2.3 Discussion
Our results did not show any evidence of early ERP components linked to the processing of spoken emotionally laden words. Notably, we did not find that negatively or positively valenced words evoked an EPN compared to neutral words. This result is in line with several previous studies, which only showed effects starting at 370 ms (Grass et al. 2016; Kanske & Kotz, 2011; Rohr & Rahman, 2015) or substantially later, only after word offset (Hatzidaki et al., 2015). In contrast, Mittermeier and colleagues (2011) reported increased early negativity, on the P2 component, for auditory words with both negative and positive connotations, but only in comparison to neutral tones (and only reported for Pz). The very early effects reported by Mettermeier (2011), replicated by Jaspers-Fayer and colleagues (2012), may nonetheless be linked to their design, which included only ten auditory words that were repeated numerous times. This may have allowed participants to anticipate the words and access meaning from their onset. In addition, as no neutral words were included, it is difficult to ascertain whether the reported effect was due to the valence of the linguistic stimuli or to differences in task demands, i.e. the ease of discriminating between two tones vs. identifying linguistic content. In line with this, previous results have shown that the N1-P2 complex is modulated by the physical characteristics of auditory stimuli but not by their linguistic (emotional) content (Grass et al., 2016). The lack of early effects in present study may be linked to the characteristics of our auditory stimuli in particular and to auditory processing of words in general, which is incremental in nature. The Mandarin words we presented were disyllabic and full lexical access was not possible until participants heard the second syllable, which on average occurred at roughly 400 msec. No stimulus was repeated such that participants could not anticipate words. Our results differ from Grass’s study (2016), which reported an enhanced frontal positivity and parieto-occipital negativity for valenced words between 370-540 msec post-stimulus, likely due to the differences in task requirement: While our study used a valence judgement task, Grass’s study employed a one-back task that encouraged participants to recall previous stimuli, leading to increased priming effect through contextual association. Valenced words have been shown to exhibit stronger inter-item associations and higher semantic relatedness (Buchanan et al., 2006), suggesting that valenced words may prime each other more effectively. This superior priming among valenced words could result in facilitated processing for valenced words during the one-back task. Instead, our findings align more closely with Hatzidaki’s (2015) study, which also used a valence judgement task and reported a late LPC between 600-950 msec post-stimulus. Although the mean amplitudes of ERPs for our Negative or Positive words were numerically larger than for Neutral words in later time windows (500-800 msec and 800-1100 msec), this pattern did not reach statistical significance.
In relation to the behavioral classification of the auditory words, our results showed that participants’ valence ratings of Mandarin words generally corresponded to the English norms, with some discrepancies. Previous studies of emotion lexicons have shown cultural differences of emotion concept, including appraisal. That is, while an emotion word may have equivalent translations in two languages, speakers of those languages may still have different evaluations of associated events or consequences with positive or negative connotation (for a review, see Pavlenko, 2008). Noting that our experimental words were translated from English to Mandarin and categorized into one of three given valence types based on an affective database of English words, our behavioral data is thus consistent with this account, reflecting that Mandarin speakers and English speakers did not always appraise lexical items that carry emotional connotations identically.
EXPERIMENT 2
As outlined in the general introduction, people tend to show enhanced recall and recognition of emotional stimuli compared to neutral ones (for a review, see Tyng et al., 2017). Such “emotion-memory” effects are less widely documented for auditory than visually presented stimuli. We capitalized on our participants having taken part in the first, auditory experiment to further examine this question. In the following experiment, stimuli were presented in written format and participants were required to identify those words that they had heard in the immediately prior experiment among an equal number of foils with positive, negative or neutral valence ratings. Participants were not informed of the second experiment prior to the completion of the first. The second experiment thus provided us with a means to examine the impact of valence and arousal on memory encoding as well as allowing a direct comparison of the electrophysiological response to emotionally laden words across auditory and written format. To our knowledge, while numerous studies have examined the ERP response to emotion words in either written or auditory format, only one study (Rohr & Rahman, 2015) to date has directly compared the two for the same set of words. Distinct patterns were reported as a function of presentation, with an enhanced negativity, specifically for negative compared to neutral words for printed words, while no reliable valence effect, but an unexpected negativity over central ROI for negative words compared to neutral words in auditory presentation.