal of the Royal Statistical Society: Series B(Methodological), 57(1), 289–300. https://doi.org/10.1111/j.2517-6161.1995.tb02031.x

Blakemore, D. (2015). Slurs and expletives: A case against ageneral account of expressive meaning. Language Sciences,52, 22–35. https://doi.org/10.1016/j.langsci.2014.06.018

Bowers, J. S., & Pleydell-Pearce, C. W. (2011). Swearing, euphe-misms, and linguistic relativity. PLoS ONE, 6(7), e22341.https://doi.org/10.1371/journal.pone.0022341

COGNITION AND EMOTION 1435

Bradley, M. M., & Lang, P. J. (2002). Measuring emotion: Behavior,feeling, and physiology. In R. D. Lane & L. Nadel (Eds.),Cognitive neuroscience of emotion (pp. 242–276). OxfordUniversity Press. https://arizona-primo.hosted.exlibrisgroup.com/permalink/f/1h28lag/TN_pq_ebook_centralEBC271199.

Brysbaert, M., & New, B. (2009). Moving beyond Kučera andFrancis: A critical evaluation of current word frequency normsand the introduction of a new and improved word frequencymeasure for American English. Behavior Research Methods, 41(4), 977–990. https://doi.org/10.3758/BRM.41.4.977

Buchanan, T. W., Etzel, J. A., Adolphs, R., & Tranel, D. (2006). Theinfluence of autonomic arousal and semantic relatedness onmemory for emotional words. International Journal ofPsychophysiology, 61(1), 26–33. https://doi.org/10.1016/j.ijpsycho.2005.10.022

Christianson, K., Zhou, P., Palmer, C., & Raizen, A. (2017). Effects ofcontext and individual differences on the processing of taboowords. Acta Psychologica, 178, 73–86. https://doi.org/10.1016/j.actpsy.2017.05.012

Citron, F. M. M. (2012). Neural correlates of written emotionword processing: A review of recent electrophysiologicaland hemodynamic neuroimaging studies. Brain andLanguage, 122(3), 211–226. https://doi.org/10.1016/j.bandl.2011.12.007

Citron, F. M. M., Weekes, B. S., & Ferstl, E. C. (2013). Effects ofvalence and arousal on written word recognition: Timecourse and ERP correlates. Neuroscience Letters, 533, 90–95.https://doi.org/10.1016/j.neulet.2012.10.054

Coulson, S., & Lovett, C. (2010). Comprehension of non-conven-tional indirect requests: An event-related brain potentialstudy. Italian Journal of Linguistics, 22(1), 107–124.

Craig, D. H., & Kinney, A. F. (2009). Shakespeare, computers, andthe mystery of authorship. Cambridge University Press.

Dhooge, E., & Hartsuiker, R. J. (2011). How do speakers resist dis-traction?: Evidence from a taboo picture-word interferencetask. Psychological Science, 22(7), 855–859. https://doi.org/10.1177/0956797611410984

Dien, J. (2009). The neurocognitive basis of reading single wordsas seen through early latency ERPs: A model of convergingpathways. Biological Psychology, 80(1), 10–22. https://doi.org/10.1016/j.biopsycho.2008.04.013

Dörre, L., Czypionka, A., Trotzke, A., & Bayer, J. (2018). The proces-sing of German modal particles and their counterparts.Linguistische Berichte, 255, 313–346. http://nbn-resolving.de/urn:nbn:de:bsz:352-2-8iyf1z1elfx0.

Eilola, T. M., & Havelka, J. (2011). Behavioural and physiologicalresponses to the emotional and taboo Stroop tasks innative and non-native speakers of English. InternationalJournal of Bilingualism, 15(3), 353–369. https://doi.org/10.1177/1367006910379263

Erdelyi, M. H. (1974). A new look at the new look: Perceptualdefense and vigilance. Psychological Review, 81(1), 1–25.https://doi.org/10.1037/h0035852

Faul, F., Erdfelder, E., Buchner, A., & Lang, A.-G. (2009). Statisticalpower analyses using G*Power 3.1: Tests for correlation andregression analyses. Behavior Research Methods, 41(4), 1149–1160. https://doi.org/10.3758/BRM.41.4.1149

