Liberty,Justice,and Victory: Abstract Nouns are Grammatically and Psychologically Feminine

Abstract

Grammatical gender generally implies disadvantages for women (e.g., the generic masculine leading to real-life invisibility of women). Diverging from this idea, we here postulate that abstract concepts (such as “justice” or “democracy”), arguably reflecting the most sophisticated expression of human thought, are associated with femininity. Two archival studies (Study 1a, b) show the greater frequency of feminine gender among abstract nouns across seven (out of nine) Indo-European languages. Studies 2 and 4 show that feminine abstract nouns are envisaged as female, both explicitly and through a voice-assignment task. Furthermore, through Implicit Association Tests (IATs), we show that abstract concepts are associated with women as a social category (Studies 3a, b). Together, our studies show an association between abstractness and grammatical gender that carries psychological meaning through the gender congruency effect. Arguably, these cognitive processes may contribute to the idealization of women as “super-human,” through the association between abstractness and super-human imagery (Study 4).

Keywords

abstractness language grammatical gender gender congruency effect superhumanization

It is often argued that grammatical gender creates disadvantages for women. In most grammatical-gendered languages, nouns carry one of two (e.g., Italian, French) or three (e.g., German, Polish) genders and this grammatical gender then affects other constituents of the sentence such as articles, verbs, adjectives and the like, making gender a salient feature of the sentence. Although grammatical gender itself is largely arbitrary (e.g., bridge is feminine in German, but masculine in Spanish), it may affect socio-cognitive processes. This is best illustrated by the “generic masculine” rule, according to which the masculine is employed when the gender of the person is not specified (e.g., “an ideal student_masc should…”) and, in its plural form, to describe groups of mixed gender (e.g., “The girl and the boy are students_masc”). Theoretically, such generic masculine forms are neutral, as they comprise both male and female exemplars; in reality, however, they tend to facilitate the retrieval of gender-congruent (male) exemplars, while making women mentally invisible (see Gygax et al., 2021; Menegatti & Rubini, 2017; Schmitz et al., 2023). Among others, this has been shown to have tangible implications for the outcome of opinion polls (e.g., Braun et al., 2005). Some researchers have taken this argument one step further, hypothesizing that gender disparity and gender prejudice are more pronounced in countries with gendered languages (DeFranza et al. 2020; Prewitt-Freilino et al., 2012; Shoham & Lee, 2018), although the latter argument rests exclusively on correlational evidence.

Diverging from this perspective, we here argue that there is one area in which grammatical gender may “favor” women, namely abstractness. Abstract nouns refer to entities that are not directly perceivable through the senses (Della Rosa et al., 2010): these include concepts such as democracy, justice, or infinity, which are arguably the most sophisticated expression of human thought. Learning of and reasoning about such abstract concepts critically depend on language (Luchkina & Waxman, 2023). Compared to concrete concepts, abstract concepts are acquired at a later age, learned mainly through verbal stimulation rather than direct experience with the object, and are much more difficult to envisage or contextualize (Della Rosa et al., 2010; Mestres-Missé et al., 2014). They generally express complex cultural ideas that transcend perceptual and bodily experiences.

Across six studies, we show that a grammatical and psychological association exists between abstractness and femininity. To address the abstractness–femininity overlap and its psychological impact, the present paper is organized into four sections, each addressing a specific research question. In the first section, we discuss the historical origins of the abstractness–femininity overlap and investigate whether abstract nouns have an above-chance likelihood of carrying feminine grammatical gender in Indo-European languages (Studies 1a and b). In the second section (Study 2), we investigate whether grammatical gender colors the perception of abstract concepts as being feminine (vs. masculine), similar to the gender-congruency effect observed in past research for concrete nouns. In the third section (Studies 3a and b), we argue that if abstract nouns carry predominantly feminine grammatical gender, and if grammatical gender evokes feminine images, then people may implicitly associate abstractness (vs. concreteness) with women (vs. men) as a social group. In the fourth and final section, we explore whether the abstractness–femininity association may also compel people to perceive women as the super-human entities.

We report all data exclusions, manipulations, and measures in the studies. Data, study materials, and supplementary materials are available on the Open Science Repository OSF: https://osf.io/8ysmx/.

Are Abstract Nouns Grammatically Feminine? Origins and Pervasiveness of Grammatical Gender Systems (Study 1)

Why would grammatical gender be associated with abstractness? The history and the development of gender systems cannot be directly observed, so the reconstruction necessarily remains speculative, but going back in time may be useful for understanding today's link between feminine grammatical gender and abstractness. Proto-Indo-European language is considered the common ancestor of Indo-European languages and was spoken as a single language from approximately 4500 BC to 2500 BC, that is, during the Late Neolithic up to the Early Bronze Age. It is believed to have developed from a two-gender system, distinguishing animate vs. inanimate, into a three-gender system (Luraghi, 2011; Matasovic, 2004), divided into animate vs. inanimate/abstract or collective vs. inanimate/concrete. Importantly, the three-gender connotation initially did not have a male/female connotation, as already argued by Brugmann (1891). In references to this author, Luraghi (2011, p. 437) states that “this system was a late development from an earlier two-gender system commonly held to be animacy-based, which morphologically consisted of […] the masculine and the neuter, while the feminine gender was later formed through the addition of a special suffix.” Similarly, Wheeler (1899, p. 529ff) argues that

the form-groups of nouns which mark the classification by gender had originally nothing to do with distinctions of sex. Their distinctive endings did not carry with them originally the suggestion of sex; thus, the so-called “feminine” suffixes […] served rather, […] to form collectives and abstracts.

If the feminine suffices initially did not imply any gender meaning at all, but referred to collectives and/or abstractness, this may account for today's co-occurrence of abstractness and feminine grammatical gender. How the extension from abstract to feminine came about, is still debated (for an overview, see Luraghi, 2009) and goes beyond the scope of the present paper.

Regardless of how exactly the switch occurred, linguists seem to agree that the suffices that originally referred to abstractness only subsequently became associated with femininity. If this argument is correct, then it would not be surprising if abstract nouns were predominantly of feminine grammatical gender in those languages that derive from Proto-Indo-European language. In line with this argument, Luraghi (2009) states that feminine nouns tend to be cognitively more complex and that the association between feminine grammatical gender and abstractness is not surprising as it “logically follows from the historical development of the feminine gender out of an (earlier) class of abstract nouns” (p. 8). This reasoning would also allow an additional prediction, namely that the abstractness–femininity association may become weaker over time as a function of subsequent gender change. For instance, it may be stronger in Latin than in Italian, or in Ancient than in Modern Greek. The first aim of the present research program was therefore to investigate, through archival analyses, whether feminine grammatical gender is over-represented among abstract nouns, whether this asymmetry holds across different (current and ancient) Indo-European languages, and whether it tends to become weaker over time.

Study 1a

The primary aim of the first study was to investigate the “feminine abstract” phenomenon in Italian. In line with prior theorizing (Brugmann, 1891; Luraghi, 2011; Matasovic, 2004), we predicted that grammatical gender would be associated with abstractness, so that abstract nouns be predominantly feminine.

Furthermore, we aimed at looking at the implications of the abstractness–femininity association on how feminine (vs. masculine) nouns are imagined and how they are acquired during socialization. If abstract nouns are learned through distinct pathways, as suggested by Della Rosa et al. (2010), and if feminine nouns are associated with abstractness, as tested here, then feminine nouns, compared to masculine ones should also be more difficult to imagine, more detached from context, less familiar, and acquired later in life and mainly through verbal stimulation rather than through direct experience.

Finally, we aimed at exploring the evolution of the abstractness–femininity association over time, specifically by analyzing Latin (i.e., the precursor of Italian and other Romance languages). We expected that in Latin, too, abstract nouns would be predominantly feminine rather than masculine, and that this effect may be even stronger than with Italian, given that Latin is closer to Proto-Indo-European. Results (see supplementary materials, which are available on the OSF: https://osf.io/8ysmx/) confirmed that the abstract–feminine association decreases with increasing temporal distance from Proto-Indoeuropean language, with the abstractness–femininity association being stronger in Latin than in Italian.

Method

To test our hypotheses, we employed the database by Della Rosa et al. (2010), which includes participants’ ratings for concreteness (directly perceivable through the senses), abstractness (not directly perceivable through the senses), imageability (easy to imagine), context availability (easy to think of a context), familiarity (familiar), age of acquisition (age at which participants learnt the word), mode of acquisition (concept entirely acquired through language rather than experience) for 417 Italian nouns, on a scale from 100 to 700. Furthermore, for analysis purposes, we dichotomized the abstractness ratings by splitting the noun list into two groups of approximately equal size (the cut-off being 300); we refer here to the continuous variable as “abstractness ratings,” and the dichotomized variable as “abstractness.”

Since grammatical gender was our main predictor, four nouns were excluded due to gender ambiguity (i.e., they can be both masculine and feminine; nomad, tourist, athlete, teacher) and one due to gender duality (i.e., grammatical gender changes from singular to plural; egg). Our final word list consisted of 412 nouns (205 abstract and 207 concrete). The majority of these nouns had unambiguous gender suffices (-a or -o), whereas a minority (26%) ended in the gender-ambiguous -e, and a negligible number (n = 4) ended in -i or -u.

