ArticlePDF Available

Abstract and Figures

The influence of stereotype and grammatical information (masculine intended as generic) on the representation of gender in language was investigated using a sentence evaluation paradigm. The first sentence introduced a role name (e.g., The spies came out …) and the second sentence contained explicit information about the gender of one or more of the characters (e.g., …one of the women …). The experiment was conducted in French, German, and English. In contrast to English, stereotypicality of role names had no influence on readers’ male biased representations in French and German, where interpretations were dominated by the masculinity of the masculine (allegedly) intended as generic.
Content may be subject to copyright.
This article was downloaded by:[Bibliothèque cantonale et universitaire - Fribourg]
On: 1 April 2008
Access Details: [subscription number 788628288]
Publisher: Psychology Press
Informa Ltd Registered in England and Wales Registered Number: 1072954
Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Language and Cognitive Processes
Publication details, including instructions for authors and subscription information:
Generically intended, but specifically interpreted: When
beauticians, musicians, and mechanics are all men
Pascal Gygax
; Ute Gabriel
; Oriane Sarrasin
; Jane Oakhill
; Alan Garnham
University of Fribourg, Fribourg, Switzerland
Norwegian University of Science and Technology, Trondheim, Norway
University of Bern, Bern, Switzerland
University of Sussex, Falmer, UK
Online Publication Date: 01 April 2008
To cite this Article: Gygax, Pascal, Gabriel, Ute, Sarrasin, Oriane, Oakhill, Jane
and Garnham, Alan (2008) 'Generically intended, but specifically interpreted: When
beauticians, musicians, and mechanics are all men', Language and Cognitive
Processes, 23:3, 464 - 485
To link to this article: DOI: 10.1080/01690960701702035
Full terms and conditions of use:
This article maybe used for research, teaching and private study purposes. Any substantial or systematic reproduction,
re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly
The publisher does not give any warranty express or implied or make any representation that the contents will be
complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be
independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings,
demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or
arising out of the use of this material.
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
Generically intended, but specifically interpreted: When
beauticians, musicians, and mechanics are all men
Pascal Gygax
University of Fribourg, Fribourg, Switzerland
Ute Gabriel
Norwegian University of Science and Technology, Trondhei m, Norway
Oriane Sarrasin
University of Bern, Bern, Switzerland
Jane Oakhill and Alan Garnham
University of Sussex, Falmer, UK
The influence of stereotype and grammatical information (masculine intended
as generic) on the representation of gender in language was investigated using a
sentence evaluation paradigm. The first sentence introduced a role name (e.g.,
The spies came out . . . ) and the second sentence contained explicit information
about the gender of one or more of the characters (e.g., . . . one of the
women . . . ). The experiment was conducted in French, German, and English.
In contrast to English, stereotypicality of role names had no influence on
readers’ male biased representations in French and German, where interpreta-
tions were dominated by the masculinity of the masculine (allegedly) intended
as generic.
In many gender marked languages, such as German or French, the gender of
a character in a text is ! in general ! explicitly given by the form of the
determiner and by the morphological form of the noun. In contrast to
Correspondence should be addressed to Pascal Gygax, De´partement de Psychologie,
Universite´ de Fribourg, Rue Faucigny 2, 1700 Fribourg, Switzerland.
This research was supported by a grant from the Swiss National Foundation (100013-
109705j) to Ute Gabriel and Pascal Gygax. We thank Emanuelle Carruzzo for her help in
conducting this study.
2008, 23 (3), 464!485
2008 Psychology Press, an imprint of the Taylor & Francis Group, an Informa business DOI: 10.1080/01690960701702035
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
English, in which readers of the sentence The football player had been training
very hard cannot be sure whether it is about a man or a woman, in Ger man
or French, the sentence Die Fussballspielerinnen hatten hart trainiert or Les
joueuses de football s’e´taient entraıˆne´es durement unequivocally signifies that
a group of women is referred to. But the situation becomes less obvious if in
the German and the French example the feminine form is exchanged for the
masculine form. Whereas femi nine plural forms refer to women only,
masculine plural forms either refer to a group of men (specific use of
masculine) or to a group of people of both sexes, to persons of unknown sex,
or where the sex is irrelevant (generic use of the masculine).
This is an
explicit grammatical rule (Acade´mie Franc
aise, 2002; Baudino, 2001; Duden,
2005) and although there exist guidelines on how to avoid the use of the
masculine only in official announcements, the masculine is still commonly
used as a generic in spoken as well as in written language.
As masculine words also have a male-specific meaning, comprehenders
have to use contextual information to correctly identify whether a word is
used in a specific or generic way. However, proponents of feminist linguistics
(e.g., Braun, 1996; Bussmann, 1995; Peyer & Wyss, 1998) express doubt that
the masculine form can be used in a way that abstracts from the gender of its
referents (i.e., in a generic way) and claim that the ambiguity about whether a
word is used as a generic or not is usually resolved to womens disadvantage:
The use of the masculine plural biases information processing, resulting in a
male oriented representation.
The question of whether the use of the masculine as generic results in a
gender-neutral or male-biased representation relates to the classical psycho-
logical question of whether language influences cognition (see Hardin &
Banaji, 1993). This issue was influenced by the work of Benjamin Lee Whorf
and Edward Sapir in the mid 20
century (i.e., the Sapir-Whorf, or Whorfian
hypothesis). The main ideas behind this issue are that (a) the structure of our
language influences the way we perceive the world (i.e., linguistic determin-
ism), and (b) since there are different languages in the world, there must be
different ways to perceive the world (i.e., linguistic relativism). Although
research on this topic has been declining in the past 20 years, recent
experimental studies have reawakened interest in this area (Li & Gleitman,
2002). Most importantly, the strong position that language entirely
determines perception has given way to a weaker argument stating that
language influences processes that are encoded through language, for
example, spatial reasoning (Levinson, Kita, Haun, & Rasch, 2002).
Note that the masculine singular (e.g., le joueur de football) does not unequivocally refer to
a man either. The masculine singular form le joueur de football, for example, refers (supposedly)
to any person who plays football, as well as to a male player. In this paper, we however only
focused on the masculine plural form, as the plural form clearly refers to a group of people.
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
Although Levinson et al.’s (2002) arguments were questioned by Li and
Gleitman (2002), they showed that different ways of verbally describing
spatial properties (e.g., using different reference frames) lead to different
perceptions of space. In terms of grammatical properties and their influence
on representation, Flaherty (2001) introduced the idea that gender systems
influence the way language users perceive the world. In her second
experiment, Flaherty (2001) asked native-English-speaking and native-
Spanish-speaking participants of different ages (5- to 7-year-olds, 8- to 10-
year-olds and adults) to assign gender and to put typical male or female
names to different objects presented in cartoons. Older Spanish participants
(8- to 10-year-olds and adults) were inclined to assign gender and names
according to grammatical gender, whereas older English participants
assigned gender according to some specific perceived gender attributes as
outlined by the participants in the first Experiment. Both 5- to 7-year-old
English and Spanish participants assigned gender according to their own sex
more than older participants. In the third Experiment, older (8- to 10-year-
old and adult) English and Spanish participants had to assign female and
male attributes to animate and inanimate objects. English participants
assigned attributes in the same way as they assigned gend er in Experiment 2,
and Spanish participants’ responses were strongly influenced by the
grammatical gender of the nouns. Flaherty’s (2001) main conclusion was
that grammatical gender enabled Spanish participants to assign gender
(Experiment 2) and assign attributes (Experiment 3) according to gramma-
tical gender. In Experiment 2, this was the case mainly for older participants
(8-year-olds and above), as younger pa rticipants (5- to 7-year-olds) had not
yet fully acquired the principles of grammatical gender. In the same vein, the
question about the effects of the masculine plural can be stated in terms of
whether (a) the use of the masculine plural leads to a gender-open
representation (i.e., no gender category is activated), (b) to a gender-spread
(i.e., both gender categories are activated) representation, or whether (c) the
use of the masculine plural leads to heightened cognitive accessibility of men
only, which would strongly support the view that language influences the way
we perceive and represent the world.
Most empirical research on the use of the masculine (intended as generic)
in German seems to support the second view: Presenting texts that contain
either the masculine plural or other linguistic forms, such as gender-balanced
forms (i.e., the joint use of the masculine and feminine, e.g., Zuschauer und
Zuschauerinnen; des spectateurs et des spectatrices) or gender-neutral forms
(e.g., Personen; des personnes), in general results in the masculine plural
condition being more strongly associated with male persons compared with
other linguistic forms (Braun, Gottburgsen, Sczesny, & Stahlberg, 1998;
Gabriel & Mellenberger, 2004; Heise, 2000, 2003; Irmen & Ko
hncke, 1996;
Rothmund & Scheele, 2004; Scheele & Gauler, 1993; Stahlberg & Sczesny,
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
2001; Stahlberg, Sczesny, & Braun, 2001). In this paper, however, we focus on
the representation of gender while participants are reading role names
written in the masculine plural form.