Fox, E., Russo, R., Bowles, R., & Dutton, K. (2001). Do threateningstimuli draw or hold visual attention in subclinical anxiety?Journal of Experimental Psychology: General, 130(4), 681–700.https://doi.org/10.1037/0096-3445.130.4.681

Frazier, L., Dillon, B., & Clifton, C. (2015). A note on interpretingdamn expressives: Transferring the blame. Language andCognition, 7(02), 291–304. https://doi.org/10.1017/langcog.2014.31

Frazier, L., Dillon, B., & Clifton, C. (2018). Together they stand:Interpreting not-at-issue content. Language and Speech, 61(2), 199–226. https://doi.org/10.1177/0023830917714608

Gisladottir, R. S., Bögels, S., & Levinson, S. C. (2018). Oscillatorybrain responses reflect anticipation during ccomprehensionof speech acts in spoken dialog. Frontiers in HumanNeuroscience, 12. https://doi.org/10.3389/fnhum.2018.00034

Gisladottir, R. S., Chwilla, D. J., & Levinson, S. C. (2015).Conversation electrified: ERP correlates of speech act recog-nition in underspecified utterances. PLOS ONE, 10(3),e0120068. https://doi.org/10.1371/journal.pone.0120068

Grosser, G. S., & Walsh, A. A. (1966). Sex differences in the differ-ential recall of taboo and neutral words. The Journal ofPsychology, 63(2), 219–227. https://doi.org/10.1080/00223980.1966.10543035

Guillet, R., & Arndt, J. (2009). Taboo words: The effect of emotionon memory for peripheral information. Memory & Cognition,37(6), 866–879. https://doi.org/10.3758/MC.37.6.866

Gutzmann, D. (2013). Expressives and beyond. An introduction tovarieties of use-conditional meaning. In D. Gutzmann & H.-M.Gärtner (Eds.), Beyond expressives: Explorations in use-con-ditional meaning (pp. 1–58). Brill.

Gutzmann, D. (2019). The grammar of expressivity. OxfordUniversity Press.

Hadley, C. B., & MacKay, D. G. (2006). Does emotion help or hinderimmediate memory? Arousal versus priority-binding mechan-isms. Journal of Experimental Psychology: Learning, Memory,and Cognition, 32(1), 79–88. https://doi.org/10.1037/0278-7393.32.1.79

Hansen, S. J., McMahon, K. L., Burt, J. S., & de Zubicaray, G. I.(2017). The locus of taboo context effects in picture naming.Quarterly Journal of Experimental Psychology, 70(1), 75–91.https://doi.org/10.1080/17470218.2015.1124895

Herbert, C., Junghofer, M., & Kissler, J. (2008). Event related poten-tials to emotional adjectives during reading. Psychophysiology,45(3), 487–498. https://doi.org/10.1111/j.1469-8986.2007.00638.x

Hinojosa, J. A., Albert, J., López-Martín, S., & Carretié, L. (2014).Temporospatial analysis of explicit and implicit processingof negative content during word comprehension. Brain andCognition, 87, 109–121. https://doi.org/10.1016/j.bandc.2014.03.008

Hinojosa, J. A., Carretié, L., Valcárcel, M. A., Méndez-Bértolo, C., &Pozo, M. A. (2009). Electrophysiological differences in the pro-cessing of affective information in words and pictures.Cognitive, Affective, & Behavioral Neuroscience, 9(2), 173–189.https://doi.org/10.3758/CABN.9.2.173

Hofmann, M. J., Kuchinke, L., Tamm, S., Võ, M. L. H., & Jacobs, A. M.(2009). Affective processing within 1/10th of a second: Higharousal is necessary for early facilitative processing of nega-tive but not positive words. Cognitive, Affective, & BehavioralNeuroscience, 9(4), 389–397. https://doi.org/10.3758/9.4.389

Ito, T. A., Larsen, J. T., Smith, N. K., & Cacioppo, J. T. (1998).Negative information weighs more heavily on the brain: Thenegativity bias in evaluative categorizations. Journal ofPersonality and Social Psychology, 75(4), 887–900. https://doi.org/10.1037//0022-3514.75.4.887

1436 S. A. DONAHOO AND V. T. LAI

Jay, T. (1992). Cursing in America: A psycholinguistic study of dirtylanguage in the courts, in the movies, in the schoolyards, and onthe streets. J. Benjamins Pub. Co.

