Access the full text.
Sign up today, get DeepDyve free for 14 days.
References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.
Keith, N., Hartwig, K., & Richter, T. (2022). Ladies First or Ladies Last: Do Masculine Generics Evoke a Reduced and Later Retrieval of Female Exemplars? Collabra: Psychology, 8(1). https://doi.org/10.1525/collabra.32964 Social Psychology Ladies First or Ladies Last: Do Masculine Generics Evoke a Reduced and Later Retrieval of Female Exemplars? a b 1 1 2 Nina Keith , Kristine Hartwig , Tobias Richter 1 2 Department of Psychology, Technical University of Darmstadt, Darmstadt, Germany, Educational Psychology, University of Würzburg, Würzburg, Germany Keywords: androcentrism, sexism, language, gender, Gendersternchen https://doi.org/10.1525/collabra.32964 Collabra: Psychology Vol. 8, Issue 1, 2022 The use of masculine generics (i.e., grammatically masculine forms that refer to both men and women) is prevalent in many languages but has been criticized for potentially triggering male bias. Empirical evidence for this claim exists but is often based on small and selective samples. This study is a high-powered and pre-registered replication and extension of a 20-year-old study on this biasing effect in German speakers. Under 1 of 4 conditions (masculine generics vs. three gender-inclusive alternatives), 344 participants listed 3 persons of 6 popular occupational categories (e.g., athletes, politicians). Despite 20 years of societal changes, results were remarkably similar, underscoring the high degree of automaticity involved in language comprehension (large effects of 0.71 to 1.12 of a standard deviation). Male bias tended to be particularly pronounced later rather than early in retrieval, suggesting that salient female exemplars may be recalled first but that male exemplars still dominate the overall categorical representations. Contemplating on the boundary of drama and play, the morphological affix -in. But Politiker can also be used Charles Dickens once opined: “Every writer of fiction, generically to refer to a politician of any gender, such as though he may not adopt the dramatic form, writes in effect in the phrase der Beruf des Politikers (the profession of a for the stage” (Crystal, 1997, p. 76). In this statement, was politician). Likewise, the plural of the male form, die Poli- Dickens referring to male writers only or to female and male tiker, may refer to either a group of only male politicians writers alike? Or is this an idle question because Dickens or generically to a mixed group of politicians of any gender had no knowledge of any significant female writers? (If so, (but not to a group of only female politicians, which would he would have been unaware of Jane Austen or the Brontё be die Politikerinnen). sisters, which may seem rather unlikely.) A common assumption is that masculine generics may As this quote illustrates, generically masculine linguistic evoke an increased cognitive accessibility of men to the expressions, that is, grammatically masculine forms that detriment of women. Consequently, the visibility of women can refer to both men and women (masculine pronoun he is decreased and, according to feminist reasoning, inequal- in the above example) may at times be ambiguous. Yet, ities that already exist between genders in society will not such forms are prevalent in many languages. In English, for be counteracted but may even be reinforced by this kind example, besides generically used masculine pronoun he, of language use (Tavits & Pérez, 2019). In line with this some personal nouns explicitly include information on a criticism, official guidelines in many languages and coun- person’s gender (e.g., queen vs. king, actress vs. actor, witch tries recommend the use of gender-inclusive (e.g., he or she) vs. wizard) while others do not (e.g., consumer, manager, or gender-neutral (e.g., singular they; spokesperson in lieu teacher, neighbor). Many other languages, including several of spokesman) linguistic expressions (for examples, see the Germanic, Slavic, and Italic languages, convey gender infor- bias-free language guidelines of the American Psychologi- mation in many personal nouns and in corresponding gram- cal Association, 2020, Ch. 5; or the guidelines on gender- matical structures, with a generic use of masculine forms neutral language of the European Parliament, 2018). Such to denote people of both genders. In German, for example, endeavors, however, often remain limited to communica- a male politician is denoted Politiker while a female politi- tion in official documents and regulated communication, cian is denoted Politikerin in singular form, marked with while everyday use of explicitly gender-inclusive language a https://www.orgpsy.psychologie.tu-darmstadt.de/ Correspondence concerning this article should be addressed to Nina Keith, Department of Psychology, Organizational and Busi- ness Psychology, Technical University of Darmstadt, Alexanderstr. 10, 64283 Darmstadt, Germany. Email: firstname.lastname@example.org- darmstadt.de b https://go.uniwue.de/richter Ladies First or Ladies Last: Do Masculine Generics Evoke a Reduced and Later Retrieval of Female Exemplars? is less common (Gabriel et al., 2018; MacArthur et al., 2020; This result is inconsistent with studies that do find such a Vervecken & Hannover, 2012). Rather, in public opinion, difference (Brauer, 2008; Gabriel, 2008; Gabriel & Mellen- the use of gender-inclusive language is often criticized for berger, 2004). This inconsistency of results may well be due being cumbersome, confusing, or even misleading (Sczesny to insufficient power. The present study uses a sample that et al., 2016; for overviews of common criticisms, see Gabriel is more than three times larger than the original one, which et al., 2018; Parks & Roberton, 1998; Vergoossen et al., allows for the detection of medium to small effects, too. 2020). Second, research on language effects such as masculine Empirically, a number of studies have shown influences generics has been criticized for using highly selective sam- of masculine generics on cognitive information processing, ples, often involving laboratory studies with student partic- using a range of methods and in several languages (for re- ipants (Tavits & Pérez, 2019). The study by Stahlberg et al. views, see Henley & Abueg, 2003; Sczesny et al., 2016). Be- (2001), too, used a selective sample, with most participants low, we will describe in more theoretical terms why we pro- being university students. The present replication used a pose masculine generics to affect cognitive accessibility of less selective sample (only 14% of participants were univer- female and male exemplars of a category. Before doing so, sity students). we will first describe the original study that the present re- Third, no direct replication of the focal study exists, al- search seeks to replicate and extend. We will also explain though the study is well-known and cited in the field. A why we deemed replication of this particular study by handful of studies exist that used a somewhat similar task Stahlberg et al. (2001, Study 2) worthwhile—a study which (i.e., to list exemplars of an occupational category). These we think used a simple but elegant design, with a realistic studies, however, do not constitute replications because and unobtrusive task. they used specific instructions and different manipulations (e.g., participants were asked to list their favorite vs. least- The Original Study by Stahlberg et al. (2001) and liked personalities, personal heros, or good prime ministers why Replication is Worthwhile of the political left vs. right; Brauer, 2008; Gabriel, 2008; Gabriel & Mellenberger, 2004; see also Sniezek & Jazwinski, The present high-powered and pre-registered study 1986). Also, as mentioned above, the studies’ results did seeks to contribute to research on cognitive effects of mas- not fully converge, which may be due to the differences culine generics by replicating and extending one particular in study designs and manipulations. Another set of studies study, namely, a roughly 20-years old study on the effects used still other methods such as sentence evaluations, rat- of masculine generics in German speakers (Stahlberg et al., ings of sentence correctness, reading times, or eye-tracking 2001, Study 2; Stahlberg & Sczesny, 2001, Study 3, is the (e.g., Gygax et al., 2008; Irmen & Kurovskaja, 2010; Irmen same study). In this study, participants (45 female and 45 & Roßberg, 2004). These studies are interesting in them- male native speakers of German) were asked to name three selves, as they demonstrate an influence of masculine famous personalities in each of four categories (e.g., generics on specific aspects of cognition with different par- singers, politicians, hosts of TV shows). Stahlberg et al. adigms and measures, but they do not constitute replica- found that significantly more (almost twice as many) female tions of the study by Stahlberg et al. (2001). Successful names were listed by participants when women and men replications, however, particularly replications of effects were explicitly referenced with the so-called capital-I (Bin- that were originally found in underpowered studies, are nen-I in German), which replaces lower-case i of feminized needed to elevate an effect from single observation to sci- forms by a capitalized generic I (e.g., PolitikerInnen to refer entific evidence (Zwaan