The strongest masculine bias position states that grammatical gender is
initially encoded and thus, the use of the masculine plural always brings
about male-biased associations no matter whether the masculine plural is
intended in a specific or a generic way. This notion of gender representations
being dominated by grammatical features is corroborated by research in
French (Cole´ & Segui, 1994) and in Italian (Bates, Devescovi, Hernandez, &
Pizzamiglio, 1996) showing that morphological gender plays an inhibitory
role when subsequent input is gender incongruent. For example, in Cole´ and
Segui (1994), participants had more trouble (i.e., took longer) in processing
(i.e., double lexical decision task) pairs of words when they constituted a
plausible semantic structure but were gender incongruent (e.g., jolie ! chat)
than when they were gender congruent (e.g., joli ! chat). In a similar vein,
and as mentioned earlier, Flaherty (2001) found that (except for 5- to 7-year-
olds) there was a strong correlation in English between assigned gend er and
assigned attributes, whereas in Spanish, gender was predominantly assigned
according to the grammatical gender of the referent noun. But with reference
to the generic interpretation of the masculine in German, the empirical
results reported in four different publications (Braun et al., 1998; Irmen &
Roßberg, 2004; Rothermund, 1998; Rothmund & Scheele, 2004) question the
generality of that notion: As will be described in what follows, non-linguistic
factors, such as gender-related context-information, might influence the
mental representation of the gender of characters in texts, and so might
linguistic factors other than masculine morphology, such as pronouns and
determiner s. In the following sections we first present research on the
influence of context information (stereotype information) and then research
on the influence of linguistic factors. The experiment reported in this paper
builds on these prior findings, exploring the interaction of both linguistic and
non-linguistic factors on readers’ gen der representation.
Carreiras, Garnham, Oakhill, and Cain (1996) conducted a series of
experiments in English and Spanish and found that not only do participants
build a representation of gender during reading, but they do so by relying on
stereotype information. In their study, English participants (Experiment 1)
had more trouble processing sentences with pronouns that mismatched the
gender-stereotype introduced by a role name. In Spanish (Experiments 2!4),
participants had more trouble processing role names when their morpholo-
gical marking or their article mismatched the stereotype of the name.
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
Garnham, Oakhill, and Reynolds (2002) found similar evidence. In their
studies using a sentence evaluation paradigm, in which participants have to
judge whether a sentence is a good continuation of the preceding text, they
found that participants had most trouble with, and took longer to respond
to, sentences that were incongruent with stereotypical gender of the role
names in the preceding text (see also Duffy & Keir, 2004; Kennison & Trofe,
2003; Sturt, 2003). In a later study Oakhill, Garnham, and Reynolds (2005)
found that participants were strongly inclined to form a representation that
was biased by stereotypic gender information. Indeed, even when the
experimenters attempted to suppress participants’ use of such information,
by encouraging them to respond strategically (as opposed to automatically),
the participants’ mental representation of gender was still stereotyped.
The proportion of members of a group judged to be male or female is also
influenced by stereotypes. Braun et al. (1998 had participants read a German
text (about 120 words) that was written like a newspaper report on a
scientific congress (Experiment 1) or on a meeting of a sports association
(Experiment 2). The task was to judge the proportion of women (or men)
attending the event. By using the masculine plural in both experiments, they
found that the expected proportion of women was higher if a female context
was established (congress of nutritional scientists vs. geo-physicists; meeting
of gymnasts vs. hockey players). In a study by Rothmund and Scheele (2004)
participants read a German text (900 words) on public baths written like an
article from a travel guide. Towards the end, participants had to complete a
sentence about conversations among the bathers, choosing among several
given alternatives. The alternatives referred to female-related topics, to male-
related topics or to topics that were not gender associated. In the first
experiment the text was on thermal spas in Budapest (male-related context),
in the second experiment it was on fun-baths at the isle of Sylt (family-related
context). The proportion of male-related completions dropped from 61% in
the first experiment to 47% in the second.
Stereotype bias in German was also shown by Irmen and Roßberg (2004,
Exp. 1). Participants read sentences about social groups denoted by
grammatically masculine nouns with different stereotypicality (Example
sentence, containing a stereotypical male noun: Mechanics have to be able to
handle tools). A second sentence contained a reference to that social group,
identifying the group members as female, male, or neither (Example
sentence, referring to women: That is why they should not have long
fingernails). Second-sentence reading times were measured and compared
with baseline reading times provided by a control group for the main task’s
second sentences. The experiment revealed that if the first sentence contained
a grammatically masculine plural noun referring to a stereotypically
feminine group (e.g., telephonists) reading time for the second sentence
was not influenced by the gender-related information and did not differ from
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
the baseline reading time. But in the two other conditions (first sentence
containing a gramm atically masculine plural noun referring to a stereo-
typically masculine group, e.g., foresters, or a stereotypically neutral group,
e.g., artists) male-related and neutral information was read faster than
female-related information and was read fast er than in the baseline
condition. Surprisingly, this latter effect was more pronounced in the
stereotypically neutral condition than in the stereotypically male conditio n.
The result that the combination of grammatical male gender and stereo-
typically female gender led to a gender-spread representation (i.e., not
facilitating any of the continuation types) suggests either that the gramma-
tical gender (masculine) does not play a dom inant role in forming a gender
representation or that the masculine form may be in certain instances
understood in a generic way. This study raised several problematic issues that
we address in this paper. First, the surprising result that there was a stronger
effect for the stereotypically neutral condition (which has not been found in
previous research) could hint at a problem with how they pretested the
stereotypicality of the role names. In their pretest they asked 30 participants
to rate the gender stereotypicality on a rating-scale. In the present paper, we
based our choice of role names on materials normed by Gabriel, Gygax,
Sarrasin, Garnham, and Oakhill (in press) who asked participants to
estimate to what extend the groups are actually (and not ‘typically’) made
up of women or men. Second, in their study, Irmen and Roßberg measured
reading times of sentences that referred to only one gender (e.g., exclusively
females: They then followed their female intuition, exclusively males: (. . . )
support from their wiv es). The problem with such exclusive target sentences is
that they only test whether the participants, when reading the sentences,
agree that the group can be made up only of women or only of men. In our
paper, we were interested in the ability of participants to maintain an open
gender representation, i.e., the group (e.g., politicians, authors, or nurses) can
comprise both men and women, which reflects more closely the idea of a
generic interpretation of the masculine plural.
Rothermund (1998) investigated the mental representation of short German
texts using a paradigm developed by McKoon and Ratcliff (1986). The texts
contained as their subjects a specific male or female designator (e.g., Herr
Almstadt or Frau Almstadt, control condition) or a masculine intended as
generic phrase in either the singular or the plural (e.g., Ein Lehrer or Lehrer).
Two test-phrases were constructed for each text, which did not appear in the
text but reflected a masculine or feminine understanding of the text. Gender-
specific associations were assessed by measuring the time it took participants
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
to reject the masculine (e.g., shaving) or feminine (e.g., making up one’s face)
distractor in a recognition task after reading the text. In this study, the
pattern of gender-specific associations for texts containing a masculine
intended as generic surprisingly depended on the grammatical number of the
texts. Reading a scenario containing a singular masculine intended as generic
increased rejection times for the masculine test-phrases, indicating a male
bias, while reading a scenario with a masculine intended as generic in the
plural increased rejection times for the feminine test -phrases, indicating a
female bias.
A plausible explanation for this somewhat puzzling effect is that in
German, determiners and pronouns in the plural (e.g., die Wissenschaftler,
sie, ihre; in English, the scientists, they, their) are identical to the feminine
singular determiner and pronoun (e.g., die Wissenschaftlerin, sie, ihre; in
English, the (female) scientist, she, her). Thus, in German, the determiner die,
used as a neutral plural determiner, could be read as the feminine singular
die. This confusion is not possible in French, as the neutral plural determiner
les is used only in the plural form. In this paper, we addressed this issue of
gender-marked versus non gender-marked determiners.