Kanske, P., & Kotz, S. A. (2007). Concreteness in emotional words:ERP evidence from a hemifield study. Brain Research, 1148,138–148. https://doi.org/10.1016/j.brainres.2007.02.044

Kanske, P., Plitschka, J., & Kotz, S. A. (2011). Attentional orientingtowards emotion: P2 and N400 ERP effects. Neuropsychologia,49(11), 3121–3129. https://doi.org/10.1016/j.neuropsychologia.2011.07.022

Kim, A., & Lai, V. (2012). Rapid interactions between lexicalsemantic and word form analysis during word recognitionin context: Evidence from ERPs. Journal of CognitiveNeuroscience, 24(5), 1104–1112. https://doi.org/10.1162/jocn_a_00148

Kissler, J., Herbert, C., Peyk, P., & Junghofer, M. (2007). Buzzwords:Early cortical responses to emotional words during reading.Psychological Science, 18(6), 475–480. https://doi.org/10.1111/j.1467-9280.2007.01924.x

Kissler, J., Herbert, C., Winkler, I., & Junghofer, M. (2009). Emotionand attention in visual word processing—An ERP study.Biological Psychology, 80(1), 75–83. https://doi.org/10.1016/j.biopsycho.2008.03.004

Klein, F., Schindler, S., Neuner, F., Rosner, R., Renneberg, B., Steil,R., & Iffland, B. (2019). Processing of affective words in adoles-cent PTSD—Attentional bias toward social threat.Psychophysiology, 56(11). https://doi.org/10.1111/psyp.13444

Knickerbocker, H., Johnson, R. L., & Altarriba, J. (2015). Emotioneffects during reading: Influence of an emotion target wordon eye movements and processing. Cognition and Emotion,29(5), 784–806. https://doi.org/10.1080/02699931.2014.938023

Kousta, S.-T., Vinson, D. P., & Vigliocco, G. (2009). Emotion words,regardless of polarity, have a processing advantage overneutral words. Cognition, 112(3), 473–481. https://doi.org/10.1016/j.cognition.2009.06.007

Kuperman, V., Estes, Z., Brysbaert, M., & Warriner, A. B. (2014).Emotion and language: Valence and arousal affect word rec-ognition. Journal of Experimental Psychology: General, 143(3),1065–1081. https://doi.org/10.1037/a0035669

Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (1990). Emotion, atten-tion, and the startle reflex. Psychological Review, 97(3), 377–395. https://doi.org/10.1037/0033-295X.97.3.377

Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (1997). Motivatedattention: Affect, activation and action. In P. J. Lang, R. F.Simons, & M. T. Balaban (Eds.), Attention and orienting:Sensory and motivational processes (pp. 97–136). LawrenceErlbaum Associates.

Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (1998). Emotion,motivation, and anxiety: Brain mechanisms and psychophy-siology. Biological Psychiatry, 44(12), 1248–1263. https://doi.org/10.1016/S0006-3223(98)00275-3

Leech, G. N., Deuchar, M., & Hoogenraad, R. (1982). Englishgrammar for today: A new introduction. Macmillan Press inconjunction with The English Association.

Mackay, D. G., Shafto, M., Taylor, J. K., Marian, D. E., Abrams, L., &Dyer, J. R. (2004). Relations between emotion, memory, andattention: Evidence from taboo Stroop, lexical decision, andimmediate memory tasks. Memory & Cognition, 32(3), 474–488. https://doi.org/10.3758/BF03195840

Madan, C. R., Shafer, A. T., Chan, M., & Singhal, A. (2017). Shock andawe: Distinct effects of taboo words on lexical decision and free

recall. The Quarterly Journal of Experimental Psychology, 70(4),793–810. https://doi.org/10.1080/17470218.2016.1167925

Mathewson, K. J., Arnell, K. M., & Mansfield, C. A. (2008). Capturingand holding attention: The impact of emotional words inrapid serial visual presentation. Memory & Cognition, 36(1),182–200. https://doi.org/10.3758/MC.36.1.182

Palazova, M., Mantwill, K., Sommer, W., & Schacht, A. (2011). Areeffects of emotion in single words non-lexical? Evidence fromevent-related brain potentials. Neuropsychologia, 49(9), 2766–2775. https://doi.org/10.1016/j.neuropsychologia.2011.06.005

Potts, C. (2005a). Conventional implicatures, a distinguished classof meanings. In G. Ramchand & C. Reiss (Eds.), The Oxfordhandbook of Linguistic Interfacess (pp. 475–501). OxfordUniversity Press.