et al., 2018; see also Cesario, 2014). to both female and male politicians), compared to the mas- Fourth, there is a renewed public interest in feminist culine generic form (e.g., Politiker). In contrast, masculine- topics around the world (e.g., Chiu, 2020; MacNicol, 2020) feminine word pairs (e.g., Politiker und Politikerinnen) did and, at least in some countries, the public debate includes not elicit significantly more female exemplars than the the potentially detrimental role of masculine generics in masculine generic form. Participants’ gender also had an ef- language use (e.g., Burgen, 2020). Empirical research that fect on the number of female names listed (i.e., less women systematically analyzes feminists’ claims using more recent were listed by male than by female participants). data may contribute scientifically to this debate. For several theoretical, practical, and methodological Fifth, two decades have passed since the publication of reasons, we deemed replication (and extension) of this the original study and there have been considerable societal study worthwhile. First, psychological research, in general, changes since then. For example, work force participation has been criticized for sometimes using small and poten- of women has increased and so has the number of female tially underpowered samples (for a meta-analysis highlight- leaders in the past decades in the United States and in Eu- ing this problem, see Stanley et al., 2018). Notwithstanding rope, although in many countries, women are still largely the strengths of the original study, its sample size was so underrepresented in leading political and economic posi- small (30 participants per condition) that the sensitivity for tions (Begeny et al., 2020; European Commission, 2019; In- comparisons of each of the two linguistic forms with mas- ter-Parliamentary Union, 2015; U.S. Bureau of Labor Statis- culine generics was rather low, allowing for a reliable detec- tics, 2020; Warner et al., 2018). Still, the increased visibility tion of large effects only (Cohen’s d > 0.75, under the as- of women in influential positions (e.g., former chancellor sumptions of a Type I error probability of .05 and a power Angela Merkel in Germany) may have increased the cogni- of .90). In fact, in Stahlberg et al. (2001), there was no dif- tive accessibility of female exemplars and decreased poten- ference between masculine-feminine word pairs and mas- tially biasing effects of masculine generics. Considering the culine generics with regard to the female exemplars listed. societal changes, it seems not only important to replicate Collabra: Psychology 2 Ladies First or Ladies Last: Do Masculine Generics Evoke a Reduced and Later Retrieval of Female Exemplars? the original effects obtained by Stahlberg et al. (2001) but plars. This increased activation of associations with male also to examine whether potential biases in the cognitive exemplars occurs irrespective of any intentions the user of accessibility exemplars are similar in magnitude to the bi- masculine generics (e.g., speaker, writer) may or may not ases found two decades ago. hold to refer to women as much as to men. This is because Finally, rather than focusing solely on a quantity effect grammatically masculine forms make masculinity particu- of masculine generics (i.e., accessibility of more men than larly salient for comprehenders who, as a consequence, be- women) as has been the focus in previous work, we sought come prone to male bias (i.e., over-association of men), to explore order effects as well. If order effects were present, when forming their mental model. that is, if there was a dynamical change in decreased ac- Empirically, a number of studies have demonstrated ef- cessibility of women in the order of exemplars recalled, this fects of masculine generics that align with this reasoning. may provide some insight as to the processes underlying the These studies used a range of experimental methods and effect of masculine generics that has been found in earlier outcome measures (e.g., reading times, reaction times, par- studies. ticipants’ use of pronouns and nouns in stories about fic- titious persons, estimations of proportions of women and Language and Cognitive Accessibility of Female men in a given population; for an overview, see Sczesny et and Male Exemplars al., 2016). In line with this research, and with the original study that we seek to replicate, we expect: Masculine generics may increase the accessibility of Hypothesis 1. Masculine generics lead to a lower cog- male exemplars via mental models that are constructed nitive accessibility of female exemplars (compared to during language comprehension (Kollmayer et al., 2018). male exemplars) than alternative, gender-inclusive Mental models (also called situation models) are cognitive forms of linguistic expression. representations of the situations and events described in a text (Johnson-Laird, 1983; van Dijk & Kintsch, 1983). These Gender Differences in Cognitive Accessibility of representations contain elements of the text, relations Female and Male Exemplars among them, as well as their perceptual and other qualities (Johnson-Laird, 1983). For instance, readers of the sentence As described above, both external factors (e.g., infor- “Three turtles rested on a floating log, and a fish swam be- mational cues such as linguistic expressions) and internal neath them” (Bransford et al., 1972, p. 195, original text factors (e.g., characteristics of comprehenders) may shape includes italics) would construct a mental model that con- mental models. With regard to cognitive accessibility of fe- tains not only three turtles, a log, and a fish, but also, for ex- male and male exemplars, one such factor may be com- ample, the relative positions of these elements, the speed at prehenders’ gender. Previous research has shown a gender which they move, possibly additional qualities of these ele- difference in androcentric bias, that is, the tendency to ments (e.g., colors and shapes of the turtles, log, and fish), “construe men as more typical than women” (Bailey et al., and, although not explicitly mentioned in the sentence, a 2018, p. 313) for human categories that technically include pond or some other stretch of water where the described both men and women. Specifically, although androcentrism scenario may plausibly take place. appears to be universal, men tend to show more andro- As this example illustrates, readers, or more generally centric bias than women, at least on some tasks (for an speaking, comprehenders, use their preexisting knowledge overview, see Bailey et al., 2018, 2020). This gender differ- (e.g., that fish swim in water and that the action of floating ence may be a result of ingroup favoritism (Tajfel & Turner, implies water) when building their mental model. Impor- 1979), which leads women more (and men less) to reduce tantly, they need not consciously or strategically decide to androcentric bias. This gender difference may also be the do so, but parts of their preexisting knowledge are activated result of differences in frequency of instantiation (i.e., the automatically by verbal cues, words, phrases or sentences phenomenon that frequent exposure to an exemplar of a in the text, as the reader moves through the text (Cook et category leads to this exemplar being judged as more typi- al., 1998; Richter & Singer, 2018). As another important cal; Barsalou, 1985). Specifically, although men and women point, the preexisting knowledge that is activated and used may principally be exposed to equal numbers of women and in mental model construction can be more or less accurate. men in the public and in the media, in more personal mat- When people are described in a text, the knowledge used ters (e.g., friendship and close work relationships, prefer- to construct mental models during comprehension may also ences in media consumption), men may still be more ex- involve social stereotypes, including stereotypic gender in- posed to men and women more to women (Bailey et al., formation (Carreiras et al., 1996; Kollmayer et al., 2018). 