Most research supports the notion that the mascul ine plural in German is
strongly associated with representation of male persons, suggesting the
grammatical (male) gender to be of cen tral importance in building a mental
representation. But so far research has only partly elucidated the interaction
between stereotyping and the masculine bias of the generically intended
masculine. The experiments by Rothermund (1998) suggested that the
occurrence of determiners and pronouns is as impor tant as the gender of
the noun, whereas Irmen and Roßberg (2004) and Braun et al. (1998) hinted
at the importance of stereotypes in building gender representations (as in
English). In this paper, we investigated these issues by directly assessing the
interaction between stereotype and masculine biases in German and French,
which we compared with English that should establish a baseline for a
stereotype bias only, as it is difficult or impossible to construct a relevant
control condition in German and French in which stereotype biases act
alone, without the effects of morphological gender marking. There are three
possible ways in which stereotype and grammatical information might
interact. First, grammatical infor mation might completely override stereo-
type information. In this case, role names in French or German presented in
the masculine plural should result in the construction of a male-biased
representation of gender whereas role names in English should result in a
stereotypical representation of gender, as in Carreiras et al. (1996) and
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
Garnham et al. (2002). Such a result would be considered as a strong support
to the Whorfian hypothesis suggesting that language influences cognition. It
would provide evidence that although role names are characterised by
specific gender stereotypes, the grammatical form in which they appear
influences their representation. Second, stereotype information may com-
pletely override grammatical information. In this case, there should be no
differences in gender representations among the three languages, as they
should all reflect stereotype information, which would be inconsistent with
the Whorfian hypothesis. Third and finally, both types of information may
influence the construction of gender representations. In this case, the only
true gender-open representation would be for stereotypical female role
names in the masculine plural form in French and Ger man, as the two
influences should cancel each other out, and for stereotypically neutral role
names in English.
If grammatical information overrides stereotype information, it is also
possible, as suggested by Rothermund (1998), that the use of the determiner
die in German lessens the impact of the masculine plural form of the role
name. This assumption reflects the previously discussed issue of the sameness
in form in German of the plural and the feminine singular, die.
English sample. Thirty-six students from the University of Sussex took
part in this experiment. Each participant was paid £4. One participant had to
be removed from the analysis, as he did not understand the instructions.
French sample. Thirty-six students from the University of Fribourg took
part in this experiment for course credits. One participant had to be removed
from the analysis, as she did not understand the instructions.
German samp le. Thirty-six students from the University of Bern took
part in this experiment for course credits.
Materials and design
In each language, 36 experimental passages were constructed. Each passage
comprised two sentences. The first sentence introduced a group of people
using a role name in the plural form, and the second sentence specified that
there were some (but not exclusively) men or women in the group (i.e., it
provided a partial constraint on the sex of the people in the group). An
example of a passage is (1a) followed by (1b):
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
(1a) The social workers were walking through the station.
(1b) Since sunny weather was forecast several of the women werent
wearing a coat.
In all languages, there were 12 stereotypically female role names, 12
stereotypically male, and 12 neutral. The roles names were chosen from a
norming study conducted by Gabriel et al. (in press). In this norming study
(conducted in French, German, and English), participants were presented
with 126 role names, for which they had to indicate the percentage of men
and women they thought occupy these roles. We selected the most female
stereotyped role names (in French, German, and English) and matched them
with similarly strong male stereotyped role names and neutral stereotypes for
our study.
The chosen stereotypes are shown in Table 1. In the example above, the
first sentenc e presents a female stereotyped role name (1a) followed by a
second sentence mentioning women (1b).
For the French and German experiments, these sentences were translated
from English. The only difference in French and German was that the role
names were in the masculine plural form, which is, as a grammatical rule, not
to be interpreted as gender marked but as generic. For the above example, the
corresponding pairs in French (2a and 2b) and Ger man (3a and 3b) a re:
(2a) Les assistants sociaux marchaient dans la gare.
(2b) Du beau temps e´tant pre´vu plusieurs femmes n’avaient pas de veste.
(3a) Die Sozialarbeiter liefen durch den Bahnhof.
(3b) Wegen der scho
nen Wetterprognose trugen mehrere der Frauen
keine Jacke.
The detailed content of the first sentences differed, but because it was
difficult to completely vary the first sentences, we used six different content
types. The first sentenc e mentioned a group of people either (1) coming out
of a place, (2) waiting somewhere, (3) going into a place, (4) being
somewhere, (5) walking, or (6) going across a place. For each content type,
there were six specific contents, and six role names were randomly assigned
to each of six content types. There were, for example, six first sentences in
which the protagonists were walking, in one of which the protagonists were
walking through the station.
The second sentences differed first and most importantly in their mention
of women or men. The second sentences qualified the ‘men’ or the ‘women’
with one of some of the, most of the, several of the, few of the, one of the or the
majority of the.
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
Role names chosen from Gabriel et al. (in press) along with the proportion of men
evaluated by each language participant group
English % German % French %
Male stereotypes
Spies 73 Spione 67 Espions 74
Golfers 73 Golfspieler 68 Golfeurs 73
Politicians 71 Politiker 69 Politiciens 72
Police officers 63 Polizisten 69 Policiers 70
Statisticians 70 Statistiker 72 Statisticiens 74
Bosses 62 Arbeitgeber 72 Patrons 74
Computer specialists 70 Informatiker 79 Informaticiens 67
Surgeons 62 Chirurgen 75 Chirurgiens 75
Technicians 72 Techniker 78 Techniciens 75
Engineers 78 Ingenieure 78 Inge´nieurs 74
Physics students 56 Physikstudenten 81 Etudiants en physique 67
Pilots 70 Flieger 76 Aviateurs 74
Mean 68 74 72
Neutral Stereotypes
Singers 53 Sa
nger 45 Chanteurs 48
Pedestrians 49 Spazierga
nger 46 Promeneurs 52
Cinema goers 51 Kinobesucher 49 Spectateurs de cine´ma 50
Concert goers 47 Konzert-Zuho
rer 47 Auditeurs de concert 51
Schoolchildren 53 Schu
ler 48 Ecoliers 53
Spectators 55 Zuschauer 41 Spectateurs 51
Neighbours 50 Nachbarn 50 Voisins 50
Swimmers 50 Schwimmer 50 Nageurs 50
Tennis players 53 Tennisspieler 52 Joueurs de tennis 54
Authors 48 Autoren 52 Auteurs 54
Musicians 54 Musiker 50 Musiciens 59
Skiers 55 Skifahrer 53 Skieurs 55
Mean 52 49 52
Female stereotypes
Beauticians 29 Kosmetiker 11 Esthe´ticiens 18
Birth attendants 29 Geburtshelfer 11 Assistants maternels 18
Fortune tellers 32 Wahrsager 24 Diseurs de bonne aventure 28
Cashiers 39 Kassierer 27 Caissiers 24
Nurses 30 Krankenpfleger 24 Infirmiers 30
Hairdressers 48 Coiffeure 21 Coiffeurs 38
Psychology students 38 Psychologiestudenten 25 Etudiants en psychologie 33
Dieticians 39 Dia
tberater 27 Die´te´ticiens 37
Dressmakers 43 Schneider/Na
her 23 Couturiers 40
Dancers 32 Ta
nzer 33 Danseurs 29
Sales assistants 34 Verka
ufer 33 Vendeurs 37
Social workers 29 Sozialarbeiter 41 Assistants sociaux 33
Mean 35 24 30
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
Each participant saw 18 continuations about women, 6 following
sentences with a female stereotyped role name, 6 following sentences with
a neutral stereotyped role name and 6 following sentences with a male
stereotyped role name, and 18 about men. Furthermore, there were three
types of content, one based on different emotions (angry, sad, happy, and
joyful), one based on different weather conditions (sunny, put some sun
cream on, cloudy, and need an umbrella) and one based on different actions
(go, have a break, leave, and rest).
We conducted a pilot study in French (N" 47: 24 male and 23 female) and
German (N" 52: 21 male and 31 female) to en sure that the passages would
be interpreted as intended, with the women mentioned in the second sentence
being taken as a proper subset of the social workers (or other group)
mentioned in the first sentence. If this interpretation is not made, and the
women are taken to be coextensive with the social workers, a negative
judgement about the passage may simply reflect the fact that, if the social
workers are all women, they should be referred to using a feminine plural
noun phrase, and if a masculine plural is used the passage is incoherent.
Participants were asked to indicate whether for example in ‘The group (in
French Le groupe and in German Die Gruppe) came out of the train. Some of
the women were wearing a coat. the group (of people) comprises only men,
both men and women (i.e., mixed) or only women. Six lists were created to
ensure that each block of first sentences (exchanging the role names by the
neutral ‘the group’) was combined with each block of the continuations.
Each participant rated three sentence pairs, resulting in 297 ratings. From
these 267 (89.9%) answers were mixed, 22 (7.4%) were only men (18 in
combination with a male continuation; 4 in combination with a female
continuation) and 8 (2.7%) only women (6 in combination with a female
continuation, 2 in combination with a male continuation). Taking together
the only men and only women answers, these were not found to be related to
the sex of respondents, x
(N" 30, df" 1)"1.6, p" .20, but to the language,
(N" 30, df" 1)" 12.7, pB .001. Such exclusive answers were more often
given in German ( f" 25) than in French (f" 5). Neither the form of the
continuation (e.g., Some of the) nor the content of the continuation (e.g.,
wearing a coat) was related to the exclusively male/female answers (all ps!
