COVIDisgust: Language processing through the lens of partisanship. Veranika Puhacheuskaya, Isabell Hubert Lyall, Juhani Järvikivi. PLoS, July 21, 2022. https://doi.org/10.1371/journal.pone.0271206
Abstract: Disgust is an aversive reaction protecting an organism from disease. People differ in how prone they are to experiencing it, and this fluctuates depending on how safe the environment is. Previous research has shown that the recognition and processing of disgusting words depends not on the word’s disgust per se but rather on individual sensitivity to disgust. However, the influence of dynamically changing disgust on language comprehension has not yet been researched. In a series of studies, we investigated whether the media’s portrayal of COVID-19 will affect subsequent language processing via changes in disgust. The participants were exposed to news headlines either depicting COVID-19 as a threat or downplaying it, and then rated single words for disgust and valence (Experiment 1; N = 83) or made a lexical decision (Experiment 2; N = 86). The headline type affected only word ratings and not lexical decisions, but political ideology and disgust proneness affected both. More liberal participants assigned higher disgust ratings after the headlines discounted the threat of COVID-19, whereas more conservative participants did so after the headlines emphasized it. We explain the results through the politicization and polarization of the pandemic. Further, political ideology was more predictive of reaction times in Experiment 2 than disgust proneness. High conservatism correlated with longer reaction times for disgusting and negative words, and the opposite was true for low conservatism. The results suggest that disgust proneness and political ideology dynamically interact with perceived environmental safety and have a measurable effect on language processing. Importantly, they also suggest that the media’s stance on the pandemic and the political framing of the issue may affect the public response by increasing or decreasing our disgust.
General discussion
Our study tested the hypothesis that the media’s stance on the pandemic may elevate or reduce participants’ disgust, which would affect word ratings and word recognition latencies. We also predicted that such person-based factors as disgust proneness and political ideology will mediate the effect. A word rating (Experiment 1) and a lexical decision (Experiment 2) study found partial support for these hypotheses. In brief, the main findings were as follows:
- More liberal participants rated the stimuli as more disgusting after being exposed to the headlines downplaying the threat of COVID-19, whereas more conservative participants gave higher disgust ratings following the headlines emphasizing it.
- More liberal participants were more extreme with their ratings and gave a broader range of responses (rating disgusting words as more disgusting and negative words as more negative, as well as rating non-disgusting words as less disgusting and positive words as more positive) than their more conservative peers regardless of the condition.
- Disgusting and negative words had a facilitatory effect for more liberal participants (shorter RTs) and an inhibitory effect for more conservative participants (longer RTs).
- More disgust-prone individuals rated everything as more disgusting than low disgust-prone ones.
- In the severe condition, low disgust-prone participants rated all the stimuli as more disgusting and negative, whereas high disgust-prone participants only rated low to moderately disgusting words as more disgusting and negative.
Affective word ratings and political ideology
As expected, political orientation had a clear impact on word ratings. As we noted in the Introduction, the perception of the severity of the virus became an identity marker for both ends of the political spectrum. Given such drastic polarization, it is not surprising that the two types of the headlines produced the exact opposite effect on the participants depending on their political ideology. More liberal participants in our study were more disgusted by the headlines downplaying the severity of COVID-19 than those emphasizing it, rating everything as more disgusting afterwards. In contrast, more conservative participants assigned higher disgust ratings following the severe headlines. We offer two possible explanations for this result, one based on direct affective evaluation and the other on the disgust system response due to stimulus habituation. According to the first explanation, the stance on the virus from the political outgroup (downplaying headlines for the liberal participants and severe headlines for the conservative ones) evoked a strong emotional response, which translated into higher ratings on the disgust scale. However, in this case, one would also expect lower ratings on the valence scale depending on the headline, and this was not what we found. That leaves us the second possibility. As the liberal narrative revolves around the costs of not treating the virus seriously enough, they may have become habituated and desensitized to it. The take on the danger of COVID-19 may thus be perceived as the “default” by them and no longer alert their disgust system, whereas headlines contradicting this view might instantly elevate their disgust levels, signaling danger. The opposite, of course, should be true for more conservative participants. Note that one of our hypotheses was that more conservative participants will discard the severe headlines as alarmist, since previous research showed that conservatives have less trust in contradictory media and firmly believe that COVID-19 does not pose big health risks [55]. The current findings suggest that this was either not the case or, if it was the case, it did not stop their disgust system from ramping up. All in all, this is in line with mounting evidence that conservatives are more prone to disgust [49–51]. Even though previous studies found conservatives to be less concerned about the pandemic and less eager to engage in social distancing than liberals [55, 70], our results show that highlighting the danger of the virus still makes conservative participants give higher disgust ratings. Whether this translates into more adherence to safety protocols is a topic for further research.
One novel finding of our study is more extreme disgust and valence ratings by more liberal participants compared to their more conservative peers regardless of the condition. Disgusting and negative words were rated as more disgusting and more negative by more liberal participants, and the opposite was true for non-disgusting and positive words. We are not aware of any research examining whether political ideology correlates with ratings’ extremity. It is entirely possible that this broader range of ratings is additionally mediated by some other personality traits and this needs to be verified by future research.