2018). In line with this reasoning, and with the findings of For example, when reading a sentence about a doctor, peo- the original study that we seek to replicate, we expect: ple may construct a mental model that includes a male doc- tor, whereas when reading a sentence about a nurse, they Hypothesis 2. There is a gender difference in cognitive accessibility of female and male exemplars. Accessibil- more likely construct a mental model that includes a female ity of female exemplars will be lower in men than in nurse (Carreiras et al., 1996). As these examples illustrate, women. cues in the text and internal factors within the compre- henders (e.g., general world knowledge, stereotypes) act in When Does Decreased Cognitive Accessibility of concert to shape their mental models. Masculine generics, Female Exemplars Occur? then, may function as a linguistic cue that activates asso- ciations with male exemplars more than with female exem- Up to this point, we have focused on theory and findings Collabra: Psychology 3 Ladies First or Ladies Last: Do Masculine Generics Evoke a Reduced and Later Retrieval of Female Exemplars? Table 1. Modifications of and Extensions to Original Study by Stahlberg et al. (2001, Study 2) Original study Present study Study 3 (Linguistic expression of gender: masculine generics, 4 x 2 design; additional level of factor Linguistic design feminine-masculine word pair, capital-I) x 2 design expression of gender: gender asterisk (*) (Participant gender: female, male) Participants 90 participants (45 women, 45 men), most of them 344 participants, 55% women, 14% university university students students Material Participants were asked to list 12 persons (3 persons in 4 Participants were asked to list 18 persons (3 categories): athletes, singers, politicians, TV hosts persons in 6 categories), the two additional categories were: actors, authors Dependent Number of women listed by participants (1) Proportion of women listed by participants variable (relative to overall number of persons listed by participants) (2) Probabilities to list a woman Statistical Between-subjects ANOVA, post-hoc comparisons (1) Between-subjects ANOVA, post-hoc analyses comparisons (2) General Linear Mixed Models (with logit-link) predicting the probability to list a woman ics or (b) gender differences in accessibility? indicating that male bias (i.e., a decreased accessibility of female exemplars) may occur; we have not yet discussed at Method what point in time during retrieval this bias occurs. Given the high degree of automaticity involved in reading and This study is a replication and extension of a study that comprehension in general and in the activation of preex- was conducted about 20 years ago (Stahlberg et al., 2001, isting knowledge during comprehension (Cook et al., 1998; Study 2) and we will describe the methods with reference to Richter & Singer, 2018), one possibility is that the bias the original study. Table 1 additionally lists the major sim- occurs early during retrieval. For example, in the task to ilarities and differences between the original study and this come up with names of persons of a certain category (e.g., replication. This study was preregistered on the platform politicians, musicians), masculine generics may immedi- of the Open Science Framework (https://osf.io/te9c3). The ately trigger associations with male exemplars, while only preregistration includes the description of the expected ef- at second thought, one may come up with names of women fects, details on the planning of the required sample size, as well. As a consequence, overall, there will be more men criteria for excluding participants, analysis plan, and other than women listed. In other words, the overall over-associ- methodological details. This method section discloses all ation of men that was found in earlier studies may in fact measures and manipulations used. Data analyses did not be the result of early processes during retrieval that do not commence until data collection was completed. The data necessarily persist throughout later stages of retrieval. and analysis scripts are available at https://osf.io/ Another possibility is that while female exemplars may rn96f/?view_only=4436a99bf52740ef84a97a4951c414eb. be less frequent and less typical in the representation of a category, they are not necessarily less salient. Female ex- Design emplars, despite being less frequent, might even be more salient because of their atypicality, as women’s minority The study used a 4 x 2 between-subjects design. The first, status can contribute to increased saliency (Heilman, 2012; experimentally manipulated, factor was Linguistic expres- Kanter, 1977). If this reasoning holds, one might expect a sion of gender. The first three levels of this factor were iden- male bias in the overall frequency of listed male vs. female tical to those used in the original study, namely, (1) Mas- exemplars but not necessarily an order effect with male ex- culine generics (e.g., grammatically masculine Politiker to emplars listed first. Rather, female exemplars may even be refer to both female and male politicians); (2) feminine- listed first due to their increased saliency. We conclude that masculine word pairs (e.g., Politikerinnen or Poli- feminine based on existing theory and research, we cannot draw defi- tiker ); and (3) Capital-I (Binnen-I in German), masculine nite conclusions regarding potential order effects during re- which replaces lower-case i of feminized forms by a capital- trieval. We, therefore, put forth open research questions: ized generic I (e.g., PolitikerInnen to refer to both female and male politicians). These forms all constitute formally cor- Open Research Question 1. When does decreased cogni- rect or at least accepted (and sometimes criticized) forms of tive accessibility of female exemplars occur (e.g., early or late in the process of retrieval)? expression of gender information that are probably known by all (Forms 1 and 2) or by most (Form 3) German speakers. Open Research Question 2. If decreased cognitive acces- As a fourth condition, we included the so-called gender as- sibility of female exemplars changes over time, does terisk (Gendersternchen), which is a relatively recent cre- this change interact with (a) effects of masculine gener- ation that places an asterisk between the grammatically Collabra: Psychology 4 Ladies First or Ladies Last: Do Masculine Generics Evoke a Reduced and Later Retrieval of Female Exemplars? male stem of a noun and its feminization (e.g., Politiker* in- ital-I condition who exclusively named women. We did so nen to refer to both female and male politicians). This form because these participants may have misread the capital I is explicitly meant to include not only women and men as a lower-case i, in which case they would have erroneously but also people who do not identify with binary gender. thought that only women are referred to in the experimen- The term Gendersternchen has attracted some attention in tal material. (Note that this exclusion works against our hy- German media, as part of the debate on gender-fair lan- potheses; we also reran analyses without this exclusion and guage use. In 2018, the term was awarded the publicly well- found the pattern of results to be the same.) known German Anglicism-of-the-year award by an inde- The resulting sample of 344 participants was aged be- pendent initiative, chaired by a German linguist of Freie tween 14 and 79 years, with 55% women and 44% men. Two Universität Berlin (Gendersternchen is an Anglicism be- participants did not disclose their gender and two partic- cause ‘Gender’ is not a German but an English term; ipants did not identify with binary gender. These partici- ‘Sternchen’, which refers to the asterisk, is a German term pants’ data were excluded from all analyses that involved which roughly translates little star or starlet). The second participant gender, resulting in a sample size of 340 for all factor was Participants’ gender, an organismic factor with analyses involving participant gender. Only 14% of partic- the two levels male and female participants. For additional ipants were university students; 44.7% indicated to hold a analyses regarding the exploratory research questions, we university degree; 81.1% reported to be employed. Accord- also included the Position at which a name was mentioned ing to a sensitivity power analysis with G*Power (Faul et al., by participants as additional within-subject factor with 2007), this sample was sufficiently large to detect a small three levels (Position 1, Position 2, Position 3). effect both for the experimentally manipulated factor (i.e., linguistic expression of gender, as stated in Hypothesis 1, f Participants, Sample Size, and Power = .18) and for the organismic factor (i.e., participant gender, as stated in Hypothesis 2, f = .15) (with a Type I error prob- The original study included 90 participants (45 women ability of α = .05 and a power of 1-β = .80). and 45 men), mostly university students. In the present study, we sought to employ a larger and more heteroge- Material and Procedure neous sample, with a higher inclusion of non-student par- ticipants. For this purpose, we approached persons at public The study was a paper-and-pencil survey that consisted places in the center of a mid-sized German city and addi- of three parts. In Part 1, participants were asked to respond tionally recruited participants via acquaintances and other to eleven questions concerning their media behavior (e.g., personal contacts of one of the authors. Participation was how often they watch TV, how often they watch videos on- voluntary and was not compensated except for the possibil- line, how often they read newspapers, etc.). This was done ity to win one of five vouchers for a movie theater. to conceal the actual study goal from participants, that is, to When determining the sample size, we sought, first, to give participants the impression that this study was about arrive at a sample that is substantially larger and less selec- media consumption (the same procedure was used in the tive than the sample of the original study and, second, to original study). Part 2 was the focal part of the study in have enough power to detect a medium to small effect. In which the manipulation of the experimental variable and particular, we sought to achieve reasonable power for analy- assessment of the dependent variable took place. Under 1 ses involving simple contrasts between experimental con- of the 4 experimental conditions, participants were asked ditions (masculine generics vs. other linguistic expressions to list those three athletes, singers, actors, TV hosts, politi- of gender). For detecting a medium to small effect for these cians, and