.50). The results of the pilot revealed a low rate of deviations from the
intended perception of the group as being gender-mixed. There was a very
slight tendency towards a male biased perception, as the response category
only men was chosen more often than the category only women. However, as
this was merely a very slight tendency, we be lieve that it was unlikely to have
any subsequent effect. In sum, the vast majority of participants in the pilot
study interpreted the passages as we had intended in that they imposed a
non-exclusive inter pretation of the second sentence: i.e., some, but not all, of
the people were women, as in the example above.
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
In our experiment each participant saw 12 continuations of each type.
Across the experiment, we created six lists to ensure that each role name was
equally often followed by men and women, and by sentences portraying
different situations. The latter was done to avoid any plausibility effect (i.e., it
may be more plausible for nurses, for example, to walk through a station than
to put some sun cream on). Eac h participant saw one list and hence saw all
possible continuations (3 groups of continuations# 2 gender types), and,
across the experiment, each role name was followed by all possible
continuations. The crucial experimental manipulations were the language
in which the study took place (French, German, English), which varied
between participants but within items, the nature of the role name (female,
neutral, male stereotyped), which varied between items, but within partici-
pants, and whether the continuation mentioned men or women (and hence
whether it matched the stereotype, if any, of the role name). This factor
varied within both items and participants.
In all experimental conditions the intended response was yes (the second
sentence is a sensible continuation of the first). To ensure that the
participants read the passages, 36 filler texts, requiring no answers were
constructed. These filler pairs were similar to the experimental ones (but
using different role names), though they were intended to elicit a clear no
answer. An (E nglish) example of such a pair is (4a and 4b):
(4a) The professors were taking a break in the sun.
(4b) Due to the bad weather the majority of the women had an umbrella.
The filler pairs were the same in each list and were randomly interspersed
with the experimental items.
The passages were presented on a Macintosh computer (Power Macintosh
4400 for German and French, an iMac for English) using the PsyScope
Software (Cohen, MacWhinney, Flatt, & Provost, 1993). Responses were
collected using a button box attached to the computer, which permits
millisecond accuracy.
The participants were tested individually in a small quiet room. Their task
was to read each passage, presented one sentence at a time, and to decide
whether the second sentence was a sensible continuation of the first one. In
French and German, we used the terms continuation possible and mo
Fortsetzung, which we judged as semantically closest to the English sensible.
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
The participants in all languages were asked to make a prompt decision,
based on their first impression and not on a prolonged reflection. A prompt
(i.e., **ready?**, **Preˆt?**, **Bereit?**) appeared on the screen before each
passage. The participants pressed the yes button to make the first sentence
appear, and then pressed the yes button again to make the second sentence
(target sentence) appear. They then had to make a prompt decision by
pressing either the yes button (i.e., I think it’s a sensible continuation) or the
no button (i.e., I don’t think it’s a sensible continuation). Participants were
asked to keep the index finger of their dominant hand on the yes button and
the index finger of their non-dominant hand on the no button.
Before the experiment, there was a trial session with 12 passages, to
familiarise the participants with the procedure.
We recorded the participants’ responses (i.e., yes or no) and the time it took
them to respond. Both measures were intended to evaluate the ease of
integration of the information in the target sentence. In essence, if
participants have trouble integrating that information, they are more likely
to respond no. In addition, if they respond yes, it should take them longer to
respond if they have trouble integrating the information. Results are
presented for the English sample first. The rationale behind this is that the
English establishes both a control for the manipulation of the stereotypi-
cality of the role names as well as a baseline for the stereotype bias. Ceteris
paribus, variations in the French and German results can be ascribed to
influences of the grammar.
Proportion of positive judgements
The proportion of positive judgements for the experimental passages is
shown in Table 2. All the data were analysed using ANOVAs considering first
participants and then items as random effects.
English experiment. There was neither a main effect of Continuation
(men vs. women mentioned), F
(1, 34)B1; F
(1, 33)B 1, nor a main effect of
Stereotype, F
(2, 68)" 1.44, p! .05; F
(2, 33)" 1.11, p! .05, but there was
an interaction between the two variables, F
(2, 68)" 15.81, pB .05; F
33)" 20.19, pB .05. As shown in Table 2, English participants’ responses
varied according to the stereotypicality of the role names. After sentences
containing female stereotyped role names, there were more positive judge-
ments when the second sentence mentioned women (.88) than when
it mentioned men (.65). Conversely, after sentences containing male
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
stereotyped role names, there were more positive judgements when the
second sentence mentioned men (.85) than when it mentioned women (.66).
After sentences containing neutral role names, there was no difference
between second sentences mentioning women (.81) and second sentences
mentioning men (.81).
French experiment. There was a significant main effect of Continuation,
with the men continuations showing a higher proportion of yes an swers
(M" .78) than the women continuations (M" .58), F
(1, 34)" 40.36, pB .05;
(1, 33)" 42.91, pB .05. There was no effect of stereotype, F
(2, 68)B 1;
(2, 33)" 1.20, p! .05, nor an interaction, F
(2, 68)" 1.18, p! .05; F
33)B 1. These results indicate that, regardless of the role name’s stereotype,
participants were biased towards a male representation.
German experiment. As in French, there was a main effect of Continua-
tion, with more positive judgements when the continuation sentences
mentioned men (M" .69) than when they mentioned women (M" .40),
(1, 35)" 55.4, pB .05; F
(1, 33" 88.20, pB .05. There was no main effect
of Stereotype, F
(2, 70)" 2.25, p! .05; F
(2, 33)" 1.45, p! .05, nor an
interaction, F
(2, 70)B 1; F
(2, 33)B 1. As in the French experiment,
participants were strongly biased towards a male representation when
presented with role names in the masculine plural form, even when the
role names were stereotypically female. In these results it is clear that the
determiner played no role in establishing a gender representation. This is
further corroborated by the close to significant Continuation (women/
men)# Language (French/German) interaction, F
(1, 69)" 3.00, p" .088;
(1, 66)" 4.66, pB .05 when analysing the French and the German data
together: German speaking participants were even more likely to produce a
male representation when the role names were in the masculine plural than
Proportion of positive judgements across languages and conditions
Language Continuation Female Male Neutral
French Men .77 .83 .73
Women .59 .58 .56
German Men .65 .69 .72
Women .40 .35 .45
English Men .65 .85 .81
Women .88 .66 .81
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
the French speaking participants, even though the ambiguity of the German
article die might have suggested the opposite effect.
Overall, these results indicate a very strong effect of the masculine
(intended as generic) biasing the participants’ gender representation towards
a male representation. In the French and the German speaking samples,
stereotypicality had no effect on the representation.
Judgement times
Only response times for positive judgements were analysed. Judgement times
that were 2.5 SD or more above each participant’s mean were replaced by the
2.5 SD cut-off (2.4 4% of French, 1.98% of German, and 1.87% of English
times were affected). In addition some participants did not produce any yes
answers in certain conditions. Such missing values were replaced by
condition means (5.45% of the data). The mean times to make positive
judgements are shown in Table 3.
English experiment. There were only small differences in decision times.
There was no effect of Continuation, F
(1, 34)" 2.33, p! .05; F
(1, 33)"
1.89, p!.05, no effect of Stereotype, F
(2, 68)B 1; F
(2, 33)B1, and no
interaction, F
(2, 68)B 1; F
(2, 33)" 3.14, p! .05.
French experiment. Although the trends in the results support the
findings for the judgements, there was only a marginally significant main
effect of Continuation (w ith responses to men being faster than those to
women), and only when participants were considered as the random effect,
(1, 34)" 4.01, p" .053, F
(1, 33)" 1.16, p! .05. There was no effect of
Mean positive judgement times (and standard deviations in parentheses) across
languages and conditions
Language Continuation Female Male Neutral
French Men 3665 (1626) 3523 (1380) 3701 (1102)
Women 3871 (1486) 3863 (1794) 3873 (1355)
German Men 3149 (1084) 3066 (1207) 3132 (1127)
Women 3464 (1098) 3471 (1081) 3479 (1305)
English Men 2749 (1004) 2810 (935) 2885 (1476)
Women 2913 (1212) 2954 (1164) 2910 (961)
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
Stereotype, F
(2, 68)B 1; F
(2, 33)B 1, nor any interaction, F
(2, 68)" B 1;
(2, 33)B 1.
German experiment. A significant Continuation effect, F
(1, 35)" 9.15,
pB .05; F
(1, 33)" 21.32, pB .05, with responses to men being faster than
those to women fully supported the results found in the judgem ents. There
was neither effect of Stereotype, F
(2, 70)B 1; F
(2, 33)B1, nor an
interaction, F
(2, 70)B 1; F
(2, 33)B 1. When the French and the German
data were analysed together, there was a main effect of Continuation, F
70)" 12.77, pB .05; F
(1, 69)" 12.69, pB .05, further supporting the male
biasing effect of the masculine plural form, but no interaction of Continua-
tion and Language, although the judged differences between women and men
continuations in the German sample were numerically bigger than in the
In this experiment, conducted in English, French, and German, we
investigated the interaction between stereotype and masculine (supposedly
generic) in the construction of a representation of gender when reading role
names (e.g., musicians). In French and German, when a group of people is
presented, it is common to use the masculine plural form (e.g., les musicians,
die Musiker) but not the feminine form, even if there is a majority of women.