Results

Italian Grammatical Gender. Abstractness was strongly associated with grammatical gender, χ²(1, N = 412) = 15.91, p < .001, d = 0.40. Approximately half of the concrete nouns were feminine (47%) and half masculine (53%), binomial, p = .404, whereas for abstract nouns, feminine gender (66%) reliably outnumbered masculine gender (34%), binomial, p < .001. However, looking at the ratings of abstract nouns, grammatically feminine nouns were no more abstract (M = 489.88, SD = 62.96) than masculine ones (M = 477.92, SD = 86.16), t(106) = −1.02, 95% CI of the difference [−32.82, 8.89], p = .308, d = −0.17. Thus, masculine nouns were less frequent among the abstract nouns, but those that were present had an abstractness level similar to the (more frequent) feminine nouns.

Independent sample t-tests were also run on the other ratings to check for grammatical gender differences across all nouns. Results showed that, in the overall sample, feminine grammatical gender was linked to lower imaginability, t(399) = 3.81, 95% CI of the difference [33.37, 102.74] p < .001, d = 0.38, lower context availability, t(410) = 2.81, 95% CI of the difference [10.56, 59.88], p = .005, d = 0.28, older age of acquisition, t(403) = −3.07, 95% CI of the difference [−55.63, −12.23], p = .002, d = −0.30, and greater acquisition through a verbal modality rather than experience, t(410) = −2.14, 95% CI of the difference [−62.48, −2.67], p = .033, d = −0.21, but there was no effect of grammatical gender on context familiarity, t(410) = 1.13, 95% CI of the difference [−8.34, 30.89], p = .259, d = 0.11. To test whether these overall gender differences were explained by the fact that grammatically feminine nouns are overrepresented among abstract ones, we ran a mediation model in JASP 0.19.3 (JASP Team, 2025) with grammatical gender as predictor, abstractness ratings as mediator, and imaginability, familiarity, context availability, and age and mode of acquisition as outcomes, using 1,000 bootstrap resamples. Results (see Figure 1) showed that, indeed, abstractness ratings mediated these effects: in other words, grammatically feminine nouns were predominantly abstract, and as such harder to imagine (b = −66.61, SE = 17.02, p < .001, LLCI = −104.49, ULCI = −36.56), less familiar (b = −17.15, SE = 4.70, p < .001, LLCI = −29.31, ULCI = −9.10), acquired later in life (b = 32.83, SE = 8.50, p < .001, LLCI = 17.85, ULCI = 52.21) and through a verbal modality (b = 49.59, SE = 12.75, p < .001, LLCI = 27.04, ULCI = 77.39), and less available in context (b = −43.33, SE = 11.11, p < .001, LLCI = −68.44, ULCI = −23.35).

Figure 1.

Mediation Model With Imaginability, Familiarity, Context Availability, Age, and Mode of Acquisition as a Function of Grammatical Gender and Abstractness Ratings (Study 1a).

Study 1b

To address the generality of the association between abstractness and femininity in a broader range of Indo-European languages, we extended the prior analysis to other languages derived from Proto-Indo-European. We chose eight currently spoken Indo-European languages representing different branches of the Indo-European language tree (e.g., Kapović et al., 2017), namely Hindi, Modern Greek, Italian, Irish, German, Albanian, Lithuanian, and Polish. We expected to find a similar pattern as that found in Study 1a, with grammatically feminine nouns being over-represented among abstract (but not concrete) nouns.

This additional study also intended to resolve a methodological limit of Study 1a. In Study 1a, abstractness ratings were taken from a prior study in which Italian raters had judged the abstractness of nouns (Della Rosa et al., 2010); given the strong association between gender and abstractness, we cannot exclude with certainty that Della Rosa's Italian raters may have been under the influence of grammatical gender, possibly judging nouns as more or less abstract based on their grammatical gender. To resolve this problem, in Study 1b, we used an English language database for which concreteness ratings (i.e., the degree to which the object is perceivable through the senses and learned through experience) were available (Brysbaert et al., 2014). In this case, ratings had been obtained in a language without grammatical gender, making it unlikely that concreteness ratings had been influenced by grammatical gender.

Method

Nouns in our sample were extracted from a corpus of 40.000 English words (Brysbaert et al., 2014), which includes concreteness ratings: participants were first provided with the definition of concrete (i.e., can be directly experienced through the five senses) and abstract nouns (cannot be directly experienced, but can be defined by other words), and were asked to rate the word on a scale from 1—abstract (language based) to 5—concrete (experience based). We selected the 125 nouns with lowest concreteness ratings and the 125 with the highest concreteness ratings. Here as well, we refer to the dichotomized variable as “abstractness,” and the continuous one as “concreteness ratings.”

The nouns were translated into eight languages, each representing a different branch of Proto-Indo-European: Albanian, German, modern Greek, Hindi, Irish, Italian, Lithuanian, and Polish. The translation procedure consisted in four steps: (1) Translations and grammatical gender were obtained through the online multilanguage dictionary Glosbe (https://it.glosbe.com/); (2) Missing information were obtained through language-specific dictionaries¹; (3) Google Translate was employed to fill any remaining gaps in translations; (4) Finally, native-language raters unaware of the hypotheses were asked to check the translations and assess gender where it was still missing. In cases in which native-language raters noted mistakes or inaccuracies in translations, we adopted the translations indicated by the rater.

The selected translation was always the first presented by the dictionary (or indicated as a substitute by the native speaker), so repetitions are present in some cases (e.g., “endlessness” and “limitlessness,” both translated as “Unendlichkeit” in German). Furthermore, two words were excluded from the dataset as they were not nouns (“someway,” “unbefitting”), and eight as the translation was plural in one or more languages (e.g., “tongs”). Finally, in some cases, neither the dictionaries nor the raters were able to provide a translation, which accounts for some missing data; this is not surprising given that some of the abstract terms were indeed unusual and of very low frequency (e.g., “utopianism”).

Results

Frequency of Feminine Nouns. For every language, we calculated frequencies of masculine vs feminine vs neuter grammatical gender and abstract vs concrete noun. Since not all languages have three genders, we decided to consider masculine and neuter as a single category, to allow 2 (grammatical gender) × 2 (abstract vs. concrete) comparisons.

To test for associations between grammatical gender and abstractness, we ran Chi-square analyses for every language. As shown in Figure 2, grammatical gender and abstractness were associated in German, χ²(1, N = 240) = 55.11, p < .001, d = 1.09, Greek, χ²(1, N = 240) = 68.50, p < .001, d = 1.26, Hindi, χ²(1, N = 236) = 15.33, p < .001, d = 0.53, Irish, χ²(1, N = 238) = 29.67, p < .001, d = 0.76, Italian, χ²(1, N = 238) = 43.12, p < .001, d = 0.94, and Polish, χ²(1, N = 239) = 26.94, p < .001, d = 0.71, but not in Albanian, χ²(1, N = 240) = 1.65, p = .199, d = 0.15, and Lithuanian, χ²(1, N = 240) = 3.11, p = .078, d = 0.23.

Figure 2.

Percentage of Feminine vs. Masculine Nouns as a Function of Abstractness and Language (Study 1b).

Then, we first ran non-parametric binomial tests to check whether the frequency of feminine gender among abstract and concrete nouns was higher or lower than chance. Among concrete nouns, masculine grammatical gender reliably exceeded chance levels for German, p < .001; Greek, p < .001; Hindi, p = .002; Irish, p < .001; Italian, p = .006; and Lithuanian, p = .003. Among abstract nouns, feminine grammatical gender reliably exceeded chance levels for German, p < .001; Greek, p < .001; Hindi, p = .027; Irish, p < .001; Italian, p < .001; and Polish, p < .001.

This pattern was confirmed by independent sample t-tests, using the Bysbaert et al. (2014) concreteness ratings as dependent variable (see supplementary materials, which are available on the OSF: https://osf.io/8ysmx/).

As in Study 1a, feminine gender was numerically over-represented among abstract nouns, but the (relatively few) masculine nouns that were present among the abstract category were no less abstract than the feminine ones (ts ≤ |.90|, ps ≥ .368).

Discussion

Three main conclusions can be drawn from the results of Studies 1a and 1b. First and foremost, the results show that feminine nouns are over-represented among abstract, but not among concrete nouns, in seven (out of nine) Indo-European languages, namely German, modern Greek, Hindi, Irish, Italian, Latin, and Polish. An opposite pattern was observed for concrete nouns in most languages included in Study 1b. Together, our findings suggest a widespread, cross-language association between femininity and abstractness, which, in line with linguistic theory (Luraghi, 2009), presumably derives from the fact that feminine (vs. masculine) gender was historically superimposed on a previously existing classification of abstract (vs. concrete) inanimate concepts. Why this asymmetry was not found in two of the Indo-European languages investigated here, namely Albanian and Lithuanian, remains to be explored in future research; however, evidence suggests that for Lithuanian, the masculine grammatical gender is overrepresented in language and adjective-noun matching follows a very complex pattern, with a dominance of masculine and neutral forms, which may explain the lack of an effect (Adamson & Šereikaitė, 2019; Bruno, 2012).

Second, the Proto-Indo-European derivation of this phenomenon, as detailed above, may also explain why the link between feminine gender and abstractness, although present in both languages, is stronger in Latin than in Italian. Finally, the mediation model suggests that because they are more likely to be abstract, feminine nouns also present those characteristics that are typical of abstract concepts (Della Rosa et al., 2010): in virtue of their overrepresentation among abstract nouns, grammatically feminine nouns are more difficult to envisage and to contextualize, they are learned later in life, and are acquired more through verbal stimulation than through direct experience. This is also in line with the concreteness effect, suggesting that abstract (vs. concrete) words are encoded and memorized less well because they are less likely to activate context information and visual imagery (e.g., Jessen et al., 2000).