authors who would come to their minds first. contrasts (d = 0.40), with a Type I error probability of α = The original study did not include actors and authors but only four categories (Table 1). We included the two addi- .05 and a power of 1-β = .80, a sample size of 312 partici- tional categories to achieve a broader and more reliable pants (78 per group) is required (according to G*Power; Faul measurement of the dependent variable. Participants were et al., 2007). This sample size is also sufficient to detect a randomly assigned to experimental conditions by placing medium to small effect (f = .19) for the overall effect of lin- the surveys in random order before distributing them. As a guistic expression of gender. We opted to recruit a larger result, experimenters were blind to conditions as well. All sample, however, to account for potential data losses (which experimenters distributed surveys of all experimental con- we expected, as described in the preregistration). ditions. In Part 3, additional variables were collected for ex- Of the originally recruited 400 participants, the data of 3 ploratory purposes and demographic data was ascertained. participants could not be used due to incompleteness. Ad- ditional 27 participants (6.75%) correctly guessed the study Measures purpose, 5 participants of the generic-masculinum condi- tion deliberately rewrote the instructions to include gram- Dependent Variable matically female forms, and 1 participant of the generic- masculinum condition erroneously thought that the generic In the original study, the dependent variable was the masculinum referred to men only (which may be a conse- overall number of women listed by participants. This vari- quence of the use of masculine generics; Miller & James, able, in the present study, has a theoretical range from 0 to 2009). As preset in our preregistration, these participants’ 18 (the empirical range was 0 to 16; M = 5.54, SD = 2.93), be- data were also deleted from further analyses, resulting in a cause participants could list a maximum of 18 persons (i.e., sample size of 364. Again as preset in our preregistration, 3 persons x 6 occupational categories). Yet, we decided not we further excluded data sets of 20 participants of the cap- Collabra: Psychology 5 Ladies First or Ladies Last: Do Masculine Generics Evoke a Reduced and Later Retrieval of Female Exemplars? Table 2. Means, Standard Deviations, and Intercorrelations of Study Variables M SD 1 2 3 4 5 6 Demographics 1 Age 39.12 16.04 2 Gender 0.56 0.50 -.12* 3 Student 0.14 0.35 -.40** .07 4 First language 0.93 0.25 -.02 .09 .04 Experimental factor 5 Gender-inclusive 0.75 0.44 -.03 .04 .02 -.02 Dependent variable 6 Proportion of women (in %) 33.37 17.19 -.03 .38** -.01 .05 .35** 7 Number of women 5.53 2.93 .02 .37** .02 .08 .34** .93** Note. 340 < N < 344 (of 344 participants, 2 did not disclose their gender and another 2 did not identify with either female or male gender, resulting in 340 participants for all analyses involving gender). Variables were coded as follows: Gender was coded 1 for women and 0 for men; Student was coded 1 for student and 0 for non-student participant; First language was coded 1 for German and 0 for other language; Experimental factor was coded 1 for gender-inclusive conditions (i.e., feminine-masculine word pairs, capital-I, or gender asterisk) and 0 for mascu- line generics. * p < .05. ** p < .01. to use the overall number of women listed by participants ings of female vs. male exemplars) as nested within as main dependent variable but the ratio of women listed participants and participants nested within occupational by participants, relative to all persons listed by participants. roles and, hence, used a Generalized Linear Mixed Model We did so for two reasons. First, we observed that not all (GLMM) with a logit link function and crossed random ef- participants were able to come up with 18 persons (for ex- fects (Dixon, 2008; Jaeger, 2008). Being based on item-by- ample, some were able to list only 2 actors or only 1 author; item analyses rather than aggregated data, the GLMM al- although the average was close to 18, at 16.50, SD = 2.59) lows for estimating the effects of the position (first, second, and the ratio (rather than the number) of women appears or third position) at which a male or female exemplar was to be the more accurate measure in these cases. Second, the listed, in addition to the effects of linguistic expression of ratio or percentage is readily interpretable. For example, a gender and participants’ gender. By including the random value of 33.3% indicates that one-third of the persons listed effects of participants and occupational roles, the GLMM were women. (We also reran all analyses with the number also accounts for the fact that both participants and the of women as dependent variable and found highly similar occupational roles are samples drawn from larger popu- results, which is not surprising because number and ratio lations, avoiding a problem known as “language-as-fixed- are partially redundant and empirically correlated at .93, p effect fallacy” (Clark, 1973) in psycholinguistics and en- < .01, see Table 2.) hancing the validity of statistical inferences (Baayen et al., 2008). For the GLMM analysis, we used the packages lme4 Additional Variables (Bates et al., 2015) and lsmeans in the R-environment (R Core Team, 2020). In the following, we will report the re- For exploratory purposes, we assessed participants’ self- sults of these two analytical strategies consecutively. For reported gender role (German version of the Bem Sex Role descriptive statistics and intercorrelations of study vari- Inventory by Troche & Rammsayer, 2011) and attitude to- ables, see Table 2. wards the use of gender-fair language (self-developed single item measure). We will not report on these variables in the Effects of Linguistic Expression of Gender and of remainder of this manuscript because they did not reveal Participant Gender on Proportion of Listed any noteworthy insights. Women Results Hypothesis 1 predicted linguistic expression of gender to affect the proportion of women listed by participants; We analyzed the data using two analytical strategies. The in particular, we expected participants of the masculine- first analytical strategy closely resembled the one used in generics condition to list less women than participants of the original study and uses classical between-subjects the three gender-inclusive conditions (i.e., feminine-mas- ANOVA as well as post-hoc comparisons between groups; culine word pair, capital-I, or gender asterisk). We further the unit of analysis was the participant and independent sought to explore whether this effect is stronger for the cap- variables were linguistic expression of gender and partici- ital-I than the word-pair conditions, as has been found in pants’ gender. The second analytical strategy treated par- the original study by Stahlberg et al. (2001). Hypothesis 2 ticipants’ binary responses to the occupational roles (list- further predicted participant gender to affect the propor- Collabra: Psychology 6 Ladies First or Ladies Last: Do Masculine Generics Evoke a Reduced and Later Retrieval of Female Exemplars? Table 3. Means and Standard Deviations of Dependent Variable (Percentage of Women Listed) in Experimental Groups as well as Post-hoc Comparisons with Masculine Generics (Experimental Condition) and With Female Participants (Participant Gender) M SD N t(df) d Overall 33.37 17.19 344 By experimental condition Masculine generics 23.04 12.66 87 — — Feminine-masculine word pairs 32.46 13.67 95 4.00(181)*** 0.71 Capital-I 40.58 18.71 72 6.93(159)*** 1.12 Gender asterisk 38.55 18.24 90 6.49(161)*** 0.99 By participant gender Female Participants 39.21 16.59 189 — — Male Participants 26.01 14.95 151 8.42(339)*** 0.84 Note. N = 340 for analyses involving participant gender. *** p < .001. tion of women listed, in that female participants would list more women than would male participants. We first tested Hypothesis 1 in a one-factorial between- subjects ANOVA , with four levels of Factor Linguistic ex- pression of gender (masculine generics, feminine-masculine word pair, capital-I, and gender asterisk). We found a sig- nificant and large main effect, F(3,340) = 20.50, p < .001, η = .15. In line with what we expected based on partial the original study by Stahlberg et al. (2001), participants of the conditions with gender-inclusive expressions listed more women than those of the masculine-generics condi- tion, t(343) = 7.17, p < .001, Cohen’s d = 0.86 . The effect is illustrated in Figure 1. Post-hoc analyses further revealed Figure 1. Percentage of Women Listed as a Function all comparisons with the masculine-generics condition to of Linguistic Expression of Gender (Masculine be substantial, with Cohen’s d effect sizes ranging from 0.71 Generics vs. Three Gender-Inclusive Forms) to 1.12 (Table 3). Among the gender-inclusive expressions, the capital-I and the gender-asterisk condition led to an Note. Femin.