This masculine plural form is supposed to result in a gender-open or gender-
spread representation. In English, gender-marked role names are rare, so
gender representations from role names should be influenced mainly by
stereotype (i.e., if beauticians are stereotyped as female, reading the role
name should lead to a female-biased gender representation). As discussed in
the Introduction, there are three ways in which stereotype and grammatical
information might interact. First, grammatical information may override
stereotype information. Second and conversely, stereotype information may
override grammatical information. Third and finally, both could influence
gender representations.
When no grammatical gender information was available, in English, the
mental representation of gender was solely based on stereotype information.
But when role names were written in the masculine plural form, the results
indicated that grammatical information overrode stereotype information in
constructing a mental representation of gender. This override happened both
in French and in German. Note that in German, the fact that the determiner
die is used in both the feminine singular and the plural did not have any
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
impact on the male bias in the representation of the people referred to by the
role names.
Three main conclusions can be drawn from this experiment. First, when
no mark of gender is provided by role names or their accompanying definite
articles, the representation of gender is based on stereotypicality. We can
draw this conclusion from our English conditions. Such a result further
supports findings such as those of Carreiras et al. (1996), Garnham et al.
(2002), and Oakhill et al. (2005), which suggest that people draw inferences
about the gender of people immediately when they encounter a stereotyped
role name. Second, when a grammatical mark of gender is provided, the
representation of gender is based on that mark of gender, and not on
stereotype information. The influence of grammatical features on gender
representation has been previously suggested by studies in different
languages (e.g., French: Cole´ & Segui, 1994; Italian: Bates et al., 1996;
Spanish: Flahert y, 2001). In a Whorfian framework, such a result provides a
strong indication that language influences cognition. Not only for neutral
but even for female stereotyped role names, the grammatical form of the role
names overrode the stereotypicality. This is not to say that stereotypicality
does not influence gender representation, but in the case where grammatical
gender is available, such stereotypicality has little influence. As in Flaherty
(2001), grammatical gender seemed to provide sufficient information for
readers to build their mental representation of gender. Third, our findings
indicate that masculine forms intended as generic are typically not
interpreted as such. In this experiment, we demonstrated that the use of
the masculine plural does not lead to a gender-open or gender-spread
representation, but to a specifically male representation. Put differently, a
German phrase such as die Musiker, which is ambiguous between a
masculine and a generic intended plural is by default interpreted as a
masculine plural, and presumably this interpretation would only be changed
if there was a specific indication that women were involved and, hence, that
the generic interpretation was intended. This finding has critical social
implications, especially in relation to the acknowledgement of women in
society. We believe that our results show that the so-called generic use of the
masculine biases gender representations in a way that is discriminatory to
women. This is especially true as French and German newspapers still
display some job announcements in the masculine form, therefore, we
believe, suggesting that women are not suitable candidates. In 2002, the
Acade´mie Franc
aise deemed it necessary to publish a document stating that
the use of the masculine as generic was totally gender-open and that us ing
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
different grammatical forms was unnecessary.
Our findings demonstrate
that the Acade´mie Franc
aise have clearly underestimated the influence of the
masculine form in leading readers towards a male representation.
Our results differ from those of Rothermund (1998), Braun et al. (1998) ,
and Irmen and Roßberg (2004, Exp. 1). First we did not find any influence of
the determiner die in German. If anything, our results suggest that the
influence of the masculine form of the role name is stronger in German than
in French. Rothermund (1998), who used a different methodology, reported
that the pattern of gender-specific associations for texts containing a
masculine intended as generic depended on the grammatical number of the
phrase. Reading a scenario containing a singular masculine intended as
generic increased rejection times for the masculine test-phrases, indicating a
male bias, while reading a scenario with a masculine intended as generic in
the plural increased rejection times for the femin ine test-phrases, indicating a
female bias. Based on the author’s explanation that this effect originates in
the German gender neutral plural determiner die having the same form as the
feminine determiner, we hypothesised a possible male bias to be weaker in
German compared to French, as the gender neutral plural determiner in
French is different from the feminine determiner. In contrast, our results
suggested a stronger male bias in German than in French. This is not to say
that the determiner die had no influence, but its infl uence, if any, was
definitely weaker than the masculine bias introduced by the use of a role
name written in the masculine plural form.
Second, in contrast to the findings of Braun et al. (1998) and Rothmund
and Scheele (2004), stereotypicality had no significant effect in our French
and German samples. This might be explained by the aforementioned
authors using a different research paradigm. First, participants read longer
texts in which various (neutral and stereotypical) role names were repeated.
Second, the dependent variable was the participants’ explicit estimate of the
gender-composition of the group introduced by the texts. In contrast, our
stimuli were shorter and we assessed reactions that were less elaborated and
less obviously related to the gender-composition which probably meant that
the influence of the stereotypes was not sufficiently strong to interact with
the grammatical bias under such experimental conditions.
Two issues already mentioned in the Introduction and Methods sections,
deserve attention. First, the decrease in positive judgements in the French
and German samples may not, as we inferred, be due to the fact that the role
The Acade´mie Franc
aise also stated that using two grammatical forms for a role name (les
musiciens et musiciennes) rendered the text difficult to read. Gygax and Gesto (2007) have replied
by showing that although the first encounters of such terms does slow down reading (taking into
account sentence length), there is a very fast habituation effect, leading to a perfectly normal
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
names in the masculine form were taken as mainly representing men. Instead,
it may be that participants interpreted phrases in the second sentences, for
example . . . some of the women . . . ’, as constraining the interpretation of
the first sentence to mean that the group must be exclusively composed of
women, i.e., they interpreted ‘the women’ as coreferential and coextensive
with the set of individuals (e.g., ‘the social workers’) introduced in the first
sentence. This would mean that in French and German participants had
trouble in processing the second sentence, and considered it a poor
continuation of the first sentence, because a group exclusively composed of
women should be introduced, in the first sentence, by a role name in the
feminine form. Thus, the negative judgements would be arising because of a
perceived grammatical problem with the texts, not because ‘the social
workers’ had been inter preted as a group of men, and that the reference to
‘women’ was unexpected.
Although this kind of interpretation might have occurred, our data, as
well as the data from the pilot study mentioned in the Method section,
render this possibility unlikely. If individuals interpret female continuations
in this sense, it is likely that they do so consistently. If so, the acceptance of
female continuations should drop to zero (or close to zero) ! at least for some
participants. But inspection of our data revealed no such case. Furthermore,
in our pilot study female continuations were only taken to indicate that the
group comprised exclusively of women in six instances (4%). This effect is
not sufficiently strong to explain the entire continuation effect found in the
French and the Ger man sample. Nevertheless, future research might benefit
from including measures to assess individual differences that may lead to an
inclusive or exclusive interpretation of phrases such as ‘some of the women’.
The second issue refers to the use of English as a comparison language for
French and German. A possible criticism of our design is that instead of
morphological differences among the three languages we studied, other
aspects either of the languages themselves or of the cultural backgrounds of
the speakers co ntributed to the effects we observed. One could for example
argue that the ‘people" male’ bias (i.e., the effect that neutral terms have a
masculine connotation; Silvera, 1980) is more deeply ingrained in the Swi ss
compared with the English culture and therefore the Swi ss participants
would in general have difficulties in responding to the female continuations.
Although we are not aware of research showing such cultural differences or
displaying other language-related differences, e.g., in the use of the role
names employed, we cannot definitely exclude such alternative explanations.
However, given the general broad similarities in European cultures, we
believe it to be rather unlikely. Note that if this was the case, our norming
study (i.e., Gabriel et al., in press, in which participants were asked to rate the
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
perceived percentage of men and women in 120 roles), should have
demonstrated a large male bias in the French and German samples, as
opposed to our English sampl e. This was, as shown in Table 1, not the case.
Furthermore, it would be difficult to empirically rule out such an alternative
explanation, as neither German nor French allows for a ‘true’ gender-neutral
control condition: all role-names, even those that do not have a feminine
specific form such as people (German: Personen, French: des personnes) or
human-beings (German: Menschen, French: des hommes) nevertheless have a
grammatical gender.
Although our sentence evaluation paradigm cannot truly distinguish
between automatic and strategic processes, previous research has suggested
that automatic processes are more appropriate for explaining the construc-
tion of representations of gender (e.g., Garnham et al., 2002). Garnham et al.