Affective word ratings and disgust proneness
Disgust proneness affected word ratings over and above the effects of political ideology. Regardless of the headline type, more disgust-prone individuals rated all the stimuli as more disgusting and negative stimuli as more negative than less disgust-prone individuals. This demonstrates that disgust ratings can serve as a good proxy for participant’s disgust and adds to the growing body of evidence regarding the effects of disgust proneness on cognition in general and language processing in particular. [40, 41] found that disgust sensitivity was positively correlated with pupil dilation during the processing of stereotype-based clashing statements, suggesting that more disgust-prone individuals may experience greater arousal when interacting with stimuli that are disgusting either physically or morally. The results of the current study further indicate that even single word processing can be significantly affected by the participant’s disgust sensitivity. In addition, disgust proneness significantly interacted with the headline type. While low disgust-prone participants rated all the stimuli as more disgusting and negative when the threat of COVID-19 was highlighted (severe headlines), high disgust-prone participants only rated low and moderately disgusting words as more disgusting and positive words as more negative in the severe condition. As we addressed in the Discussion after Experiment 1, this may be due to the ceiling effect since ratings assigned by high-disgust prone participants to extremely valenced stimuli were very close to the top of the disgust scale and the bottom of the valence scale.
Political ideology vs disgust proneness in lexical access: The role of the pandemic
Our findings from the lexical decision experiment partially corroborated and extended the results for French by [46]. The authors found that disgusting words had a facilitatory effect for lexical recognition in less disgust-prone participants and an inhibiting effect in more disgust-prone participants. Our study, however, found that political ideology was more predictive of RTs than disgust sensitivity. Even though the general direction of the effect was the same (more liberal participants patterned like less disgust-prone ones), only political ideology significantly improved the model’s fit when both factors were examined together. Overall, disgusting and negative words had a facilitatory effect on word recognition for more liberal participants and an inhibitory effect for more conservative participants. To the best of our knowledge, the interaction between political ideology and lexical decision times has not yet been researched. One possible explanation for the dominant effect of political ideology in our study is a big political component pertinent to the ongoing pandemic from its very beginning. [55] suggested that political ideology was uniquely predictive of the participant’s COVID-19 behavior even when controlling for such variables as belief in science and COVID-related anxiety. Thus, it may be that political views have temporarily become a more salient marker of the behavioral immune system response than disgust sensitivity per se. This is, of course, a speculative idea that needs to be addressed by further research. One way to verify this would be to conduct the same study during the pandemic and post-pandemic.
Differential effects of traits and states on lexical access
We did not find an effect of dynamically changing disgust levels (induced by headlines) on lexical access. Even though the headlines successfully affected participants’ ratings in Experiment 1, they did not have an effect on RTs in Experiment 2—neither by themselves nor in interaction with person-based factors. Unlike headlines, however, political ideology was found predictive of word recognition latencies. Why would that be the case? Previous research has found political views to be just one manifestation of a cognitive and affective make-up and to have a robust correlation with threat perception [74]. It is thus not surprising that aligning with a particular political ideology may make disgust-related concepts in long-term memory more or less accessible (see [75] for converging findings with threat-related concepts). Thus, our results suggest a difference between fluctuating states (i.e., the participant’s emotional response to a particular set of headlines) and stable traits (i.e., aligning with more conservative or more liberal ideology) in affecting the ease of accessing disgusting and negative words. One alternative possibility to consider is that an exposure to COVID-related news may need to be longer to see an effect on lexical decision (we only showed a handful of headlines that the participants could switch through at their own pace). This could be tested by future research.
Limitations of present research
Our study had several limitations that need to be noted. First and foremost, we did not collect participants’ socioeconomic status, belief in science, self-perceived likelihood of contracting COVID-19, or COVID-19 related anxiety. Second, a convenience sample of university students produced a slightly skewed distribution of gender, age, and political ideology (most participants were young and more liberal females), which may have affected the results. That said, within the range of scores obtained in this experiment, the distribution was very close to normal.
One other concern needs to be addressed. Since our headlines reported on the pandemic, it is important to make sure that word recognition latencies were not affected by the presence of words directly related to the pandemic and to disease in general. As no lists of pandemic-related words exist, it is difficult to estimate how many words in the final dataset satisfied this criterion. Using our best judgment, we counted 6 out of 99 words that were disease-related, with the disgust indexes given in brackets: “unhealthy” (-0.7), “germ” (0.9), “sickening” (1.24), “deadly” (0.8), “parasite” (1.5), “disease” (1). Three of those words occurred in the severe headlines in full (“sickening”, “disease”, “deadly”) and one in part (“bloodthirsty”–“blood”). As one can see, the words were relatively dispersed on the disgust scale. To make sure the results of Experiment 2 were not contaminated by this overlap, we reran the models without these four words. While disgust proneness was no longer significant, political ideology remained significant. This, once again, testifies to the stability of the effect of political ideology.
All in all, our studies found that not only do headlines about the pandemic affect the participants’ disgust levels but that they also interact with a range of person-based factors, namely how prone the participant is to disgust and what political ideology they align with.