To sum up, Studies 1a and 1b provide evidence for the hypothesized abstractness–femininity association, yet, they do not speak to any psychological meaning of this association. In the following studies, we aimed at answering the question: why should the grammatical gender of abstract nouns be psychologically relevant?

Does Grammatical Gender of Abstract Nouns Accord Psychological Meaning? (Study 2)

An extensive line of research suggests that grammatical gender accords psychological meaning to concrete objects (Samuel et al., 2019; Stahlberg et al., 2007). For instance, bridge may elicit masculine qualities such as sturdiness in the minds of Spanish speakers, but feminine qualities such as elegance in the minds of German speakers. This phenomenon, denoted gender-congruency effect,² is believed to occur through three distinct routes: (a) the greater accessibility of gender-congruent exemplars during retrieval, (b) the activation of gender-stereotypes that are projected onto objects, and (c) the transmission of gendered images through cultural artifacts.

The first process concerns the fact that grammatical gender determines, to some degree, which examples come to mind. For instance, priming studies suggest that same-gender nouns become more accessible than opposite gender nouns (e.g., Bates et al., 1996). The greater accessibility of gender-congruent exemplars has tangible social consequences for opinion polls, which are often framed in generic masculine forms that favor the retrieval of male exemplars (Stahlberg et al., 2001; Stahlberg & Sczesny, 2001, Study 2). The common process likely to underline these phenomena is selective retrieval, during which grammatical gender channels the search for corresponding gender of the exemplar to be retrieved.

The second process involves the projection of stereotypically masculine or feminine characteristics onto objects, animals, or concepts on the basis of grammatical gender (for an overview, see Bender et al., 2016a). In these cases, the arbitrary grammatical gender of nouns carries over to the object itself. For instance, in Konishi’s (1993) classical study, Spanish and German speakers reacted to words whose grammatical gender differed across the two languages. Participants judged words as higher in potency when they carried masculine grammatical gender. Alternatively, studies may include semantic similarity judgments of word pairs that share or not the same gender. For instance, Vigliocco et al. (2005) found such a carry-over effect in Italian speakers, but not in German speakers. Thus, the overall evidence in this line of research is relatively weak, especially in languages with more than two genders (such as German).

Gender congruency emerges more clearly from studies using the voice-assignment task. Here participants are told that their opinion is needed for the development of a cartoon in which common objects come to life, by indicating whether a male or female voice is better suited to represent that object (e.g., Sera et al., 1994, Exp. 2). In languages with grammatical gender, voice assignment is greatly influenced by the gender of the word (for a review, see Bender et al., 2016a). Although grammatical gender is not the only variable at play (e.g., Mullen, 1990; Sera et al., 1994), it tends to spread to semantically genderless objects (namely nouns, both animate and inanimate, that are not inherently featured by a sex), so that the gender of the selected voice corresponds to the grammatical gender of the word.

One limit of this research is that, to our knowledge, almost all prior studies have investigated gendered images related to concrete nouns such as animals, plants, and artifacts. Among the few exceptions are studies investigating allegorical concepts such as state, innocence, peace, or honor, finding overall relatively small grammatical gender effects in voice assignment (Beller et al., 2015), or only finding an effect when allegories are paired to gender-congruent personified representations (Bender et al., 2016b). On the one hand, one may argue that it is easier to project gender onto concrete objects than onto highly abstract concepts, because abstract nouns are overall much harder to imagine (Della Rosa et al., 2010). On the other hand, the very fact that abstract concepts are not directly accessible to our senses and become comprehensible mainly through linguistic or visual metaphors (Landau et al., 2014) may actually promote personified (and possibly also gendered) mental representations of such concepts. Cultural artifacts such as paintings and sculptures fulfill exactly this function and constitute the third route through which grammatical gender may transmute into gender-congruent mental representations. For instance, Liberty is feminine in numerous gendered languages, including Latin, Italian, French, German, Spanish, ancient and modern Greek, and Polish, and it is usually also represented as a woman in Western cultures. Thus, one way in which grammatical gender can affect our cognition and imagination, are cultural artifacts informed by grammatical gender. However, how strong the direct language vs. the indirect cultural effect is on (social) cognition is still debated (Beller et al., 2015; Bender et al., 2018; Samuel et al., 2019).

Importantly, with the exception of personified allegorical concepts (e.g., Beller et al., 2015; Bender et al., 2016b; Segel & Boroditsky, 2011), relatively little is known about the gender congruency effect for abstract concepts. One study in Polish language (Maciuszek et al., 2019, Study 1) found no evidence for gender-congruency effect for abstract nouns, but, as mentioned by the authors, the experimental task (triadic similarity judgments) may have invited participants to use specific semantic criteria; moreover, the majority of the stimuli employed (e.g., sound, shape, color) were actually rather concrete according to the Brysbaert et al. (2014) database. Hence, the literature still lacks conclusive evidence on the applicability of the gender congruency effect to abstract nouns. A central aim of this research program was therefore to establish whether grammatical gender “rubs off” on abstract nouns as it does on concrete ones, and whether this association between abstractness and femininity spills over to the perception of abstract concepts as having “feminine” properties, much as it does for concrete nouns.

To address the question whether the grammatical gender of abstract nouns carries over to semantics, we investigated whether grammatically feminine abstract nouns are perceived as more feminine than masculine ones, in line with the gender congruency effect. The hypothesis was tested in two ways. First, blind raters indicated whether they pictured the abstract nouns included in Study 1a as feminine or masculine. Second, for a subsample of these abstract nouns, we also generated opposite-gender synonyms, which were then rated through the same procedure. In both cases, we expected grammatically feminine (vs. masculine) nouns to be perceived as representing more feminine (vs. masculine) concepts.

Study 2

Method

Raters and Word Stimuli. Twelve collaborators (seven women, five men), whose age ranged from 19 to 30 years old (M = 25, SD = 3.56), took part in this study, which was run on Qualtrics. Raters were blind to the hypotheses and had the possibility to complete the task in more than one session, within 10 days.

Due to technical issues, stimuli consisted of 204 of the 205 abstract nouns included in Study 1, together with 31³ opposite-gender synonyms, namely synonyms of a given noun that had the alternative grammatical gender (e.g., concept, that in Italian is grammatically masculine, and idea, that is grammatically feminine). Since we selected the first synonym presented by the dictionaries, two repetitions are present (i.e., “guilt,” “passion”). Thus, our final word sample of original nouns + opposite-gender synonyms consisted of 235 nouns.

Study Variables. Collaborators first rated the degree to which they considered each word (including both the abstract nouns and the 31 synonym pairs) to be feminine or masculine on a 0-to-100 slider (“based on the first images or associations that come to your mind when thinking about the concept they represent”). To avoid response biases, the order of presentation of nouns was randomized and the endpoint labels (“masculine” and “feminine”) were counterbalanced across raters. To reduce the salience of the grammatical gender, nouns were presented without the article so that grammatical gender was only revealed by the suffix. Subsequently, raters were shown the 31 noun + opposite-gender synonyms word pairs and asked to indicate how much the meaning of the two overlapped, on a slider from 0 to 100. Again, the order of the labels “overlapping” and “non-overlapping” was counterbalanced to avoid bias. Collaborators then indicated their gender, age, whether they were Italian native speakers, and what other languages they knew. Finally, they were asked to indicate the criteria on which they based their ratings, the options being “representation (e.g., artistic),” “association (e.g., social/cultural),” “grammatical gender,” and “other,” and they had the possibility to leave comments to the questionnaire or give examples of their reasoning in rating the words.

Results

Femininity–Masculinity Ratings of Nouns. Since there was a reasonable agreement among raters (α = .66), ratings were averaged across raters. Grammatically feminine nouns evoked more feminine images (M = 55.11, SD = 9.78) than grammatically masculine nouns (M = 42.55, SD = 8.95), t(233) = −9.75, p < .001, d = −1.32, 95% CI [−1.62, −1.03]. Comparing means to the scale midpoint (50), one-sample t-tests revealed that grammatically feminine nouns were reliably above, t(149) = 6.40, p < .001, d = 0.52, 95% CI [0.35, 0.69], and grammatically masculine nouns below the midpoint, t(84) = −7.68, p < .001, d = -0.83, 95% CI [−1.08, −0.58], suggesting that even in absolute terms nouns were perceived in line with their grammatical gender.

Opposite-Gender Synonyms. To test whether the same concept was perceived as more feminine when expressed through a feminine noun, we used the opposite-gender synonym pairs as word unit and calculated the mean femininity ratings separately for the masculine and the feminine word of the pairs (Figure 3). A paired-sample t-test showed that even similar concepts were perceived as more feminine when expressed through a grammatically feminine (M = 55.88, SD = 7.33) vs. masculine noun (M = 40.90, SD = 7.83), t(30) = −8.33, p < .001, d = −1.50, 95% CI [−2.01, −0.97]. Specifically, ratings for both feminine nouns, t(30) = 4.47, p < .001, d = 0.80, 95% CI [0.39, 1.20], and masculine nouns, t(30) = −6.47, p < .001, d = −1.16, 95% CI [−1.62, −0.70], differed significantly from the scale mid-point. Furthermore, the difference in femininity of grammatically masculine and feminine synonyms did not correlate with the perceived overlap between the two nouns, r(31) = .24, p = .196.