(2002) presented people with sentences that did not contain any morpho-
syntactic cues to gender. Their sentences comprised elements that could be
defined as pieces of probabilistic information (Garnham et al., 2002, p. 445)
about the gender of the characters. Their results indicated that people
nevertheless constructed a mental representation of gender. Oakhill et al.’s
(2005) results also suggest an automatic element to the representation of
stereotyped gender information, since those authors found that such
information was difficult to suppress, even when it was detrimental to
performance. Our English results corroborated their findings, suggesting that
our participants’ representation processes were similar. Nevertheless, whether
our participants behaved strategically or automatically, our results still
provide evidence that people have difficulty in mapping women onto role
names in the masculine plural form, regardless of the role name’s stereo-
typicality. Therefore, our main point about the biasing influence of the
masculine form holds true, whether the processes under investigation were
automatic or not.
To conclude, we believe that our results on the overriding effect of
grammar over stereotypicality, taken together with previous research on the
automatic representation of gen der, demonstrate that people construct
representations of gender, and that they base their representations on
grammar when available, and on stereotype information when grammatical
cues are not available. Future research, however, might investigate the
influence of other variables (e.g., additional gender-related context informa-
tion or additional gender-marked grammatical features) that could moderate
the strong masculine bias imposed by the masculine form in French and
Manuscript received January 2007
Revised manuscript received September 2007
First published online February 2008
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
Acade´mie Franc
aise (2002). Fe´minisation des noms de me´tiers, fonctions, grades et titres. Retrieved
May 24, 2005, from
Bates, E., Devescovi, A., Hernandez., A., & Pizzamiglio, L. (1996). Gender priming in Italian.
Perception and Psychophysics, 58, 992!1004.
Baudino, C. (2001). Politique de la langue et diffe´rence sexuelle: La politisation du genre des noms de
me´tiers. Paris: L’Harmattan.
Braun, F. (1996). Das grosse I und seine Schwestern-eine kritische Bewertung [The capital I and its
sisters ! a critical evaluation]. Deutschunterricht, 48(1), 54!62.
Braun, F., Gottburgsen, A., Sczesny, S., & Stahlberg, D. (1998). Ko
nnen Geophysiker Frauen sein?
Generische Personenbezeichnungen im Deutschen [Can geophysicians be women? Generic
terms in German]. Zeitschrift fu
r germanistische Linguistik, 26, 265!283.
Bussmann, H. (1995). Das Genus, die Grammatik und-der Mensch: Geschlechterdifferenz in der
Sprachwissenschaft [Gender differences in linguistics]. In H. Bussmann & R. Hof (Eds.), Genus.
Zur Geschlechterdifferenz in den Kulturwissenschaften (pp. 114!160). Stuttgart, Germany:
Carreiras, M., Garnham, A., Oakhill, J., & Cain, K. (1996). The use of stereotypical gender
information in constructing a mental model: Evidence from English and Spanish. Quarterly
Journal of Experimental Psychology, 49(3), 639!663.
Cohen, J.D., MacWhinney, B., Flatt, M., & Provost, J. (1993). PsyScope: A new graphic interactive
environment for designing psychology experiments. Behavioral Research Methods, Instruments
and Computers, 25, 257!271.
Cole´, P., & Segui, J. (1994). Grammatical incongruence and vocabulary types. Memory and
Cognition, 22, 387!394.
Duden (2005) Die Grammatik (7th ed.) [The German Grammar]. Mannheim, Germany:
Duffy, S.A., & Keir, J.A. (2004). Violating sterotypes: Eye movements and comprehension processes
when text conflicts with world knowledge. Memory and Cognition, 32, 551!559.
Flaherty, M. (2001). How a language gender system creeps into perception. Journal of Cross-
Cultural Psychology, 32(1), 18!31.
Gabriel, U., Gygax, P., Sarrasin, O., Garnham, A., & Oakhill, J. (in press). Au-pairs are rarely male:
Role names’ gender stereotype information across three languages. Behavior Research Methods,
Instruments and Computers.
Gabriel, U., & Mellenberger, F. (2004). Exchanging the generic masculine for gender-balanced
forms: the impact of context valence. Swiss Journal of Psychology, 63(4), 273!278.
Garnham, A., Oakhill, J., & Reynolds, D. (2002). Are inferences from stereotyped role names to
characters’ gender made elaboratively? Memory and Cognition, 30(3), 439!446.
Gygax, P., & Gesto, N. (2007). Lourdeur de texte et fe´minisation: Une re´ponse a` l’Acade´mie
aise. L’Anne´e Psychologique, 107, 233!250.
Hardin, C., & Banaji, M. R. (1993). The influence of language on thought. Social Cognition, 11,
Heise, E. (2000). Sind Frauen mitgemeint? Eine empirische Untersuchung zum Versta
ndnis des
generischen Maskulinums und seiner Alternativen [Are women included? An empirical study of
the generic masculine and its alternatives]. Sprache und Kognition, 19(1-2), 3!13.
Heise, E. (2003). Auch einfu
hlsame Studenten sind Ma
nner: Das generische Maskulinum und die
mentale Repra
sentation von Personen [Sensitive students are men too: The generic masculine
and the mental representation of persons]. Verhaltenstherapie und psychosoziale Praxis, 35(2),
Irmen, L., & Ko
hncke, A. (1996). Zur Psychologie des ‘generischen Maskulinums [The psychology
of the generic masculine]. Sprache und Kognition, 15(3), 152!166.
Downloaded By: [Bibliothèque cantonale et universitaire - Fribourg] At: 12:58 1 April 2008
Irmen, L., & Roßberg, N. (2004). Gender markedness of language: The impact of grammatical and
nonlinguistic information on the mental representation of person information. Journal of
Language and Social Psychology, 23, 272!307.
Kennison, S.M., & Trofe, J.L. (2003). Comprehending pronouns: A role for word-specific gender
stereotype information. Journal of Psycholinguistic Research, 32, 355!378.
Levinson, S.C., Kita, S., Haun, D.B.M., & Rasch, B.H. (2002). Returning the tables: Language
affects spatial reasoning. Cognition, 84(2), 155!188.
Li, P., & Gleitman, L. (2002). Turning the tables: language and spatial reasoning. Cognition, 83(3),
McKoon, G., & Ratcliff, R. (1986). Inferences about predictable events. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 12, 82!91.
Oakhill, J., Garnham, A., & Reynolds, D. (2005). Immediate activation of stereotypical gender
information. Memory and Cognition, 33(6), 972!983.
Peyer, A., & Wyss, E.L. (1998). JazzmusikerInnen-weder Asketen noch Mu
esli-Fifis’. Feministische
Sprachkritik in der Schweiz, ein U
berblick [‘Jazz-musicians ! neither ascetics nor cereals-picker’
! Feministic language critic in Switzerland, a review]. Germanistische Linguistik, 139!140, 117!
Rothermund, K. (1998). Automatische geschlechtsspezifische Assoziationen beim Lesen von Texten
mit geschlechtseindeutigen und generisch maskulinen Text-Subjekten [Automatic gender-
specific associations to texts containing gender-specific and masculine generic text subjects].
Sprache und Kognition, 17(4), 183!198.
Rothmund, J., & Scheele, B. (2004). Personenbezeichnungsmodelle auf dem Pru
fstand. Lo
glichkeiten fu
r das Genus-Sexus-Problem auf Textebene. [Putting gender-neutral reference
terms to the test: Constructive solutions to the problem of grammatical vs. referential gender on
the textual level]. Zeitschrift fu
r Psychologie, 212(1), 40!54.
Scheele, B., & Gauler, E. (1993). Wa
hlen Wissenschaftler ihre Probleme anders aus als
WissenschaftlerInnen? Das Genus-Sexus-Problem als paradigmatischer Fall der linguistischen
tsthese [Do male and female scientists differ in their selection of research problems?
The problem of gender-sex bias as a paradigmatic example of linguistic relativity]. Sprache und
Kognition, 12(2), 59!72.
Silvera, J. (1980). Generic masculine words and thinking. Womens Studies International Quarterly,
3, 165!178.
Stahlberg, D., & Sczesny, S. (2001). Effekte des generischen Maskulinums und alternativer
Sprachformen auf den gedanklichen Einbezug von Frauen [Effects 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.
Stahlberg, D., Sczesny, S., & Braun, F. (2001). Name your favorite musician: Effects of masculine
generics and of their alternatives in German. Journal of Language and Social Psychology, 20(4),
Sturt, P. (2003). The time-course of the application of binding constraints in reference resolution.
Journal of Memory and Language, 48, 542!562.
... Instead, reading or hearing generic masculine forms causes men to be more strongly represented than women (for reviews, see Gabriel et al., 2018;Stahlberg et al., 2007). For example, after reading generic masculine forms, participants were faster and more accurate to react to male compared to female exemplars and subgroups (e.g., Garnham and Yakovlev, 2015;Gygax et al., 2008Gygax et al., , 2012Irmen and Roßberg, 2004). Analogue results have been observed when participants estimated the proportion of women compared to men in a group (Braun et al., 1998;Hansen et al., 2016), named exemplars from occupational groups ; see also Gabriel & Mellenberger, 2004), reported the gender of previously imagined people (Hamilton, 1988;Gastil, 1990), decided on first names (Heise, 2000;Kaufmann & Bohner, 2014;Vervecken et al., 2013) or selected pictures (Bailey & LaFrance, 2017;Schneider & Hacker, 1973). ...