-Masc. = Feminine-Masculine. Error bars represent standard errors. Dashed line denotes parity (50%). equally high proportion of women listed by participants, t(161) =0.81, p = .419, with both conditions differing from the feminine-masculine word-pair condition, t(256) = 3.45, 50% (for all comparisons, p < .0001). p < .001. In sum, masculine generics affected the proportion We tested Hypothesis 2 in a two-factorial between-sub- of women listed by participants in the expected way, with jects ANOVA, with participant gender as second factor, in the effect being more pronounced for the capital-I and gen- addition to the experimental factor Linguistic expression der-asterisk conditions than for the word-pair condition. of gender. As in the one-factorial ANOVA reported above , On an additional note, in all experimental conditions, we found a significant and large effect of the experimental the proportion of women listed was substantially below a factor, F(3,332) = 23.99, p < .001, η = .18. We also partial hypothetical parity of 50%. This is illustrated by the parity found a significant and large effect of participant gender, line in Figure 1 (all values and standard errors fall substan- F(1,332) = 70.92, p < .001, η = .18, but no interaction partial tially below the parity line). This observation was also con- of the two factors, F(3,332) = 1.18, p = .316, η = .01. firmed in one-sample t-tests that tested the values in exper- partial In line with what we expected based on the original study imental conditions against the hypothetical parity value of by Stahlberg et al. (2001), female participants listed more 1 A two-factorial ANOVA with the second factor participant gender (see test of Hypothesis 2) yielded the same results. We still report the results for the one-factorial ANOVA because they are based on the full sample available (because two participants did not disclose their gender and another two did not identify with binary gender, all analyses involving participant gender are based on a reduced sample of 340 instead of 344 participants). Collabra: Psychology 7 Ladies First or Ladies Last: Do Masculine Generics Evoke a Reduced and Later Retrieval of Female Exemplars? women than male participants did, with a Cohen’s d effect tests of interaction terms, |z| < 1.33, p > .183). We also esti- size of 0.84 (Table 3). mated two simpler models (Models 1a and 2a) in which lin- Taken together, our results replicate those found by guistic expression of gender was included as a two-level fac- Stahlberg et al. (2001). In particular, (1) the linguistic ex- tor, masculine generics vs. all other conditions combined. pression of gender affected the extent to which women were Again, Model 2a that included the interaction terms did not listed by participants, with masculine generics leading to exhibit a better model fit than Model 1a that included only the lowest percentage of women listed; (2) this effect was the main effects, χ (df = 7) = 7.10, p = .409 (Likelihood-Ra- more pronounced for the capital-I and for the gender aster- tio Test) and none of the interaction terms was significant isks conditions (the latter was not included in the original (for all tests of interaction terms: |z| < 1.84, p > .067). We study) than for the feminine-masculine word-pair condi- therefore conclude with regard to Open Research Question tion; (3) there was a large effect of participant gender (fe- 2 that there is no evidence that the time course of retrieving male participants listed more women than did male partici- female exemplars is affected by linguistic expression of gen- pants); and (4) the two factors did not interact. der or by participants’ gender (i.e., no interaction effect). As an additional comparison between the original and To answer Open Research Question 1, we proceeded with the present study, we compared the absolute numbers of the more parsimonious Model 1, which was structured as women listed as a function of experimental condition, follows: based on the four occupational categories included in the original study. We did so by calculating 95% confidence in- tervals around the mean number of women listed in condi- tions and by determining whether the values reported in the original study fell within or outside these confidence in- In this model, β represents the fixed effect estimate of tervals, with the following results. In the masculine-gener- the intercept. The predictors X , X , and X are dummy- 1 2 3 ics (95% CI [2.11, 2.91]) and the capital-I (95% CI [4.24, coded variables representing the levels of linguistic expres- 5.12]) conditions, the values of the original study fell within sion of gender that compare the reference category mas- the confidence intervals (2.37 and 4.72, respectively). In the culine generics with masculine-feminine word pairs (X ), condition of feminine-masculine word pairs (95% CI [3.20, capital-I (X ), and gender asterisk (X ), respectively. The 2 3 3.96]), however, the value of the original study fell sub- coefficients β , β , and β represent the fixed effects of 1 2 3 stantially below the lower bound of the confidence interval these predictors. The predictor X is participants’ gender (2.67), indicating a significant difference (i.e., in the origi- (dummy-coded, male participants as reference category), nal study, there were fewer women listed in the word-pairs and the coefficients β is the fixed effect of this predictor. condition than in the present study). The predictors X and X are dummy-coded variables repre- 5 6 senting the position on list that compare Position 2 to Posi- Additional Analyses of Order Effects tion 1 (X ) and Position 3 to Position 1 (X ). The error terms 5 6 and represent the random effects (random in- In addition to testing the hypotheses on the effects of tercepts) of participants and occupations, respectively. linguistic expression of gender and of participant gender, we explored at what point in time the decreased cognitive Table 4 (left-hand side) displays the parameter estimates accessibility of female exemplars occurs (e.g., early or late of this model. The main effect of linguistic expression of in the process of retrieval; Open Research Question 1) and gender predicted by Hypothesis 1 was again significant, χ whether the time course of retrieving female exemplars (df= 3) = 76.39, p < .001. Moreover, the main effect of par- changes as a function of linguistic expression of gender ticipants’ gender predicted by Hypothesis 2 was again sig- or of participant gender (i.e., interaction effect; Open Re- nificant, χ (df= 1) = 60.74, p < .001. Additionally, a main search Question 2). To examine these research questions, effect of position emerged, χ (df= 2) = 68.95, p < .001. Fe- we first estimated a model that contained only the main ef- male exemplars were more likely to be listed at the first po- fects of linguistic expression of gender, of participants’ gen- sition (Probability = .390, SE = .042), compared to the second der, and of position as fixed effects, plus the random effects position (Probability = .289, SE = .037) or the third position (random intercepts) of participants and occupations (Model (Probability = .272, SE = .036). The effects of the three fac- 1) (9 parameters, AIC = 6678, BIC = 6738, deviance = 6660). tors (linguistic expression of gender, participants’ gender, We then estimated a model that additionally included the and position on list) on probabilities are shown in Figure two-way interactions of these predictors (Model 2) (26 pa- 2. As this figure illustrates, the probabilities varied widely, rameters, AIC = 6699, BIC = 6871, deviance = 6646). Despite ranging from .17 (Position 3 for male participants in mascu- being far more complex than Model 1, Model 2 did not ex- line-generics condition) to .58 (Position 1 for female partic- hibit a better model fit than Model 1, χ (df = 17) = 13.66, p ipants in capital-I condition). With few exceptions, proba- = .691 (Likelihood-Ratio Test). Moreover, none of the inter- bilities were far below parity (i.e., far below .5). action terms in Model 2, including the interactions of po- To explore whether the effects of the independent vari- sition with linguistic expression of gender and with partic- ables generalize across occupations, we ran a variant of ipant gender, was significantly different from zero (for all 2 We used the values reported in the original study because the study’s raw data, collected some 20 years ago, were no longer available. Collabra: Psychology 8 Ladies First or Ladies Last: Do Masculine Generics Evoke a Reduced and Later Retrieval of Female Exemplars? Table 4. Parameter Estimates for the Generalized Linear Mixed Model (With Logit-Link) Examining the Effects of Linguistic Expression of Gender, Participant Gender and Position on the Probability to List a Woman Model 1 Model 1 (random slopes) Effect Parameter Est. (SE) z Est. (SE) z Fixed effects β -1.340 -6.59*** -1.414 -3.72*** Intercept (0.204) (0.381) Linguistic Expression of Gender Feminine-masculine word pairs vs. masculine β 0.494 4.65*** 0.522 3.92*** generics (X ) (0.106) (0.133) β 0.932 8.22*** 0.961 4.92*** Capital-I vs. masculine generics (X ) (0.113) (0.195) β 0.804 7.51*** 0.806 5.01*** Gender asterisk vs. masculine generics (X ) (0.107) (0.161) β 0.616 8.01*** 0.654 4.60*** Participants’ gender (female vs. male) (X ) (0.077) (0.142) Position β -0.470 -6.54*** -0.438 -1.84 Position 2 vs. Position 1 (X ) (0.072) (0.239) β -0.558 -7.56*** -0.528 -1.73 Position 3 vs. Position 1 (X ) (0.074) (0.306) Random effects (variances) Participants: Random intercept 0.168 0.196 Occupations: Random intercept 0.191 0.803 Random Slope X 0.032 Random Slope X 0.143 Random Slope X 0.078 Random Slope X 0.082 Random Slope X 0.309 Random Slope X 0.527 Model fit AIC 6678 6594 BIC 6738 6833 Deviance 6660 6522 Note. N = 5606. N = 340. Linguistic expression of