Figure 3.

Femininity Ratings for OGS (Study 2).

Discussion

Extending prior research on concrete nouns, the above study confirms the gender congruency effect for abstract nouns. Note that prior research on this issue had been rather inconclusive, in part because of the limited attention to abstract nouns, in part because the few existing studies failed to provide evidence for the gender-congruency effect for abstract nouns (see Maciuszek et al., 2019). In the present research, the effect was rather robust, suggesting that grammatical gender affects the mental representations of nouns, even when they are abstract. Furthermore, 10 out of 12 raters reported having based their ratings of femininity on social or cultural associations and were apparently unaware that grammatical gender may have played a role. However, given the explicit and transparent nature of the task, we cannot exclude that demand characteristics may have been at play. Also, Study 2 only speaks to the association between grammatical and semantic gender, but is silent as to the association between abstractness and semantic gender. To fill this gap, the next studies will therefore address the following question: Given the disproportional presence of grammatically feminine nouns among abstract concepts, are such abstract concepts associated with women above and beyond grammatical gender?

Is Abstractness Associated With Women as a Social Category? (Study 3)

In Studies 3a and 3b, we tested whether abstract (vs. concrete) nouns are more strongly associated with women (vs. men), extending our previous findings in three ways: First, different from Study 2 that assessed the association between grammatical gender and masculinity/femininity, we here investigate the association between abstractness and gender. Given that abstract nouns are overwhelmingly feminine (Study 1) and that grammatical gender of abstract concepts is associated with femininity vs. masculinity (Study 2), one may suspect that people implicitly associate abstractness with women. This prediction rests on the assumption that the repeated co-occurrence of concepts leads to the formation of automatic associations between them. Thus, through repeated pairings, abstractness may come to be associated with women much in the same way in which space becomes associated with gender (Suitner et al., 2017), time (Fuhrman & Boroditsky, 2007), or numbers (Fias & Fischer, 2005), or in which taste becomes associated with colors (Higgins & Hayes, 2019). Second, Study 3 investigates associations of abstract concepts with gender categories (men vs. women) rather than with gendered properties (masculinity vs. femininity). Third, contrary to Study 2 employing a blatant measure, we used an Implicit Association Test (IAT; Greenwald et al., 1998) in Studies 3a and b to assess the implicit (rather than reasoned) association between abstractness and gender categories.

Together, the following two studies (3a and 3b) aimed at testing whether women as a social group are implicitly associated with abstract concepts. Using an IAT, we predicted that women (vs. men) would be associated more strongly with abstract (vs. concrete) concepts (for previous use of the IAT in the context of the gender congruency effect, see Maciuszek et al., 2019).

Study 3a

Method

Participants. Italian participants who were also native Italian speakers were recruited through Prolific Academic and paid 1.35£ for their participation in the study. After exclusion of one participant whose reaction times to the IAT were too slow, our final sample consisted of 159 participants, including 78 women, 80 men, and one non-binary person (M_age = 28.52, SD_age = 8.87).

IAT. The IAT was built through iatgen (Carpenter et al., 2019) in the Shiny version. Abstract and concrete nouns were treated as categories/attributes (abstract being the positive attribute and concrete being the negative attribute), whereas faces of men and women were treated as targets (women being target A and men being target B). This way, positive D-scores would indicate a greater association of women with abstract nouns and men with concrete nouns, whereas negative scores would indicate the opposite. Participants were instructed to classify the pictures as “man” or “woman,” and the words as “abstract” and “concrete,” as fast as they could but making the least possible mistakes, by pressing the “E” and “I” keys on their keyboard. They were also informed they could take breaks between the different sections of the IAT.

Visual Stimuli. For visual stimuli, photos of White men and women with neutral facial expressions were extracted from the Chicago Face Database (Ma et al., 2015).

Word Stimuli. Stimuli were selected from the Brysbaert database (Brysbaert et al., 2014) through six steps. First, we extracted words that were 100% known with frequency > 500 (both provided by the Brysbaert et al. database).

Second, we split the database into concrete (concreteness ratings ≥ 4) and abstract nouns (concreteness ratings ≤ 3), randomized the words within the two lists, and selected 100 concrete and 101 abstract nouns to evaluate and translate.

Third, we excluded (a) nouns related to people (e.g., “wizard”) or body parts (e.g., “body”); (b) human-related activities (e.g., “bowling”); (c) plural nouns (e.g., “politics”), (d) group nouns (e.g., “team”); (e) profanities, swearwords, and pornography references (e.g., “motherfucker”); (f) culturally bound words (e.g., “penny”); (g) loanwords (e.g., “privacy”); and (h) opaque/ambiguous suffixes (e.g., -e, -a for masculine nouns, -o for feminine nouns).

Fourth, nouns were translated into Italian through the same dictionary employed in Study 2; we selected alternative translations when the primary translation listed in the dictionary: (a) was a compound noun (e.g., “scholarship,” namely “borsa di studio”); (b) had multiple meanings (e.g., the Italian word for “blade” also indicates an animal); (c) was plural; (d) was archaic; (e) was a vezzeggiativo (Italian endearment suffix; e.g., “bunny”) or had a diminutive suffix; (f) was already present among our stimuli (e.g., “heaven” and “paradise,” which have the same translation). If no alternative translation met these criteria, the noun was excluded. The translations were checked by an English–Italian bilingual speaker who was blind to the hypotheses, with the instruction to identify any imprecise translations, which were deleted.

Fifth, we excluded additional nouns that met all criteria to achieve a similar number of stimuli per intersection between grammatical gender and abstract vs. concrete concept. Our final 111 stimuli (28 abstract feminine, 27 abstract masculine, 28 concrete feminine, 28 concrete masculine) are available in the online supplementary materials (https://osf.io/8ysmx/).

Sixth, specifically for the IAT, we included only words that were between 6 and 9 letters long and grouped them randomly into three sets of five words each, evenly split between masculine and feminine, concrete and abstract. Each set (five masculine abstract, five masculine concrete, five feminine abstract, and five feminine concrete) was matched in valence, using ratings by Warriner et al. (2013), so that for the four categories, the averages of valence were within a range of 1.5 points (on a scale going from 1 to 9).

Experimental Manipulation. To test whether the association between womanhood and abstractness would be driven by grammatical gender, participants were randomized into a balanced condition or a representative condition. In the balanced condition, the IAT word stimuli included an even amount of masculine and feminine words. In the representative condition we followed the same procedure, but rather than having five masculine and five feminine abstract nouns in each set, we selected additional abstract feminine stimuli (see step 6) to achieve the 80–20 gender ratio that emerged in Study 1b⁴: therefore, we selected only two abstract masculine nouns and eight abstract feminine nouns (rather than five and five).

Results

The IATs were processed through iatgen, through which we obtained the D-scores. To test whether participants showed a stronger association for women-abstractness than for women-concreteness, we performed a one-sample t-test on the D-scores. The D-scores (M = 0.14, SD = 0.37) indicated that, overall, participants associated women more with abstract than concrete words, t(158) = 4.85, p < .001, d = 0.39, 95% CI [0.22, 0.55]. This was true for both the balanced, t(80) = 3.37, p = .001, d = 0.38, 95% CI [0.14, 0.60] (M = 0.14, SD = 0.38), and the representative condition, t(77) = 3.48, p < .001, d = 0.39, 95% CI [0.16, 0.62] (M = 0.14, SD = 0.35), with no difference between the two conditions, t(157) = 0.08, p = .933, d = 0.01, 95% CI [-0.30, 0.32]. Unexpectedly but interestingly, analyses run separately by participant gender revealed that the effect was driven primarily by men (M = 0.34, SD = 0.30), t(79) = −10.15, p < .001, d = 1.14, 95% CI [0.85, 1.42], while it was absent for women (M = −0.06, SD = 0.30), t(77) = −1.83, p = .071, d = −0.21, 95% CI [−0.43, 0.02].

Study 3b

Study 3a revealed an implicit association between women and abstract concepts, suggesting that the abstractness–femininity association may generalize to womanhood. However, unexpectedly, this effect was only present in men. Therefore, in Study 3b we tested whether this implicit association was explained by higher levels of benevolent sexism (Glick & Fiske, 1996): perceiving women as associated with abstract (e.g., virtues) and men with concrete concepts is in line with the complementary gender differentiation that is part of benevolent sexism, a stereotypical view that is most commonly held by men (Glick & Fiske, 1996).

Method

Participants. Italian participants who were also native Italian speakers were recruited through Prolific Academic and paid 1.50£ for their participation in the study. After data cleaning by the iatgen Shinyapp (see Carpenter et al., 2019), our sample consisted of 197 participants, including 96 women, 98 men, and three non-binary people (M_age = 30.40, SD_age = 8.98).

Procedure. We employed the three IAT versions included in the balanced condition from Study 3a. Additionally, benevolent and hostile sexism were assessed through the Ambivalent Sexism Inventory (Glick & Fiske, 1996), using a scale from 1—Not at all agree to 7—Completely agree (α_HS = .90; α_BS = .85). Order of IAT and sexism questionnaires was randomized.