Potts, C. (2005b). The logic of conventional implicatures. OxfordUniversity Press.

Potts, C. (2007). The expressive dimension. Theoretical Linguistics,33(2), 165–198. https://doi.org/10.1515/TL.2007.011.

Pratto, F., & John, O. P. (1991). Automatic vigilance: The attention-grabbing power of negative social information. Journal ofPersonality and Social Psychology, 61(3), 380–391. https://doi.org/10.1037/0022-3514.61.3.380

Rastle, K., Harrington, J., & Coltheart, M. (2002). 358,534 non-words: The ARC Nonword Database. The Quarterly Journal ofExperimental Psychology Section A, 55(4), 1339–1362. https://doi.org/10.1080/02724980244000099

Reilly, J., Kelly, A., Zuckerman, B. M., Twigg, P. P., Wells, M., Jobson,K. R., & Flurie, M. (2020). Building the perfect curse word: Apsycholinguistic investigation of the form and meaning oftaboo words. Psychonomic Bulletin & Review, 27(1), 139–148.https://doi.org/10.3758/s13423-019-01685-8

Robertson, M. M. (2000). Tourette syndrome, associated con-ditions and the complexities of treatment. Brain, 123(3),425–462. https://doi.org/10.1093/brain/123.3.425

Russell, J. A. (1980). A circumplex model of affect. Journal ofPersonality and Social Psychology, 39(6), 1161–1178. https://doi.org/10.1037/h0077714

Russell, J. A., & Barrett, L. F. (1999). Core affect, prototypicalemotional episodes, and other things called emotion:Dissecting the elephant. Journal of Personality and SocialPsychology, 76(5), 805–819. https://doi.org/10.1037/0022-3514.76.5.805

Schacht, A., & Sommer, W. (2009a). Time course and task depen-dence of emotion effects in word processing. Cognitive,Affective, & Behavioral Neuroscience, 9(1), 28–43. https://doi.org/10.3758/CABN.9.1.28

Schacht, A., & Sommer, W. (2009b). Emotions in word and faceprocessing: Early and late cortical responses. Brain andCognition, 69(3), 538–550. https://doi.org/10.1016/j.bandc.2008.11.005

Scott, G. G., Keitel, A., Becirspahic, M., Yao, B., & Sereno, S. C.(2018). The Glasgow norms: Ratings of 5,500 words on ninescales. Behavior Research Methods. https://doi.org/10.3758/s13428-018-1099-3

Scott, G. G., O’Donnell, P. J., & Sereno, S. C. (2014). Emotion wordsand categories: Evidence from lexical decision. CognitiveProcessing, 15(2), 209–215. https://doi.org/10.1007/s10339-013-0589-6

Segalowitz, S. J., & Lane, K. C. (2000). Lexical access of functionversus content words. Brain and Language, 75(3), 376–389.https://doi.org/10.1006/brln.2000.2361

COGNITION AND EMOTION 1437

Siegrist, M. (1995). Effects of taboo words on color-naming per-formance on a Stroop test. Perceptual and Motor Skills, 81(3suppl.), 1119–1122. https://doi.org/10.2466/pms.1995.81.3f.1119

Simons, M., Tonhauser, J., Beaver, D., & Roberts, C. (2010). Whatprojects and why. Semantics and Linguistic Theory, 20, 309.https://doi.org/10.3765/salt.v20i0.2584

Stephens, R., Atkins, J., & Kingston, A. (2009). Swearing as aresponse to pain. NeuroReport, 1. https://doi.org/10.1097/WNR.0b013e32832e64b1

Tanaka-Ishii, K., & Terada, H. (2011). Word familiarity and fre-quency: Word familiarity and frequency. Studia Linguistica, 65(1), 96–116. https://doi.org/10.1111/j.1467-9582.2010.01176.x

Thomas, L., & LaBar, K. (2005). Emotional arousal enhances wordrepetition priming. Cognition & Emotion, 19(7), 1027–1047.https://doi.org/10.1080/02699930500172440