... All materials, data, and analyses for both experiments are available at The study employed a modified version of the materials used by Gygax et al. (2008). Specifically, participants evaluated 72 sentence pairs, 36 target pairs and 36 filler pairs. ...
... Specifically, participants evaluated 72 sentence pairs, 36 target pairs and 36 filler pairs. In the target pairs, the first sentence described a group of people using a category with a roughly balanced gender stereotype (taken from Gygax et al., 2008;Kennison and Trofe, 2003). The categories were occupational (e.g., pharmacists) or non-occupational (e.g., neighbors). ...
Full-text available
In many languages, masculine language forms are not only used to designate the male gender but also to operate in a generic fashion. This dual function has been found to lead to male biased representations when people encounter the generic masculine. In German, the now predominant substitute is the gender star form (e.g., Athlet*innen). In two experiments, we examined gender representations elicited when reading the gender star form (vs. generic masculine vs. pair forms). We found that, following the generic masculine, continuations about men (vs. women) were more frequently and more quickly judged to be compatible, replicating the male bias, even though participants were informed about the generic intention. Following the gender star form, a female bias in judgments (both Studies) and speed (only Study 2) occurred, which was somewhat smaller. Representations were most balanced when both male and female forms were mentioned.
... Linguistic research has repeatedly demonstrated that masculine generics in German show a masculine bias (e.g. Gabriel et al., 2008;Gygax et al., 2008;Irmen & Kurovskaja, 2010;Koch, 2021;Misersky et al., 2019;Stahlberg & Sczesny, 2001). That is, grammatically masculine role-nouns such as Anwalt 'lawyer' can refer to men and women but may favour an interpretation in which mostly or only men are considered as potential referents (e.g. ...
Conference Paper
Full-text available
Linguistic research has repeatedly demonstrated that masculine generics in German show a masculine bias (e.g. Gabriel et al., 2008; Gygax et al., 2008; Irmen & Kurovskaja, 2010; Koch, 2021; Misersky et al., 2019; Stahlberg & Sczesny, 2001). That is, grammatically masculine role-nouns such as Anwalt ‘lawyer’ can refer to men and women but may favour an interpretation in which mostly or only men are considered as potential referents (e.g. Misersky et al., 2019). However, research on this matter faces two major issues. First, previous studies have used numerous of such masculine generics and their feminine counterparts to gain insights into their semantics without accounting for language external but potentially confounding influences such as stereotypicality. Second, the majority of studies finds the aforementioned masculine bias; however, very few studies offer a theoretical account of its underlying representations (e.g. Irmen & Linner, 2005). To this date, no investigation has been made to discover whether there is a connection between the male bias and the representation of masculine generics in the mental lexicon. The present paper offers a solution to both issues. First, the role nouns under investigation are those for which stereotypicality ratings are available (Gabriel et al., 2008). Language external factors are incorporated in the analysis by such ratings. Second, using the general ideas of distributional semantics (Harris, 1954) as well as naïve discriminative learning (e.g. Baayen & Ramscar, 2015) and linear discriminative learning (e.g. Baayen et al., 2019) the underlying nature of masculine generics and counterparts is investigated. The proposed analysis aims at exploring the following question: How semantically similar are masculine generics and their explicitly masculine and feminine counterparts when taking into account stereotypicality? The following method was employed to tackle this question. An 830,000 sentence (1.7 million words) corpus of contemporary German was created using the Leipzig Corpora Collection (Goldhahn et al., 2012). The corpus included 113 target word pairs which were based on the set of words provided by Gabriel et al. (2008). All target word occurrences were checked for their usage, i.e. whether they were generically or explicitly intended, and annotated accordingly. The corpus was then used to train semantic vectors based on the Rescorla-Wagner equation (Wagner & Rescorla, 1972) as implemented by naïve discriminative learning. Finally, the resulting semantic vectors were then used to train an implementation of linear discriminative learning. Taking a closer look at the semantic vectors as created by the naïve discriminative learning algorithm, one finds that for the singular generic and explicit masculine role nouns are semantically closest with a mean cosine similarity value of approx. 0.98. Explicit masculine and feminine role nouns are less similar (approx. 0.94), and masculine generic and explicit feminine are least similar (0.93). These differences are highly significant (Mann-Whitney-Wilcoxon, p < 2.16e-16) and even more pronounced in the plural. To account for potential influences of stereotypicality, measures derived from the linear discriminative learning implementation were modelled by the stereotypicality measure as provided by Gabriel et al. (2008). As a result one finds that generic and explicit masculines are highly similar in terms of their semantic activation diversity and semantic neighbourhood size, while explicit feminines show higher activation diversities and denser neighbourhoods. Our results indicate that there is a male bias in masculine generics in German as they exhibit highly similar semantic vectors as well as highly similar levels of semantic activation diversity and semantic neighbourhood size, even when controlled for their stereotypicality. Consequently, generic and explicit masculines show similar underlying representations, while the representations of explicit feminines are less similar. Thus, even though the use of masculine generics might be intended as generic, their resonance with the lexicon, that is more specifically their similarity with explicit masculines, leads to an overall biased association towards male referents.
... Studies employing ERPs (Event Related Potentials, a measure used to determine the difficulty of processing certain stimuli) found that comprehending linguistic information consistent with stereotypical gender-expectations (e.g., feminine pronouns with the role descriptor nurse) is more fluent than comprehending inconsistent gendered information (e.g., masculine pronouns with nurse, see Misersky et al. 2019). Role nouns seem to be infused with gendered stereotypes even in the absence of grammatical cues denoting the gender of the referent (Gygax et al. 2008), so that English-speaking participants are more likely to associate mathematician with men than with women, even though the role-noun is not gendered in English (Misersky et al. 2014). ...
Full-text available
Biases in cognition are ubiquitous. Social psychologists suggested biases and stereotypes serve a multifarious set of cognitive goals, while at the same time stressing their potential harmfulness. Recently, biases and stereotypes became the purview of heated debates in the machine learning community too. Researchers and developers are becoming increasingly aware of the fact that some biases, like gender and race biases, are entrenched in the algorithms some AI applications rely upon. Here, taking into account several existing approaches that address the problem of implicit biases and stereotypes, we propose that a strategy to cope with this phenomenon is to unmask those found in AI systems by understanding their cognitive dimension, rather than simply trying to correct algorithms. To this extent, we present a discussion bridging together findings from cognitive science and insights from machine learning that can be integrated in a state-of-the-art semantic network. Remarkably, this resource can be of assistance to scholars (e.g., cognitive and computer scientists) while at the same time contributing to refine AI regulations affecting social life. We show how only through a thorough understanding of the cognitive processes leading to biases, and through an interdisciplinary effort, we can make the best of AI technology.
... Also, as mentioned above, the studies' results did not fully converge, which may be due to the differences in study designs and manipulations. Another set of studies used still other methods such as sentence evaluations, ratings of sentence correctness, reading times, or eye-tracking (e.g., Gygax et al., 2008;Irmen & Kurovskaja, 2010;Irmen & Roßberg, 2004). These studies are interesting in themselves, as they demonstrate an influence of masculine generics on specific aspects of cognition with different paradigms and measures, but they do not constitute replications of the study by . ...
Full-text available
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.
Es gibt ein Argument, das in keiner Diskussion über das Gendern fehlt: Behauptet wird, dass Menschen vorrangig an Männer denken, wenn sie Formulierungen im generischen Maskulinum lesen/hören. Empirische Untersuchungen hätten zweifelsfrei erbracht, dass generische Maskulina überwiegend „innere Bilder“ von männlichen Akteuren erzeugten (Braun et al., 2005; Heise, 2000; Irmen & Köhncke, 1996; Irmen & Linner, 2005; Irmen & Roßberg, 2004; Klein, 1988; Rothermund, 1998; Rummler, 1995; Scheele & Gauler, 1993). Die Studien sollen belegen, dass das generische Maskulinum de facto nicht generisch verstanden wird, sondern zu einem male bias führt – einer Dominanz von Männern in der Vorstellungswelt der Rezipienten.