Results

The D-scores (M = 0.13, SD = 0.41) indicated that, overall, participants associated women more with abstract than concrete words, t(197) = 4.40, p < .001, d = 0.31, 95% CI [0.17, 0.46], but consistently with Study 3a, the effect was driven primarily by men (M = 0.31, SD = 0.38), t(97) = 8.21, p < .001, d = 0.83, 95% CI [0.60, 1.06], while it was absent for women (M = −0.06, SD = 0.36), t(95) = −1.53, p = .131, d = −0.16, 95% CI [−0.36, 0.05].

Contrary to our hypotheses, however, after accounting for gender, neither benevolent nor hostile sexism predicted the women–abstractness association (ps > .547, see supplementary materials for full model, which are available on the OSF: https://osf.io/8ysmx/). Therefore, even though men (M = 2.89, SD = 1.35) reported greater hostile (but not benevolent) sexism than women (M = 1.88, SD = 1.02), t(183) = 5.94, p < .001, d = .85, 95% CI [0.55, 1.14], this did not explain why men associate abstract concepts more to womanhood, and concrete concepts more to manhood, while women do not (although interestingly, men and women show opposite correlations between sexism and d-scores; see supplementary materials, which are available on the OSF: https://osf.io/8ysmx/).

Discussion

If abstract nouns are predominantly feminine in Indo-European languages (as shown in Studies 1a and 1b), and if grammatical gender affects mental representations through the gender congruency effect (as shown in Study 2), then people should associate women with abstractness. This prediction was supported by Studies 3a and 3b, adding to the literature by showing that women, as a social category, are implicitly associated with abstract concepts, at least by men. Why only men, but not women, associate abstractness with womanhood remains an open question for future research. Our hypothesis that differences in sexism may drive this effect was not supported (Study 3b), in line with an environmentalist model of IATs by which the IAT, and implicit bias more generally, does not reflect individual attitudes, but rather cultural associations that individuals have been exposed to (e.g., Karpinski & Hilton, 2001; Payne & Hannay, 2021). Nevertheless, we can envisage at least two alternative interpretations for the fact that only men associated women with abstractness, based, respectively, on ingroup bias and gender categorization. On the one hand, men may value concreteness more than women and therefore associate men with concreteness and women with abstractness more than women do. On the other hand, the presence of the abstractness–womanhood association in men, but not women, may reflect the fact that, compared to women, men are more likely to construe gender as a dichotomous (Bosson & Michniewicz, 2013) and essentialist, natural category (Morgenroth et al., 2021; Şahin & Soylu Yalcinkaya, 2021),

Another limit of Studies 3a,b lies in the use of the classic IAT that, in the absence of raw data (which is our case), does not allow to distinguish whether the bias was driven by the women–abstract association, or the men–concrete association. Only future studies will be able to disentangle the two, for example, through single-target IATs.

Study 4: Abstract Nouns Are Superhuman

Given that women are associated with abstractness, at least by men (Studies 3a and b), one may wonder whether there are psychological consequences that may derive from this association. In particular, may this association promote the idea of women as super-human beings?

Attitudes toward women tend to be greatly polarized. From a historical and cultural perspective, women in the Western world have been represented as either whores or saints, a traditional divide that can be traced back to Greek and Roman times. Over the centuries, this polarized view of women in Christianity, which is well symbolized by the figure of Mary Magdalena, has remained prevalent in the arts, in fiction (Brown, 2006), theater (Bienias, 2011), music (Richter, 2017), and in cinematography (Sabine, 2013). Psychological research confirms this attitudinal polarization and synthesizes it in the Madonna-whore belief (Bareket et al., 2018): a strict dichotomy, according to which women are classified as either Madonnas (chaste, pure, communal, lacking agency, and worthy of love) or Whores (promiscuous, unfit for motherhood and care, sexually objectified, and recipients of sex). This subtyping of the female universe—into those to be put on a pedestal and those equated to objects or subhuman beings—was shown to be correlated, in both men and women, to system justification, hostile and benevolent sexism, sexist double standard, and sexual objectification (Bareket et al., 2018; Kahalon et al., 2019). Over the past decades, social–psychological research has paid much more attention to the latter than to the former, showing that, when sexualized, women tend to be dehumanized and associated with either animals or objects (e.g., Rudman & Mescher, 2012; Vaes et al., 2011).

By contrast, relatively little is known about the super-humanization of women. Throughout European history, women and girls have often been seen as possessing super-natural powers, and they seem over-proportionally represented among oracles (having prophetic abilities) and in witchcraft (practicing evil spells), including witch trials in Colonial America (see Salem witch trials, Reis, 1997). At times, prominent women (possibly including the mathematician and philosopher Hypatia) were accused of witchcraft (see Janowitz, 2001).

From a psychological perspective, women are often attributed exceptional moral and intellectual skills (for an overview, see Eagly & Mladinic, 1994). This phenomenon, which later became known as “women-are-wonderful effect,” emerges mainly on stereotypically feminine characteristics (Glick et al., 2004; Krys et al., 2018). It can be observed, for example, in the myth of the superwoman who successfully manages multiple roles in society (Nicolson, 2003; Whitty, 2001), in the widely held belief that women are better than men at multi-tasking (Szameitat et al., 2015), and in the belief, within benevolent sexism, that women possess a unique “purity” (Glick & Fiske, 1997). Thus, with the exception of fictional superheroes (which outnumber female superheroines by a ratio of at least 2 to 1; Miller et al., 2016), women are often attributed super-human abilities: even scientists may refer to women in management as potential superheroes (Adams, 2016).

The main aim of Study 4 was to investigate the super-humanization of female nouns. We predicted that abstract and feminine nouns would be perceived as more super-human than concrete and masculine nouns, respectively; thus, abstract feminine nouns would be those with higher super-humanization ratings. Two additional pilots for this study, showing similar patterns, are available in the online supplementary materials (https://osf.io/8ysmx/).

As a secondary aim, we also intended to replicate the results of Study 2 with a subtler task. Even though only one rater reported having based her ratings on grammatical gender, the task employed in Study 2 explicitly prompted a decision related to masculinity vs. femininity, thereby making gender particularly salient (for a methodological critique, see Samuel et al., 2019). Thus, we aimed to solve this issue by investigating the gender congruency effect through a voice-assignment task, that is somewhat more implicit.

Study 4

Method

Participants and Word Stimuli. Raters were recruited through Prolific Academic. After exclusion of seven participants who failed at least one attention check, 95 native Italian speakers (46 women, 48 men, one non-binary person), whose age ranged from 21 to 65 years old (M = 23.64, SD = 10.23) took part in this study, which was run on the online survey platform Qualtrics.

Word stimuli were selected following the same procedure as Study 3a, but omitting step 6, which was not relevant to the present study.

Procedure. Raters were told that our research group was designing an educational cartoon to teach elementary school children the meaning of certain words that would be represented by cartoon characters. For this reason, raters were told that the aim of this data collection was to select the design and voice actor for these characters, which were to represent both concrete (e.g., frog, clock) and abstract (e.g., luck, oblivion) words. To avoid fatigue, each rater was asked to judge only a subset (about 28) of our stimuli. For this and the subsequent task concrete and abstract words were presented together in random order. Participants were first asked to imagine what these characters would look like and decide if they should be voiced by a female or male voice actor, that is, if a male or female voice would be better suited for the cartoon character, on a Likert scale from 1—voice actor to 6—voice actress (labels were counterbalanced across participants).

To assess super-humanization, two measures were used (voice super-humanization and categorization of abstract nouns). In the former, participants were asked whether they imagined the voice to be sub-human (described as “a voice similar to animal sounds, or synthesized and distorted, as if it was less developed than human language”), human (described as “a realistic voice, like that of an ordinary person”), or super-human (described as “a surreal voice, similar to one a super-human entity might have, as if it was superior to human language”). In the categorization task, participants were asked to classify all abstract nouns (randomly ordered) into six categories, based on how they imagined the character: “object” and “mechanical being (e.g., robot, android)” which represented mechanization, “force of nature (e.g., hurricane, plant)” and “animal” which represented animalization,⁵ “person” which represented humanization, and “superhuman entity (e.g., spirit, deity)” which represented super-humanization.

Finally, participants reported gender, age, and whether they spoke additional languages fluently.

Results

Perceived Femininity. A 2 (grammatical gender) × 2 (concrete vs abstract) ANOVA showed a significant effect of grammatical gender, F(1,107) = 80.60, p < .001, η_p² = .43, with grammatically feminine nouns (M = 3.88, SD = 0.51) being perceived as more feminine than masculine ones (M = 3.01, SD = 0.53). Furthermore, abstract nouns (M = 3.56, SD = 0.66) were perceived as more feminine than masculine ones (M = 3.34, SD = 0.68), F(1,107) = 5.20, p = .025, η_p² = .05. The interaction was not significant, F(1,107) = 2.70, p = .103, η_p² = .03, suggesting that the gender congruency effect was similarly present for concrete and abstract nouns, contrary to previous evidence on Polish using a different task (Maciuszek et al., 2019). Therefore, this study confirmed the gender congruency effect for abstract nouns found in Study 2 and moreover shows that it is of similar magnitude as that found for concrete nouns.