gender: Three dummy-coded predictors, with masculine generics as reference category (coded as 0); Participants’ observations participants gender: dummy-coded, male participants as reference category (coded as 0); Position: dummy-coded, Position 1 as reference category (coded as 0). *** p < .001, + p < .10 (two-tailed). Model 1 that additionally included random slopes of lin- only at a liberal level of α < .10. This result indicates that guistic expression of gender, of participants’ gender, and of the random variation of the position effect between differ- position (i.e., the effects of these factors were assumed to ent occupations was large, leading to a decreased reliabil- vary across occupations). In comparison with the baseline ity of the fixed effect that reflect the mean effect of posi- Model 1 described above, this model additionally includes tion across occupations. In other words, the position effects the random slopes and differ between occupations, suggesting that the position ef- in the linear combination. fects might not generalize to other occupations. The parameter estimates of this model are displayed in Table 4 (right-hand side). In this model, the fixed effects of Discussion linguistic expression of gender and of participants’ gender The present high-powered and pre-registered study remained significant (p <.001), despite the increased stan- sought to replicate the detrimental effect of masculine dard errors. However, including the random slopes of occu- generics on the accessibility of female exemplars of a given pations increased the standard errors for the fixed effects occupational category—an effect that was found about two of position to an extent that these effects were significant Collabra: Psychology 9 Ladies First or Ladies Last: Do Masculine Generics Evoke a Reduced and Later Retrieval of Female Exemplars? Figure 2. Probabilities to List a Woman as a Function of Linguistic Expression of Gender (i.e., Masculine Generics vs. Three Gender-Inclusive Forms), Participants’ Gender (Male vs. Female Participants), and Position in List (Positions 1, 2, or 3) (Back-Transformed from the Logits Estimated in the GLMM, Model 1) Note. Error bars represent standard errors. Dashed line denotes parity (probability of .5). decades ago in a sample of university students (Stahlberg et Theoretical Contributions al., 2001, Study 2; Stahlberg & Sczesny, 2001, Study 3). We The present research informs theory in several ways. tested the effect using an almost four times larger and less First, the convergence of findings between the present selective participant sample (only 14% university students) study and the 20-years-old one suggests that the effects and found virtually the same results. This convergence of of masculine generics are based on highly automatized findings is remarkable, given the societal changes that have processes during text comprehension, as recipients use lin- occurred in the past decades in terms of, for example, in- guistic cues when constructing their mental model. These creases in workforce participation of women and visibility processes appear to be independent of societal changes re- of female political leaders. Apparently, a seemingly minor garding the public visibility of influential men and variation of linguistic expression, namely, whether mascu- women—or the societal changes may simply be still too in- line generics or explicitly gender-inclusive forms are used, cremental to evoke measurable changes in highly automa- can have strong effects on recipients’ cognitive inclusion of tized cognitive information processing. female exemplars. This convergence of findings did not only Second, not only linguistic cues (i.e., masculine generics) occur for the experimental effect (i.e., same pattern of main effects in the present and in the original study) but also can produce considerable effects on cognitive inclusion of in absolute terms at least in 2 of the 3 experimental con- women; the effect was just as strong for a participant char- acteristic, namely, their gender, and this occurred indepen- ditions. That is, for both the experimental condition that dently of the linguistic cue (i.e., no interaction between the produced the lowest inclusion of women (i.e., masculine two factors). Notwithstanding the large effects produced generics) and the highest inclusion of women (capital-I), by the linguistic cue (masculine generics) and participant the number of women listed by participants in the present characteristic (participant gender), under none of the con- study and by participants two decades ago did not differ sta- ditions, a numerical 50%-parity was achieved (i.e., even un- tistically. Although not a true experimental test, this con- der the condition with the highest proportion of listed vergence in absolute numbers indicates that the changes women, the average proportion was only slightly above regarding the visibility and increased power of women in 40%, see Table 3 and one-sample t-tests). Thus, if we define society did not lead to an overall increased cognitive acces- male bias as any deviation from 50%, male bias was present sibility of women. Only for masculine-feminine word pairs, under all conditions and only the degree of male bias varied our results differed from the original study. In our replica- tion, significantly more women were listed in this condition by condition. For example, being a female participant re- than in the masculine generics condition, whereas this con- duced, but did not eliminate male bias. Similarly, gender- inclusive linguistic cues reduced, but did not eliminate male trast was not significant in the original study. Accordingly, bias. These findings once again underscore the prevalence the number of women listed in this condition was signifi- and persistence of male bias, in terms of a male dominance cantly larger in the present study than in the original study. in cognitive representations and accessibility, as suggested A likely explanation for this deviation is the relatively small by various lines of research (Bailey et al., 2018; Tavits & sample size of the original study, which led to relatively low Pérez, 2019). statistical power and potentially instable estimates of the Third, the order effect we found in additional analyses population means. indicated that male bias dynamically changed during re- Collabra: Psychology 10 Ladies First or Ladies Last: Do Masculine Generics Evoke a Reduced and Later Retrieval of Female Exemplars? trieval, in that it was least pronounced early during re- generics from language use. Also, it should be kept in mind trieval. That is, the probability to list a woman was highest that avoiding masculine generics in favor of more gender- for the first position and then the probability dropped for inclusive forms may also have undesirable side effects. For Positions 2 and 3. This pattern was consistent across lin- example, for some forms of feminizations and gender-in- guistic cues (i.e., masculine generics vs. gender-inclusive clusion (e.g., explicitly referencing both male and female forms) and participant gender (i.e., no interaction of posi- engineers), research indicates that explicit female referenc- tion with these factors). This position effect suggests that ing may negatively affect the perceived social status and female exemplars do exist in the minds of many respon- evaluation of the referenced person and decrease the per- dents and that these exemplars are, in principle, cognitively ceived difficulty of an occupation (Formanowicz et al., 2013; accessible. However, if at all, these salient exemplars come Gabriel et al., 2018; Vervecken & Hannover, 2015). Also, in to one’s mind first, whereas later retrieval of female exem- some languages (e.g., Italian), feminized forms may be pe- plars becomes far less likely. For example, when asked to jorative. As a recent English-language example, Hollywood list three politicians, many Germans may come up with the star Cate Blanchett, who headed the Venice film festival in name of longtime chancellor Angela Merkel but a second 2020, was cited for saying that she would rather be called an or even third women would not easily come to their mind actor than an actress because of the pejorative sense often (to rule out the possibility that the position effect was ac- implied in the word actress (Agence France-Presse, 2020). tually driven by the category of politicians and chancellor Limitations and Directions for Future Research Merkel’s saliency, we reran analyses to the exclusion of this category and found the pattern of results to remain un- We would like to point out two limitations. First, this re- changed). This saliency interpretation of the position effect search was conducted in one particular context and tested is consistent with tokenism theory, according to which “so- one particular form of masculine generics in German lan- cial group members who are numerically underrepre- guage. Obviously, we do not know to what extent these find- sented” (Watkins et al., 2019, p. 334), such as female lead- ings generalize to other contexts, including, among others, ers, become all the more salient (Heilman, 2012; Kanter, other occupational roles, other forms of text and discourse, 1977). as well as other forms of masculine generics. Given the The position effect may also be a manifestation of an an- substantial size of effects in the present study and other drocentric bias to conceive