Full-text available
Der Band vereint neueste, empirisch abgesicherte Forschung zu den Zusammenhängen zwischen Genus, Sexus und der sozialen Kategorie Gender. Reflektiert werden sprachliche Erscheinungen auf der Wortebene, im Bereich der Syntax und Textkohärenz und soziopragmatische sowie diskurstheoretische Fragen. Die wortzentrierten Beiträge umfassen Untersuchungen zu genderstereotypen Bedeutungsangaben in Wörterbüchern, zur in-Movierung bei Anglizismen und zur Reichweite des Genus-Sexus-Prinzips bei Tierbezeichnungen. Syntaktische Fragestellungen zielen auf anaphorische Wiederaufnahmephänomene und deren innersprachliche und pragmatische Bedingungen. Dies betrifft hybride Nomina und Epikoina (das Kind > es/?er; das Mädchen > es/sie) ebenso wie Koreferenzphänomene auf der Phrasenebene (Herr Meier und Frau Schmid > die beiden Lehrer/*Lehrerinnen). Soziopragmatische Aspekte werden anhand dialektaler Genus-Sexus-Diskordanzen vom Typ das Emma beleuchtet; auch syntaktische Serialisierungspräferenzen (Mann und Frau, Mama und Papa) variieren entlang sozio-pragmatischer Faktoren. Auf Diskurse rekurriert die Untersuchung zu typischen Argumentationsmustern gegen genderneutrale Sprache. Die Einstellungen nicht-binärer Personen zu gendersensiblem Sprachgebrauch spiegeln aktuelle sprachkritische Diskurse. Der Band ist eine hervorragende Informationsquelle für alle, die qualitativ hochwertige linguistische Argumentation und empirische Daten zu den komplexen Zusammenhängen zwischen Sprache und Geschlecht suchen.
French, psycholinguistic studies have shown that the generic use of the masculine gender results in a male bias in mental representations of gender. To counteract this effect, French speakers have begun using various gender-fair forms like joggeur·euses, joggeuses et joggeurs or un groupe de jogging. With the aim of investigating the influence of four different gender-fair forms, as compared to the masculine one, on perceived gender proportions of a role noun, we analysed 1,018 native French speakers’ estimated percentage of women in 22 role nouns. The results showed that all kinds of gender-fair forms significantly increased the estimated percentage of women in comparison to the masculine form and that no gender-fair form was significantly more efficient than any other. In addition, we found a somewhat surprising result: the more positive a participant’s attitudes towards gender-fair language were, the lower was their estimated percentage of women. We believe that participants who estimated lower percentages of women are the ones who believe gender-fair language is necessary and therefore hold more positive attitudes. Thus, the decrease in the estimated percentages would rather be the cause of the positive attitudes than an effect of them. To conclude, the results present strong arguments in favor of gender-fair language and underline the necessity of further investigations into the role of attitudes in relation to perceived gender ratios and gender-fair language.
Full-text available
There is empirical evidence in different languages on how the computation of gender morphology during psycholinguistic processing affects the construction of sex-generic representations. However, there are few experimental studies in Spanish and there is no empirical evidence that analyzes the psycholinguistic processing of morphological innovations used as non-binary forms (-x; -e) in contrast to the generic masculine variant (-o). To analyze this phenomenon, we designed a sentence comprehension task. We registered reading times, precision and response times. The results show the specialization of non-binary forms as generic morphological variants, as opposed to the generic masculine. The non-binary forms consistently elicited a reference to mixed groups of people and the response times indicated that these morphological variants do not carry a higher processing cost than the generic masculine. Contrary to what classical grammatical approaches propose, the generic masculine does not function in all cases as generic and its ability to refer to groups of people without uniform gender seems to be modulated by the stereotipicality of the role names.
Full-text available
In Polish, it is obligatory to mark feminine or masculine grammatical gender on second-person singular past tense verbs (e.g., Dostałaś list ‘You received-F a letter’). When the addressee’s gender is unknown or unspecified, masculine but never feminine gender marking may be used. The present self-paced reading experiment aims to determine whether this practice creates a processing disadvantage for female addressees in such contexts. We further investigated how men process being addressed with feminine-marked verbs, which constitutes a pragmatic violation. To this end, we presented Polish native speakers with short narratives. Each narrative contained either a second-person singular past tense verb with masculine or feminine gender marking, or a gerund verb with no gender marking as a baseline. We hypothesised that both men and women would read the verbs with gender marking mismatching their own gender more slowly than the gender-unmarked gerund verbs. The results revealed that the gender-mismatching verbs were read equally fast as the gerund verbs, and that the verbs with gender marking matching participant gender were read faster. While the relatively high reading time of the gender-unmarked baseline was unexpected, the pattern of results nevertheless shows that verbs with masculine marking were more difficult to process for women compared to men, and vice versa. In conclusion, even though masculine gender marking in the second person is commonly used with a gender-unspecific intention, it created similar processing difficulties for women as the ones that men experienced when addressed through feminine gender marking. This study is the first one, as far as we are aware, to provide evidence for the male bias of second-person masculine generics during language processing.
Heated societal debates in various countries concern the use of gender-fair language, meant to replace the generic use of grammatically masculine forms. Advocates and opponents of gender-fair language disagree on – among other things – the question of whether masculine forms leave women underrepresented in people's minds. We investigated the influence of linguistic form on the mental representations of gender in French. Participants read a short text about a professional gathering and estimated the percentages of men and women present at the gathering. Results showed higher estimates of the percentage of women in response to two gender-fair forms relative to the masculine form. Comparisons with normed data on people's perception of real-world gender ratios additionally showed that the gender-fair forms removed or reduced a male bias for neutral- and female-stereotyped professions, respectively, yet induced a female bias for male-stereotyped professions. Thus, gender-fair language increases the prominence of women in the mind, but has varying effects on consistency, i.e., the match with default perceptions of real-world gender ratios.
Conducted 4 experiments with 168 undergraduates to examine the conditions under which readers appear to infer predictable outcomes. Three retrieval paradigms were used: immediate recognition test, cued recall, and priming in word recognition. Findings indicate that on immediate test, responses to a word representing the implicit outcome were slow, but on delayed test these responses were slow or inaccurate only when primed by an explicitly stated word. However, the word expressing the predictable outcome did function as an effective recall cue. It is concluded that readers encode inferences about predictable outcomes into memory only minimally but that they can make use of a cue word that represents the inference both at the time of an immediate test and in delayed cued recall. (22 ref) (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Two experiments (N=48 each) were conducted to investigate gender-specific elements in the mental representation of short German texts. The texts contained a specific male or female designator (e.g., Mr. Smith, Mrs. Meyer) or a masculine generic phrase (CM) in either singular or plural usage (e.g., the student, the students) as text-subject. Two testphrases were constructed for each text, which did not appear in the text but reflected a masculine or feminine understanding of the text. Gender-specific associations were measured via the time that was required to reject the masculine and feminine test-distractors in a subsequent recognition task. Reading the texts with a specific male or female designator as text-subject increased the rejection time for the gender-congruent testphrases. For texts containing a GM as text-subject, the pattern of gender-specific associations was dependent on the grammatical numberof the GM-phrase. Reading a scenario containing a GM in the singular increased rejection times for the masculine test-phrases, while reading a scenario with a GM in the plural increased rejection times for the feminine test-phrases.
It has been alleged that, in appropriate verbal contexts, man and he are generic, i.e. that the words include women as well as men, as for example in, Man is mortal, or One must watch his language. Many feminists argue for the elimination of this generic use of man and he and the substitution of such non-male words as people and they. Others argue on various grounds that these changes are unnecessary. This paper isolates the issues involved in such arguments and provisionally concludes that a reduction in the generic use of man and he would result in a long term reduction in sexist thinking. Recent feminist research on man and he is carefully reviewed. In its final section, the paper develops the implication that women experience more alienation than men in the presence of the generic man and he.
Three experiments investigated how grammatical gender and gender stereotypicality influence the way person information is mentally represented. Participants read sentences about social groups denoted by nouns with different grammatical gender and stereotypicality. A following sentence contained a reference to the social group that qualified the group members as female, male, or neither one. Experiment 1 tested grammatically masculine nouns; Experiment 2 tested gender-balanced forms, composed of the masculine and the feminine or neither one; and Experiment 3 tested nouns without gender inflection. Stereotypicality varied within studies. Second sentence reading times differed depending on the fit between grammatical gender and stereotypicality of the first sentence’s subject and the subsequent information’s gender-relatedness. Both grammatical gender and stereotypicality contribute biological gender information to mentally represented person information. Strong grammatical input may override stereotypicality’s influence. The feminine’s influence seems to be weaker than the masculine’s. Results are discussed in the framework of the scenario mapping and focus approach.
The influence of a gender system in a language on perception was examined in a cross-cultural study. Participants were from two language groups, one with a gender system, Spanish, and the other with a limited gender system, English. In each language group, participants were from three age groups: 5-7 years old, 8-10 years old, and adult. In one experiment, participants were asked to put a typical male or female name to 20 objects. In another experiment, participants were asked to assign attributes to the objects. Language gender tags influenced the Spanish adults and the 8- to 10-year-olds in their choice of gender assignment, whereas perceived attributes influenced the younger Spanish children and English speakers (both adults and children). It appears that in a language with a grammatical gender system, such as Spanish, the gender system creeps into perception after the gender tags have been acquired.