Super-Humanization. To test whether abstract feminine nouns were primarily categorized as super-human, we first ran a word type (abstract feminine vs. abstract masculine vs. concrete feminine vs. concrete masculine) by category (subhuman vs. human vs. superhuman) chi² analysis. Results showed a significant association between the two variables, χ²(6, N = 2636) = 216.74, p < .001, d = .60 (Figure 4, upper panel). Concrete nouns were imagined as more subhuman, but there was no difference between the two grammatical gender forms, χ²(1, N = 1330) = 1.39, p = .238, d = .07. Abstract nouns were categorized more often as super-human, although in this case feminine abstract nouns were imagined as super-human more frequently than masculine abstract ones, χ²(1, N = 1306) = 5.34, p = .020, d = .13.

Figure 4.

Super-Humanization as a Function of Grammatical Gender and Abstractness (Study 4).

Looking at the third task, namely the categorization of abstract nouns as objects, animals or natural forces, humans, or super-human entities, results showed a similar pattern (Figure 4, lower panel). Again, the association was significant, χ²(3, N = 1306) = 21.81, p < .001, d = .26. In particular, there was no difference in the categorization of masculine and feminine abstract nouns as human, χ²(1, N = 1306) = 1.14, p = .286, d = .06; but, masculine abstract nouns were associated with a dehumanized imagery more frequently than feminine ones, χ²(1, N = 1306) = 12.38, p < .001, d = .20, whereas feminine abstract nouns were associated with a super-human imagery more frequently than masculine ones, χ²(1, N = 1306) = 6.24, p = .013, d = .14.

To explore whether these results may be driven by potential differences in valence, we verified whether there was a difference in valence between masculine and feminine, concrete and abstract nouns: independent-samples t-tests showed this was not the case (ps ≥ .096). In line with this, valence did not correlate with concreteness ratings, r(111) = .04, p = .714. Nevertheless, valence correlated with perceived femininity, r(111) = .30, p = .001.

Discussion

Study 4, once again, confirmed the gender congruency effect for abstract nouns; it also showed that the magnitude of this effect resembles gender-congruency of concrete ones. Additionally, abstract feminine nouns were generally not perceived as more feminine than concrete ones, suggesting again that we are dealing with an epiphenomenon: any difference between abstract and concrete nouns has an effect only in virtue of the grammatical overlap between abstractness and feminine grammatical gender.

Most importantly, however, Study 4 provides first evidence that abstract nouns are super-humanized, and that this is even stronger for grammatically feminine nouns. Therefore, there is preliminary evidence that the association between women and abstractness may play a part in the super-humanization of women, possibly upholding the Madonna–whore dichotomy. Importantly, as in Study 3a and 3b, the results on super-humanization were not confounded by the valence of the stimuli themselves: valence was not predicted by grammatical gender, nor did it correlate with concreteness ratings.

General Discussion

Across six studies, we discuss the prevalence and the psychological implications of the association between abstractness and feminine grammatical gender in Indo-European languages. These questions were addressed in a four-phase research project. In the first phase (Studies 1a and b) we tested the underlying assumption that abstract nouns carry feminine grammatical gender far more frequently than would be expected by chance, an asymmetry that holds for the majority of Indo-European languages investigated here, regardless of whether they have two (masculine, feminine) or three genders (masculine, feminine, neutral). Why two of the nine languages do not show this pattern remains a question for future (linguistic) inquiry.

In the second phase, we explored the psychological consequences of this grammatical regularity, finding a gender-congruency effect for abstract nouns, similar to that found in past research for concrete nouns. Grammatically feminine [vs. masculine] abstract nouns were found to be associated with femininity [vs. masculinity] by raters asked to imagine these concepts (Study 2). This was true even when judging synonyms of opposite grammatical gender (e.g., idea_f - concept_m; fate_{f −} destiny_m; importance_{f −} value_m; reason_{f −} intellect_m). Similarly, naïve participants in Study 4 over-proportionally selected female voices to represent grammatically feminine abstract concepts and male voices to represent grammatically masculine abstract concepts. In these studies, the gender-congruency effect emerged robustly despite the fact that abstract nouns were presented without articles, which are arguably the strongest marker of grammatical gender. Besides confirming our contention that the majority of abstract nouns will be seen as (conceptually) feminine, these results also extend prior research that had not produced conclusive evidence for the gender–congruency effect for highly abstract concepts. With the exception of personified allegorical concepts (e.g., Beller et al., 2015; Bender et al., 2016b; Segel & Boroditsky, 2011), research on abstract concepts has failed to find evidence for gender congruency (Maciuszek et al., 2019, Study 1). In contrast, our studies 2 and 4 provide coherent evidence that the grammatical gender of abstract nouns carries psychological meaning, such that grammatical gender is projected onto gendered perceptions of the concept itself.

In Studies 3a and b, we took this argument one step further by testing whether the above association would also be reflected in an implicit psychological association between abstractness and woman as a social category. Both studies confirmed the existence of the abstractness–women association, although it was reliable only for male participants, possibly due to additional biases, such as a more binary vision of gender among men.

Finally, Study 4 showed that abstract nouns are associated with super-human imagery, and that this tendency is particularly strong for grammatically feminine abstract nouns. Importantly, these effects were not confounded by valence, in line with the fact that the abstract–feminine association encompasses both positive and negative values. For instance, the four cardinal virtues (prudence, justice, fortitude, and temperance) are grammatically feminine in many Indo-European languages, and they also tend to be depicted as women in the arts. But so are the vices (e.g., pride, greed, wrath, envy), for which, again, female allegorical figures abound in Western arts. Thus, the super-humanization of women seems to encompass the good and the bad, possibly due to the prevalence of feminine gender among abstract concepts, which include virtues and vices.

Together, our studies suggest that abstract concepts, including “super-human” qualities, are associated with femininity both grammatically and psychologically. Although it is premature to speculate about the applied implications of this phenomenon, we can envision two possible cascading effects of this linguistic phenomenon on attitudes toward women. The first possibility is that the association between abstractness and femininity may reinforce sexist attitudes by contributing to a dichotomized view of women as either saints or whores—by perpetuating the notion that womanhood is, in itself, idealized and superhuman. Furthermore, it is conceivable that it hinders women's participation in any activity that requires concreteness and action, such as the workplace. The second possibility is that, to the contrary, this association may actually counter sexist attitudes: particularly in light of the Madonna–whore dichotomy, associating womanhood with super-humanness may inhibit the view of women as sexual objects, possibly limiting societal phenomena such as the sexualization and sexual objectification of women. Future studies should address these, and other, possible consequences of this linguistic phenomenon on attitudes toward women, for example, by testing whether the association between women and abstractness (Studies 3a and 3b) correlates positively with the belief in the Madonna–whore dichotomy or whether the over-representation of feminine nouns among abstract ones in a language correlates with country-level ambivalent sexism or gender inequality.

Limitations and Future Development

Being the first to investigate the abstractness–femininity association, this research necessarily has a number of limitations. First, although suggesting a possible link between abstractness, grammatical gender, and the social construal of gender, our studies did not determine whether the gender congruency effect for abstract nouns is driven primarily by the linguistic path or by the cultural representations path (e.g., through cultural artifacts such as paintings and statues). Future research should disentangle the two, by testing whether the abstractness–femininity association is stronger for abstract concepts (such as virtues and vices) that are widely represented as women in cultural imagery.

Second, although providing first suggestive evidence for the role of language in the super-humanization of women, the present work did not determine any specific forms of super-humanization as they occur in real life. As suggested above, future studies should investigate this issue further, for instance by experimentally testing whether the abstractness–femininity association may trigger or sustain idealizing conceptions of women as Madonna and the like.

Third, when investigating the psychological implications of the abstract = feminine phenomenon, we here focused on super-humanization, but did not explore other consequences. Abstract constructs such as democracy, oppression or injustice are generally learned through linguistic (rather than sensorial or motor) experiences (Barca et al., 2020). The fact that these complex concepts carry feminine gender may well have other, and possibly more positive, implications for the social construal of gender, as well as for the construal of the concepts in question. Future research may explore some of these non-obvious implications of the abstractness–femininity link.

Finally, our research project focused exclusively on Indo-European languages, while ignoring grammatical-gender languages of other language families, such as Arabic. In the absence of empirical evidence, we suspect that the abstract = feminine phenomenon will emerge only in those languages in which grammatical gender became superimposed on a previously existing word classes that distinguished concrete from abstract or collective nouns (as is the case for Indo-European languages).

Conclusions

Although much is known about the effect of grammatical gender on the mental representations of women and men, one regularity of Indo-European languages has largely been overlooked, namely the fact that abstract concepts tend to be grammatically feminine.

We explored some of the potential consequences of this unique association between abstractness and feminine gender by showing that feminine abstract nouns, which greatly outnumber masculine ones, are indeed seen as psychologically feminine, resulting in an implicit association of women with abstractness (at least by male participants). Although the association of women with the most abstract concepts of human thought (such as justice or liberty) may give rise to optimistic interpretations, it may also imply risks, ultimately contributing to the polarized view of women as either sub- or superhuman.

Footnotes

Acknowledgments

We are grateful to Eleni Lipourli, Magdalena Formanowicz, Anisa Shyti, Rachel O’Driscoll, and Namrata Goyal for providing help with the translations of the stimuli in Study 1b, Elizabeth Maria Doerr for providing help with the translation of the stimuli in Study 4, and the raters who collaborated with us in Study 2.

ORCID iDs

Carmen Cervone

Andrea Scatolon

Valentina Cimmino Picone

Anne Maass

Ethical Approval and Informed Consent

Studies involving human participants were approved by the New York University Abu Dhabi Institutional Review Board [HRPP-2023-214].