of men as being more typical available evidence (from different paradigms in different exemplars of a human category than women (Bailey et al., languages), the ubiquity of linguistic gender asymmetries 2018, 2020). Specifically, after listing a woman, people may across many languages (Sczesny et al., 2016), and the uni- consciously or unconsciously be motivated not to list a sec- versality of androcentrism (Bailey et al., 2018, 2020), we ond or even a third woman, because a two-women-one-man would expect similar effects of masculine generics in a or even a three-women-no-man list would somewhat just range of situations. Moreover, the effects of linguistic ex- not feel right (i.e., not representative for the category as a pression of gender and of participants’ gender were signif- whole), whereas a one-women-two-men list feels appropri- icant and strong in the GLMM, too, that included random ate (i.e., representative for a human category such as politi- effects of participants and occupations, accounting for the cians, authors, or athletes). Obviously, based on the present fact that not only participants but also the occupations were data we can only speculate about the causes of the posi- drawn from larger populations. Nevertheless, future re- tion effect we found and we encourage future research to search may investigate systematically what context charac- replicate the effect and to identify its underlying mecha- teristics may modulate effects of masculine generics. For nisms. In that regard, it must be noted that we also found example, future research may systematically vary the in- a considerable variation between occupational categories, cluded occupations with regard to how many female exem- which, when included as a random slope, greatly increased plars plausibly exist or are visibly represented in public me- the standard errors of the position effect and lowered its dia. If only few or less representative female exemplars exist reliability. Thus, position effects affecting the probability in a category, accessibility of female exemplars may become of naming male or female exemplars seem to depend con- even more unlikely (Stahlberg & Sczesny, 2001, Study 2). siderably on the particular occupational category which is Second, it remains unclear to what extent our findings provided as additional retrieval cue. In other words, the are manifestations of conscious/controlled vs. unconscious/ position effects that we found might not generalize to occu- automatic processes. On the one hand, based on research pations other than those used in the present study. in reading, it seems safe to assume that processes involved in text comprehension are highly automatized (e.g., Cook Practical Implications et al., 1998; Richter & Singer, 2018). Unlike many other Practical implications of these findings are quite studies in the field, due to its simplicity, the paradigm in- straightforward: If we agree upon the goal to increase cog- troduced by Stahlberg et al. (2001) is likely to induce little nitive inclusion of women and to decrease male bias, a di- strategic processing. On the other hand, the task we used, rect recommendation would be to avoid masculine generics namely, to list three exemplars of a given category, is a and to replace them by gender-inclusive linguistic expres- task that is potentially controllable and strategic. It is pos- sions. This seems all the more appropriate when consider- sible that the gender-inclusive forms triggered participants’ ing that cognitive biases may contribute to biased behavior motivation to deliberately and consciously come up with (Bailey et al., 2018). However, our results suggest that parity at least a few female exemplars, at least in the beginning will still not be achieved even when abandoning masculine of retrieval, as the position effect suggests. Gender-exclu- Collabra: Psychology 11 Ladies First or Ladies Last: Do Masculine Generics Evoke a Reduced and Later Retrieval of Female Exemplars? sive masculine generics, in contrast, might not motivate ex- Acknowledgements plicit retrieval of women, with the consequence of male bias We acknowledge support by the Deutsche Forschungs- remaining uncorrected and particularly strong under this gemeinschaft (DFG – German Research Foundation) and condition. the Open Access Publishing Fund of Technical University of Conclusion Darmstadt. Our findings clearly indicate that masculine generics de- Contributions crease cognitive accessibility of female exemplars. If we Contributed to conception and design: NK, KH, TR agree that this form of male bias is undesirable, then the conclusion is straightforward, namely, that masculine Contributed to acquisition of data: KH generics should be avoided and replaced by gender-inclu- Contributed to analysis and interpretation of data: NK, KH, TR sive forms of linguistic expression—even though the former Drafted and/or revised the article: NK, TR may be considered by some to be grammatically correct and Approved the submitted version for publication: NK, TR the latter cumbersome. Referring back to the opening quote from a speech by Charles Dickens: It is not the point that Competing Interests we cannot be sure what writers of what gender Dickens was referring to. The point is that, irrespective of Dickens’ in- The authors declare no conflict of interest. tentions, when listening to or reading his speech, chances are that many people will mentally include only few female Data Accessibility Statement writers. The preregistration of this study can be found at: https://osf.io/te9c3 The data and analysis scripts are available at Author Note https://osf.io/ rn96f/?view_only=4436a99bf52740ef84a97a4951c414eb Portions of these findings were presented as a poster th at the 11 conference of the Division of Work, Organiza- Submitted: December 04, 2021 PST, Accepted: February 10, tional, and Business Psychology of the German Psycholog- 2022 PST ical Society (DGPs) in Braunschweig, Germany. The poster was awarded the conference’s Best Student Poster award (awardee: KH). This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International License (CCBY-4.0). View this license’s legal deed at http://creativecommons.org/licenses/by/4.0 and legal code at http://creativecom- mons.org/licenses/by/4.0/legalcode for more information. Collabra: Psychology 12 Ladies First or Ladies Last: Do Masculine Generics Evoke a Reduced and Later Retrieval of Female Exemplars? REFERENCES Agence France-Presse. (2020, September 3). Cate Cesario, J. (2014). Priming, replication, and the hardest Blanchett says she would rather be called an actor than science. Perspectives on Psychological Science, 9(1), an actress. https://www.theguardian.com/film/2020/s 40–48. https://doi.org/10.1177/1745691613513470 ep/03/cate-blanchett-says-she-would-rather-be-calle Chiu, B. (2020, March 8). 2020s mark a new wave of d-an-actor-than-an-actress feminist mobilization. Forbes. https://www.forbes.co American Psychological Association. (2020). Publication m/sites/bonniechiu/2020/03/08/2020s-mark-a-new-w manual of the American Psychological Association (7th ave-of-feminist-mobilization/?sh=23490f2e485e ed.). https://doi.org/10.1037/0000165-000 Clark, H. H. (1973). The language-as-fixed-effect fallacy: Baayen, R. H., Davidson, D. J., & Bates, D. M. (2008). A critique of language statistics in psychological Mixed-effects modeling with crossed random effects research. Journal of Verbal Learning and Verbal for subjects and items. Journal of Memory and Behavior, 12(4), 335–359. https://doi.org/10.1016/s00 Language, 59(4), 390–412. https://doi.org/10.1016/j.j 22-5371(73)80014-3 ml.2007.12.005 Cook, A. E., Halleran, J. G., & O’Brien, E. J. (1998). What Bailey, A. H., LaFrance, M., & Dovidio, J. F. (2018). Is is readily available during reading? A memory‐based man the measure of all things? A social cognitive view of text processing. Discourse Processes, 26(2–3), account of androcentrism. Personality and Social 109–129. https://doi.org/10.1080/0163853980954504 Psychology Review, 23(4), 307–331. https://doi.org/1 1 0.1177/1088868318782 Crystal, D. (1997). The Cambridge encyclopedia of Bailey, A. H., LaFrance, M., & Dovidio, J. F. (2020). language (2nd ed.). Cambridge University Press. Implicit androcentrism: Men are human, women are Dixon, P. (2008). Models of accuracy in repeated- gendered. Journal of Experimental Social Psychology, measures designs. Journal of Memory and Language, 89, 103980. https://doi.org/10.1016/j.jesp.2020.10398 59(4), 447–456. https://doi.org/10.1016/j.jml.2007.1 1.004 Barsalou, L. W. (1985). Ideals, central tendency, and European Commission. (2019). 