When unpleasantness meets feminines: a behavioural study on gender agreement and emotionality

ABSTRACT

The emotional connotation of words is known to affect word and sentence processing. However, the when and how of the interaction between emotion and grammar are still up for debate. In this behavioural experiment, 35 female university students read noun phrases (NPs) composed by a determiner and a noun in their L1 (Spanish), and were asked to indicate if the NPs were grammatically correct (el_masc camarero_masc) or not (*la_fem tornillo_masc; i.e. a gender agreement task). The type of gender (arbitrary/natural), the emotionality (unpleasant/neutral), and the gender class (feminine/masculine) of the nouns were manipulated. We found an overall grammaticality effect, responses being faster in grammatically correct trials than in incorrect ones. However, the effects of emotionality and gender class varied depending on gender type. For arbitrary gender, the grammaticality effect was greater in feminine nouns than in masculine nouns and independent of emotionality. For natural gender, the grammaticality effect interacted with gender class and emotionality, this effect only emerging in unpleasant stimuli for feminine nouns. Our results reveal that it is possible to find emotional effects at the behavioural level in an intrinsically grammatical task. Yet, these effects depend on gender properties like the type of gender and the gender class.

KEYWORDS:

• Gender agreement
• emotionality
• gender asymmetries
• performance

Acknowledgements

The authors thank two anonymous reviewers for their comments on an earlier version of this manuscript.

Disclosure statement

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

Notes

1 Five male participants signed up to take part in the study. Since this number was quite small in comparison to the number of female participants and considering that differences in emotional processing between men and women have been found in previous research (e.g., Pinheiro et al., 2017), these participants were excluded from the final sample.

2 As for frequency, for example, the unmarked status of the masculine gender (along with other sociolinguistic factors) makes the masculine form of nouns much more frequent than the feminine one.

3 Although it was possible to separate the arbitrary gender NPs using only two lists (each noun is necessarily either masculine or feminine), we decided to separate these NPs into four lists instead. Note that, due to this specific distribution of items across lists, each participant saw half of the arbitrary gender items and just one version (masculine or feminine) of the natural gender items. Since items were distributed across conditions ensuring that half were grammatical/neutral/masculine and half were ungrammatical/unpleasant/feminine, each participant read 9 (arbitrary gender) and 7 (natural gender) items from each of the eight conditions (see Table 1). This allowed for the experimental lists to be balanced across both types of gender.

4 In prior literature on the interaction between grammaticality and emotionality, a strong effect of grammaticality is consistently observed. The effect of emotionality, if found, is usually weaker or restricted to its interaction with grammaticality, and no prior research has added the effect of gender class to the equation. Therefore, even if we could expect the magnitude of these two effects to diverge across participants, we also expect for these variables to have a smaller impact than grammaticality in the random structure of our models. Thus, we prioritised maintaining random slopes for our critical variables in this order (from more to less priority): grammaticality, gender, and emotionality.

5 Following open science practices the raw data and the R script used are available at https://osf.io/63eck/.

6 Note, however, that the difference between the match and mismatch conditions in neutral nouns was near significance (p = 0.069). This effect could be underpowered due to the limitations on the number of trials per condition. Nonetheless, the significant three-way interaction and the statistics for the other pair-wise comparisons indicate that the grammaticality effect is substantially stronger for unpleasant than for neutral nouns.

7 Note that mean response times for the unpleasant match feminine condition were very similar for arbitrary and natural gender (see Supplementary Materials).

Additional information

Funding

This study was supported by grant PID2019-110583GB-I00 funded by MICIU/AEI/10.13039/501100011033 and grant ED431B 2022/19 funded by the Autonomous Government of Galicia (Xunta de Galicia).