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability

Data, study materials, and supplementary materials are available on the OSF: .

Notes

Author Biographies

Carmen Cervone is a postdoctoral research fellow at the University of Padua. Her research focuses on the psychological processes that maintain social inequalities: in particular, her interests include the impact of sexist language on gender inequality and gender-based violence, as well as the moralization of economic inequality.

Ilaria Motta is a researcher in the fields of addiction and mental health. Holding a Master's degree in community psychology, she works in the research and development unit of CEIS Formazione, the training and research spin-off of the CEIS Group in Modena. Her work is grounded in the organization's 40+ years of direct field experience in addressing social distress.

Andrea Scatolon is a research fellow at the University of Padua (Italy). Her research interests include framing strategies (a) to reduce meat consumption, (b) to increase the consumption of seasonal and local products, and (c) to increase tax compliance. Her interests also include the moralization of the need to reduce economic inequality and non-binary gender inclusion.

Valentina Cimmino Picone is a PhD candidate in social sciences at the Department of Philosophy, Sociology, Education, and Applied Psychology (FISPPA), University of Padua. Her research focuses on the psychotherapy process, with particular interest in identification dynamics within interpersonal relationships and their somatic underpinnings.

German-born and raised, US-trained, and Italian by choice, Anne Maass is currently a professor of social psychology at NYU in Abu Dhabi, UAE, and a professor emerita at the University of Padua, Italy, where she had spent most of her career. She has been associate editor of the Journal of Personality and Social Psychology and chief editor of the European Journal of Social Psychology. Her main research interests concern the intersection between language and social cognition and the psychology of economic inequality.

References

Adams

R. B.

(2016). Women on boards: The superheroes of tomorrow? The Leadership Quarterly, 27(3), 371–386. https://doi.org/10.1016/j.leaqua.2015.11.001

Adamson

Šereikaitė

(2019). Representing gender and defaults: Evidence from Lithuanian. Glossa: A Journal of General Linguistics, 4(1), 121. https://doi.org/10.5334/gjgl.860

Barca

Mazzuca

Borghi

A. M.

(2020). Overusing the pacifier during infancy sets a footprint on abstract words processing. Journal of Child Language, 47(5), 1084–1099. https://doi.org/10.1017/s0305000920000070

Bareket

Kahalon

Shnabel

Glick

(2018). The madonna-whore dichotomy: Men who perceive women's nurturance and sexuality as mutually exclusive endorse patriarchy and show lower relationship satisfaction. Sex Roles, 79, 519–532. https://doi.org/10.1007/s11199-018-0895-7

Bates

Devescovi

Hernandez

Pizzamiglio

(1996). Gender priming in Italian. Perception & Psychophysics, 58(7), 992–1004. https://doi.org/10.3758/BF03206827

Beller

Brattebø

K. F.

Lavik

K. O.

Reigstad

R. D.

Bender

(2015). Culture or language: What drives effects of grammatical gender? Cognitive Linguistics, 26(2), 331–359. https://doi.org/10.1515/cog-2014-0021

Bender

Beller

Klauer

K. C.

(2016a). Crossing grammar and biology for gender categorisations: Investigating the gender congruency effect in generic nouns for animates. Journal of Cognitive Psychology, 28(5), 530–558. https://doi.org/10.1080/20445911.2016.1148042

Bender

Beller

Klauer

K. C.

(2016b). Lady Liberty and Godfather Death as candidates for linguistic relativity? Scrutinizing the gender congruency effect on personified allegories with explicit and implicit measures. Quarterly Journal of Experimental Psychology, 69(1), 48–64. https://doi.org/10.1080/17470218.2015.1021701

Bender

Beller

Klauer

K. C.

(2018). Gender congruency from a neutral point of view: The roles of gender classes and conceptual connotations. Journal of Experimental Psychology: Learning, Memory, and Cognition, 44(10), 1580–1608. https://doi.org/10.1037/xlm0000534

10.

Bienias

(2011). Virgins or whores? The image of a renaissance woman in parlour games in William Shakespeare’s Hamlet and Othello. Acta Philologica, 39(39), 29–36.

11.

Bosson

J. K.

Michniewicz

K. S.

(2013). Gender dichotomization at the level of ingroup identity: What it is, and why men use it more than women. Journal of Personality and Social Psychology, 105(3), 425–442. https://doi.org/10.1037/a0033126

12.

Braun

Sczesny

Stahlberg

(2005). Cognitive effects of masculine generics in German: An overview of empirical findings. Communications, 30(1), 1–21. https://doi.org/10.1515/comm.2005.30.1.1

13.

Brown

(2006). Tis pity she's a whore: The revision of Mary Magdalene in contemporary fiction. Papers on Language and Literature, 42(3), 291–316.

14.

Brugmann

(1891). Zur Frage der Entstehung des grammatischen Geschlechtes. Beiträge zur Geschichte der deutschen Sprache und Literatur (PBB), 15, 523–531.

15.

Bruno

(2012). The assignment of grammatical and inherent gender to English loan words in Lithuanian discourse. The ITB Journal, 13(1), 4. https://doi.org/10.21427/D7174F

16.

Brysbaert

Warriner

A. B.

Kuperman

(2014). Concreteness ratings for 40 thousand generally known English word lemmas. Behavior Research Methods, 46(3), 904–911. https://doi.org/10.3758/s13428-013-0403-5

17.

Carpenter

T. P.

Pogacar

Pullig

Kouril

Aguilar

LaBouff

Isenberg

Chakroff

(2019). Survey-software implicit association tests: A methodological and empirical analysis. Behavior Research Methods, 51(5), 2194–2208. https://doi.org/10.3758/s13428-019-01293-3

18.

DeFranza

Mishra

(2020). How language shapes prejudice against women: An examination across 45 world languages. Journal of Personality and Social Psychology, 119(1), 7–22. https://doi.org/10.1037/pspa0000188

19.

Della Rosa

P. A.

Catricalà

Vigliocco

Cappa

S. F.

(2010). Beyond the abstract—concrete dichotomy: Mode of acquisition, concreteness, imageability, familiarity, age of acquisition, context availability, and abstractness norms for a set of 417 Italian words. Behavior Research Methods, 42(4), 1042–1048. https://doi.org/10.3758/BRM.42.4.1042

20.

Eagly

A. H.

Mladinic

(1994). Are people prejudiced against women? Some answers from research on attitudes, gender stereotypes, and judgments of competence. European Review of Social Psychology, 5(1), 1–35. https://doi.org/10.1080/14792779543000002

21.

Fias

Fischer

M. H.

(2005). Spatial representation of numbers. In The handbook of mathematical cognition (pp. 43–54). Psychology Press.

22.

Fuhrman

Boroditsky

(2007). Mental time-lines follow writing direction: Comparing English and Hebrew speakers. In Proceedings of the annual meeting of the cognitive science society (Vol. 29, No. 29) (pp. 1007–1011). Cognitive Science Society.

23.

Glick

Fiske

S. T.

(1996). The ambivalent sexism inventory: Differentiating hostile and benevolent sexism. Journal of Personality and Social Psychology, 70(3), 491–512. https://doi.org/10.1037/0022-3514.70.3.491

24.

Glick

Fiske

S. T.

(1997). Hostile and benevolent sexism. Psychology of Women Quarterly, 21(1), 119–135. https://doi.org/10.1111/j.1471-6402.1997.tb00104.x

25.

Glick

Lameiras

Fiske

S. T.

Eckes

Masser

Volpato

Manganelli

A. M.

Pek

J. C. X.

Huang

Sakalli-Uğurlu

Castro

Y. R.

D'Avila Pereira

M. L.

Willemsen

T. M.

Brunner

Six-Materna

Wells

(2004). Bad but bold: Ambivalent attitudes toward men predict gender inequality in 16 nations. Journal of Personality and Social Psychology, 86(5), 713–728. https://doi.org/10.1037/0022-3514.86.5.713

26.

Greenwald

A. G.

McGhee

D. E.

Schwartz

J. L.

(1998). Measuring individual differences in implicit cognition: The implicit association test. Journal of Personality and Social Psychology, 74(6), 1464. https://doi.org/10.1037/0022-3514.74.6.1464

27.

Gygax

Sato

Öttl

Gabriel

(2021). The masculine form in grammatically gendered languages and its multiple interpretations: A challenge for our cognitive system. Language Sciences, 83, 101328. https://doi.org/10.1016/j.langsci.2020.101328

28.

Haslam

(2006). Dehumanization: An integrative review. Personality and Social Psychology Review, 10(3), 252–264. https://doi.org/10.1207/s15327957pspr1003_4

29.

Higgins

M. J.

Hayes

J. E.

(2019). Learned color taste associations in a repeated brief exposure paradigm. Food Quality and Preference, 71, 354–365. https://doi.org/10.1016/j.foodqual.2018.08.013

30.

Janowitz

(2001). Magic in the Roman world: Pagans, Jews and Christians. http://ci.nii.ac.jp/ncid/BA53085843.

31.

JASP Team . (2025). JASP (Version 0.19.3) [Computer software].

32.

Jessen

Heun

Erb

Granath

Klose

Papassotiropoulos

Grodd

(2000). The concreteness effect: Evidence for dual coding and context availability. Brain and Language, 74(1), 103–112. https://doi.org/10.1006/brln.2000.2340

33.