2019 report on equality frequency of instantiation as determinants of graded between women and men in the EU. [Online document]. structure in categories. Journal of Experimental Publications Office of the European Union. https://do Psychology: Learning, Memory, and Cognition, 11(4), i.org/10.2838/395144 629–654. https://doi.org/10.1037/0278-7393.11.1-4.6 European Parliament. (2018). Gender-neutral language in the European Parliament [Online document]. https://w Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). ww.europarl.europa.eu/cmsdata/151780/GNL_Guideli Fitting linear mixed-effects models using lme4. nes_EN.pdf Journal of Statistical Software, 67(1), 1–48. https://do Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). i.org/10.18637/jss.v067.i01 G*Power 3: A flexible statistical power analysis Begeny, C. T., Moss-Racusin, C. A., Ryan, M. K., & program for the social, behavioral, and biomedical Ravetz, G. (2020). In some professions, women have sciences. Behavior Research Methods, 39(2), 175–191. become well represented, yet gender bias https://doi.org/10.3758/bf03193146 persists—perpetuated by those who think it is not Formanowicz, M., Bedynska, S., Cisłak, A., Braun, F., & happening. Science Advances, 6(26), eaba7814. http Sczesny, S. (2013). Side effects of gender-fair s://doi.org/10.1126/sciadv.aba7814 language: How feminine job titles influence the Bransford, J. D., Barclay, J. R., & Franks, J. J. (1972). evaluation of female applicants. European Journal of Sentence memory: A constructive versus interpretive Social Psychology, 43(1), 62–71. https://doi.org/10.100 approach. Cognitive Psychology, 3(2), 193–209. http 2/ejsp.1924 s://doi.org/10.1016/0010-0285(72)90003-5 Gabriel, U. (2008). Language policies and in-group Brauer, M. (2008). Un ministre peut-il tomber enceinte ? favoritism: The malleability of the interpretation of L’impact du générique masculin sur les generically intended masculine forms. Social représentations mentales. L’Année Psychologique, Psychology, 39(2), 103–107. https://doi.org/10.1027/18 108(2), 243–272. https://doi.org/10.4074/s000350330 64-93184.108.40.206 Gabriel, U., Gygax, P. M., & Kuhn, E. A. (2018). Burgen, S. (2020, January 19). Masculine, feminist or Neutralising linguistic sexism: Promising but neutral? The language battle that has split Spain. The cumbersome? Group Processes & Intergroup Relations, Guardian. https://www.theguardian.com/world/2020/j 21(5), 844–858. https://doi.org/10.1177/13684302187 an/19/gender-neutral-language-battle-spain Carreiras, M., Garnham, A., Oakhill, J., & Cain, K. Gabriel, U., & Mellenberger, F. (2004). Exchanging the (1996). The use of stereotypical gender information in generic masculine for gender-balanced forms – the constructing a mental model: Evidence from English impact of context valence. Swiss Journal of Psychology, and Spanish. Quarterly Journal of Experimental 63(4), 273–278. https://doi.org/10.1024/1421-0185.6 Psychology, 49(3), 639–663. https://doi.org/10.1080/7 3.4.273 Collabra: Psychology 13 Ladies First or Ladies Last: Do Masculine Generics Evoke a Reduced and Later Retrieval of Female Exemplars? Gygax, P., Gabriel, U., Sarrasin, O., Oakhill, J., & Richter, T., & Singer, M. (2018). Discourse updating: Garnham, A. (2008). Generically intended, but Acquiring and revising knowledge through discourse. specifically interpreted: When beauticians, In M. F. Schober, D. N. Rapp, & M. A. Britt (Eds.), musicians, and mechanics are all men. Language and Routledge handbooks in linguistics. The Routledge Cognitive Processes, 23(3), 464–485. https://doi.org/1 handbook of discourse processes (pp. 167–190). 0.1080/01690960701702035 Routledge/Taylor & Francis Group. Heilman, M. E. (2012). Gender stereotypes and Sczesny, S., Formanowicz, M., & Moser, F. (2016). Can workplace bias. Research in Organizational Behavior, gender-fair language reduce gender stereotyping and 32, 113–135. https://doi.org/10.1016/j.riob.2012.11.0 discrimination? Frontiers in Psychology, 7, 25. http 03 s://doi.org/10.3389/fpsyg.2016.00025 Henley, N. M., & Abueg, J. (2003). A review and Sniezek, J. A., & Jazwinski, C. H. (1986). Gender bias in synthesis of research on comprehension of the English: In search of fair language. Journal of Applied masculine as a generic form in English. Sociolinguistic Social Psychology, 16(7), 642–662. https://doi.org/10.1 Studies, 4(2), 427–454. https://doi.org/10.1558/sols.v4 111/j.1559-1816.1986.tb01165.x i2.427 Stahlberg, D., & Sczesny, S. (2001). Effekte des Inter-Parliamentary Union. (2015). Women in generischen Maskulinums und alternativer parliament: 20 years in review [Online document]. Sprachformen auf den gedanklichen Einbezug von Inter-Parliamentary Union. http://archive.ipu.org/eng Frauen [Effects of the generic masculinum and of lish/perdcls.htm#wmn-year alternative forms of expression on the mental inclusion of women]. Psychologische Rundschau, 52(3), Irmen, L., & Kurovskaja, J. (2010). On the semantic 131–140. https://doi.org/10.1026//0033-3042.52.3.13 content of grammatical gender and its impact on the representation of human referents. Experimental Psychology, 57(1), 367–375. https://doi.org/10.1027/16 Stahlberg, D., Sczesny, S., & Braun, F. (2001). Name 18-3169/a000044 your favorite musician: Effects of masculine generics and of their alternatives in German. Journal of Irmen, L., & Roßberg, N. (2004). Gender markedness of Language and Social Psychology, 20(4), 464–469. http language: The impact of grammatical and s://doi.org/10.1177/0261927x01020004004 nonlinguistic information on the mental representation of person information. Journal of Stanley, T. D., Carter, E. C., & Doucouliagos, H. (2018). Language and Social Psychology, 23(3), 272–307. http What meta-analyses reveal about the replicability of s://doi.org/10.1177/0261927x04266810 psychological research. Psychological Bulletin, 144(12), 1325–1346. https://doi.org/10.1037/bul0000169 Jaeger, T. F. (2008). Categorical data analysis: Away from ANOVAs (transformation or not) and towards Tajfel, H., & Turner, J. C. (1979). An integrative theory logit mixed models. Journal of Memory and Language, of intergroup conflict. In W. G. Austin & S. Worchel 59(4), 434–446. https://doi.org/10.1016/j.jml.2007.1 (Eds.), The social psychology of intergroup relations (pp. 1.007 33–47). Brooks/Cole. Johnson-Laird, P. N. (1983). Mental models. Cambridge Tavits, M., & Pérez, E. O. (2019). Language influences University Press. mass opinion toward gender and LGBT equality. Proceedings of the National Academy of Sciences, Kanter, R. M. (1977). Men and women of the corporation. 116(34), 16781–16786. https://doi.org/10.1073/pnas.1 Basic Books. Kollmayer, M., Pfaffel, A., Schober, B., & Brandt, L. Troche, S., & Rammsayer, T. H. (2011). Eine Revision (2018). Breaking away from the male stereotype of a des deutschsprachigen Bem Sex-Role Inventory [A specialist: Gendered language affects performance in revision of the German Bem Sex-Role Inventory]. a thinking task. Frontiers in Psychology, 9, 985. Klinische Diagnostik und Evaluation, 4(3), 262–283. MacArthur, H. J., Cundiff, J. L., & Mehl, M. R. (2020). U.S. Bureau of Labor Statistics. (2020). Labor force Estimating the prevalence of gender-biased language statistics from the current population survey. https://w in undergraduates’ everyday speech. Sex Roles, 82(3), ww.bls.gov/cps/cpsaat11.htm 81–93. https://doi.org/10.1007/s11199-019-01033-z van Dijk, T. A., & Kintsch, W. (1983). Strategies of MacNicol, G. (2020, December 4). 50 years on, the discourse comprehension. Academic Press. Feminist Press is radical and relevant. The New York Times. https://www.nytimes.com/2020/12/04/us/50-y Vergoossen, H. P., Renström, E. A., Lindqvist, A., & ears-the-feminist-press-gloria-steinem-florence-how Sendén, M. G. (2020). Four dimensions of criticism e.html against gender-fair language. Sex Roles, 83(5–6), 328–337. https://doi.org/10.1007/s11199-019-0110 Miller, M. M., & James, L. E. (2009). Is the generic 8-x pronoun he still comprehended as excluding women? The American Journal of Psychology, 122(4), 483–496. Vervecken, D., & Hannover, B. (2012). Ambassadors of gender equality? How use of pair forms versus Parks, J. B., & Roberton, M. A. (1998). Contemporary masculines as generics impacts perception of the arguments against nonsexist language: Blaubergs speaker. European Journal of Social Psychology, 42(6), (1980) revisited. Sex Roles, 39(5/6), 445–461. https://d 754–762. https://doi.org/10.1002/ejsp.1893 oi.org/10.1023/a:1018827227128 R Core Team. (2020). R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org/ Collabra: Psychology 14 Ladies First or Ladies Last: Do Masculine Generics Evoke a Reduced and Later Retrieval of Female Exemplars? Vervecken, D., & Hannover, B. (2015). Yes I can! Effects Watkins, M. B., Simmons, A., & Umphress, E. (2019). It’s of gender fair job descriptions on children’s not black and white: Toward a contingency perceptions of job status, job difficulty, and perspective on the consequences of being a token. vocational self-efficacy. Social Psychology, 46(2), Academy of Management Perspectives, 33(3), 334–365. 76–92. https://doi.org/10.1027/1864-9335/a000229 https://doi.org/10.5465/amp.2015.0154 Warner, J., Ellman, N., & Boesch, D. (2018). The women’s Zwaan, R. A., Etz, A., Lucas, R. E., & Donnellan, M. B. leadership gap: Women’s leadership by the numbers (2018). Making replication mainstream. Behavioral [Online document]. Center for American Progress. htt and Brain Sciences, 41, e120. https://doi.org/10.1017/s ps://cdn.americanprogress.org/content/uploads/2018/ 0140525x17001972 11/19121654/WomensLeadershipFactSheet.pdf Collabra: Psychology 15 Ladies First or Ladies Last: Do Masculine Generics Evoke a Reduced and Later Retrieval of Female Exemplars? SUPPLEMENTARY MATERIALS Peer Review History Download: https://collabra.scholasticahq.com/article/32964-ladies-first-or-ladies-last-do-masculine-generics-evoke-a- reduced-and-later-retrieval-of-female-exemplars/attachment/83011.docx?auth_token=sWwlThtafvsRPI1s0A2U Collabra: Psychology
Collabra Psychology – University of California Press
Published: Feb 28, 2022
Keywords: androcentrism; sexism; language; gender; Gendersternchen
Access the full text.
Sign up today, get DeepDyve free for 14 days.