Trauer, S. M., Andersen, S. K., Kotz, S. A., & Müller, M. M. (2012).Capture of lexical but not visual resources by task-irrelevantemotional words: A combined ERP and steady-state visualevoked potential study. NeuroImage, 60(1), 130–138. https://doi.org/10.1016/j.neuroimage.2011.12.016

Trauer, S. M., Kotz, S. A., & Müller, M. M. (2015). Emotional wordsfacilitate lexical but not early visual processing. BMCNeuroscience, 16(1). https://doi.org/10.1186/s12868-015-0225-8

Van Lancker, D., & Cummings, J. L. (1999). Expletives:Neurolinguistic and neurobehavioral perspectives on swear-ing. Brain Research Reviews, 31(1), 83–104. https://doi.org/10.1016/S0165-0173(99)00060-0

Van Petten, C., Kutas, M., Kluender, R., Mitchiner, M., & McIsaac, H.(1991). Fractionating the word repetition effect with event-related potentials. Journal of Cognitive Neuroscience, 3(2),131–150. https://doi.org/10.1162/jocn.1991.3.2.131

Vinson, D., Ponari, M., & Vigliocco, G. (2014). How does emotionalcontent affect lexical processing? Cognition and Emotion, 28(4), 737–746. https://doi.org/10.1080/02699931.2013.851068

Wabnitz, P., Martens, U., & Neuner, F. (2012). Cortical reactions toverbal abuse: Event-related brain potentials reflecting the pro-cessing of socially threatening words. NeuroReport, 23(13), 774–779. https://doi.org/10.1097/WNR.0b013e328356f7a6

Wabnitz, P., Martens, U., & Neuner, F. (2016). Written threat:Electrophysiological evidence for an attention bias toaffective words in social anxiety disorder. Cognition andEmotion, 30(3), 516–538. https://doi.org/10.1080/02699931.2015.1019837

Wang, L., & Bastiaansen, M. (2014). Oscillatory brain dynamicsassociated with the automatic processing of emotion inwords. Brain and Language, 137, 120–129. https://doi.org/10.1016/j.bandl.2014.07.011

Wang, W., Chen, L., Thirunarayan, K., & Sheth, A. P. (2014). Cursingin English on Twitter. Proceedings of the 17th ACM Conferenceon Computer Supported Cooperative Work & Social Computing -CSCW ‘14, 415–425. https://doi.org/10.1145/2531602.2531734

Warriner, A. B., Kuperman, V., & Brysbaert, M. (2013). Norms ofvalence, arousal, and dominance for 13,915 English lemmas.Behavior Research Methods, 45(4), 1191–1207. https://doi.org/10.3758/s13428-012-0314-x

Wentura, D., Rothermund, K., & Bak, P. (2000). Automatic vigi-lance: The attention-grabbing power of approach- and avoid-ance-related social information. Journal of Personality andSocial Psychology, 78(6), 1024–1037. https://doi.org/10.1037/0022-3514.78.6.1024

Yao, Z., Yu, D., Wang, L., Zhu, X., Guo, J., & Wang, Z. (2016). Effectsof valence and arousal on emotional word processing aremodulated by concreteness: Behavioral and ERP evidencefrom a lexical decision task. International Journal ofPsychophysiology, 110, 231–242. https://doi.org/10.1016/j.ijpsycho.2016.07.499

Zhang, M., Ge, Y., Kang, C., Guo, T., & Peng, D. (2018). ERP evi-dence for the contribution of meaning complexity underlyingemotional word processing. Journal of Neurolinguistics, 45,110–118. https://doi.org/10.1016/j.jneuroling.2016.07.002

Zhang, M., & Guo, T. (2014). Event-related brain potentials differ-entiate three types of emotional words categorized from lin-guistic perspective. Journal of Neurolinguistics, 31, 17–27.https://doi.org/10.1016/j.jneuroling.2014.06.001

Zhang, D., Nie, A., Wang, Z., & Li, M. (2019). Influence of lag lengthon repetition priming in emotional stimuli: ERP evidence.Journal of Clinical Laboratory Analysis, 33(1), e22639. https://doi.org/10.1002/jcla.22639

1438 S. A. DONAHOO AND V. T. LAI