Kahalon

Bareket

Vial

A. C.

Sassenhagen

Becker

J. C.

Shnabel

(2019). The Madonna-whore dichotomy is associated with patriarchy endorsement: Evidence from Israel, the United States, and Germany. Psychology of Women Quarterly, 43(3), 348–367. https://doi.org/10.1177/0361684319843298

34.

Kapović

Ramat

A. G.

Ramat

(2017). The Indo-European languages. Routledge. https://doi.org/10.4324/9781315678559

35.

Karpinski

Hilton

J. L.

(2001). Attitudes and the implicit association test. Journal of Personality and Social Psychology, 81(5), 774. https://doi.org/10.1037/0022-3514.81.5.774

36.

Konishi

(1993). The semantics of grammatical gender: A cross-cultural study. Journal of Psycholinguistic Research, 22(5), 519–534.

37.

Krys, K., Capaldi, C. A., van Tilburg, W., Lipp, O. V., Bond, M. H., Vauclair, C.‐M., Manickam, L. S. S., Domínguez‐Espinosa, A., Torres, C., Lun, V. M.‐C., Teyssier, J., Miles, L. K., Hansen, K., Park, J., Wagner, W., Yu, A. A., Xing, C., Wise, R., Sun, C.‐R.,…Ahmed, R. A. (2018). Catching up with wonderful women: The women-are-wonderful effect is smaller in more gender egalitarian societies. International Journal of Psychology, 53(S1), 21–26. https://doi.org/10.1002/ijop.12420

38.

Landau

M. J.

Robinson

M. D.

Meier

B. P.

(2014). Metaphor research in social psychology: Current issues and future directions. In Landau

Robinson

M. D.

Meier

B. P.

(Eds.), The power of metaphor: Examining its influence on social life (pp. 269–286). American Psychological Association.

39.

Luchkina

Waxman

(2023). Talking about the absent and the abstract: Referential communication in language and gesture. Perspectives on Psychological Science, 19(6), 978–992. https://doi.org/10.1177/17456916231180589

40.

Luraghi

(2009). The origin of the feminine gender in PIE. Grammatical Change in Indo-European Languages. Amsterdam, Philadelphia. https://doi.org/10.1075/cilt.305.04lur

41.

Luraghi

(2011). The origin of the Proto-Indo-European gender system: Typological considerations. Folia Linguistica, 45(2), 435–463. https://doi.org/10.1515/flin.2011.016

42.

D. S.

Correll

Wittenbrink

(2015). The Chicago face database: A free stimulus set of faces and norming data. Behavior Research Methods, 47(4), 1122–1135. https://doi.org/10.3758/s13428-014-0532-5

43.

Maciuszek

Polak

Świa¸tkowska

(2019). Grammatical gender influences semantic categorization and implicit cognition in Polish. Frontiers in Psychology, 10, 2208. https://doi.org/10.3389/fpsyg.2019.02208

44.

Matasović

(2004). Gender in Indo-European. Winter.

45.

Menegatti

Rubini

(2017). Gender bias and sexism in language. Oxford Research Encyclopedia of Communication. https://doi.org/10.1093/acrefore/9780190228613.013.470

46.

Mestres-Missé

Münte

T. F.

Rodriguez-Fornells

(2014). Mapping concrete and abstract meanings to new words using verbal contexts. Second Language Research, 30(2), 191–223. https://doi.org/10.1177/0267658313512668

47.

Miller

M. K.

Rauch

J. A.

Kaplan

(2016). Gender differences in movie superheroes’ roles, appearances, and violence. Ada: A Journal of Gender, New Media, and Technology, No. 10. https://doi.org/10.7264/N3HX19ZK

48.

Morgenroth

Sendén

M. G.

Lindqvist

Renström

E. A.

Ryan

M. K.

Morton

T. A.

(2021). Defending the sex/gender binary: The role of gender identification and need for closure. Social Psychological and Personality Science, 12(5), 731–740. https://doi.org/10.1177/1948550620937188

49.

Mullen

M. K.

(1990). Children’s classifications of nature and artifact pictures into female and male categories. Sex Roles, 23(9–10), 577–587. https://doi.org/10.1007/bf00289769

50.

Nicolson

(2003). Having it all?: Choices for today's superwoman. John Wiley & Sons.

51.

Payne

B. K.

Hannay

J. W.

(2021). Implicit bias reflects systemic racism. Trends in Cognitive Sciences, 25(11), 927–936. https://doi.org/10.1016/j.tics.2021.08.001

52.

Prewitt-Freilino

J. L.

Caswell

T. A.

Laakso

E. K.

(2012). The gendering of language: A comparison of gender equality in countries with gendered, natural gender, and genderless languages. Sex Roles, 66(3-4), 268–281. https://doi.org/10.1007/s11199-011-0083-5

53.

Reis

(1997). Damned Women: Sinners and Witches in Puritan New England. Cornell University Press. ISBN 978-0-8014-8611-1.

54.

Richter

T. L.

(2017). Witches, whores, and virgin martyrs: Female roles in seventeenth century opera [Doctoral thesis]. University of California, Los Angeles.

55.

Rudman

L. A.

Mescher

(2012). Of Animals and Objects. Personality and Social Psychology Bulletin, 38(6), 734–746. https://doi.org/10.1177/0146167212436401

56.

Sabine

(2013). Veiled desires: Intimate portrayals of nuns in postwar Anglo-American film [Doctoral thesis]. Fordham Univ Press.

57.

Şahin

Soylu Yalcinkaya

(2021). The gendered brain: Implications of exposure to neuroscience research for gender essentialist beliefs. Sex Roles, 84(9), 522–535. https://doi.org/10.1007/s11199-020-01181-7

58.

Sá-Leite

A. R.

Luna

Fraga

Comesaña

(2020). The gender congruency effect across languages in bilinguals: A meta-analysis. Psychonomic Bulletin & Review, 27, 677–693. https://doi.org/10.3758/s13423-019-01702-w

59.

Samuel

Cole

Eacott

M. J.

(2019). Grammatical gender and linguistic relativity: A systematic review. Psychonomic Bulletin & Review, 26(6), 1767–1786. https://doi.org/10.3758/s13423-019-01652-3

60.

Schmitz

Schneider

Esser

Schmitz

Schneider

Esser

(2023). No genericity in sight: An exploration of the semantics of masculine generics in German. Glossa Psycholinguistics, 2(1). https://doi.org/10.5070/G6011192

61.

Segel

Boroditsky

(2011). Grammar in art. Frontiers in Psychology, 1, https://doi.org/10.3389/fpsyg.2010.00244

62.

Sera

M. D.

Berge

C. A.

Del Castillo Pintado

(1994). Grammatical and conceptual forces in the attribution of gender by English and Spanish speakers. Cognitive Development, 9(3), 261–292. https://doi.org/10.1016/0885-2014(94)90007-8

63.

Shoham

Lee

S. M.

(2018). The causal impact of grammatical gender marking on gender wage inequality and country income inequality. Business & Society, 57(6), 1216–1251. https://doi.org/10.1177/0007650317696231

64.

Stahlberg

Braun

Irmen

Sczesny

(2007). Representation of the sexes in language. Psychology Press eBooks. http://psycnet.apa.org/psycinfo/2007-01308-006.

65.

Stahlberg

Sczesny

(2001). Effekte des generischen Maskulinums und alternativer Sprachformen auf den gedanklichen Einbezug von Frauen [Effect of the generic use of the masculine pronoun and alternative forms of speech on the cognitive visibility of women]. Psychologische Rundschau, 52(3), 131–140. https://doi.org/10.1026/0033-3042.52.3.131

66.

Stahlberg

Sczesny

Braun

(2001). Name your favorite musician. Journal of Language and Social Psychology, 20(4), 464–469. https://doi.org/10.1177/0261927x01020004004

67.

Suitner

Maass

Ronconi

(2017). From spatial to social asymmetry: Spontaneous and conditioned associations of gender and space. Psychology of Women Quarterly, 41(1), 46–64. https://doi.org/10.1177/0361684316676045

68.

Szameitat

A. J.

Hamaida

Tulley

R. S.

Saylik

Otermans

P. C.

(2015). Women are better than men”–public beliefs on gender differences and other aspects in multitasking. Plos One, 10(10), e0140371. https://doi.org/10.1371/journal.pone.0140371

69.

Vaes

Paladino

Puvia

(2011). Are sexualized women complete human beings? Why men and women dehumanize sexually objectified women. European Journal of Social Psychology, 41(6), 774–785. https://doi.org/10.1002/ejsp.v41.6

70.

Vigliocco

Vinson

D. P.

Paganelli

Dworzynski

(2005). Grammatical gender effects on cognition: Implications for language learning and language use. Journal of Experimental Psychology: General, 134(4), 501–520.

71.

Warriner

A. B.

Kuperman

Brysbaert

(2013). Norms of valence, arousal, and dominance for 13,915 English lemmas. Behavior Research Methods, 45(4), 1191–1207. https://doi.org/10.3758/s13428-012-0314-x

72.

Wheeler

B. I.

(1899). The origin of grammatical gender. The Journal of Germanic Philology, 2(4), 528–545. http://www.jstor.org/stable/27699089

73.

Whitty

M. T.

(2001). The myth of the superwoman: Comparing young men’s and women’s stories of their future lives. Journal of Family Studies, 7(1), 87–100. https://doi.org/10.5172/jfs.7.1.87